The ECS is one of the largest marginal seas in the Northwest Pacific Ocean. The submarine landform is divided from west to east into the East China Sea shelf, East China Sea slope, Okinawa trough, Kyushu-Ryukyu Island slope, and Ryukyu Island arc (Liu et al., 2003, 2020). The hydrodynamic of the ECS is complex with Changjiang (Yangtze River) diluted water, East China Sea coastal current, Kuroshio, Taiwan Warm Current, and Tsushima warm current (Su, 1998; Zhou et al., 2015). The interaction of these currents has a dominant influence on the regional marine environment. Seasonal changes in different water masses and the mixing of riverine runoff and coastal currents with the intrusion of saline currents significantly affect the redox conditions of the ECS (Sun et al., 2017, 2019). Generally, the strong oxidizing seawater distributes discontinuously in the ECS, in contrast to the weak oxidizing condition of the aquatic environment in spring. Two anoxic zones located at the north and the south Yangtze River estuary are developing, particularly in the summer, and can even merge into one large anoxic zone, which endangers both fish and benthos (Wei et al., 2015).

Qiandaohu Lake is located in eastern China and is fed from the East China Sea, in the low reach of the Changjiang River Basin. There are nearly 30 tributaries entering the lake, among them, Xin’an River is the largest water inflow, accounting for approximately 60% of the total runoff (Li et al., 2011). Geologically, the Xin’an River Basin is located on the eastern margin of the Yangtze Plate, the upper reaches of the lake basin are the Neoproterozoic Jiangnan paleo island arc, and the lower reaches are the Early Paleozoic Huaiyushan-Tianmushan passive marginal basin (Pan et al., 2009). The carbonate rocks and clastic rocks of the early Paleozoic are exposed, and intrusive rocks and volcanic rocks of the Yanshan and Jinning periods have been developed (Tang et al., 2023).

A series of NE-SW trending to NS-trending fault deformation belts divides the CIOB into the eastern and the western parts. The eastern CIOB is older (60-90 Ma), characterized by compressional structures, and presents a north-south trending fault zone, while the western is younger (30 Ma), including a NE-SW trending fault zone (Mukhopadhyay, 1998). Many of the isolated seamounts and seamount chains distributed parallel to the north-south trend in the basin are the products of a collision between the Indian plate and the Eurasian plate during the period of 65-49 Ma (Raju and Ramprasad, 1989; Dyment, 1993; Mukhopadhyay et al., 1997; Rajendran and Rao, 2000; Das et al., 2007). The Wharton Basin is located to the east of the CIOB, separated by the Ninety-East Ridge (NER). Based on the Fe-Mn crusts, the sources of detrital material change along the NER and reflect, from north to south, the decreasing influence of the Ganga River system and volcanic arcs located to the east, with the increasing influence of sediment derived from Australia to the south (Hein et al., 2016). The inhomogeneous growth of seamounts in the basin and the large amounts of volcanic-hydrothermal materials in volcanic rocks and sedimentary layers indicate the existence of volcanic hydrothermal activities in the plate (Iyer, 1991 and 2005; Mukhopadhyay, 1998; Iyer et al., 2007). Low-oxygen North Atlantic deep water (NADW) at 2,000–3,800 m and oxygen-rich Antarctic bottom water (AABW) below 3,800m converge to form circumpolar deep water. The redox condition at the sediment-water interface in the Indian Ocean is mainly regulated by export productivity and ventilation of abyssal water mass (Nambiar et al., 2022; Piotrowski et al., 2009).

The modern fish teeth samples from the ECS include a Miichthys miiuy, a Cynoglossus semilaevis, a Scomber japonicus, and three Larimichthys crocea (numbered D₁, D₂, D₃, and D₄–D₆, respectively). Among them, Miichthys miiuy (D₁), Scomber japonicus (D₃), and Larimichthys crocea (D₄–D₆) are normal offshore bottom fishes. They live in a muddy and sandy environment and will migrate from deep water to the shore for long-distance reproductive migration during the reproductive season every year. By contrast, Cynoglossus semilaevis (D₂) are usually inert with slow action and only move in a small range. In addition, two Mylopharyngodon piceus (numbered QDH₁ and QDH₂) were collected from Qiandaohu Lake. They are bottom-living fish in fresh water, with reproductive migration generally from the lower reaches of rivers and their tributaries or affiliated lakes up to the middle and upper reaches.

