Early Permian zircon ages from the P. confluens and P. pseudoreticulata spore-pollen zones in the southern Bonaparte and Canning basins, northwestern Australia

Abstract The Pseudoreticulatispora confluens–P. pseudoreticulata spore-pollen zonal datum typically coincides with the end of widespread Permian glacial deposits in Western Australia. Although previously attributed to the mid-Sakmarian, chemical abrasion isotope dilution thermal ionisation mass spectrometry (TIMS) dating of zircons from volcanic tuffs in the Ditji Formation of the Bonaparte Basin and the Grant Group in the Canning Basin point to an Asselian age of about 295.25 Ma for this datum. All dated zircons from the Ditji Formation came from petroleum well cuttings but the accompanying palynology was mostly from sidewall cores; however, all Grant Group samples were from conventional core. TIMS dates from the Ditji Formation range in age from 295.2 to 292.7 Ma whereas the only productive tuff from the Grant Group yielded a 296.26 Ma date. By comparison, there are no zircon dates to constrain the onset of glacial deposition in Australia. The Bonaparte Basin ages overlap with those for the Edie Tuff (296.1–294.5 Ma) in Queensland’s Galilee Basin, approximately 2000 km to the southeast, which also lies close to the base of the P. pseudoreticulata Zone. To date the only fossil group within the P. confluens Zone in Western Australia to provide independent age control, albeit loosely, are goniatites from the northern Perth Basin (Uraloceras irwinense and Juresanites jacksoni) that have consistently been attributed to the Sakmarian; these require a reassessment of their affinity with Russian faunas and therefore to global stratotypes. The position of the Carboniferous–Permian boundary is elusive in Australia and will remain so until additional volcanic tuffs containing young datable zircons are found; however, spore-pollen and zircon dates from Namibia place this boundary within the P. confluens Zone.


Introduction
The Australian Carboniferous-Permian spore-pollen zonation is based mostly on endemic species but is widely used to constrain stratigraphic correlations within and between the many basins of this age across the continent (Figure 1; e.g. Backhouse, 1998;Jones & Truswell, 1992;Kemp et al., 1977;Laurie et al., 2016;Mory & Backhouse, 1997). However, use of this zonation is confined largely to the subsurface as outcrops typically are too oxidised to preserve palynomorphs. Even where macrofossils are recorded, links between them and the spore-pollen zonation can be tenuous, as are relationships to the international chronostratigraphic scheme (Gradstein et al., 2020), particularly for the mid-Pennsylvanian-lowermost Cisuralian. Within this interval, Australian marine faunas, where present, yield unclear ages owing to endemism, low diversity and sporadic distributions influenced by adverse climates-most notably the Late Paleozoic Ice Age (LPIA). Similarly, the associated spore-pollen zonation relies on relatively few species that yield only broad age control. The age of Asselian faunas has been especially difficult to confirm as conodonts, fusulinids and critical ammonoid species are unknown across most of Gondwana and its periphery (Archbold, 2001). Consequently, existing biostratigraphic schemes, even those based on marine organisms, have provided only tenuous or ambiguous age control. There has been a concerted effort to remedy this for the Guadalupian and Lopingian in Australia, mostly from the eastern part of the continent, using chemical abrasion isotope dilution thermal ionisation mass spectrometry (TIMS) dating of zircons from volcanic tuffs intercalated within palynomorph-bearing strata Phillips et al., 2018a). However, Pennsylvanian-lower Cisuralian strata have received relatively little attention. Volcanic tuffs are rare in this interval, apart from within the Pennsylvanian of New South Wales where the SHRIMP dating of Roberts et al. (1995aRoberts et al. ( , 1995bRoberts et al. ( , 1996 and Birgenheier et al. (2009) provided limited links to the spore-pollen zonation; however, their dates have yet to be revisited using more precise methods such as TIMS.
In Western Australia (WA), the only published Permian tuffs dates are Guadalupian from the Kennedy Group (Lever & Fanning, 2004) and Liveringa Group Mory et al., 2017) of the Southern Carnarvon and Canning basins, respectively. Possible Permian tuffs from the Artinskian Irwin River Coal Measures in the Perth Basin yielded only reworked pre-Permian zircons (Vladimir I. Davydov, pers comm.). The present study focusses on tuffs from the Ditji Formation in the southern Bonaparte Basin (Figures 2 and 3), first reported by Gorter et al. (2008), and a recently located tuff in the Grant Group of the southern Canning Basin from mineral exploration core hole Fortescue RUD0007.
Here we present laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) and TIMS zircon dates from volcanic tuffs within the Grant Group of the southern Canning Basin and the Ditji Formation in the southern Bonaparte Basin (Figures 3 and 4; Gorter et al., 2008). The former unit in the sampled core hole lies entirely within the P. confluens spore-pollen Zone, whereas the latter lies immediately above in the lower P. pseudoreticulata Zone. That level typically coincides with the end of  Backhouse, 1993b;Backhouse & Mory, 2020;Mory & Backhouse, 1997;Mory et al., 2008). In earlier literature (summarised in Backhouse & Mory, 2020, p. 5-9) this datum equates with the top of palynological Unit II in WA or the top of Stage 2 in eastern Australia.

