Lithogeochemical and sulfide trace-element systematics across the Permian–Triassic boundary, Perth Basin, Western Australia: constraints on the shallow marine environment during the end-Permian mass extinction

Abstract Sedimentary pyrite trace-element composition is an established proxy for determining paleo-ocean geochemistry and atmospheric oxygen concentrations through deep time. However, its applicability over shorter time-scales (i.e. <20 Ma) is not well known. To test this, we targeted fine-grained pyrite in the Hovea Member of the Kockatea Shale (Perth Basin, Western Australia), which encompasses the late Permian inertinitic interval and the end-Permian to Early Triassic sapropel, and spans approximately 10 million years. The end-Permian mass extinction (EPME) was the largest extinction event in Earth history, and its greatest effect is documented in the marine environment. Samples were collected from two oil exploration wells—Redback-2 and Hovea-3—spaced ∼20 km apart. In the two boreholes, a change in depositional facies (i.e. between the inertinite and sapropel) occurs below the Permian–Triassic boundary and records the transition from a marginal marine to a shelf environment. This transition is highlighted by several lithogeochemical indicators (e.g. negative shift δ13C values and Corg reduction; increases in Ca, Fe and P), which are themselves tied to fundamental changes in modal mineralogy between the two zones. Importantly, the sapropel also records a major increase in iron sulfide burial over that in the inertinite. LA-ICPMS analyses of pyrite demonstrate that trace-element abundance is highest in samples below the facies transition, and in places reaches a few percent, particularly of Ni (4 wt%), Co (1.5 wt%) and As (2.8 wt%). Moreover, these and other trace elements decrease by an order of magnitude in concert with the negative shift in δ13C values in the sapropel zone. Various whole-rock based paleosalinity indicator ratios (e.g. B/Ga) indicate that the areas of the Perth Basin intersected by Redback-2 and Hovea-3 were not fully connected to the open ocean at the time of the EPME, which leads us to conclude that the very high trace-element values in the sedimentary sulfides are reflective of regional environmental shifts rather than a global signal. Nonetheless, a geochemical contribution from a distant igneous province, such as the Siberian Traps Large Igneous Province, cannot be ruled out. Our work underscores the strength of sedimentary pyrite as a robust paleoenvironmental proxy in the marine environment and highlights the need for further investigation of pyrite trace-element profiles across the mass extinction interval in other sedimentary sequences around the globe. KEY POINTS LA-ICPMS-based geochemistry of sedimentary pyrite from the Hovea Member of the Kockatea Shale is considered within a lithochemostratigraphic context. The overall interpretation of the results involves a change in depositional setting from the marginal in the late Permian brackish waters to shelfal marine and loss of oxygen in the Early Triassic Perth Basin.


Introduction
During the last 10 years, LA-ICPMS-based geochemical studies identified temporal trends between the trace-element contents of sedimentary pyrite and some geological and biological processes, such as variation of oxygen levels in the atmosphere-ocean system (Cannell et al., 2022; years (Gregory, 2020;Large et al., 2014Large et al., , 2015Large et al., , 2020Large et al., , 2022Long et al., 2016;Mukherjee & Large, 2016). In more focused research, Gregory (2020) and Gregory et al. (2015) demonstrated that the trace-element contents of syn-sedimentary or early diagenetic pyrite from the same sequences could vary by more than an order of magnitude over time intervals shorter than 100 million years. In this work, we evaluate the applicability of the sedimentary pyrite proxy on a time-scale shorter than 20 Ma. For these purposes, we targeted shales across the Permian-Triassic stratigraphic boundary (PTB).
The PTB is associated worldwide with the most severe mass extinction in Earth's history and a significant change in the carbon cycle (Baud et al., 1989), both of which are thought to have been triggered by numerous climatic and tectono-magmatic drivers (Black et al., 2014;Clarkson et al., 2015;Chen et al., 2015;Ernst, 2014;Ernst & Youbi, 2017;Grice et al., 2007;Metcalfe et al., 2013;Payne & Kump, 2007). This mass extinction event has been given different names by different authors, but for simplicity and clarity, we will refer to it as the end-Permian mass extinction (EPME). The determinative factors causing the EPME are thought to have involved significant disturbance to the balance of atmospheric gases essential for life on Earth, such as oxygen, methane, carbon dioxide and sulfur dioxide, and an increase in toxic metals in the ocean. Pyrite, ubiquitously present in both deep and shallow marine sedimentary rocks of the upper Permian, has been cited as an indicator of widespread ocean anoxia at the EPME, which early researchers believed to be the root cause of the event (Cao et al., 2009;Isozaki, 1997;Song et al., 2012). However, more recent research has focused on establishing a causative link between the emplacement of the Siberian Traps Large Igneous Province (STLIP) and the EPME (Ernst & Youbi, 2017;Ernst et al., 2021). Regardless of the difference in proposed triggers (i.e. anoxia vs asphyxiation), it is likely that significant chemical changes in the marine environment and atmosphere associated with the EPME can be identified using the major, minor and trace-element compositions of coeval sediments and syn-sedimentary pyrite.

