The Carnian Pluvial Episode (CPE) was a time of global environmental perturbations with significant biological turnover both in the ocean and on land (e.g., Simms and Ruffell, 1989; Dal Corso et al., 2020). The CPE was initially recognized in many European sedimentary basins as drastic lithologic changes, such as the demise of carbonate platforms and increased siliciclastic input (e.g., Gattolin et al., 2015). Multiple negative shifts of δ ¹³C are recognized across the CPE, suggesting a link with volcanic activity of Wrangellia LIPs (e.g., Dal Corso et al., 2018). Recent studies revealed that the intensification of the hydrological cycle also occurred not only in the western Tethys realm but in worldwide sedimentary basins (e.g., North China Plate; Lu et al., 2021; Boreal realm; Mueller et al., 2016; Gondwana region; Bernardi et al., 2018).

In contrast, the terrestrial records of the Pangean inland region are still controversial. Paleoclimate proxies, such as clay mineral profiles and paleosols, suggest only minor climate change with a generally arid to semi-arid background climate across the CPE from the United Kingdom (Baranyi et al., 2019a) and from Spain (Barrenechea et al., 2018), while data from the Germanic Basin and the Danish Basin did not show any humidity signal across the CPE (Visscher et al., 1994; Lindström et al., 2017). Enhanced continental weathering related to a more humid climate is only suggested for the early stages of the CPE in the Julian 2 (Baranyi et al., 2019b). Such observed discrepancies in the environmental signal across the CPE seem to be related to the contrasts in the climate of different regions and the limited age constraint of the terrestrial strata.

Abyssal deep-sea sediment provides a relatively continuous and consistent archive (e.g., Ikeda and Tada., 2014). An increase in the relative abundance of smectite among Fe-bearing minerals was detected in Japanese pelagic deep-sea chert across Julian/Tuvalian boundary (Nakada et al., 2014). Studies of Cenozoic sediments show that the terrigenous materials supplied to abyssal plains are mostly eolian dust generated from continental arid regions (e.g., Asahara, 1999). It is also noteworthy that a relatively thick, muddy silicious claystone interval (CS-1; Sugiyama, 1997) is observed within the smectite interval (Nakada et al., 2014). Along with the presence of CS-1, the thickness of intercalated mudstones drastically increased above the CS-1, and this can be correlated at least regionally (Sugiyama, 1997; Sugiyama et al., 2001). Because chert mainly consists of radiolarian tests and eolian mudstone (e.g., Hori et al., 1993), such an increase in mudstone may be related to increased eolian supply and/or decreased productivity (Sugiyama, 1997). The paleoenvironmental significance of these lithologic changes in the Carnian bedded cherts remains unclear.

In this study, the Carnian bedded cherts from the type section of Japanese radiolarian zones were investigated in order to elucidate the influence of the CPE on the Pangean interior. We seek to extract terrestrial environmental conditions by investigating the geochemistry of the intercalated mudstone of the bedded cherts, which concentrate terrigenous materials much higher than chert layers. The terrigenous content fluctuation was reconstructed throughout the Carnian by multivariate statistical analyses of major element composition. The increase of terrigenous materials flux implying reinforcement of eolian supply occurred. Our results thus firstly provide evidence that extensive aridification occurred in the inland of the Pangea across the CPE.

Early Triassic to Early Jurassic bedded chert is exposed in the Inuyama area, central Japan, as a part of the Mino-Tamba belt, Jurassic accretionary complex (Figure 1). The bedded cherts and the overlying Middle Jurassic clastic rocks are repeatedly exposed as a chert-clastic sequence truncated by bedding-parallel thrusts (Figure 1B; Yao et al., 1980; Matsuda and Isozaki, 1991). Since this bedded chert had an extremely slow sedimentation rate of 1–3 mm/kyr, and lacks coarse-grained terrigenous material for over 60 Myr, it is considered to have been deposited in a distal setting (Matsuda and Isozaki, 1991; Hori et al., 1993). The latitude of the depositional site was estimated as 12.3° ± 5.6°N or S and 19.0° ± 10.5°N or S for 237 Ma and 223.5 Ma, respectively based on the inclination of magnetic component of hematite (Figure 1A; Oda and Suzuki, 2000).

Bedded cherts are composed of rhythmic alteration of chert layers (cm scale) and intercalated mudstones. Chert layers are mainly composed of microcrystalline quartz and biogenic siliceous tests, whereas intercalated mudstones are mainly composed of clay minerals derived from eolian dust brown from the continental region (Hori et al., 1993). The relationship of chemical composition of chert layers and intercalated mudstones has been recognized as silica dilution by radiolarian test through the observation of negative correlation of SiO2 with other elements (e.g., Hori et al., 2000).