Regional geology
In eastern Australia three distinct Carboniferous and four Permian LPIA glacial pulses have been identified (Birgenheier et al., 2009;Fielding et al., 2008aFielding et al., , 2008bFielding et al., , 2022 but their veracity in WA is uncertain (Haig et al., 2014) where a hiatus, attributed to major ice sheets inhibiting    Table S2). deposition, spans much of the Pennsylvanian (Backhouse & Mory, 2020). This break was followed by deposition of widespread glacial facies during the latest Pennsylvanian to early Cisuralian, with subsequent minor and less continuous Artinskian-Kungurian (late Cisuralian) glacial deposits attributed to sea ice in WA (Haig et al., 2014) possibly coeval with the end of the P2 glacial pulse in eastern Australia. The end of the main glacial phase in WA is typically regarded to be within the Sakmarian based on regional correlations (e.g. Haig et al., 2014) and goniatite faunas from the Perth Basin considered close to Russian assemblages, even though there are no species in common (Leonova, 1998(Leonova, , 2011(Leonova, , 2018. Differences in the durations and timing of Permian glacial pulses between the east and west of the Australian continent are attributed to a temperate Tethyan oceanic influence in the west compared with a combination of alpine glaciation and direct seaway connection from the South Pole in the east (Brakel & Totterdell, 1990).
In WA, Carboniferous-Permian glacial facies extend across all Paleozoic basins and arguably represent the first time when there was little, if any, difference in facies between basins. These facies are assigned to the lowermiddle Kulshill Group in the Bonaparte Basin, the Grant Group in the Canning Basin, the Lyons Group in the Northern and Southern Carnarvon basins, and the Nangetty Formation in the Perth Basin (Mory et al., 2008). Directly overlying these strata in all WA basins are marine facies typically including carbonate, indicative of a warmer climatic phase than the glacially influenced interval (Figure 4; Haig et al., 2014).

Bonaparte Basin stratigraphy
The sampled wells from the southern Bonaparte Basin (Figures 3 and 4; Table 1) illustrate a 385-km traverse along the southwestern margin of the basin encompassing glacial-post-glacial, dominantly siliciclastic facies of the Pennsylvanian-Cisuralian Kulshill Group. As revised by Gorter et al. (2008), this succession transitions from dominantly glacio-fluvial fining-up cycles, each up to 90 m thick, within the Kuriyippi Formation, to glacio-marine facies of the Treachery Formation showing basal sub-glacial valleys eroded into the underlying strata and evidence of freshwater influx during deglaciation. The overlying Quoin Formation demonstrates waning glacial conditions in a tidal environment in response to deglaciation of the hinterland. The succeeding Ditji Formation contains dominantly marine facies with thin volcanic tuffs up to 85 cm thick, overlain by largely fluvial facies of the Keyling Formation, which is the uppermost unit of the Kulshill Group. Initial correlations tied to seismic profiles by Gorter et al. (2008) show distinct incisions or erosional surfaces at most formation boundaries. The base of the Ditji Formation shows a flooding event coincident with the Pseudoreticulatispora confluens-P. pseudoreticulata palynological datum. Gorter et al. (2008, p. 95) interpreted a distinctive volcanic ash bed in the Ditji Formation, based on wireline logs showing consistent high radioactivity, low resistivity, low sonic velocity and high density, and traced 36 km south of the Barnett oilfield and Blacktip gasfield to Weasel 1. The temporal relationship they suggested with an intrusive dolerite dated at 293 ± 3 Ma by K-Ar in Kulka 1, 620 km northeast in the Goulburn Graben in the Arafura Basin (Diamond Shamrock, 1985, Appendix 4) remains speculative, even though the updated K-decay constant from Renne et al. (2011) indicates an age of 295.5 ± 3.0 Ma. Demonstration of a genetic relationship between the Arafura Basin intrusion and the Bonaparte Basin tuffs requires zircon Lu-Hf isotope analyses from both areas but first requires the recovery and dating of zircons from the Arafura Basin. Such analyses could also confirm that the tuffs within the Ditji Formation in Intrepid 1, 375 km westnorthwest of the Turtle oilfield, are from the same source, but is beyond the scope of the present study. Other wells apart from those sampled for this study probably also contain tuffs in at least the Ditji Formation based on interpreted wireline log responses ( Figure 3) but cannot be easily substantiated given the lack of core.

