Biotic and environmental changes in the Panthalassa Ocean across the Norian (Late Triassic) impact event

Abstract

Stratigraphic records of impact ejecta preserved in a pelagic deep-sea setting occur within Upper Triassic successions of the subduction-generated accretionary complexes of central Japan. A significant biotic turnover in radiolarians occurred during the ~ 300 kyr time interval after the impact event, which is characterized by a remarkable reduction in the burial flux of radiolarian silica. However, the nature of the environmental conditions at this time remains unclear. To investigate the environmental changes that triggered a decline in radiolarian burial flux after the impact event, geochemical proxies (major, trace, and rare earth elements) were applied to the middle–upper Norian (Upper Triassic) bedded chert succession of the Mino Belt, central Japan. A progressive environmental deterioration is evident from (1) a post-impact shutdown of burial flux of primary and silica- and apatite-secreting organisms; and (2) a subsequent abrupt increase in chemical weathering associated with a sustained reduction in the burial flux of radiolarian silica. No significant redox changes were observed across the impact event. The continental weathering proxies suggest a transient increase in weathering intensity occurred during the decline of radiolarian burial flux, likely in response to a short-term warm and humid period. Our results delineate a remarkable record of progressive environmental changes in the Panthalassa Ocean after this large impact event.

1 Introduction

An Upper Triassic (Norian) impact event has been inferred from anomalous concentrations of platinum-group elements (PGEs) and a negative osmium (Os) isotope excursion, in addition to occurrences of microspherules and Ni-rich magnetite, in a claystone layer in an Upper Triassic bedded chert succession in the Sakahogi section, central Japan (Onoue et al. 2012; Sato et al. 2013; Sato et al. 2016) (Fig. 1). Previous paleomagnetic studies of the Triassic bedded chert succession in the Sakahogi section suggest that these sediments accumulated in a pelagic, open-ocean setting within a low- to mid-latitudinal zone of the Panthalassa Ocean (Uno et al. 2015) (Fig. 2). The late middle Norian age of the claystone layer (Onoue et al. 2016a; Yamashita et al. 2018) suggests that the PGE anomalies originate from an extraterrestrial source, related to an impact event that formed the 90 km-diameter Manicouagan crater in Canada at 215.5 Ma (Clutson et al. 2018) (Fig. 2). Studies of PGEs and Os isotopes have revealed that the anomalously high PGE abundances in the lower sublayer claystone resulted from a large chondritic impactor with a diameter of 3.3–7.8 km (Sato et al. 2013; Sato et al. 2016).

Fig. 1

Map showing the study area. a, b Geographic maps of the study area in the Inuyama area, Mino Belt, central Japan. c The Sakahogi section along the middle reaches of Kiso River

Fig. 2

The paleogeography in the Late Triassic. Approximate location of the Manicouagan crater and the inferred depositional area of the bedded chert in the Mino Belt in low-latitude zones of the Panthalassa Ocean (Uno et al. 2015). The red circles represent the approximate paleo-locations of the Late Triassic sites. The orange arrows are the simulated surface winds from Kutzbach and Gallimore (1989)

Onoue et al. (2016a) showed that extinctions of middle Norian radiolarian species occurred in a stepwise fashion in the ~ 1 Myr interval above the ejecta horizon, which was associated with the radiation of late Norian radiolarians (Fig. 3a). Furthermore, high-resolution paleontological and geochemical data also revealed that two paleoenvironmental events occurred during the initial phase of the radiolarian turnover interval. The first event (E1) involved post-impact shutdown of primary productivity reflected by a remarkable decline in the amount of biogenic silica. The second event (E2) was characterized by a large and sustained reduction in the burial flux of radiolarian silica and the proliferation of siliceous sponges. This E2 event lasted until ~ 300 kyr after the impact. Although the relatively long period of the E2 interval (~ 300 kyr after the impact) largely excludes the possibility that the decline was triggered by instantaneous environmental stresses (e.g., extended darkness, global cooling, or acid rain) that would have been caused by a bolide impact, the primary cause of this decline remains uncertain.

Fig. 3

Biostratigraphy and chemostratigraphy of the bedded cherts in the Sakahogi section (Onoue et al. 2016a; Sato et al. 2016). a Stratigraphic profiles of organic carbon isotopes, mass accumulation rates of biogenic silica, and radiolarian biostratigraphy. The mass accumulation rate of biogenic silica was estimated by assuming a constant sedimentation rate of 1.1–1.6 mm kyr⁻¹ in the middle–upper Norian chert succession (Onoue et al. 2016a), which is an alternative to a previous model assuming different sedimentation rates between chert and claystone (Hori et al. 1993). b Stratigraphic profiles of PGE concentrations in the claystones of the studied section. NH52-R2 indicates the lower sublayer of claystone (ejecta layer) with microspherules and Ni-rich magnetite

In this study, we examined variations in the compositions of major, trace, and rare earth elements (REE) in bedded cherts in the Sakahogi section to identify the environmental changes responsible for the decline in radiolarian burial flux after the middle Norian impact event. The bedded cherts were composed originally of biogenic silica, apatite, barite, clastic lithogenic materials, and hydrogenous materials such as Fe–Mn oxides. The geochemistry of the cherts has been extensively studied in the context of the depositional environment and global environmental change (e.g., Murray et al. 1992; Hori et al. 1993; Murray 1994; Kato et al. 2002; Takiguchi et al. 2006; Hori et al. 2007). The stratigraphic variations of major and trace elements in Triassic bedded cherts from Japanese accretionary complexes are widely used as proxies to trace changes in (1) hinterland components, (2) degree of chemical weathering in hinterland regions, (3) paleoproductivity, and (4) oceanic redox conditions. In this study, we assessed temporal variations in the compositions of major, trace, and rare earth elements (REE) in the Triassic bedded cherts to investigate the biotic responses to environmental changes that occurred after the impact event in the Panthalassa Ocean.