Bedded cherts are considered to be formed by a diagenetic segregation process as silica dissolution and transportation (Tada, 1991). The lithology of bedded cherts in the Inuyama area is mainly divided into two types: F type and B type which denote “fine grained quartz” and “biogenic clasts”, respectively (Imoto 1983; Sugiyama, 1997). F-type cherts show a few siliceous biogenic tests scattered in the fine-grained quartz matrix under microscopic observation with very thin (ca. 0.5–2 mm) intercalated mudstone layers and laminations. B-type cherts show concentrated biogenic tests under microscopic observation and are usually associated with thick (ca. 2–20 mm) intercalated mudstones. The difference in their formation mechanism may be related to changes in silica and terrigenous contents of the precursor sediment of bedded chert, formed by cyclic changes of sedimentary environments and/or diagenetic conditions (Sugiyama, 1997).

In this study, Section R and Section Q of Sugiyama (1997) in the Inuyama area are selected for detailed geological mapping and geochemical study in order to reconstruct paleoclimatic conditions of the entire Carnian stage (Figures 1, 2). Section R is the type section of Japanese radiolarian zone TR5A, which contains the most complete and unaltered sequence of the TR5A zone (Sugiyama, 1997). Section Q is located to the west of Section R and is one of the most complete and unaltered sequences of TR5B zone in Inuyama area (Sugiyama, 1997).

The depositional age of the Carnian bedded chert is constrained by radiolarian and conodont biostratigraphy, cyclostratigraphy, and carbon isotope stratigraphy (Sugiyama, 1997; Ikeda and Tada., 2014; Nakada et al., 2014; Yamashita et al., 2018; Tomimatsu et al., 2021). Sugiyama (1997) correlated interval (ca. 20 m thick) from the upper part of the radiolarian zone TR4B (Spongoserrula delhi) to the lower part of TR6A (Capnodoce–Trialatus) based on the first occurrence (FO) of the Carnian radiolaria Praeheliostaurus laevis, the FO of Carnian conodont Neogondolella foliata, and the FO of Norian radiolaria Capnodoce gilis in Sections N, R and Q in Inuyama area. Nakada et al. (2014) assigned the Julian/Tuvalian boundary within the Elbistanium gracile radiolarian zone between their samples R3-1 and R5-3. Tomimatsu et al. (2021) detected two negative excursions in carbon isotopes, NCIE-α and NCIE-β, in the TR5A zone within the Nicoraella? budaensis conodont zone in Section N, and correlated them to the global carbon isotopic excursion across the CPE.

We conducted a detailed three-dimensional outcrop observation and constructed a bed-by-bed scale continuous lithostratigraphy to avoid missing or duplicating the strata (Figure 3). The lithological column was made from the lower part of Section R to the upper part of Section Q, in ascending order. The F-type red chert is dominant with a white chert bed WC-2 of Sugiyama (1997) at the base. About 900 cm above WC-2, the chert changes to ca. 300 cm purple laminated interval. Above the purple chert interval, ca. 200 cm B-type chert is intercalated with the ca. 20 cm siliceous claystone bed CS-1 of Sugiyama (1997) at the base. The CS-1 between F-type chert to B-type chert is widely recognized, not only in Inuyama area, but also in Hisuikyo area, which is 25 km north to Inuyama area (Sugiyama, 1997; Sugiyama et al., 2001).

Eighty-one intercalated mudstone samples were collected from Sections R and Q to cover the Carnian interval (Figure 2). Fresh samples were handpicked for geochemical analyses. After cleaning in an ultrasonic bath, the samples were crushed using a tungsten carbide mortar and pestle then pulverized in an agate ball mill. The samples were heated at 550°C for 4 h and 950°C for 2 h to decompose organic matter and carbonate material respectively prior to analysis of major elements and loss on ignition was measured.

Major elements (SiO₂, TiO₂, Al₂O₃, Fe₂O₃, MnO, MgO, CaO, Na₂O, K₂O, and P₂O₅) were analyzed by X-ray fluorescence (ZSX Primus II, Rigaku Co., Japan) at Waseda University. To minimize matrix effects and mineral effects, the fused disk method was used with a sample to lithium borate flux ratio of 1:10. Calibration lines were created using reference samples GSJ (JDo-1, JSd-1, JSd-2, JSd-3, JSI-2, JLk-1, JCh-1, JA-2, JA-3, JB-1a, JB-2, JB-3, JG-1a, JG-2, JG-3, JGb-1, JP-1, JR-1, JR-2). The accuracy of calibration lines for all elements was better than 0.6%. The sample R39, which is located on ca. 60 cm below the bottom of CS-1, was hand-picked and measured twice because it showed very high CaO and P₂O₅ concentrations.