Canning Basin stratigraphy
Recent revisions of the Permian stratigraphy of this basin (Backhouse & Mory, 2020;Mory, 2010) emphasise the variation in lower Permian glacial and deltaic facies within the Grant Group, and suggest all formations identified within the group are of too limited extent to be useful for regional correlations. Directly overlying the group are warmer water facies of the Nura Nura Member at the base of the Poole Sandstone, which is a lateral equivalent to the Ditji Formation based on the contained palynoflora ( Figure 2). The only known Cisuralian tuff in the basin known to date is from the Grant Group about 180 m below the Poole Sandstone in Fortescue RUD0007 where the group directly overlie Precambrian gabbro.

Spore-pollen zonation
The Granulatisporites confluens Oppel-zone, established by Foster and Waterhouse (1988) from Calytrix 1 in the Canning Basin (Figure 1), is widely recognised across Australia and other Gondwanan basins including within Namibia (Stephenson, 2009). As Granulatisporites is a noncheilocardioid spore from the European Pennsylvanian with a sculpture of discrete grana only, Backhouse (1991) and Playford and Dino (2002) re-assigned the eponymous species to Pseudoreticulatispora, a Gondwanan genus within the cheilocardioid lineage, characteristic of Permian strata, and emended the zone's name accordingly, but the definition of the zone remains largely unchanged. This reassignment is in accord with Price and Foster (in Price, 1983, p. 169), who previously included forms with a reticulate distal sculpture and a proximal sculpture that may be similar to the distal sculpture, or with 'free clavae, verrucae or grana' in Pseudoreticulatispora-this is an accurate description of the sculptural arrangement on this species. By comparison, for the corresponding spore-pollen assemblage in Namibia, Stephenson (2009) followed the initial designation of the species by Archangelsky and Gamerro (1979) as Converrucosisporites confluens. Although present throughout the Asselian-Artinskian Vittatina costabilis Zone in South America (e.g. Souza et al., 2021), few other of the species identified there are present in Australian assemblages in spite of seemingly significant commonality at a generic level.
The upper limit of the Pseudoreticulatispora confluens Zone marks a significant palynofloral boundary defined by the first appearance of Pseudoreticulatispora pseudoreticulata, the eponymous species for the succeeding zone. Other species also appearing at or near the base of this zone include Diatomozonotriletes townrowii, Laevigatosporites vulgaris (L. colliensis of Backhouse, 1991), Procoronaspora spinosa and Tiwariasporites simplex. In the Bonaparte Basin, reliable samples (such as sidewall or conventional cores) from the two zones are typically separated by 100 m, whereas this gap is as little as 11 m in the Canning Basin (Backhouse & Mory, 2020). Nevertheless, this zonal boundary appears to lie close to the contact between the Quoin and Ditji formations ( Figure 4; Supplemental data, Tables S2 and S3).