Two sediment cores (GC04 and GC11) from the CIOB and one sediment core (GC03) from the Wharton Basin were collected using a gravity piston corer by the China Ocean Mineral Resources R & D Association (COMRA) during the 30th and 34th Dayang Cruises in 2014 and 2015.

GC04 was collected at 80.10° E, 17.54° S at a water depth of 4,826 m, 350 cm long, homogeneous, and was light reddish brown. In total, 14 subsamples were collected at intervals of 20–30 cm. GC11 was collected at 80.49° E, 19.54° S with a water depth of 5,106 m, 200 cm long, and was homogeneously reddish brown. In total, 21 subsamples were collected at 10 cm intervals. GC03 was collected at 92.41° E, 13.35° S with a water depth of 5072 m, 200 cm long, and was a heterogeneous texture with the upper part a yellowish-brown and the lower part a reddish-brown. In total, nine subsamples were collected at 10 cm intervals in the top 90 cm. The samples were cryopreserved at -20°C after collection and transported back to the laboratory for subsequent tests. In this study, 15 samples of debris of deposited fish teeth in three cores were identified and sorted under a binocular microscope.

An OLYMPUS CX43 microscope is used for transreflection photography of fish teeth at 5× magnification. After mounting in epoxy and polishing, samples were put into the TESCAN MIRA3 field emission microscope to take backscattered electron images (BSE).

Fossil and modern fish teeth show distinct morphological characteristics respectively. The teeth of the Mylopharyngodon piceus from Qiandaohu Lake are approximately 5 mm in width and length showing a nightcap style ( Figure 2b ). In contrast, the teeth of marine fish from the ECS are tapered, and smaller with a length of approximately 1 mm ( Figure 2a ). The fossil fish teeth show similar morphology and structure to the modern marine fish teeth but were much smaller with only 100–250 μm and/or less. Beside cones, the fossil fish teeth also showed ellipses, spindles, long stripes, and other shapes ( Figure 2c ). All fish teeth have a two-layer structure of enamel and dentin. The dentin is inside the fish teeth with obvious pores, indicating a relatively loose texture. The enamel is a thin light-colored dense outer layer of the fish teeth ( Figure 2 ), which exhibits the typical growth structure clearly.

The marine fish teeth and lake fish teeth in this study had similar Raman spectra with four vibration modes of apatite [PO₄]^3-, indicating that the main component is hydroxyphosphate. The enamel showed a generally higher peak intensity than that of dentin ( Figure 3 ). The main peak of marine fish teeth enamel was 1,030 cm^-1 ( Figures 3a, b ), representing the A-type substitution of carbonate (that is, the substitution of hydroxyl), while the main peak of lake fish teeth enamel was 1070 cm^-1 ( Figures 3e, f ), representing the B-type substitution of carbonate (that is, the substitution of phosphate). There were both A-type and B-type substitutions in the dentin of the marine and lake fish teeth.

A similar double-layer structure in fossil fish teeth was indicated, and both the enamel and dentin of fossil fish teeth are composed of hydroxyapatite. Moreover, the x-ray diffraction (XRD) of the enamel also showed a higher peak than that of dentin, indicating higher crystallinity of enamel (Liao et al., 2019a).

Elements mapping of the living fish teeth (sample D₅ as an example) showed that the major elements vary in the profile ( Figure 4 ). The contents of fluorine (F), natrium (Na), phosphorus (P), calcium (Ca), and ferrum (Fe) increased from the root to the tip and decreased from the enamel to the central dentin. In contrast, the contents of magnesium (Mg) increased from the root to the tip, and from the lateral enamel to the medial dentin. Sr and sulfur (S) were evenly distributed in the dentin and enamel of fish teeth without an obvious tendency.