2 Geological setting and stratigraphy

The Mino Belt consists of Jurassic accretionary complexes in central Japan. The accretionary complexes in the Mino Belt consist of two coherent units (i.e., the Samondake and Kamiaso units) and five melange units (i.e., the Sakamoto-toge, Funafuseyama, Kuze, Nabi, and Kanayama units). The coherent units consist of imbricate thrust sheets of sedimentary sequences that largely retain their primary stratigraphic coherency (e.g., Matsuda and Isozaki 1991; Isozaki 2014).

The study area in the Mino Belt is located in the Inuyama area, central Japan (Fig. 1). This area is in the southern part of the coherent Kamiaso Unit of the Mino Belt, which strikes E–W to NE–SW. The Kamiaso Unit in this area consists of thrust sheets of sedimentary sequences containing Triassic to Lower Jurassic bedded cherts and overlying Middle Jurassic clastic rocks (i.e., ocean plate stratigraphy; Isozaki 2014). The ocean plate stratigraphy is interpreted to have accumulated in a pelagic, deep-sea setting below the carbonate compensation depth and within a tapering wedge of distal, trench-fill turbidites at a subduction zone, respectively (Matsuda and Isozaki 1991). The absence of carbonate rocks and coarse-grained terrigenous material in the bedded cherts suggests that its primary depositional site was deeper than the carbonate compensation depth, and far beyond the transport distance of terrigenous clastic grains. Previous paleomagnetic studies of the Triassic bedded chert succession in the Mino Belt suggested that these sedimentary rocks accumulated in a pelagic, open-ocean setting within a low- to mid-latitudinal zone of the Panthalassa Ocean (Ando et al. 2001; Uno et al. 2015).

The Triassic to Jurassic cherts occur in 4 distinct thrusts sheets named CH-1, -2, -3, and -4 in structurally ascending order (Yao et al. 1980). The Sakahogi section consists of a ~ 26 m-thick sequence of bedded cherts in the CH-2 thrust sheet. Individual chert beds in the Sakahogi section range in thickness from 1 to 10 cm, and are typically red to greenish gray, or occasionally white in color. The claystone layer in which PGE anomalies have been documented ranges in thickness from 4 to 5 cm and extends laterally for at least 90 m (Sato et al. 2016). The claystone comprises lower and upper sublayers (Onoue et al. 2012; Sato et al. 2013). The lower sublayer (NH52-R2) (Fig. 3b) contains microspherules within the clay mineral matrix (mainly illite), cryptocrystalline quartz, and hematite. The upper sublayer (NH52-R3 to NH52-R7) is composed of undisturbed clay minerals (illite) and cryptocrystalline quartz. The upper sublayer is further divided into five sub-categories based on variations in lithology, SiO₂ content, and relative abundance of biotic components (Sato et al. 2013). The SiO₂-rich sub-categories, NH52-R4 and NH52-R6, contain small amounts of radiolarians, sponge spicules, and conodonts.

Biostratigraphic and magnetostratigraphic studies (Sugiyama 1997; Onoue et al. 2012; Uno et al. 2015; Yamashita et al. 2018) have revealed that the claystone layer occurs in upper middle Norian bedded chert. Given that the average sedimentation rate of the middle Norian chert, estimated from the measured thickness of the chert and the duration of deposition (Onoue et al. 2016a), is 1.1 mm kyr^–1, deposition of the claystone layer occurred ~ 1 Myr before the middle–upper Norian boundary (~214 Ma; Ogg 2012). Although the ages of substage boundaries in the Norian remain unclear due to a lack of reliable radio-isotopic dates and uncertainties in the magnetostratigraphic correlation between the Newark astrochronological polarity time scale (APTS) and marine-zoned strata (Ogg 2012), Onoue et al. (2016a) estimated the average sedimentation rate of the middle Norian chert based on the measured thickness (3.8 m) of the chert and the duration between the lower/middle (217.42 Ma) and middle/late Norian (213.97 Ma) boundaries (Ogg 2012).

3 Methods/Experimental

3.1 Major and trace elements analyses

To understand the environmental changes that triggered the decline in radiolarian burial flux after the middle Norian impact event, fifty-seven chert samples were collected across the ejecta layer for whole-rock geochemical analysis (Fig. 3). We also collected 6 claystone samples from the lower and upper sublayers. Veins and strongly recrystallized/weathered parts of the samples were avoided to minimize the effects of diagenetic and metamorphic overprinting on the sediment geochemistry. The chert and claystone samples were crushed, and the resulting fragments were carefully handpicked. These fragments were then powdered in an agate mortar and ball mill.

Major element (Ti, Al, Fe, Mn, K, and P) abundances were measured using a Rigaku ZSX Primus II X-ray fluorescence (XRF) spectrometer at the University of Tokyo, Japan, following the methods described by Kato et al. (1998) and Yasukawa et al. (2014). After drying the powdered samples at 110 °C for ~ 12 h, loss-on-ignition (LOI) values were calculated from the weight loss after ignition at 950 °C for > 6 h. Fused glass beads for XRF analysis were made from a mixture of 0.400 g of ignited sample powder and 4.00 g of lithium tetraborate (Li₂B₄O₇) flux at ~ 1190 °C for 7 min in a Pt crucible. The standard data were generally within 3% (relative % difference) of accepted values for the reference basalt JB-2 (Imai et al. 1995), issued by the Geological Survey of Japan (GSJ).

Major (Mg, Ca, and Na), trace, and REE abundances were determined using an inductively coupled plasma quadrupole mass spectrometer (ICP–QMS; Agilent 7500c) at the University of Tokyo, following procedures described by Kato et al. (2005, 2011) and Yasukawa et al. (2014). After drying powdered samples at 110 °C for ~ 12 h, 0.100 g chert and 0.050 g claystone samples were dissolved by HNO₃–HF–HClO₄ digestion in tightly sealed Teflon PFA vessels, and then heated for several hours on a hot plate at 130 °C. The dissolved samples were progressively evaporated at 110 °C for 12 h, 160 °C for 3 h, and 190 °C until dryness. The residues were then dissolved in 4 mL of HNO₃ and 1 mL of HCl, and the solution was diluted to 1:1000 for the cherts and 1:2000 for the claystones (on a weight basis). Standard analyses were generally within 5% of the accepted values for JB-2 (Imai et al. 1995; Makishima and Nakamura 2006; Lu et al. 2007).