Previous age allocations
The ages of Cisuralian faunal elements (especially brachiopods, bivalves, and bryozoans) and palynological assemblages in WA rely greatly on their stratigraphic position alongside goniatites. Unfortunately, the stratigraphic record of the earliest western Australian Permian goniatite faunas is patchy with none known from the main glacial facies ($lower 'P1' of Fielding et al., 2008aFielding et al., , 2008bFielding et al., , 2022, and Middle-Late Pennsylvanian faunas are completely unknown (e.g. Jones et al., 2000), even in other parts of the continent. Previous age determinations of lowermost Permian glacial strata in WA have largely depended on marine macrofossils and stratigraphic evidence, but these have  proved ambiguous or speculative owing to low species diversity and endemism (Skwarko, 1993), and the presence of fluvial facies. By comparison, strata post-dating the main glacial succession contain diverse faunas allowing more robust age assignations and are typically regarded as late Sakmarian and younger (e.g. Archbold et al., 1993Archbold et al., , 1998Haig et al., 2014). Dating subsurface Permian strata has relied greatly on palynology, but this is rarely possible for outcrops, which typically are strongly oxidised, thereby compromising preservation of these fossils. Thus, the relationship between palynological zones and macrofossil assemblages mostly relies on imprecise stratigraphic correlations that are rarely more detailed than by formation unless fully cored sections are available. As there are no outcrops of lowermost Permian marine facies in the Bonaparte Basin, the following discussion draws on evidence from farther south in the Perth and Canning basins, albeit reliant on longrange correlations. Of all fossil groups present in the lowermost Permian glacial successions, only goniatites appear to provide independent evidence of age; however, their record is poor, and no Australian phylogenetic series can be established (Haig et al., 2014). The oldest Permian faunas in WA are from the Woolaga Limestone Member in the middle of the Holmwood Shale within the northern Perth Basin and contain just two goniatite species, Uraloceras irwinense and Juresanites jacksoni. The Sakmarian age attributed to these species is based on 'the primitive nature of its [J. jacksoni] suture', and 'Uraloceras irwinense is closely comparable with several species of Uraloceras from the Sakmarian and Artinskian of the Ural mountains' (Teichert & Glenister, 1952, pp. 14-15). Whereas this age mostly has been favoured in the literature (e.g. Glenister et al., 1973Glenister et al., , 1990Glenister & Furnish, 1961), Archbold (1995, p. 96) indicated that 'discussion of the age of [U.] irwinense are equivocal'. Furthermore, Glenister et al. (1993, pp. 55-56) noted that that the 'possibility of an Asselian age cannot be rejected completely' for the Holmwood Shale fauna and that 'Early Permian ammonoid faunas are strikingly provincial'. Notwithstanding changes in generic assignations (J. jacksoni was previously placed in Metalegoceras, and U. irwinense in Svetlanoceras), a Sakmarian age continues to be allocated to this stratigraphic level (e.g. Haig et al., 2014Haig et al., , 2022Playford, 2021). The generic assignations are seemingly unhelpful, as Juresanites and Svetlanoceras are present in Asselian strata, whereas Uraloceras and Metalegoceras first appear in the Sakmarian (Leonova, 2011(Leonova, , 2018. In WA, the two goniatite species are from the middle of the Holmwood Shale where they are loosely associated with the uppermost part of the P. confluens spore-pollen Zone (Backhouse, 1992(Backhouse, , 1993a(Backhouse, , 1993b. By comparison, Playford (2021, p. 58) suggested the associated palynoflora from this level (the Woolaga Limestone Member) could be assigned to either the P. confluens Zone or the P. pseudoreticulata Zone. There are no goniatites known from coeval strata in any other western Australian basin (see lists in Skwarko, 1993). Archbold (1982) appears to have made the first suggestion of an Asselian age for the uppermost part of the Permo-Carboniferous glacial succession for the Perth Basin, but not in other basins within WA. Foster and Waterhouse (1988) favoured a mid-to late Asselian age for faunas associated with the P. confluens Zone in the Grant Group in the Canning Basin based on loose macrofaunal associations. By comparison, Taboada et al. (2015) opted for a Sakmarian age for material from the same stratigraphic level and subbasin. About 200 m stratigraphically higher than that level is a goniatite fauna from the type area of the Nura Nura Member (in the Fitzroy Trough of the Canning Basin near 18.0856 S, 124.4094 E) at the base of the Poole Sandstone, which directly overlies the Grant Group. Leonova (1998Leonova ( , 2011 and Boiko et al. (2008) (Glenister et al., 1993;Glenister & Furnish, 1961;Miller, 1936;Teichert, 1942). In contrast, subsurface sections assigned to the Poole Sandstone and Nura Member yield palynomorphs of the P. pseudoreticulata Zone, here considered to be late Asselian-mid-Sakmarian ( Figure 2).

Methodology
Intervals judged to contain volcanic tuff beds in the Ditji Formation, based on wireline log correlations following the initial interpretation by Gorter et al. (2008), were sampled from available cuttings samples held at the Department of Mines, Industry Regulation and Safety (DMIRS) core library. These samples were collected from just below the interpreted depths owing to the likelihood of down-hole caving (Table 1; Figure 4). By comparison, the only tuffs known from the Grant Group are in Fortescue RUD00007, which was sampled from the core held by DMIRS.
Of the 23 samples processed, 16 yielded sufficient zircon for screening using cathodoluminescence (CL) images and LA-ICPMS (Table 1). This step helps discriminate young volcanic zircons from older ones, especially those of detrital origin, and thus the selection of grains likely to provide depositional ages using U-Pb isTIMS. Ten of the 16 samples with suitable zircons were analysed in this fashion. Appendix S1 of the supplemental data files outlines the LA-ICPMS and TIMS methods employed at Boise State University. Appendix S2 lists revised formation picks for wells initially considered to have intersected the upper Kulshill Group including the Ditji Formation. Appendix S3 summarises the re-evaluation of legacy palynological interpretations from company reports for the sampled wells.