Regional and local geology of the EPME in the Perth Basin
The Hovea Member in the northern Perth Basin, Western Australia ( Figure 1), was targeted for this study as the only marine sequence on the Australian continent that includes the PTB. Two oil-exploration wells-Redback-2 and Hovea-3-were selected for sampling, as they contain well-preserved intersections of the Hovea Member. Figure 1. Perth Basin's current location, onshore in Western Australia and offshore in the Indian Ocean (simplified from Geoscience Australia). The inset shows the Gondwana land, paleo-oceans, and the Perth Basin paleo-location based on tectonic reconstructions (Blakey & Ranney, 2018) at the joint point between Antarctica, Australia and Great India.
Based on the completion reports for Redback-2 and Hovea-3 and other researchers (Jones, 2011), the Hovea Member represents the earliest sedimentation product of the Kockatea Shale. It was deposited during post-rift thermal subsidence and marine transgression, and contains two distinct facies: a lower package of inertinitic mudstones (INI) and an upper package of sapropelic (SPI) mudstones. The stratigraphic framework of the Hovea Member has been defined by studies of palynofacies (R. Purcell in Origin Ltd, 2003, 2010D. Mantle in Lounejeva et al., 2021) and associated invertebrates (Shi et al., 2022), as well as a Re-Os age of 253.5 ± 1.4 Ma for the INI top shale in the Hovea-3 borehole, which is equivalent to the late Permian Changhsingian stage (Georgiev et al., 2020). The lower INI contains abundant late Permian palynoflora (D. parvithola Zone) and brachiopods, whereas the SPI contains palynofacies belonging to the Earliest Triassic (K. saeptatus Zone), as well as C. perthensis bivalves and other invertebrates from the brachiopod, ammonoid, microconchid and spinicaudatan groups. The PTB in the Hovea Member is located in the lower part of the SPI, between the first appearance of the Triassic Griesbachian bivalves and the last appearance of the Australian endemic P. microcorpus Zone palynoflora and associated lingulid species of the latest Permian Changhsingian (Metcalfe et al., 2008(Metcalfe et al., , 2015Shi et al., 2022). The EPME, which occurs below the PTB and straddles the SPI/INI transition, contains abundant black wood fragments, degraded organic material and some marcasite (Lounejeva et al., 2021;Shi et al., 2022).
The most prominent chemostratigraphic changes occurring at the top of the INI, which is within the EPME, are the negative shift or excursion of d 13 C (organic and kerogen) values, recognised worldwide and linked to the end-Permian mass extinction interval (EPMEI) and the STLIP (Korte & Kozur, 2010;Metcalfe et al., 2013), and a positive spike in d 34 S values of marcasite-pyrite (Lounejeva et al., 2021) that has allowed geochemical correlation of the Hovea Member P. microcorpus Zone (Grice et al., 2007;Lounejeva et al., 2021;Metcalfe et al., 2015;Nabbefeld et al., 2010;Sial et al., 2020;. Multiple bulk-rock geochemical studies of the Hovea Member have revealed contrasting geochemical changes across the EPMEI, suggesting a discontinuous sedimentary record or geochemical mixing around the SPI/INI interface. Some studies infer anoxia well before the EPMEI, indicating a change to euxinia in the photic zone in the early Triassic (Georgiev et al., 2020;Grice, Cao, et al., 2005;, whereas some others strengthen the temporal link to different stages of the STLIP activity (Sial et al., 2020). Pyrite is commonly the main host for several redox-sensitive elements (e.g. Se, Co, As, Ni, Cu, Bi, Ag); thus, we investigated whether pyrite in the Hovea Member retained a record of the inferred changes. A full data set of the trace elements in sedimentary pyrite (TESPy) from Redback-2 and Hovea-3 boreholes as determined by LA-ICPMS along with the sediment bulk geochemistry.