4 Results

4.1 Major and trace elements

Major and trace element data are listed in Tables 1, 2, and 3. Previous sedimentological and geochemical studies revealed that the chert beds in the Sakahogi section consist of SiO₂-diluted siliceous shale beds, reflecting fluctuations in radiolarian test abundances (Hori et al. 1993; Takiguchi et al. 2006; Sato et al. 2013). In fact, the chert bed samples contain > 90 wt.% SiO₂, which means the other major and trace elements are diluted by a large amount of biogenic silica. In particular, abundances of the major lithogenic elements, such as Al₂O₃, K₂O, and TiO₂, show a negative correlation with SiO₂ contents. This is attributed to dilution with nearly pure biogenic silica (e.g., radiolarians), as exemplified by the SiO₂–Al₂O₃, SiO₂–K₂O, and SiO₂–TiO₂ plots (Fig. 4). We assumed the SiO₂/Al₂O₃ ratio of terrigenous material in the Sakahogi section was 4.1, which is that of a sample with the highest Al₂O₃ content in the section (Fig. 4) and comparable to the ratio (~ 4.3) of upper continental crust (UCC; Rudnick and Gao 2014).

Table 1 Major element compositions of chert and claystone samples in the Sakahogi section

Table 2 Trace element compositions of chert and claystone samples in the Sakahogi section

Table 3 Rare earth element compositions of chert and claystone samples in the Sakahogi section

Fig. 4

Plots of SiO₂ vs. Al₂O₃, K₂O, and TiO₂ for the chert and claystone samples. Chert (red open circles) and lower (blue open circles) and upper (blue filled circles) claystone samples exhibit negative correlations, indicating dilution with biogenic silica (e.g., radiolarians). TE terrigenous material in the Sakahogi section, UCC upper continental crust, PAAS post-Archean average Australian shale. Data sources: UCC (Rudnick and Gao 2014); PAAS (Taylor and McLennan 1985)

The data for the chert beds demonstrate that Al shows an excellent positive correlation with high-field-strength (HFS) elements such as Ti, Nb, Ta, and Th (Fig. 5). Considering the fact that Al, Ti, Nb, Ta, and Th are not significantly mobilized during post-depositional processes, including diagenesis and weathering, these correlations indicate that the majority of these elements are of detrital origin. The Ti/Al, Nb/Al, Ta/Al, and Th/Al ratios are identical to those of average UCC (Rudnick and Gao 2014) (Fig. 5). The present study also revealed that K and Rb correlate well with Al, Ti, Nb, Zr, Hf, and Th, but their Al-normalized ratios are different from those of UCC. In addition, Zr/Hf and Nb/Ta values obtained from chert and claystone samples show a strong linear correlation, and are comparable to those of average UCC (Fig. 6).

Fig. 5

Plots of Al vs. Ti, Nb, Ta, Th, K, and Rb for the chert samples. Ti/Al, Nb/Al, Ta/Al, and Th/Al ratios are identical to those of average UCC (Rudnick and Gao 2014), whereas K/Al and Rb/Al ratios are significantly higher than those of UCC

Fig. 6

Plots of Zr vs. Hf and Nb vs. Ta for chert (open red circles) and claystone (filled blue circles) samples. Compilation of modern deep-sea sediments (gray field; Plank and Langmuir 1998) are also shown for comparison

To avoid the significant dilution effect by biogenic SiO₂, concentrations of the elements were normalized to Al concentrations and compared with those of the UCC (Rudnick and Gao 2014) to obtain enrichment factors. The enrichment factor is defined as follows:

$$ {\mathrm{X}}_{\mathrm{EF}}=\left({\mathrm{X}}_{\mathrm{sample}}/{\mathrm{Al}}_{\mathrm{sample}}\right)/\left({\mathrm{X}}_{\mathrm{UCC}}/{\mathrm{Al}}_{\mathrm{UCC}}\right) $$

where X and Al are the concentrations of element X and Al, respectively. Furthermore, to minimize the possibility of diagenetic redistribution of elements from layers with lower SiO₂ content to adjacent layers with higher SiO₂ content (Tada 1991), we excluded the elemental data for non-detrital elements from SiO₂-poor parts of the claystone samples (NH52-R2, -R3, -R5, and -R7) when reconstructing the stratigraphic variations in enrichment factors.

Figures 7 and 8 show stratigraphic variations of enrichment factors for terrigenous elements, including HFS elements (Ti, Ta, Nb, Hf, and Zr) and alkali elements (K and Rb). Ti_EF, Nb_EF, and Ta_EF values are relatively constant throughout the studied section and close to a UCC value of 1. However, the lower sublayer claystone (NH52-R2) exhibits higher Ti_EF (~ 1.5) than those of other upper claystone and chert beds. Changes in Zr_EF and Hf_EF are also parallel to the UCC line in the studied interval. K_EF and Rb_EF vary between 1.12–1.75 and 1.6–2.4, respectively (Fig. 8). The K_EF and Rb_EF values decreased during the E2 event.

Fig. 7

Stratigraphic profiles of major and trace elements in the studied section. Enrichment factors (EF) were calculated relative to UCC (Rudnick and Gao 2014). Gray lines indicate UCC values. Data for non-detrital elements in SiO₂-poor parts of the claystone samples (NH52-R2, -R3, -R5, and -R7) are excluded from the stratigraphic profiles

Fig. 8

Stratigraphic profiles of K_EF, Rb_EF, and CIA in the studied section. Enrichment factors (EF) were calculated relative to UCC (Rudnick and Gao 2014). K_EF and Rb_EF decrease with increasing CIA values during the E2 event

Vanadium, U, and Mo form highly soluble ions under oxygenated conditions, but under anoxic conditions, they are insoluble and in a lower valency state (Calvert and Pedersen 1993; Calvert and Pedersen 2007). Enrichment factors of these redox-sensitive elements are generally parallel to the UCC line in the studied interval (Fig. 7). Similarly, there are no significant changes in the enrichment factors of Ni and Zn (Table 2), which form highly insoluble sulfides under suboxic and/or anoxic bottom water conditions (Calvert and Pedersen 1993).