Bonaparte Basin
No core is available from the Ditji Formation in any of the 23 petroleum exploration wells that intersected this unit (Appendix S3). The core library permits the removal of no more than 20 g from each archived cuttings sample, so we amalgamated samples (mostly over 10 or 15 m intervals) to avoid excessive processing and to provide sufficient material for analysis, even though these intervals may not be ideal to differentiate individual volcanic tuffs. Even so, sample sizes were small, which may partly explain the low number of overlapping ages in some of the analyses (summarised in Figure 5, based on Appendix S1). The TIMS results for most samples indicate that zircons have caved from higher stratigraphic levels, i.e. contradictory ages that are too young based on regional correlations (Figure 4), and/or have been reworked based on markedly older ages. Thus, it is possible that some samples fortuitously include zircons from the same event reworked into different stratigraphic levels, but 'reunited' during drilling owing to down-hole caving. By comparison, the larger errors associated with LA-ICPMS results typically do not allow differentiation of caved vs unworn reworked zircons. All zircon dates considered reliable lie within or adjacent to the Ditji Formation, which has yielded spore-pollen of the P. pseudoreticulata Zone based on palynological reports (Foster, 1984(Foster, , 1985(Foster, , 1986Purcell, 2001;Purcell & Hooker, 2000) and reviewed by Backhouse (2021; Supplemental data Appendix S3). By comparison, all other samples provide ambiguous zircon dates unlikely to be close to depositional ages deduced from MSWD analyses.

Barnett 1
Of the nine samples from this well, five yielded young zircons suitable for TIMS ( Figure 5), two had insufficient zircons to proceed beyond LA-ICPMS or the age was out-ofplace and therefore indicated likely reworking, and two had no zircons. The five successive samples over 1026 À 1300 m analysed by TIMS reveal ages that become progressively older down-hole: 294.16 ± 0.86 (n ¼ 2; 1026 À 1044 m), 294.44 ± 0.25 (n ¼ 3; 1080-1095 m), 295.13 ± 0.12 (n ¼ 5; 1095-1110 m), 295.22 ± 0.13 (n ¼ 5; 1130-1145 m) and 300.59 ± 1.08 (n ¼ 2; 1285-1300 m). The uppermost of these five samples is from the base of the Keyling Formation, the next three are from the Ditji Formation, and the deepest is from low within the Quoin Formation. However, regional correlations ( Figure 2) point to the two youngest zircons in the uppermost sample (289.94 ± 0.99) being from a higher level within the well, whereas two slightly older zircons (294.16 ± 0.86 Ma) were probably reworked from the immediately underlying Ditji Formation. It is uncertain if the 300.59 ± 1.08 Ma date from the Quoin Formation is near the depositional age or is from reworked zircons.
Considering all the youngest TIMS zircon dates <297 Ma (n ¼ 23) in the three samples spanning 1080-1145 m from this well there may be as many as 6 or 7 individual tuffs within the section. Given the overlapping individual dates from this interval and the likelihood of mixing owing to caving and reworking in our amalgamated samples, the simplest interpretation is that only three tuffs can be discriminated, i.e. one from each sample. An older cluster (at ca 296.2 ± 0.13 Ma based on three grains from 1080 to 1110 m) is attributed to reworking. The age of this cluster overlaps with the tuff from Fortescue RUD00007 in the southern Canning Basin (296.26 ± 0.12 Ma) suggesting derivation from the same eruptive event, but such speculation requires confirmation from Lu-Hf isotopic analysis.

Berkley 1
Of the four samples processed from this well, only three yielded suitable zircons ( Figure 5). The youngest zircons in the highest sample (430-440 m) from the Ditji Formation yielded a TIMS date of 292.74 ± 0.19 Ma (n ¼ 4). In the underlying sample from 470 to 480 m, the youngest zircon is 293.47 ± 0.19 Ma, but its significance is unclear, given it is from a single grain. Whereas both dates are somewhat younger than all the other TIMS dates not discounted, as from cavings, they are not young enough to be explicitly interpreted as such. The other productive sample from 630 to 650 m close to the Quoin Formation-Treachery Shale contact (631.2 m) provided a 297 ± 4 Ma LA-ICPMS age based on eight grains; however, TIMS was not attempted, given the likelihood of reworking/caving providing too broad a spectrum of dates.

Blacktip 1
The single sample from the upper part of the Ditji Formation in this well (2608-2640 m) yielded five zircons with a mean LA-ICPMS age of 293 ± 7 Ma. However, the three youngest zircons analysed by TIMS were non-concordant or with a large error and possibly reworked, so the well was omitted from Figure 4.

Intrepid 1
Of the five samples from this well, just three (all from the Ditji Formation) yielded sharp zircon grains unlikely to be reworked. Two of these (1070-1080 m and 1120-1130 m) yielded Artinskian LA-ICPMS ages (284 ± 6 Ma, n ¼ 4 and 287 ± 5 Ma, n ¼ 5, respectively) presumably owing to Pb loss. The intermediate sample (1080-1090 m) yielded 14 young zircons when analysed by LA-ICPMS with a mean of 293 ± 3 Ma, but of the nine grains subjected to TIMS, five appeared reworked, and one may be from higher in the well. The remaining three grains yielded a mean of 294.72 ± 0.15 Ma (Figure 5), within the range of the dates considered reliable from Turtle 1 and Barnett 1, over 370 km to the southeast (Figure 4).