Some mudstone samples were observed using scanning electron microscopy (SEM-EDS, Hitachi S-3400N, Japan) at the Waseda University with an accelerating voltage of 15 kV. Soft X-ray photography was employed using a Naomi NX digital X-ray Transmission System (RF Co., Ltd, Japan) with tube voltage of 50 kV and tube current of 40 mA at the Waseda University.

The result of major element composition of the intercalated mudstone samples and data normalized by PAAS (Post Archean Australian Shale; Taylor and McLennan, 1985) are shown in Table 1 and Figure 4, and Supplementary Figure S1. To avoid the false correlation effect of silica dilution, TiO₂ was used to normalize the major element concentrations. The elemental abundance of the intercalated mudstones relative to TiO₂ is generally similar to PAAS, except for CaO, Na₂O, MnO (Figure 4). The value of MnO/TiO₂ of the studied samples is higher than that of PAAS, whereas CaO/TiO₂ and Na₂O/TiO₂ are distinctively depleted.

The presence of apatite has a significant effect on the CaO and P₂O₅ concentration. Figure 5 shows the Back Scatter Electron (BSE) image, Energy dispersive spectrometry, and scatter plot of CaO and P₂O₅ of the intercalated mudstones. Apatite fragments were observed in SEM. The concentration of CaO and P₂O₅ shows a clear positive correlation along with the P₂O₅/CaO of 0.75, the ratio of apatite (e.g. Takiguchi et al., 2006).

To evaluate the possible provenance of the studied materials, provenance discrimination diagram of Roser and Korsch (1988) were adopted (Figure 6). Almost all samples are plotted in the “Quartzose sedimentary provenance”.

To identify groups of elements with the same behavior, Q-mode cluster analysis were performed on the bulk chemical composition of the studied mudstones (Figure 7). Results of a cluster ordination of the different variables point to the different grouping of elements associated with siliciclastics (SiO₂, K₂O, MgO, Al₂O₃, Fe₂O₃, TiO₂, and Na₂O) and apatite elements (CaO and P₂O₅) along with MnO, which solely consists of a cluster.

To investigate latent information of the bulk chemical composition of the studied samples, PCA was performed (Table 2; Figure 8). The p-values of the Shapiro-Wilk test are 0.35 for clr (SiO₂), 0.44 for clr (TiO₂), 0.14 for clr (Al₂O₃), 0.45 for clr (Fe₂O₃), 0.86 for clr (MnO), 0.70 for clr (MgO), 0.09 for clr (CaO), 0.59 for clr (Na₂O), 0.04 for clr (K₂O), and 0.84 for clr (P₂O₅).

PC1 accounts for 38.9% of the variance of the clr-transformed dataset. PC1 has positive loading from Al₂O₃, K₂O, Na₂O, MgO, SiO₂, TiO₂, and Fe₂O₃, whereas it has negative loading from MnO, P₂O₅, and CaO. Elements positively loaded on PC1 belong to the siliciclastic cluster of above mentioned Q-mode cluster analysis. On the contrary, the elements belonging to the other two clusters (the cluster of manganese and apatite minerals) show negative loadings on PC1.

Stratigraphic variations of PC1 and Al₂O₃/TiO₂ ratio are shown in Figure 9. The PC1 and Al₂O₃/TiO₂ show a weak positive correlation (R = 0.3). Al₂O₃/TiO₂ ratio increased across the CS-1, and then decreased across the B type/F type boundary but this trend is within the fluctuation throughout the studied section. The stratigraphic variation of PC1 along with the other elemental percent data is shown in Supplementary Table S1.

PC2, which is statistically orthogonal with PC1, accounts for 23.1% of the variance of the clr-transformed dataset. PC2 is characterized by high positive loading from P₂O₅, CaO, and Na₂O, and negative loading from MnO and Fe₂O₃.

The depositional site of the studied bedded chert is located on the deep-sea plane, which was isolated from continental fluvial transportation by trench (Figure 1; Matsuda and Isozaki, 1991). Thus, terrigenous materials contained by the bedded cherts are considered to supplied from aeolian transportation (Asahara, 1999). The provenance discrimination diagram of Roser and Korsch (1988) suggests that the provenance of the studied samples is from “Quartzose sedimentary provenance” (Figure 6). This result is concordant with the supposed transportation mode of the bedded cherts, that is, the terrigenous materials were blown in from continental interiors (Hori et al., 1993).