Si, Ca, Sr, P, and Ba are widely used as proxies for burial flux of biogenic material (Hollis 2003; Hollis et al. 2003; Takiguchi et al. 2006), based on their distribution in modern marine sediments (e.g., Schroeder et al. 1997). Stratigraphic variations of enrichment factors for these elements are shown in Fig. 7. Si_EF abruptly decreased during the initial E1 interval where the burial flux of silica-secreting radiolarians was markedly reduced (Onoue et al. 2016a). The trend in Ca_EF is similar to those of P_EF and Sr_EF, and there are strong positive correlations between Ca_EF and P_EF (r = 0.93) (Fig. 9). As with the Si_EF trend, there is an abrupt decrease in Ca_EF, P_EF, and Sr_EF during the initial E1 event, and these values began to rise after the E2 event. Ba_EF values also decreased sharply during the E1 interval, and were relatively constant after the E2 interval.

Fig. 9

Correlation coefficients between the enrichment factors for the chert and claystone samples

4.2 Rare earth elements

Table 3 lists the REE concentrations of the chert and claystone samples in the studied section. Enrichment factors of the total REE concentrations (∑REE_EF) have positive correlations with Ca_EF and P_EF and are also positively correlated with some trace elements, such as Zr_EF, Hf_EF, and Th_EF (Fig. 9). Figure 10 shows the REE patterns of chert and claystone beds across the impact ejecta layer normalized to UCC (Rudnick and Gao 2014). Enrichment of the middle REEs (MREEs; Sm, Gd, Tb, and Dy) exhibits stratigraphic changes. The chert beds throughout the whole section are relatively enriched in MREEs, whereas REE patterns for the claystones (NH52-R2 to R7) are relatively flat with nearly UCC values (Fig. 10). To calculate the relative MREE enrichments, UCC-normalized MREE/MREE* ratios were used (Chen et al. 2015):

$$ \mathrm{MREE}/{\mathrm{MREE}}^{\ast }=\frac{2\times \left(\mathrm{avg}.\mathrm{MREE}/\mathrm{avg}.{\mathrm{MREE}}_{\mathrm{UCC}}\right)}{\left(\mathrm{avg}.\mathrm{LREE}/\mathrm{avg}.{\mathrm{LREE}}_{\mathrm{UCC}}\right)+\left(\mathrm{avg}.\mathrm{HREE}/\mathrm{avg}.{\mathrm{HREE}}_{\mathrm{UCC}}\right)} $$

Fig. 10

UCC-normalized REE patterns of the chert and claystone samples. a Chert layers in the E2 interval, b claystone layers in the E1 interval, and c chert layers before the impact event

where LREE and HREE represent light REEs (La, Ce, Pr, and Nd) and heavy REEs (Er, Tm, Yb, and Lu) respectively.

The MREE/MREE* ratios across the ejecta layer show a decrease in MREE enrichment after the impact event. The average MREE/MREE* ratio in cherts below the ejecta layer is 1.35, whereas in the E1 interval it is 1.12. The MREE/MREE* ratios rapidly increase in the chert beds of the E2 interval, indicating a pronounced decline in MREE enrichment during the E1 interval.

Ce and Eu are the only two REEs that have multiple valence states, which results in fractionation that can be quantified by Ce and Eu anomalies (German and Elderfield 1989; German and Elderfield 1990; Holser 1997). Cerium and Eu anomalies are calculated as:

$$ \mathrm{Ce}/{\mathrm{Ce}}^{\ast }=2{\mathrm{Ce}}_N/\left({\mathrm{La}}_N+{\Pr}_N\right) $$

$$ \mathrm{Eu}/{\mathrm{Eu}}^{\ast }=2{\mathrm{Eu}}_N/\left({\mathrm{Sm}}_N+{\mathrm{Gd}}_N\right) $$

The subscript “N” denotes UCC-normalized values. Ce/Ce* values exhibit positive anomalies and vary from 1.10–1.63 throughout the studied section (Fig. 11). Ce/Ce* values fluctuate during the radiolarian turnover interval (Fig. 11), but are essentially uniform after this interval. There are no Eu anomalies relative to UCC through the studied section (Fig. 10). The average Eu/Eu* value of the studied section is 0.98 ± 0.02.

Fig. 11

Stratigraphic profiles of ∑REE_EF, MREE/MREE*, Ce/Ce*, and Eu/Eu* in the studied section. Enrichment factors (EF) were calculated relative to UCC (Rudnick and Gao 2014). SiO₂-poor parts of claystone samples (NH52-R2, -R3, -R5, and -R7) are excluded from the stratigraphic profiles

5 Discussion

On the basis of high-resolution profiles of biogeochemical signatures (Si, Ba, Ca, and P), redox-sensitive elements, REE, and Chemical Index of Alteration (CIA; Nesbitt and Young 1982) values obtained from the Triassic bedded cherts, we assessed the environmental changes that triggered a decline in radiolarian burial flux after the Norian impact event. Here, we consider three of the main environmental changes (paleoproductivity, oceanic redox conditions, and provenance) before discussing the triggers for the decline in radiolarian burial flux based on these controls.

5.1 Paleoproductivity

Changes in marine primary productivity may have played an important role at the base of the marine food chain after the middle Norian impact event. A post-impact reduction in primary productivity has been suggested based on the negative δ¹³C_org excursion observed during the E1 event (Onoue et al. 2016a). To evaluate productivity estimates based on δ¹³C_org data, we used Ba_EF as another proxy for primary productivity. It is generally considered that barite precipitation occurs in decaying particulate organic matter while it sinks to the seafloor (Dehairs et al. 1980; Bishop 1988; Dymond and Collier 1996). Enhanced Ba_EF values below high-productivity areas support this assumption (Nürnberg et al. 1997). Given that pore water in a sediment column is generally saturated with respect to barite, the barite associated with productivity can be preserved after burial. Therefore, the Ba_EF record of the bedded cherts can be used as a proxy for productivity variations in the surface ocean (e.g., Zachos et al. 1989; Dymond and Collier 1996; McManus et al. 1998; Algeo et al. 2011). Our data show that Ba_EF in the studied section decreases in the E1 interval and recovers in the E2 interval. The stratigraphic profile of Ba_EF mimics that of δ¹³C_org, which supports the previous interpretation of δ¹³C_org records that marine primary productivity decreased in the E1 interval and recovered in the E2 interval (Onoue et al. 2016a).