Turtle 1
A single sample from within the Ditji Formation (1225-1255 m) yielded three TIMS dates with a mean of 294.54 ± 0.27 Ma ( Figure 5) consistent with dates from much the same level in Barnett 1, 14 km to the southeast.

Canning Basin
Of the four potential tuffs sampled from Fortescue RUD0007 in the southern part of the Canning Basin (Figure 1), only one yielded zircon suitable for TIMS analysis. The 296.26 ± 0.25 Ma age is from three zircon grains ( Figure 5). The tuff lies at 389.6-389.67 m below ground level within mudstone, sandstone and diamictite of the Grant Group. It is possible that the Poole Sandstone overlying the Grant Group is present in the uppermost part of the section or nearby but cannot be assessed, as core recovery commenced at 333.5 m, and no higher samples are available; in addition, no wireline logs were run in this borehole. All nine palynology samples (335-507.8 m) above and below the tuff yielded palynomorphs of the P. confluens Zone (Backhouse, 2020).

Discussion
All of the Bonaparte Basin samples are cuttings from petroleum exploration wells that inevitably incorporate grains and rock fragments caved from higher levels during drilling. No conventional cores are available from the volcanic tuffs targeted for analysis from this basin or, in the case of sidewall cores, too little remains as exploration companies  (Ludwig, 2003). Errors on individual dates are at 2r. Weighted mean dates are shown and represented by the grey boxes, with errors on the weighted mean dates at the 95% confidence interval. A few older dates that plot off scale are not shown (see Supplemental data, Appendix 1c).
preferentially use them for palynology to avoid down-hole contamination. As well as reworking from older levels during deposition, down-hole caving is potentially significant in assessing the veracity of our zircon dates. The probability that each well intersected several tuffs cannot be resolved from our data owing to the inevitable down-hole dispersion of grains and rock fragments during drilling, insufficient resolution of wireline logs and the possibility of higher-density minerals taking longer to reach the surface within the mud system-difficulties perhaps compounded by the amalgamation of individual cuttings samples.
Seven TIMS dates from the Ditji Formation in the Bonaparte Basin that are not obviously reworked span 295.2-292.7 Ma with 2r errors of 0.12-0.27 Ma, i.e. early Sakmarian-late Asselian (Figure 4). Of these dates, the two from Berkley 1 (293.5 and 292.7 Ma) are Sakmarian, whereas those from Barnett 1, Intrepid 1 and Turtle 1 are Asselian (295.2-294.2 Ma). Although the difference implies that at least part of the interval 325-536 m in Berkley 1 could be a younger unit, there are insufficient data to support such an interpretation. The 296.0 Ma date from the upper Ditji Formation in Blacktip 1 seemingly represents reworking. These dates point to a late Asselian age for the base of the Ditji Formation and the base of the associated P. pseudoreticulata spore-pollen Zone. A date of 295.25 Ma is suggested for that level based on the cluster of younger TIMS dates from that formation, and thereby supports the Asselian age proposed for the P. confluens Zone by Foster and Waterhouse (1988), in contrast to later suggestions of a mid-Sakmarian age (e.g. Backhouse, 1991;Mory, 2010;Mory et al., 2008;Mory & Backhouse, 1997).
In the Galilee Basin of northern Queensland, the Edie Tuff Member (upper Joe Joe Group) in OEC Glue Pot Creek 1 has yielded 295.65 ± 0.07 Ma and 296.09 ± 0.07 Ma TIMS dates (Phillips et al., 2018a). By comparison, GSQ Muttaburra 1, 155 km to the west, yielded dates of 294.80 ± 0.12 Ma and 294.91 ± 0.15 Ma from the same unit (Nicoll et al., 2015, p. 215;Phillips et al., 2018aPhillips et al., , 2018b. The dates from the latter well come from 38 m below an APP2.1 palynology assemblage (Nicoll et al., 2015, p. 215). The lower part that assemblage equates with the P. pseudoreticulata Zone (e.g. Laurie et al., 2018), so the TIMS dates from the Queensland sections appear to be consistent with those from the Bonaparte Basin.
Beyond Australia, outcrop of the Ganigobis Shale Member (Dwyka Group) along the Fly River in Namibia yielded SHRIMP dates of 302 ± 3 Ma and 299.2 ± 3.2 Ma (in Bangert et al., 1999) at a level equivalent to the P. confluens Zone (Stephenson, 2009). These dates are close to Griffis et al.'s. (2021) TIMS date of 299.31 ± 0.35 Ma from a sample collected just over 1 km downstream from Stephenson's (2009) section. Given the low dips along that river, both localities most likely represent much the same stratigraphic level. Unfortunately, the 295.84 ± 0.47 Ma age Griffis et al. (2021) obtained from a sample of the Owl Gorge Member (Prince Albert Formation) 320 km farther south has no associated palynology. Similarly, the palynoflora associated with dates of 296.77 ± 0.04 and 296.14 ± 0.09 Ma from volcanic tuffs in the Mengkarang Formation of West Sumatra reported by van Waveren et al. (2021) was considered 'unlike those typical of the Asselian-Sakmarian of Gondwanan areas' (Crippa et al., 2014, p. 215). U-Pb dating in the Paran a Basin of Brazil (summarised by Souza et al., 2021) suggests that the Vittatina costabilis Zone, which contains P. confluens, spans the Gzhelian to mid-Artinskian, but none of the cited TIMS zircon dates are older than Asselian. Kavali et al. (2022) indicate that their Palynoassemblage II in Argentina contains species typical of the P. confluens Zone in Australia and that it may be as old as 310.63 ± 0.1 Ma (Moscovian) based on a 206 Pb/ 238 U age reported by Gulbranson et al. (2010). It is unclear how this age may be applicable in Australia, given contradictory spore-pollen ranges between the two countries possibly owing to the influence of facies, reworking and preservation issues (Backhouse & Mory, 2020, p. 36).
Although Fang et al. (2021) placed the base of the Sakmarian at 294.1 ± 0.2 Ma, the International Commission on Stratigraphy endorses an age of 293.52 ± 0.17 Ma for that level (Gradstein et al., 2020). Nevertheless, there may be some elasticity in the absolute age of this stage boundary, which could diminish the magnitude of the discrepancy in ages assigned to the top of the P. confluens Zone in WA based on associated zircon dating and goniatite affiliations.
In the northern Perth Basin, the stratigraphic association of a limited goniatite assemblage, to which a Sakmarian age is attributed (Leonova, 2011(Leonova, , 2018, with spore-pollen of the P. confluens Zone (Backhouse, 1992(Backhouse, , 1993a(Backhouse, , 1993b is seemingly at odds with the Asselian TIMS zircon ages from close to the top of this zone in the Bonaparte and Galilee basins. The difference may be a function of facies controls or disparate biogeographic provinces, but the consistency in ages across the north of the continent suggests otherwise. By comparison, the Sakmarian age attributed to goniatites associated with a P. pseudoreticulata palynoflora in the Canning Basin by Haig et al. (2014 and references therein) is consistent with the 291.62 Ma date from a tuff within the zone in the Gunnedah Basin of New South Wales (Nicoll et al., 2016, p. 35). A reassessment of Australian goniatite affiliations with Russian material, and therefore the global stratotypes, is required for the specimens associated with the P. confluens Zone.
The seemingly abrupt end to glacial-deglacial conditions close to the P. confluens-P. pseudoreticulata spore-pollen datum across WA may not be a coincidence in that land plants are highly susceptible to climatic changes. This level seems tantalisingly close to the Carboniferous-earliest Permian marine Biodiversification Event (placed at 294.8 Ma by Fan et al., 2020;ca  However, further and more detailed studies are needed within Australia to establish if the events are related.