Bulk analyses
The sediment chemical compositions have been obtained via ICP-AES and ICP-MS instrumental analyses (ALS Minerals Laboratory, four-acid digestion method ME-MS61), achieving an analytical precision of $5% on duplicates. Sulfur and carbon content was determined using an ELTRA CS-2000 elemental analyser.

LA-ICPMS analysis
We followed the methodology for LA-ICPMS sedimentary pyrite geochemistry as a proxy to paleo-ocean environment fully described by Large and co-authors (Gregory, 2020;Gregory et al., 2015Gregory et al., , 2020Large et al., 2014Large et al., , 2022Stepanov et al., 2020). This approach involves textural screening of suitable sedimentary pyrite material, preparation of polished surfaces, in situ pyrite and surrounding matrix analysis via LA-ICPMS, raw data processing and statistics.
The trace-element contents of pyrite have been analysed across multiple sessions by laser ablation-inductively coupled-mass spectrometry (LA-ICPMS) at CODES, University of Tasmania. The suite of elements likely incorporated into pyrite structure analysed for this study includes Co, Ni, Pb, Cu, Zn, As, Se, Ag, Sb, Bi, Tl, Au, Mn and Cd, whereas Al, Si, Ca, Mg and a set of other elements have been analysed for other minerals for signal monitoring and the matrix deconvolution. LA-ICPMS mass spectrometry allows the signal of each element during analysis to be observed to determine whether the element belongs to pyrite or to a neighbouring mineral. The primary calibration standards include the in-house reference materials STDGL2b2 (Danyushevsky et al., 2011) and STDGL3 (Belousov et al., 2014(Belousov et al., , 2023 for quantification of siderophile and chalcophile elements, and the USGS reference material GSD-1G (Jochum, 2014), for quantification of lithophile elements. A natural pyrite standard PPP-1  was used to quantify sulfur. Instrumental specifications are provided in the Supplemental data.
The raw results have been reduced using the inhousedeveloped method (Stepanov et al., 2020), which is based on the Excel spreadsheets and the Basic script templates for mass balance and Fe-S linear regression. In addition, matrix-pyrite deconvolution is possible using the Laser Ablation Data Reduction (LADR) software from Norris Scientific (https://norsci.com/?p=ladr). The ioGAS software from IMDEX (https://iogas.imdexlimited.com) has been used for basic statistics and presentation of the data.

Sampling, lithology and mineralogy
For this study, we targeted pyrite-bearing mudstones in Redback-2 (3788-3935 m depth) and Hovea-3 (1968Hovea-3 ( .6-1995 boreholes, stored at the Geological Survey of Western Australia's Core Repository in Carlisle, WA. A total of 48 samples were collected with their distribution shown in Figures 2 and 3, with a complete list of samples, methods and results in the Supplemental data. The rock chips were sectioned, placed in 25 mm round mounts of epoxy resin, polished with and examined by reflected light optical microscopy to reveal pyrite morphology in its textural context and guide LA-ICPMS analysis. Rock powder for whole-rock analyses was micro-drilled and milled from offcuts. A general description of lithofacies and mineralogy has been adopted from the Redback-2 and Hovea-3 well-completion reports (Origin Ltd, 2003, 2010, as well as the study by Jafary Dargahi and Rezaee (2013). X-ray diffraction was used to determine the composition of siderite nodule and scanning electron microscopy (SEM) with energy dispersive x-ray spectrometry (EDS) for some mineral identification and backscattered electron imaging.
In both cores, the Permian dark grey and black fine mudstones, which are interbedded with silty sandstones, and in places bioturbated, are composed mostly by clays, with variable content of quarts and pyrite. The Triassic sediments are dark brown, brittle, microlaminated siliceous, fossiliferous and calcareous mudstones where siderite is present as thin layers, calcite mostly as nodules and pervasive cement or streak veins; dolomite presence has been inferred from the rhombic shape. The SPI/INI interface position is in agreement with the uppermost record of inertinite in the completion reports, i.e. at RB2-3804.14/3804.15 and at HO3-1980.95 m depth ( Figure 2).