Phosphorous is also a geochemical proxy for paleoproductivity in pelagic deep-sea sediments (e.g., Murray and Leinen 1993; Algeo et al. 2011). Phosphorous is transferred to the sediment mainly as organically bound P, most of which is subsequently liberated through re-mineralization of organic matter; long-term retention of P in sediment requires adsorption onto ferric oxyhydroxide and subsequent mineralization as authigenic phosphates (Algeo and Ingall 2007). For our samples, a strong positive correlation between P_EF and Ca_EF (r = 0.93) implies that the apatite species is the main host phase for P. P₂O₅/CaO ratios of Triassic conodonts, or biogenic apatite, obtained from a bedded chert succession in Inuyama area range from 0.71 to 0.77 with an average of 0.74 (Takiguchi et al. 2006). The P₂O₅/CaO obtained from our samples is almost identical (P₂O₅/CaO = 0.72) to those of conodonts. These data suggest that biogenic apatite such as conodonts is preserved in our samples as the main carrier of Ca and P. Thus, P_EF cannot be used as a proxy for primary productivity in the studied section, but instead P_EF and Ca_EF reflect biogenic apatite (i.e., conodonts) accumulation relative to terrigenous accumulation in the bedded cherts.

The stratigraphic variations in Ca_EF and P_EF indicate that the accumulation of biogenic apatite decreased during the E1 interval, whereas its accumulation recovered after the E2 interval (Fig. 7). Strontium is geochemically similar to Ca and may replace Ca in apatite, which is supported by the positive correlation between Sr_EF and Ca_EF (r = 0.76). Significant negative excursions in Sr_EF in the E1 interval may also reflect a decrease in biogenic apatite accumulation.

Additional information regarding changes in biogenic apatite accumulation may be derived from REE data. The UCC-normalized REE patterns of the bulk chert samples show slight MREE enrichment (Fig. 10), which is probably due to the presence of fossil biogenic apatite that recrystallized during diagenetic processes (Reynard et al. 1999; Chen et al. 2015). The claystones in the E1 interval have similar LREE and HREE abundances, but slightly different MREE abundances, resulting in a flatter pattern in that interval. Relative depletion of MREEs in the claystone samples is probably due to the paucity of conodont fossils, which is consistent with the interpretation that the decrease in Ca_EF and P_EF was associated with decreased accumulation of biogenic apatite during the E1 interval.

The oxygen isotopic composition of conodont apatite derived from Late Triassic deep-marine sediment sections in the western Tethys has revealed that conodont δ¹⁸O values reflect surface water temperatures and that these conodonts lived in near-surface waters (Rigo and Joachimski 2010; Rigo et al. 2012). Thus, significant negative excursions in Ca_EF and P_EF in the E1 interval may reflect a decrease in biogenic apatite accumulation by conodonts in near-surface waters in a pelagic realm of the Panthalassa Ocean. δ¹³C_org and Ba_EF data also imply that primary productivity in the pelagic Panthalassa was also decreased during the E1 interval. We infer that the decimation of conodonts in this oceanic region was likely to have been related to the collapse of primary productivity. Following the resurgence in primary productivity after the E1 event, the biogenic apatite content proxies had recovered to pre-impact values by the first chert bed overlying the claystone layer, which suggests conodonts had begun to recover after the E1 event.

In contrast, radiolarian burial flux did not recover until the E2 event, which was 300 kyr after the impact event (Onoue et al. 2016a). To account for the sustained reduction in the burial flux of radiolarian silica during the E1 and E2 events, the factors controlling radiolarian burial flux (e.g., availability of dissolved silica and seawater temperature; De Wever et al. 2014) are discussed below.

5.2 Redox change

Elements such as Mn, V, U, and Mo are useful to constrain ancient oceanic redox conditions (Tribovillard et al. 2006), because their valency can vary as a function of the prevailing redox potential (Calvert and Pedersen 1993; Calvert and Pedersen 2007). We discuss the sedimentary redox changes through the studied section using the following proxies for redox conditions.

Manganese is one of the most commonly used geochemical proxies for redox conditions in the oceanic environment. It forms insoluble Mn(III) or Mn(IV) hydroxides or oxides (e.g., MnO₂) that are deposited rapidly in particulate form (Calvert and Pedersen 1993; Sholkovitz et al. 1994). However, under anoxic conditions, Mn is reduced to Mn(II) and forms soluble cations (e.g., Mn²⁺ and MnCl⁺). Consequently, a small enrichment factor for Mn (Mn_EF) in marine sediments suggests reducing depositional conditions near the surface of the sediments. In the studied section, Mn_EF generally fluctuates (0.55–1.95) around a UCC value of 1, except for several high-Mn_EF peaks (Mn_EF > 2, Fig. 7). Although it is possible that these high-Mn_EF peaks could reflect diagenetic enrichment of Mn, their origin remains unclear. Compared with the general trend of Mn_EF in Upper Triassic bedded cherts in the Sakahogi section (Nozaki et al. 2019), the Mn_EF values in the studied section are lower than the average value of the entire Norian Stage (Mn_EF = 3.00; n = 35). However, significant redox changes across the ejecta layer cannot be recognized from Mn.