Sources of Asselian zircons
During the Carboniferous-Permian, the western margin of Australia underwent continental fragmentation with many blocks dispersed into present Southeast Asia (Metcalfe, 2013), but this episode left scant evidence of volcanism in WA. This is unlike the eastern margin of the continent where a subducting continental collision zone generated extended periods of volcanism and deposited tuffs in the subsiding marginal sediments, especially within the developing Bowen-Gunnedah-Sydney basins (Jessop et al., 2019;Rosenbaum, 2018). High atmospheric circulation from such eruptive episodes in eastern Gondwana could have carried detritus, including zircons, to WA basins, even though those volcanic centres lie 2000-3000 km to the east. Major eastern volcanic sequences from Queensland, preserved as the Edie Tuff Member in Muttaburra 1 or within the Jochmus Formation in Glue Pot Creek 1 as well as the Camboon Volcanics (Phillips et al., 2018a), are possible sources for zircons in the Ditji Formation in the Bonaparte Basin and possibly the Grant Formation of the Canning Basin, given the age similarities. Other potential source areas include the Arafura region (Gorter et al., 2008), volcanism along the margins of other blocks that once lay along the northern margin of east Gondwana (Metcalfe, 2013) and Sumatra (van Waveren et al., 2021), or perhaps the Simao Block in southwestern China (Li et al., 2012). Further analyses, notably Lu-Hf, which could confirm such speculation, are beyond the scope of the present study.
Asselian ages from provenance studies of Triassic and Jurassic strata from the Canning and Roebuck basins of the North West Shelf include rare reworked Asselian zircons (Lewis & Sircombe, 2013;Thomas, 2012). However, the origin of these zircons is unclear, and the lower precision of SHRIMP analyses (with errors up to two orders of magnitude greater than TIMS) renders some uncertainty to their significance. Those and other provenance studies (e.g. Craddock et al., 2019;Mor on et al., 2019;Veevers & Saeed, 2008) show a similar frequency distribution in Precambrian ages from our Bonaparte Basin samples-the so-called 'Gondwanan' zircon signature, which ultimately originated from Antarctica and thereabouts-but there are noteworthy differences from the Paleozoic LA-ICPMS zircon age distribution. In particular, our samples yielded few Devonian or Cambrian-Ordovician zircons ( Figure 6). This difference suggests that Silurian and Carboniferous-lowermost Permian zircons came from the edge of the Kimberley Basin, especially as the Treachery Shale sits directly on basement in Intrepid 1 and Berkley 1; however, the ultimate origin of these zircons may have been from well beyond the Bonaparte Basin. By comparison, the paucity of Devonian ages is consistent with the apparent lack of volcanic rocks of that age in other western Australian basins. Transportation by ice during the LPIA best explains the minimal abrasion of our zircons, and probably incorporated several routes from different sediment sources now difficult to pinpoint.