Microfossils and the EPMEI
Abundant remnants of sea bottom creatures have been observed under reflected light in the Hovea-3 Permian sediments from 1983.2 to 1991 m depth. The fossils are best preserved in the middle of the interval but frequently broken next to enclosing sandstones. The only foraminifer identified by Patrick Quilty (personal communication) with little doubt is Nodosaria tereta Crespin 1958 (current accepted name Protonodosaria tereta Crespin 1958) ( Figure 5k) in HO3-1987.84. Some other species have been only tentatively identified, including bryozoans, sponge spicules (diactinal monaxons [simple and pointed at both ends]), foraminifera (Nodosaria ragatti, Tolypammina or Calcitirnella) and algae (Solenopora). Scarce pyrite in a subinterval 1983.20-1987.84 m is present as very fine framboids spread in a siliceous clastic matrix or inside the bryozoan cavities. In contrast, no marine fauna has been observed in the studied samples from Redback-2 Permian INI sandstone, except a few nodules of pyritised fossils (RB2-3814.5 m; also see anomalous samples).
In RB2, the previously defined Indeterminate interval about 5 m between 3804 and 3808.9 m at the top of the 35 m-thick INI, contains rare Protohaploxypinus microcorpus Zone palynomorphs with abundant black wood, inertinite, degraded amorphous organic matter, large siderite nodules and small-sized pyrite framboids (Lounejeva et al., 2021). This has recently been considered as the latest Permian mass extinction interval (Shi et al., 2022), and so we refer to this interval as the EPMEI. Despite only $50 km separating the boreholes, the Hovea Member and, consequently, the extinction interval in the Hovea-3 borehole are thinner, <3 m, and should be constrained between the last bryozoans at 1983.50 m and the first Triassic bivalves at 1980.85. This is within the P. microcorpus Zone (1981-1984.2 m) and only $10 m above the Dongara sediments, appearing at 1992.5 m depth .
Further, we correlate the data by a relative depth assigning 'level zero' at RB2-3806.5 and HO3-1980.95 m corresponding to the prominent negative shift in d 13 C values and the lowest organic carbon content in both boreholes (Lounejeva et al., 2021;.

Bulk sediment geochemical composition
The full data set of bulk geochemistry (Supplemental data) includes the results of 43 bulk analyses of major and trace elements for Redback-2 sediments, and 42 results for Hovea-3 sediments for rare earth elements (REEs), boron and an additional analysis for the bulk sulfur isotope values (d 34 S), total carbon and sulfur. The results of basic statistics, enrichment factors and reference values for the Average Black Shale and Post-Archean Australian Shale are provided in the Supplemental data, and geochemical parameters most relevant for this study are shown in Figure 4.
In both boreholes, the major-element composition of the Permian inertinitic interval is distinct from the Triassic sapropel, whereas the EPME contains the various swings and spikes in both isotopic and elemental ratios. A double or triple increase above the EPMEI in content of elements with affinity to pyrite (Fe, S, Ag, Bi, Cd, Cu, Mo, Te, Zn, Sb, Tl and TOC), carbonates and phyllosilicates (Ca, Sr, Mn, La, U, B) contrasts with the decrease in immobile elements (Hf, Zr, Th, Ti, Ta, Nb), Mg and K. Calcium increases by an order of magnitude, from an average of 0.5 wt% in the upper Permian black shales to 9.5 wt% within and above the EPMEI, in the Triassic calcareous and fossiliferous mudstones. Accordingly, Ca is followed by Sr and, to less extent, by Mn. Sodium content is significantly lower in Hovea-3 (0.2 wt%) than in RB2 (0.5 wt%). Pyrite-like elements correlate well with each other, REEs in HO3 correlate with Ca, P, Y and U (R 2 > 0.9), and alkalis (Li, Cs) correlate with the post-transition metals (Al, Ga, In).