The main element compositions of both fossil and modern fish teeth are CaO and P₂O₅, and there is a significant positive correlation between them ( Figure 5a ). Generally, the fossil fish teeth are characterized by higher contents but narrower fluctuations of CaO (44.32%–51.96%) and P₂O₅ (36.03%–40.73%) than the modern fish teeth (29.00%–48.49% for CaO and 24.77%–42.36% for P₂O₅) as shown in Table 1 . CaO/P₂O₅ in the fish teeth was broader than the stoichiometric ratio of apatite (10/6) but was basically within the range of CaO/P₂O₅ in fossil fish teeth and mammal bones samples studied previously [from 1.2 to 2.0, Nemliher et al. (2004)]. CaO/P₂O₅ in the fossil fish teeth varied in the range of 1.19 to 1.39, while CaO/P₂O₅ in the modern fish teeth varied in the range of 1.00 to 1.27. The higher CaO/P₂O₅ in the fossil fish teeth may indicate the influence of the fossilization and diagenesis with the replacement of orthophosphate-ion by carbonate-ion in the apatite crystal lattice (Nemliher et al., 2004).

Except for CaO and P₂O₅, F had the highest content in the fossil fish teeth, ranging from 0.94% to 3.12%. It was much higher than the modern fish teeth with 0.00%–1.05%. Furthermore, the fossil fish teeth were also distinctly more enriched in Na₂O, FeO, and SiO₂ ( Table 1 ), which may be due to the porous particles with inclusions of terrigenous material after deposition (Nemliher et al., 2004). The lowest CaO/P₂O₅ was observed in GC03 (as shown in Table 1 ), together with significantly higher contents of Al₂O₃, SiO₂, and K₂O than GC04 and GC11, suggesting intense influence from terrigenous materials in core GC03 as well. Inversely, the modern fish teeth had higher MgO, K₂O, and chlorine (Cl) ( Table 1 ). As essential mineral elements for fish life, teeth, bone, and scales are important sinks of these elements (Bijvelds et al., 1998).

Generally, most trace elements were detected in fossil fish teeth and varied greatly. Some trace elements including beryllium (Be), germanium (Ge), cadmium (Cd), indium (In), thallium (Tl), bismuth (Bi), and most REEs (except lanthanum, cerium, yttrium) were below the limit of detection in both the marine and lake fish teeth. Vanadium (V), nickel (Ni), arsenic (As), stannum (Sn), and thorium (Th) only in the marine fish teeth while Hf was detected only in the lake fish teeth ( Table 2 ).

Due to the extremely low Nd content in the modern fish teeth, the ¹⁴³Nd/¹⁴⁴Nd ratio was available only in the fossil fish teeth. The ⁸⁷Sr/⁸⁶Sr values of the modern fish teeth are shown in Table 5 and 6 exhibits the ¹⁴³Nd/¹⁴⁴Nd ratio and ⁸⁷Sr/⁸⁶Sr values for the fossil fish teeth.

The ⁸⁷Sr/⁸⁶Sr of marine fish teeth was less radiogenic but fluctuated in a wider range (from 0.70633 to 0.70913, with an average of 0.70832) than that of the lake fish teeth (from 0.70918 to 0.70973, with an average of 0.70953). Both the Sr content and ⁸⁷Sr/⁸⁶Sr isotopic ratio in different marine fish varied in a wide range, as shown in Table 5 and Figure 8 . The Cynoglossus semilaevis (D₂) had the highest Sr content and range of variation, and also had the highest average ⁸⁷Sr/⁸⁶Sr ratio, while the Miichthys miiuy (D₁) and Larimichthys crocea (D₄) had the lowest Sr contents and the least radioactive but biggest fluctuation of ⁸⁷Sr/⁸⁶Sr. The values of the Scomber japonicus (D₃) were located in the transition region.