Carboniferous-Permian boundary in Australia
The position of the Carboniferous-Permian boundary has not been satisfactorily located in Australia, largely because there are no Middle-Late Pennsylvanian faunas (e.g. Jones et al., 2000), and Asselian faunas are of low diversity with highly provincial species (e.g. Archbold, 2000). Similarly, palynological assemblages contain species at best endemic to Gondwanan terranes but typically are more restricted in their distribution.
Suggestions that the M. tentula Zone spans the Carboniferous-Permian boundary in eastern Australia (e.g. Bodorkos et al., 2016;Phillips et al., 2018b) were speculative and conflict with dates from Namibia and possibly some from the Bonaparte Basin. Our samples show that the P. confluens Zone seemingly extends into the Gzhelian based on the 300.59 ± 0.17 Ma date within the Quoin Formation at 1285-1295 m in Barnett 1; however, this date is based on just two zircons, and the possibility of reworking/down-hole caving imparts considerable uncertainty. At present, the best evidence relevant to Australia for the position of the Carboniferous-Permian boundary is from zircon dates along the Fly River in Namibia (Bangert et al., 1999, Griffis et al., 2021Stephenson, 2009), which places it within the P. confluens Zone. In Australia, by comparison, this boundary has previously been placed anywhere between the top of this zone (e.g. Balme, 1980;Kemp et al., 1977) to somewhere below (Archbold et al., 2004;Backhouse & Mory, 2020), with the most judicious placement being 'at or near the base of Unit II/Stage 2' (Archbold, 1982, p. 267;Archbold, 1984).

Conclusions
The age of the Pseudoreticulatispora confluens-P. pseudoreticulata spore-pollen datum, which typically coincides with the upper limit of the main phase of glacial sedimentation in WA, is placed close to 295.3 Ma based on zircons dated by TIMS from the Ditji Formation in the Bonaparte Basin, i.e. within the Asselian, the earliest stage of the Permian. Although based on petroleum well cuttings, which inherently incorporate material caved from higher stratigraphic levels, the TIMS ages are remarkably consistent (295.2-294.5 Ma with errors generally less than 0.15 Ma) in the five sampled wells spanning almost 400 km along the southern margin of the basin. This age range overlaps with that for the Edie Tuff (296.1-294.8 Ma) in Queensland's Galilee Basin approximately 2000 km to the southeast, which is also considered to lie close to the base of the P. pseudoreticulata Zone . Previously, there have been few suggestions of Asselian ages in WA based on fossil assemblages, although Glenister et al. (1993) did not entirely dismiss this possibility. By comparison, the Asselian age Foster and Waterhouse (1988) suggested for the macrofossils and palynoflora from Calytrix 1 in the Canning Basin largely has been overlooked. Goniatites, which Leonova (1998Leonova ( , 2011 attributes to the Sakmarian, from the Woolaga Limestone Member within the middle of the Holmwood Shale in the northern Perth Basin, are loosely associated with the uppermost P. confluens Zone (Backhouse, 1992(Backhouse, , 1993a(Backhouse, , 1993b)-a reassessment of this material is required. The position of the Carboniferous-Permian boundary, which lies within the Pseudoreticulatispora confluens sporepollen Zone based on data from Namibia (Griffis et al., 2021;Stephenson, 2009), is elusive in Australia and will remain so until more volcanic tuffs with datable zircons are found. Other future work could include Lu-Hf analyses, particularly of the reworked 296.2 Ma zircons from the Bonaparte Basin and the 296.3 Ma zircons from a thin tuff 1000 km to the south within the Grant Group in southern Canning Basin to see if they share a common origin or not.