A major increase in several ratios argued by other authors as indicative of the paleoenvironment (Dymond et al., 1992;McKay & Algeo, 2013;Rem ırez & Algeo, 2020;Tribovillard et al., 2006) also occurs above the EPMEI. In particular, the increase in ratios indicative of redox (S/Fe, Fe/Al, Mo/Al, Cu/Zn, La/Ce, Mo/U), productivity (TOC, $2.2-4.5 wt%; P/Ti, 548-1742 mg/g; Ag, 0.18-0.43 mg/g; Ag/Ba), water depth (Mn/Fe, Mn/Ti), salinity (B/Ga, Sr/Ba) and hotand-arid climate (Sr/Cu) correlate with each other in Triassic samples. The average Ba content ($534 mg/g) in Permian samples decreases ($312 mg/g) in the Triassic samples and has a negative correlation with organic carbon, sulfur and silver contents (Figure 4). Other insights from the bulk geochemistry come from REEs in HO3, and siderite nodules and some enrichment factors in RB2. The HO3 sediment PAAS-normalised REY patterns (Figure 4) are almost flat, but with a moderate change noticeable around the SPI/INI boundary. Below 1980.95 m, a Ce positive anomaly (calculated following Tostevin, 2021) is barely perceptible, but in the overlying 5 cm, a pronounced Ce anomaly and a positive Y anomaly appear within a fractionated HREE-enriched pattern that is strongly correlated with P and Ca; these subsamples also contain four to four times more Ni, Co, Sb, Mo and Mn than the underlying samples. The overlying 5 cm still preserve a subtle Y anomaly but a Ce anomaly is reduced, and REEs correlate with the highest Ca (24%), Sr, total carbon and Y/Ho ($42).
Aluminium contents vary in RB2 from 6 to 10 wt%. The Al-normalised patterns of the Hovea Member sediments are compared with the Average Black Shale (ABS, Ketris & Yudovich, 2009) by calculating enrichment factors (EF element ¼ R SAMPLE / R ABS .). The comparison reveals that alkalis (Li, Rb, Cs), REE and Pb are moderately above (EF ¼ 2-4), and P, Cd, Te, Sb and immobile elements are well below the ABS (EF < 0.4). The bulk Ni content throughout the RB2 sequence is within the reference range for shales (40-80 mg/g, EF$1); however, a few Permian horizons are exceptionally enriched in arsenic (EF ABS >5) and, to less extent, Ni and Co, whether compared with the ABS, the Post-Archean Australian Shale (PAAS; Nance & Taylor, 1976) or the Cariaco Basin shelf shale (Martinez et al., 2010).

Pyrite types
Representative types of pyrite analysed during this study are shown in Figure 5 (photos of each sample, including some microfossils, are shown in the Supplemental data). We analysed single framboids, framboidal aggregates (or clusters) and euhedral disseminated crystals <30 mm ( Figure  5a-d). The euhedral marcasite-pyrite intergrowths in the d 13 C-EPMEI interval and a few pyrite nodules (Figure 5g, i, j) have been characterised.

LA-ICPMS results
The calculated contents of Mn, Co, Ni, As, Mo, Cu, Zn, Cd, Mo, Te, Pb, Se, Bi, Tl and Ag in the analysed pyrites from both boreholes are shown in Figures 6 and 7. Using major and minor element changes, the EPMEI was split into upper and lower, and the low-profile interval 3823.8-3835 m in the RB2 INI interval separated to assess general trends ( Table 1).
The downhole trends of TESPy are similar in Redback-2 and Hovea-3. In both boreholes, there is an interval in the late Permian inertinitic interval below the d 13 C anomaly where pyrite is enriched in elements with pyrite affinity, namely As, Ni, Co, Mo, Tl, Ag and Sb. In RB2, this interval comprises at least 15 m of pyritic mudstones and black shales (from 3806.5 down to 3821 m depth) that correlates with about 5 m (1983.5-1988.5 m) of fossils and trace-element-rich pyrite in HO3 mudstones. In both boreholes, the concentration of these elements decreases by several orders of magnitude in the Triassic sapropelic pyrite, whereas, Mn, Cu, Zn and Cd increase significantly in the Triassic interval.

Discussion
Pyrite trace-element geochemistry across the decline in d 13 C values The most noticeable feature of TESPy patterns in both boreholes (Figure 7), during the EPMEI in the lowermost Triassic is a drop by an order of magnitude from the high concentration of Ni-Co-As in the range of thousands of micrograms per gram in the latest Permian. The TESPy concentrations in the Triassic sediments are twice as high in HO3 as in RB2 but still within the Paleozoic-Mesozoic range . This main change coincides with (1) the global organic carbon cycle perturbation, (2) a change in regional depositional environment and (3) the rock source and the nature and availability of organic matter.