Vanadium and U enrichments in sediments are useful proxies for reducing depositional conditions, ranging from moderately to strongly reducing (Sadiq 1988; Algeo and Maynard 2004; Tribovillard et al. 2006). Vanadium and U reduction in seawater occurs under Eh conditions close to that required for the reduction of Fe(III) to Fe(II), which is lower than that for Mn reduction (Takeno 2005). In oxic seawater, V is present as soluble V(V) in the quasi-conservative form of vanadate oxyanions (HVO₄^2– and H₂VO₄^–). When conditions change from oxic to mildly reducing, V(V) converts to V(IV) and forms the vanadyl ion (VO²⁺), related hydroxyl species (VO(OH)₃^–), and insoluble hydroxides (VO(OH)₂) (Breit and Wanty 1991; Wanty and Goldhaber 1992). Under such reducing conditions, soluble U(VI) is reduced to insoluble U(IV). In the studied section, V_EF values show constant low values across the ejecta layer (0.88–1.67) and are similar to UCC values throughout the studied section (Fig. 7). U_EF is higher than that of UCC, but its values are low (1.02–2.73) compared with U_EF in modern anoxic–sulfidic basins (~ 5; Algeo and Tribovillard 2009) and a period of anoxia in the Anisian reported from the Sakahogi section (Nozaki et al. 2019). The V and U data demonstrate that the redox conditions in the studied section were stable and oxic across the impact event. We further examined the redox conditions based on the enrichment of Mo, but Mo_EF is constantly low or similar to UCC values in the studied section, except for one outlier at 90.0 cm (Fig. 7).

The Ce anomaly is also a common tracer for redox conditions in pelagic sediments (e.g., Kato et al. 2002; Fujisaki et al. 2016; Nozaki et al. 2019). Given that Ce is relatively insoluble in seawater following oxidation from Ce(III) to Ce(IV), it responds to redox changes in seawater (Sholkovitz et al. 1992). Authigenic carbonate and phosphate minerals are considered to record seawater REE chemistry (Toyoda and Tokonami 1990; Kamber and Webb 2001). However, in the studied section, the chert samples consist of a mixture of biogenic silica and detrital materials. Therefore, Ce_EF in the chert samples may reflect both seawater and detrital components, whereas in the claystones could be dominated by components from detrital material rather than from seawater. Ce/Ce* values in the studied section range from 1.10 to 1.63, and generally show a relatively constant value of ~ 1.4 (Fig. 11). The slightly positive Ce anomaly (Ce/Ce* > 1) throughout the studied section could be attributable to the provenance of detrital components (discussed in next subsection), as reflected by the different K/Al and Rb/Al ratios as compared with average UCC (Fig. 5). Hence, our results suggest that Ce/Ce* values of the studied section mainly record the detrital signature. However, the Ce/Ce* variation in this study is much narrower than that during the late Permian oceanic anoxia event (Kato et al. 2002) and the period of acidic conditions at the end-Triassic (Hori et al. 2007), which again implies that there was no significant redox change at this pelagic site due to the Late Triassic impact event.

Our analysis of redox-sensitive elements cannot detect any redox change in the equatorial region of the Panthalassa Ocean across the Late Triassic impact event. Redox conditions along the Panthalassan coast of the western Pangean supercontinent have been studied throughout the Upper Triassic to Lower Jurassic interval in the Black Bear Ridge section, British Columbia (Sephton et al. 2002; Wignall et al. 2007; Onoue et al. 2016b) (Fig. 2). The redox-sensitive elements show there was no redox change in the middle to late Norian (Wignall et al. 2007). In the western Tethys Ocean, Upper Triassic carbonate sequences of the Lagonegro Basin, southern Italy (Fig. 2) have been examined for redox changes using Mo and U enrichments and Ce/Ce* anomalies. Redox changes in the middle to upper Norian interval have not been detected (Casacci et al. 2016). In summary, the Norian impact did not appear to induce redox changes in the Panthalassa and western Tethys oceans, although this interpretation requires further geochemical study at other middle to late Norian sites.

5.3 Changes in dust provenance or hydrological cycle

The HFS (Ti, Zr, Hf, Nb, and Ta) and alkali (K and Rb) elements are largely contained in detrital phases in the Triassic bedded chert sequences of the Inuyama area (Hori et al. 2000; Soda and Onoue 2019). These detrital components were transported to a pelagic setting in a low-latitude region of the Panthalassa Ocean (Uno et al. 2015), mainly as eolian dust or aerosols (Hori et al. 1993). According to model simulations of the surface winds (Kutzbach and Gallimore 1989), dust delivered to the depositional area of the bedded cherts in the Inuyama area largely originated from a low-latitudinal zone of western Pangea (Nakada et al. 2014; Ikeda et al. 2017) (Fig. 2).

The enrichments of HFS elements are relatively constant throughout the studied section and close to a UCC value of 1. This suggests that UCC-like terrigenous material was the source for these elements throughout the whole section. Given that different terrestrial reservoirs have characteristic Zr/Hf and Nb/Ta ratios, and there is no significant fractionation of these ratios by continental weathering or sedimentation processes, Nb/Ta and Zr/Hf can be used to constrain the source of these elements in sediments (Pfänder et al. 2007; Berndt et al. 2011). In the Sakahogi section, both the chert and claystone samples have typical UCC ratios (Pfänder et al. 2007), with Zr/Hf = 40.1 ± 1.9 (chert; n = 57) and 39.5 ± 2.4 (claystone; n = 6), and Nb/Ta = 14.5 ± 0.4 and 14.4 ± 0.2 (Table 2), respectively. In addition, Zr/Hf and Nb/Ta values in the Sakahogi section show a strong linear correlation (R² > 0.99, Fig. 6) compared with the values in modern deep-sea sediments (e.g., Plank and Langmuir 1998). These data suggest that typical UCC-like terrigenous material was the principal detrital components throughout the Sakahogi section. Alternatively, if the values of Zr/Hf and Nb/Ta were similar in wide area of hinterland in Pangea, the potential change in provenance of eolian dust cannot be ruled out. Actually, changes in dust provenance that occurred during Norian in Fundy Basin, North America, are suggested based on reappearance of eolian deposits on various timescale (Kent and Olsen 2000). Notably, the lower sublayer claystone (NH52-R2) exhibits lower Zr_EF, Hf_EF, Nb_EF, and Ta_EF, and higher Ti_EF than those of the other upper claystone samples and cherts. This is most likely explained by contamination of ejecta materials in the lower sublayer.