The ⁸⁷Sr/⁸⁶Sr ratios of the fossil fish teeth ranged from 0.70808 to 0.71075, with an average of 0.70942, close to that of the lake fish teeth but significantly more radioactive than that of the marine fish teeth. Among the fossil fish teeth from different cores, ⁸⁷Sr/⁸⁶Sr can be ranked as GC03 < GC04 < GC11 from the point of view of minimum, maximum, and average values. The ϵNd of the fossil fish teeth from the three cores ranged from -8.1 to -2.2 with an average of -4.7, in a sequence of GC04 (-4.2) > GC03 (-4.3) > GC11(-5.7). The fossil fish teeth in the GC03 showed the widest ϵNd range (-7.0–2.2), while the ϵNd of GC04 (-6.7–2.2) and GC11 (-8.1–3.9) varied in a relatively narrow range, as shown in Table 6 .

Except for Mo, W, Zn, and Rb (the contents of which were significantly higher in lake fish teeth), the contents of most elements were higher in the marine fish teeth and further increased by different degrees in the fossil fish teeth after deposition. To evaluate the aggregation capability of the trace elements in the fish teeth during different processes, including the life cycle and post-deposition, M_i/W_i and F_i/M_i were introduced in this study for comparison. M_i/W_i represents the ability of fish teeth to absorb and accumulate trace elements from the water in which they live, while F_i/M_i represents the degree of trace elements diffusion and adsorption to fish teeth after death. Both the amount and the magnitude of trace element enrichment were considered to evaluate the enrichment degree of trace element during the life and after deposition. Then, trace elements were divided into three categories based on their geochemical properties.

It should be noted that F in fossil fish teeth was much higher than that in modern fish teeth with 0.00%–1.05%, which can be attributed to the transformation of hydroxyapatite into hydroxyfluorapatite after deposition in which the hydroxide is replaced by F. This transformation was indicated to be the main factor accelerating the adsorption of Nd, resulting in 3 to 6 orders of magnitude more Nd in fossil fish teeth than in living fish teeth (Wright et al., 1984; Shaw and Wasserburg, 1985; Staudigel et al., 1985). The extreme enrichment of REE in bio-apatite is related to ion exchange with REE³⁺+Na⁺→2Ca²⁺; REE³⁺+Si⁴⁺→Ca²⁺+P⁵⁺. It has been shown that biological apatite is composed of elongated nanocrystals arranged with hexagonal hydroxyapatite cross sections, with large porosity and extremely high specific surface area, which is conducive to REE adsorption. In addition, Si-rich regions appear along the nanoboundary, showing element exchange along the mineral nanocrystal boundary, confirming the presence of Na, Si, Ce, and Y in the biological apatite nanocrystals, and indicating that atoms heavier than Ca (Ln, such as Ce and Y) are substituted into the lattice, with weaker peaks associated with the exchange of lighter Na atoms (Liao et al., 2019a).

In general, the occurrence of trace elements in fish teeth is influenced by the geochemical properties of elements, the crystal structure of bio-apatite, and the redox conditions in different processes. This can be shown in a two-stage three-classification model, as shown in Figure 9 .

As a highly enriched trace element in modern fish (M_i/W_i=293.6, Figure 6 ), Sr is indicated to be mainly absorbed by the fish teeth during life and is slightly enriched by recrystallization exchange after deposition (with the smallest F_i/M_i of 1.13, Figure 6 ). Previous research has proved that the deposition of some trace elements in otolith is related to the history of fish migration in addition to their contents in water and the environmental temperature (Beck et al., 1992; Izzo et al., 2018; Hüssy et al., 2021; Arai and Kimura, 2022). In this study, the much higher Sr content in the marine fish teeth (1425.18 ppm) than lake fish teeth (455.10 ppm) is consistent with the large difference of Sr content in seawater (7.88 ppm) and in rivers (0.09 ppm) (Anadón et al., 2002). In addition, the Sr/Ca ratio was significantly positively correlated with Sr content in fish teeth (R² = 0.87) ( Figure 5b ), but differences in Sr/Ca and ⁸⁷Sr/⁸⁶Sr were also found among different modern marine fish species ( Figure 8 ). It is reasonable to assume that element composition (especially Sr and Ca) was also recorded in the fish teeth, similar to those in the otolith (Avigliano et al., 2018; Bock et al., 2017). This means that in the process of material exchange between fish and the external environment, some trace elements are continuously ingested into the fish, and after a series of metabolisms and circulations, finally also deposited in fish teeth.