Constraints on the depositional setting and paleoenvironment of the Hovea Member
A general deepening of the Perth Basin in the Early Triassic, coincident with rifting off the western margin of the Yilgarn Craton, has been identified previously using several lines of structural and geophysical evidence (Jones, 2011). The macro-and micro-textural features of the sapropel (e.g. visible bivalves and brachiopods, abundant spiny acritarchs, elevated organic carbon, etc.) provide strong evidence for a marine environment from the onset of the Triassic. In addition, the prevalence of the microlaminated carbonate texture in the sapropel (likely microbial in origin), the presence of frequent diagenetic pyrite nodules and paleoenvironment indicators are consistent with a deeper, saline, anoxic and productive environment (this work; Georgiev et al., 2020;Lounejeva et al., 2021;.
However, ascertaining the depositional environment of the Permian sections of the Hovea Member (i.e. the inertinitic interval) has proven to be less straightforward. The overall sedimentological and structural development of the Perth Basin was multi-faceted, with differential deepening across the graben at various times (e.g. Haig et al., 2017;Song & Cawood, 2000). This was especially so in the initial stages of basin opening during the late Permian (Dillinger et al., 2022). Consequently, the depositional environment of the inertinitic interval of the Hovea Member, is argued, with some authors favouring a fully marine section and others postulating a brackish to freshwater regime (e.g. Jafary Dargahi & Rezaee, 2013;Georgiev et al., 2020;Metcalfe et al., 2008Metcalfe et al., , 2013Shi et al., 2022;. We consider the various lines of evidence presented in this and previous studies and discuss them by individual wells below. In Hovea-3, the SPI/INI interface has been recognised at 1980.95 m depth, and the inertinitic interval below contains abundant microplankton (mostly freshwater algae), which  used to interpret a nearshore environment. More recently, Georgiev et al. (2020) postulated shallowing toward the INI top based on the recognition of specific terrestrial organic matter in the Hovea-3 core. Our interpretation is a predominantly brackish environment for the inertinitic interval based on intermediate values of Sr/Ba (0.2-0.5) in both boreholes and B/Ga (4-6) in Hovea-3, between the thresholds for paleosalinity defined by Wei and Algeo (2020). We place greater emphasis on the B/Ga ratio owing to the potential spurious influence of carbonate on Sr/Ba ratios in a given section ( Figure 4).
As for Redback-2, rare marine invertebrates and palynofacies indicative of nearshore marine have been reported for the Redback-2 3804-3806 m depth interval (Shi et al., 2022;Origin Ltd, 2010), i.e. only in the higher horizons of the inertinitic interval and above the decline in d 13 C values (Lounejeva et al., 2021). Nevertheless, down to 3827 m depth, the organic matter is destroyed, and palynofacies are very rare and poorly preserved, rendering palynological interpretation difficult and varying between questionable estuarine and terrestrial, although marine influence has been suggested based on the presence of rare spinose acritarchs (Lounejeva et al., 2021;Origin Ltd, 2010). Therefore, we draw on the interpretation of Jafary Dargahi and Rezaee (2013) who suggested a slight sea-level rise culminated at $3812-3814 m depth and followed by a sharp shallowing upward 3808 m, based on gamma-ray response from pyritic mudstone and black shale dominating the entire inertinitic interval. Jafary Dargahi and Rezaee (2013) defined pyritic mudstones as clays with abundant pyrite nodules and veins of likely diagenetic origin, indicating a reducing environment, and the black shale as organic-rich clays slightly silty with some pyrite indicating calm anaerobic environment. This conflicts with palynology, but given the presence of pyritised nodules of sponge spicules localised in this study in RB2-3814.5, we also consider deepening. However, our lithogeochemical and mineralogical data are inconsistent with a 'normal marine' environment, i.e. a low carbon and sulfur profile (TOS$0.5 wt%, TOC$2 wt%), highly variable but generally high TOC/TOS ratios ($3.4), eventual enrichment of the bulk in d 34 S WR values À45 to À29‰ VCDT at $3813 to 3811 m depth) suggesting some restriction, brackish water indications and presence of unusual framboidal aggregates extremely enriched in As, Ni and Co, and deteriorated aggregates enriched in Se and Pb. Trace-element anomalism in pyrite aggregates from Redback-2 Ni-Co-As-rich pyrite (pyrite pearls and pyritised fossils).