Concentrations of K and Rb correlate well with those of Al and HFS elements, which indicate that these alkali elements are of terrigenous origin. Potassium, Rb, and Al contents are attributed to aluminosilicates, especially clay minerals. Given that Al is resistant to chemical weathering (Nesbitt and Young 1982), Al-normalized K and Rb can be used as chemical weathering indicators (Tanaka and Watanabe 2015; Grygar et al. 2019; Liu et al. 2019). In the present study, the enrichment factors of K and Rb show similar trends. The stratigraphic variations in K_EF and Rb_EF show an obvious decreasing trend during the E2 event (Fig. 8), which suggest the intensity of aluminosilicate chemical weathering increased during the E2 event.

To assess the degree of chemical weathering in hinterland regions using major elements, we calculated the Chemical Index of Alteration (CIA; Nesbitt and Young 1982). Values of CIA predict the extent of decomposition of feldspar minerals, which are the most abundant mineral group in the UCC. Before calculating CIA values, apatite-derived CaO concentrations were corrected using the procedure of Price and Velbel (2003). This correction estimates CaO in the silicate phase (CaO*) by subtracting CaO in the phosphate phase predicted from the P₂O₅ concentration and stoichiometry of apatite:

$$ {\mathrm{CaO}}^{\ast }=\mathrm{molCaO}-\left[\left(10/3\right)\times \mathrm{mol}{\mathrm{P}}_2{\mathrm{O}}_5\right] $$

CIA was then calculated as:

$$ {\mathrm{CIA}}^{\ast }={\mathrm{Al}}_2{\mathrm{O}}_3/\left({\mathrm{Al}}_2{\mathrm{O}}_3+{\mathrm{Na}}_2\mathrm{O}+{\mathrm{K}}_2\mathrm{O}+\mathrm{CaO}\ast \right)\times 100 $$

CIA values vary between 75–82 in the studied section (Fig. 8). CIA values are relatively constant below the ejecta layer, slightly decrease in the E1 interval, and then increase in the E2 interval (especially in sample for NHR48). These stratigraphic variations in CIA values indicate intensified chemical weathering on continents during the E2 interval, implying that the climate changed to relatively warm and humid conditions after the impact event. K_EF and Rb_EF values decrease with increasing CIA values, indicating that K and Rb were released from primary minerals, possibly micas and feldspars in felsic rocks, during continental chemical weathering. Another possibility is that the increased CIA in the E2 interval might have reflected a change in dust provenance associated with a shift in regional precipitation pattern. If a precipitation decreased in a previously wet (i.e., intensively weathered) region, the region could become a new source of high CIA dust to the Sakahogi section. However, if that was the case, the direction and intensity of prevailing winds, or the atmospheric circulation pattern, could have also changed, as well as the changes in dust provenance and hydrological cycle. Thus, this scenario should be carefully evaluated by further investigations for other marine and terrestrial records of the corresponding period.

The low-latitudinal zone of western Pangea might have been the major source of detrital elements to the deep-sea sediments of the equatorial Panthalassa (Nakada et al. 2014; Ikeda et al. 2015). Sedimentological studies in this region (e.g., the Petrified Forest National Park in Arizona) suggest an arid to semi-arid climate with significant seasonality during the Norian (Prochnow et al. 2006; Cleveland et al. 2007). However, the mean annual precipitation (MAP) inferred from a geochemical weathering index (the chemical index of alteration minus potassium index; Nordt et al. 2015) from the Chinle Formation in the Petrified Forest National Park (Fig. 2) suggests that short-term wetter periods could have occurred in the middle Norian (218–213 Ma; Ramezani et al. 2011; Ramezani et al. 2014). According to the age model of Nordt et al. (2015), these middle Norian climate shifts occurred close to the time of formation of the Manicouagan impact crater at 215.5 Ma (Jaret et al. 2018). A palynological record from the Chinle Formation (Baranyi et al. 2017) also demonstrated that gradual aridification during the Norian was interrupted by at least two short-lived wetter climatic periods in the middle Norian, as inferred from an increase in hygrophyte vegetation. Given that the timing of this vegetation change in a wetter climatic period is very close to the age range of the Manicouagan impact event, Baranyi et al. (2017) suggested that the impact might have contributed to the vegetation change.

These studies have shown that the Manicouagan impact event might have coincided with middle Norian climate and vegetation changes during a short-lived wet period, which is consistent with our results showing intensified chemical weathering of the hinterland for ~ 300 kyr after the impact event. However, in the case of the Chicxulub impact event at the Cretaceous–Paleogene boundary, the environmental effects of climatically active gases (e.g., sulfur (di)oxide and carbon dioxide) released from the impact site were short-lived (years to hundreds of years; Artemieva et al. 2017), and cannot explain such a relatively long-term (~ 300 kyr) duration of intensified chemical weathering. The existing data imply no direct causality between the Manicouagan impact event and the change to a wetter climate. However, the change to a warm, humid climate is an important factor in considering the controls on radiolarian burial flux, which will be discussed in the following section.

5.4 Decline of radiolarian burial flux

The E1 event is considered to have lasted 10⁴–10⁵ years after the impact event, based on Os isotope studies (Sato et al. 2013; Onoue et al. 2016a). Our geochemical data demonstrate that this interval represents the duration required for the restoration of burial flux by primary and silica- and apatite-secreting organisms after the middle Norian impact. A decline in radiolarian and conodont burial flux during the E1 event might be explained by the post-impact shutdown of primary productivity as discussed above. Our data also reveal that primary productivity and conodont burial flux had recovered to pre-impact values by the beginning of the E2 event. However, radiolarian burial flux did not recover until the E2 event, which was ~ 300 kyr after the impact (Onoue et al. 2016a). The mass accumulation rates (MAR) of radiolarian silica decreased from 0.1 to 0.02 g cm⁻² kyr⁻¹ across the impact and continued to decrease during the E2 interval (Fig. 3a).