The major elemental compositions of the studied intercalated mudstone samples reflect the varying proportions of lithogenic, biogenic, and hydrogenic components in the sediments. The results of Q-mode cluster analysis point to the different grouping of “manganese” (Mn), “apatite” (Ca, P), and “terrestrial” elements (SiO₂, K₂O, MgO, Al₂O₃, Fe₂O₃, TiO₂, and Na₂O) (Figure 7).

Mn in our samples is enriched relative to PAAS (Figure 4). MnO along with Fe₂O₃ shows large negative loadings in PC2. Mn hydroxides or oxides (e.g., MnO₂) are deposited rapidly in particulate form under oxic conditions, whereas under suboxic and anoxic conditions Mn is reduced to divalent (Mn²⁺) and forms soluble cations (Sholkovitz et al., 1992; Calvert and Pedersen, 1993). It is thus suggested that the enrichment of Mn may reflect an oxic depositional environment of this section. Alternatively, additional Mn may form under surface oxidized Mn-rich zones during diagenesis (e.g., Pedersen et al., 1986).

The variation of CaO and P₂O₅ in our dataset is controlled by apatite content because of the strong positive correlation of CaO and P₂O₅ with the P₂O₅/CaO ratio of 0.27–0.69 similar to apatite (Figure 5). Takiguchi et al. (2006) interpreted the main carrier of apatite in this succession was authigenic in origin and formed during early diagenesis based on their P₂O₅/CaO ratio lower than that of conodont apatite (∼0.74; e.g., Ruttenberg and Berner, 1993; Van Cappellen and Berner, 1991). The variation of CaO, P2O5, and MnO characterize PC2, which contains 23.1% of the total variance of the studied dataset. CaO and P₂O₅ show large positive loading whereas MnO shows negative loading, indicating the opposite influence of phosphate and manganese. This relation agrees with the chemistry variation under manganese redirection (Dickens and Owen, 1994; Betzler et al., 2016; Bialik et al., 2020).

The total concentrations of “manganese” and “apatite” elements are generally less than 5%, except for the extremely low PC1 samples at ca. 20 cm below the thick claystone CS-1 (Sample Ra39-1, Ra39-2, Ra40). These levels correspond to the unusually high values of CaO and P₂O₅ contents with “apatite” elements content of ca. 7–18%, which can be recognized as an event bed (Figure 5 and Figure 9). The P₂O₅/CaO ratio of these three samples are within 0.6–0.7, lower than that of conodont apatite (Takiguchi et al., 2006). This unusually high apatite concentration may be caused by a pelagic marine environmental shift which led to precipitate abiotic apatite in the sediments, but this topic remains outside the scope of this paper.

For “terrestrial” elements, although biogenic Si is the main component of radiolarian chert, Si in intercalated mudstone is considered to be mainly derived from siliciclastic materials due to migration of biogenic silica to adjacent chert during diagenesis segregation (e.g., Tada, 1991). During this diagenetic segregation, opal-A in siliceous fossils are extensively dissolved and reprecipitated as tiny opal-CT particles within the pore space mainly in higher dissolved Si area, and thus Si migrated from proto-mudstone intervals to adjacent proto-chert intervals (e.g., Tada, 1991). We cannot rule out contamination of biogenic Si in the intercalated mudstones, but the contribution should be negligible because only well-sorted mudstones were selected for the geochemical analysis.

PC1 gradually increases below CS-1 from -1.5 to 1, except for the extremely negative values ca. 20 cm below the CS-1 (Sample Ra39-1, Ra39-2, Ra40). The PC1 score suddenly increases to 3.4 at the CS-1 horizon and remains constantly high values around 1.5 at the following ca. 1 m thick B-type bedded chert interval. The PC1 values decrease across the B type and F type boundary, and then seem to maintain a steady-state with higher variations than Julian to the end of Tuvalian.

Since PC1 has positive loading on all the “terrestrial” elements and negative loading on “manganese” and “apatite” elements, PC1 can be regarded as a latent variable that denotes the degree of relative contribution of terrigenous materials against marine-derived components. Besides, PC1 scores do not show a negative correlation with Al₂O₃/TiO₂ ratio (R = 0.3). Al₂O₃/TiO₂ ratio is known to reflect biological productivity on the abyssal fine-grained marine sediments since excess Al is scavenged by the biological pump (Murray and Leinen, 1996; Dymond et al., 1997). The studied bedded cherts are estimated to be deposited at the low latitude of abyssal deep-sea Panthalassa (Ando et al., 2001; Uno et al., 2015) thus the fluctuation of Al₂O₃/TiO₂ ratio of the current dataset is expected to reflect productivity. Therefore, the non-negative correlation observed between PC1 and Al₂O₃/TiO₂ suggests that the changes in PC1 are not mainly regulated by biological productivity fluctuations.