Transitional metal contents in this late Permian interval are generally high (100-4000 mg/g), but some framboids and nodules in the inertinitic interval 3812-3814 m depth may contain up to 4.7 wt% Ni, 1.6 wt% Co, and 2.8 wt% As. Wellpreserved metal-rich framboids are shown in Figure 5h. The large diameter (35 ± 10 mm) of framboids is compatible with either dysoxic or anoxic sedimentary environments, whereas the lenticular shape of the aggregate and lack of deformation of the adjacent strata suggest free growth in Figure 7. Main trend of framboidal and disseminated pyrite geochemistry across the EPMEI and the declining d 13 C values (relative depth 0) preceding the Permian-Triassic boundary in Hovea Member sediments, Kockatea Shale, WA. In both boreholes, Hovea 3 and Redback 2, the late Permian pyrite preferentially concentrates Ni, Co, As and other elements in a shallow depositional environment with fluctuating oxygen minimum from oxic to anoxic, whereas the early Triassic is dominated by redox-sensitive elements like Cu and Mn proper for a deeper anoxic marine. unconsolidated mud. A closer look at the Ni-As-Co distribution in secondary electrons (Figure 8) confirmed the absence of gersdorffite (NiAsS) and revealed that metals are relatively enriched on framboidal edges and joints composed of likely secondary diagenetic pyrite. The level of trace-metal concentration is above the Proterozoic and Phanerozoic average pyrite but comparable with the Archean diagenetic pyrites .
Se-Pb-rich pyrite aggregates (mottled pyrite, RB2-3808.13). Abnormally high contents of Pb (0.4-2.4 wt%) and Se (0.05-0.2 wt%), correlate positively with each other, and contain elevated Cu ($0.14 wt%), Ni (0.6 wt%) and As (0.4 wt%), and depleted Bi and Te contents. The Se-Pb aggregates have been found at RB2-3808.13, $2 m below the shift in d 13 C values and within the interval of shallowwater environment interpretated from the gamma-resonance study. Pyrite in these aggregates acquired angular shapes, and space between the pyrite crystals is rather filled with iron hydroxide than with sulfide and degraded organic matter (Figure 5g; Supplemental data).
Manganese-rich pyrite-marcasite from the EPMEI. The LA-ICPMS analysis revealed a highly variable compositions of pyrite-marcasite euhedral crystals from the EPMEI in both boreholes ( Figure 5i; Table 2; 36 analyses). Marcasite traceelement contents are dominated by Mn (0.02-2.2 wt%; median ¼ 3500 mg/g), with much lower amounts of other elements, including Co and Ni. A characteristic of the Hovea Member, Mn-rich marcasite in mudstone matrix and apatite within siderite nodules, could be an indirect indication of acidified atmospheric aerosol. Formation of marcasite along with pyrite favoured by pH drop has been considered an indicator of potential acidification (Lounejeva et al., 2021) whereas atmospheric dust has been found to be an essential source of P, REE, Mn-Fe oxyhydroxide particles in both coastal and open ocean waters (Bayon et al., 2004;Greaves et al., 1999;Richon et al., 2018). Nenes et al. (2011) argued that atmospheric acidification of aerosols is the prime mechanism producing soluble phosphorus from soil-derived minerals and that these processes could promote formation of siderite-apatite nodules. Whether acidification is related to the STLIP activity and the hypothesised associated acid rains (Black et al., 2014), ocean acidification (Clarkson et al., 2015) or a marginal low dynamic environment is not yet clear. To understand whether the presence of iron bisulfides depleted in trace elements within the mass extinction interval is incidental, an ongoing study will include analysis of pyrite and marcasite reported from the deeper marine Permian-Triassic sediments, e.g. the abyssal black shales at the Ubara section in Japan   Regional tectonic arrangements at the end of Permian To reconcile the inconsistencies between interpretation of the depositional setting and geochemical parameters for the inertinitic interval, we consider the regional tectonics and invoke intensification of weathering and fast burial. Large et al. (2017) concluded that shales with high Ni, As, Mo or Co, which are several orders of magnitude higher than in this study, require a tectonic configuration that maximises weathering rates. The Permian Hovea Member metal-rich pyrite strata are coeval with sandstones containing Precambrian zircons likely sourced from the Yilgarn Craton (Cawood & Nemchin, 2000). Thus, the Perth Basin of the late Permian time could involve weathering of Ni-Corich Archean komatiites of the Yilgarn Craton and fluvial input carrying oxy-anionic metals such as As, whereas eventual shoaling of anoxic waters could promote formation of pyrite enriched in both transitional and oxi-anionic metals. In contrast, the Triassic sequences, deposited mostly in a deeper marine oxygen-depleted (euxinicanoxic) environment, lacking metal-enriched pyrite or Precambrian detritus could be sourced from the Ni-poor Pinjarra Orogen (Veevers, 2006;Veevers & Tewari, 1995).