Previous studies have suggested that radiolarian abundance is mainly controlled by the availability of dissolved silica and variations of temperature in the surface water (Anderson et al. 1989; Abelmann and Gowing 1996; Boltovskoy et al. 2010; De Wever et al. 2014). Recently, Ikeda et al. (2017) suggested that the burial flux of biogenic silica (BSi) over a time scale longer than the residence time of dissolved silica in the ocean (generally < 100 kyr for the Phanerozoic) can be proportional to the silica input flux to the ocean that is controlled by global continental silicate weathering flux. Thus, the decline of radiolarian burial flux for ~ 300 kyr after the impact might be explained by suppressed continental silicate weathering. However, this is inconsistent with the inferred climatic changes toward a wetter condition and enhanced weathering in Pangea discussed above. In addition, MAR of radiolarian silica during the E2 event was an order of magnitude lower than the calculated BSi burial flux for the middle Norian (Ikeda et al. 2017). This discrepancy also suggests a fundamental gap between our observation and a simple mass balance model of oceanic dissolved silica during the E2 event, though we need further investigation to confirm the temporal variations in the radiolarian burial flux with a reliable sedimentation rate for the middle Norian chert. Changes in dissolved silica concentrations are also not sufficient to explain the decline of radiolarian silica burial flux during the E1 and E2 events, because silica-saturated conditions (≥ 110 mg/L SiO₂) with respect to biogenic silica presumably existed in the Late Triassic (Racki and Cordey 2000; Grenne and Slack 2003). Furthermore, Si_EF during the E2 event is comparable to that in the other intervals before and after the impact, implying no significant change in the availability of silica in the studied section (Fig. 7). The δ¹³C_org data suggest there was a resurgence in primary productivity after the E1 event (Fig. 3a, Onoue et al., 2016a). However, the recovery of radiolarian burial flux was significantly delayed (Fig. 3a). Taking these into consideration, other factors such as sea surface temperature must be invoked to account for the sustained reduction in the burial flux of radiolarian silica.

Relatively little information is available regarding the effects of seawater temperature variations on living radiolarians (Anderson et al. 1989; Anderson et al. 1990; Matsuoka and Anderson 1992). Experimental and observational studies have shown that the mean longevity and growth rate decrease markedly when temperature is raised. For example, laboratory culturing data for three tropical and subtropical radiolarian taxa (Spongaster tetras, Didymocyrtis tetrathalamus, and Dictyornyne truncatum) suggest that the maximum growth and longevity (mean longevity of 22–23 days) are achieved at a temperate condition of 21–28 °C, but skeletal growth and survival were remarkably suppressed at temperatures above 32 °C (longevity of 2 days), and no individuals survived at temperatures above 36 °C (Anderson et al. 1989; Anderson et al. 1990; Matsuoka and Anderson 1992). Furthermore, kinetic studies have shown that an increased temperature is the major factor that affects silica dissolution rates in the ocean (Erez et al. 1982; Ragueneau et al. 2000), which might enhance the in situ dissolution of radiolarian opal in surface waters.

Even if Norian radiolarians had different temperature tolerances, the consequences of climate warming would have affected the production of radiolarian opal in the near-surface waters. Our weathering proxy data (K_EF, Rb_EF, and CIA values) suggest acceleration of the hydrological cycle and an increase in continental chemical weathering rates, which is consistent with a greenhouse climate (e.g., Knobbe and Schaller 2018). Thus, we interpret that radiolarian silica burial flux did not recover until the E2 event due to an increased sea surface temperature. This interpretation is speculative—the existing data are unable to prove direct causality between greenhouse climate and the impact event, and the sea surface temperatures during the E1 and E2 events have yet to be determined by conodont δ¹⁸O_apatite data (e.g., Rigo and Joachimski 2010; Rigo et al. 2012; Trotter et al. 2015; Knobbe and Schaller 2018). Furthermore, relationships to other environmental factors related to biogenic opal production, such as dissolved Al concentrations and microbial degradation of organic matter (e.g., Van Bennekom et al. 1988; Nelson et al. 1995; Bidle and Azam 1999; Ragueneau et al. 2000), must be considered.

6 Conclusions

Stratigraphic profiles of major, trace, and rare earth elements in middle to upper Norian (Late Triassic) bedded cherts of the Mino Belt, central Japan, were constructed in order to assess the environmental changes that triggered a decline in radiolarian burial flux in the Panthalassa Ocean after the Norian impact event. Based on our geochemical data, paleoenvironmental changes across the Norian impact event can be summarized as follows.

• 1. Productivity and burial flux estimates based on δ¹³C_org values and enrichments of Ba, Ca, P, and REEs suggest that there was a post-impact shutdown of productivity by primary and silica- and apatite-secreting organisms. Primary productivity and biogenic apatite (i.e., conodont) burial flux had recovered to pre-impact levels 10⁴–10⁵ years after the impact event, but radiolarian burial flux did not recover for ~ 300 kyr after the impact.

• 2. Redox-sensitive elements (Mn, V, U, and Mo) and Ce/Ce* anomalies exhibit limited fluctuations throughout the studied section, which indicate that there was no significant redox change in the pelagic realm of the Panthalassa Ocean across the Late Triassic impact event. Therefore, the change in redox conditions was not related to the decrease in radiolarian burial flux after the impact.

• 3. Weathering proxies, such as CIA values and enrichments of K and Rb, suggest that intense chemical weathering of the hinterland occurred during the decline in radiolarian burial flux after the Norian impact event. A short-lived wet period near the time of the Norian impact has been documented from the low-latitudinal zone of the western Pangean region (e.g., the Petrified Forest National Park in Arizona), which might have been the major source of detrital elements in the studied section. These data appear to reflect acceleration of the hydrological cycle and an increase in continental chemical weathering rates, which might have been induced by a greenhouse climate after the impact.

• 4. Radiolarian burial flux did not recover for ~ 300 kyr after the impact. We hypothesize that radiolarian burial flux decreased in response to an increased sea surface temperature due to a greenhouse climate, but this needs to be verified by independent data (e.g., conodont δ¹⁸O_apatite).