In addition, lithology of the bedded chert suddenly changes from F-type to B-type just above CS-1, with corresponding increased mudstone thickness (Sugiyama, 1997) and an increase in PC1 (Figure 9). Two possible factors that regulate the type of bedded chert (F type or B type) that is, the changes in burial flux of biogenic silica and/or siliciclastics (Sugiyama, 1997). Because of the increased Al₂O₃/TiO₂ ratio across the CS-1, a coincident increase in PC1 score may reflect increased terrigenous supply rather than decreased productivity (Figure 9).

Equally, considering the negligible contribution of “manganese” and “apatite” elements in samples other than Ra39 and Ra40, the increased mudstone bed thickness across CS-1 likely reflects increased flux of terrigenous material, probably eolian dust in the equatorial pelagic Panthalassa (Hori et al., 1993).

The Carnian Pluvial Episode has been characterized by abrupt siliciclastic input in Tethyan regions (Schlager and Schollnberger, 1974; Simms and Ruffell, 1989; Simms and Ruffell, 1990). This study suggests an increased terrigenous input also occurred in the abyssal deep-sea Panthalassa across the CPE. The increases in siliciclastic input at continental and coastal regions are considered to have been a consequence of enhanced hydrological cycling and consequent increased runoff. However, as discussed above, the terrigenous input in the abyssal plane reflects increasing eolian supply (Hori et al., 1993). Since this eolian dust is likely blown from arid continental areas, the increasing terrigenous supply detected in this study may imply the extensive aridification of the Pangean supercontinent.

This idea seems to contradict the existence of smectite across the CPE which was interpreted as evidence of continental humidification (Nakada et al., 2014). However, a megamonsoonal climate was known to prevail during the Triassic owing to the configuration of supercontinent-superocean (e.g., Kutzbach and Gallimore, 1989). Observational and climate modeling studies at the present day have indicated the humidification of tropical regions and aridification of subtropical regions under global warming (i.e., “Dry gets drier, wet gets wetter”; Held and Soden, 2006). The expansion of arid regions and associated aridification of previously humid regions could account for production of eolian dust containing smectite. The arid and seasonal influences around Carnian are detected in some Gondwana successions (Ruffell et al., 2016).

Expansion of arid regions could have also increased the emission of eolian dust with smectite across CS-1. Eolian deposits temporarily appeared in the Fundy (Canada), Iberian (Spain) and Argana (Morocco) basins during the Carnian, possibly near the end of the CPE, based on the limited biostratigraphy of terrestrial vertebrates and plant fossils (Arche and López-Gómez, 2014; Sues and Olsen, 2015). Interestingly, the onset of the smectite-rich interval slightly preceded the increased mudstone bed thickness across the CS-1 (Figure 10; Sugiyama, 1997; Sugiyama et al., 2001; Nakada et al., 2014). This temporal relation is consistent with the formation of smectite during the CPE and subsequent increase in eolian dust flux owing to the expansion of arid regions in the previously humid regions. Besides, weathering of mafic volcanic sources could also account for the smectite.

Increasing wind gustiness by global cooling and increased latitudinal temperature difference is also known as a main cause of increased eolian flux increase on glacial-interglacial timescales (e.g., McGee et al., 2010). Across the CPE, high-resolution sea surface temperature profile derived from oxygen-isotope of conodont apatite show a possible two-pulse warming event in the Nanpanjiang Basin in South China Block (Sun et al., 2016), while other low-resolution records from western Tethys show only one pulse (e.g., Hornung et al., 2007). In the South China Block, the onset of the CPE, namely NCIE-1, coincided with the first pulse of warming of ∼4°C, and followed by a short cooling phase across the Julian/Tuvalian boundary, and then by a prolonged and more intense second phase of warming (Figure 10; Sun et al., 2016). According to the biostratigraphy and carbon isotope correlation (Figure 10; Tomimatsu et al., 2021), the CS-1 level can be correlated with this short cooling phase across the Julian/Tuvalian boundary. Thus, cooling and associated wind gustiness, in addition to the expansion of arid regions, could have also enhanced the eolian dust emission recorded as the lithologic change in this study section (Sugiyama, 1997).