Does the Perth Basin contain a global geochemical record of the EPME linked to STLIP activity?
The Hovea Member sediments display the prominent negative carbon isotopic excursion of 7‰ VPDB across the PTB (Lounejeva et al., 2021) coinciding with the regional rearrangement. The d 13 C (organic and carbonate) excursions are characteristic of the Paleozoic-Mesozoic carbon cycle perturbation worldwide (Korte & Kozur, 2010) and have been linked to massive release of methane and carbon dioxide owing to Siberian Traps volcanic intrusion into coal measures (Shen et al., 2011) and carbonates (Burgess et al., 2017). Spikes of high metal contents preceding the EPMEI are evident in the Hovea Member bulk sediments and pyrite. High toxic metal contents, i.e. Hg, Co and As, were also linked to STLIP ash loading and were postulated in northern Pangaea as the cause for the extinction of siliceous sponges in the ocean (Grasby et al., 2015). An exceptionally high Ni, Co and As content in the Hovea Member upper Permian sedimentary pyrite can also be explained by regional weathering input and diagenetic overprint rather than a global mechanism of metal delivery. However, a temporal link with a distant Ni source cannot be excluded, as global dispersion and loading of Ni-rich aerosol particles into the Panthalassic Ocean (Li et al., 2021) have been proposed and related to proliferation of methanogen Methanosarcina, which resulted in an abrupt anoxia expansion and the EPME (Rothman et al., 2014). Moreover, the highest content of Hg $437 ng/g and the first decline d 13 C org per 3‰ VPDB have been identified in Hovea-3 sediments at 1983.48 m depth (Sial et al., 2020), at the inception of the EPMEI. These changes occur above the last record of sponges and bryozoans at HO3-1983.67 and a 5‰ VPDB decline of d 13 C values within the next 2.5 m at the top of the inertinitic interval.
The long-term anoxia in Neo Tethys preceding the EPME, which has been inferred based on measurements of framboidal pyrite diameters (5 ± 2 mm) in Hovea-3, may be interpreted as 'deep marine' environments (Bond & Wignall, 2010). The smaller framboids in Hovea 3 sediments below the boundary, observed and analysed in this study, are however localised within the 7 m interval along with bryozoans and other fauna fossils or within a silty sandstone with some iron oxides contradicting the interpretation of long-term anoxia. Our observations rather support shoaling or short-term upwelling of deeper anoxic/sulfidic waters seeding some pyrite into sediments deposited near shore, a mechanism proposed by several authors to explain similar contradictions found around the PTB elsewhere (Kershaw & Liu, 2015;Shen et al., 2011). In contrast, the RB2 sediments contain framboidal pyrite of highly variable size formed likely in an oxic-dysoxic marine-influenced sediments and probably eventually separated from the rest of the basin.

Conclusions and implications
Whole-rock and pyrite trace element data record distinct changes in samples bridging the transition from the Permian to the Triassic in the Perth Basin. The pyrite successfully records ocean chemistry changes over an interval of less than 20 Ma years. Sedimentary pyrite and whole-rock chemistry from this interval may be used as a proxy for a shallow marine environment during the EPME. Special care must be taken regarding petrography and geochemistry in all pyrite proxy studies (i.e. lessons learned from Large et al., 2014, Gregory, 2020Gregory et al., 2015); virtually no sedimentary pyrite will ever have more than 1 wt% Co, Ni or As. Any pyrite with such concentrations is not sedimentary but may be diagenetic, if not hydrothermal. The whole-rock and pyrite geochemistry of the Permian/Triassic boundary section in borehole Redback-2, Perth Basin, supports a change from a shallow and relatively oxygenated latest Permian to a deeper early Triassic anoxic (euxinic) marine depositional environment. The shallow deposition setting in the Perth Basin was metal-enriched well before the late Permian extinction onset. Manganese, copper, zinc and cadmium deconvolution from the matrix should be considered with caution. constructive and critical comments from Ian Metcalfe and an anonymous reviewer.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by an Australian Research Council Discovery (ARC) grant to R. Large [DP150102578]. K. Grice acknowledges the ARC for DORA and DP grants for this work.