In addition, the observed lower ⁸⁷Sr/⁸⁶Sr of marine fish teeth (0.70832) than that of lake fish teeth (0.70953) ( Figure 8 ) is consistent with the difference of ⁸⁷Sr/⁸⁶Sr between seawater [modern seawater average ⁸⁷Sr/⁸⁶Sr=0.70917 (McArthur et al., 2020)] and inland freshwater [world river average ⁸⁷Sr/⁸⁶Sr=0.71190 (Palmer and Edmond, 1989)], while the ⁸⁷Sr/⁸⁶Sr of the lower reaches of Yangtze River (Datong station) was 0.71063 (Luo et al., 2014). Both the ⁸⁷Sr/⁸⁶Sr of the lake fish teeth and marine fish teeth were lower than the ambient waters, and the ⁸⁷Sr/⁸⁶Sr ratios of all the fish teeth were lower than that of modern seawater ( Figure 8 ). It was noted that the Sr/Ca and ⁸⁷Sr/⁸⁶Sr of D₁ and D₄ overlapped greatly, while D₂ exhibited significantly higher Sr/Ca and ⁸⁷Sr/⁸⁶Sr, and D₃ was located in the transition region of Sr/Ca-⁸⁷Sr/⁸⁶Sr ( Figures 5b , 8 ). The significantly different Sr/Ca-⁸⁷Sr/⁸⁶Sr among these marine fish teeth should be closely related to their different migration habit. As mentioned, D₁, D₃, and D₄ usually migrate from deep water to nearshore for relatively long distances, thus recording a chemical composition and ⁸⁷Sr/⁸⁶Sr ratios in a wide range. In comparison, due to its lack of nearshore migration, D₂ has narrow ⁸⁷Sr/⁸⁶Sr ratios. A trigonometric diagram of Sr/Ca- ⁸⁷Sr/⁸⁶Sr-Sr was then created to distinguish different aquatic environments, as shown in Figure 8 .

Previous studies have pointed out that andesite on Taiwan Island (⁸⁷Sr/⁸⁶Sr=0.70489) and basalt on the coast of Zhejiang and Fujian provinces (⁸⁷Sr/⁸⁶Sr=0.70805) are important material sources for the ECS (Mi et al., 2017). Moreover, the Okinawa Trough, located at the eastern boundary of the ECS, is tectonically active with frequent volcanism. The acidic volcanic rock in the north-central trough has ⁸⁷Sr/⁸⁶Sr ranging from 0.70406 to 0.70544 (Huang et al., 2006). These potential sources of Sr isotopes may result in the less radioactive ⁸⁷Sr/⁸⁶Sr composition of the regional seawater of the ECS and are recorded in the fish teeth during migration. ⁸⁷Sr/⁸⁶Sr in the lake fish teeth varied in a narrow range but was higher than the ⁸⁷Sr/⁸⁶Sr of carbonate rocks in the Yangtze River Basin, varying from 0.70800 to 0.70900 (Chetelat et al., 2008), also suggesting that the weathering of exposed marine carbonate rocks in the area around Qiandaohu Lake may result in relatively low ⁸⁷Sr/⁸⁶Sr in the lake water and subsequent adsorbed to fish teeth.

In comparison, the correlation between Sr/Ca-Sr was more significant in the fossil fish teeth (R² = 0.98, Figure 5b ), and similar consistent characteristics could be found in ⁸⁷Sr/⁸⁶Sr ( Figure 8 ). In the case of the main component remaining unchanged under an open system, a strong replacement of Sr after deposition accompanies the significantly strong correlation between Sr/Ca-Sr. However, it is interesting that a negative correlation of Sr-⁸⁷Sr/⁸⁶Sr was observed for the fossil fish teeth. It was proved that Sr changed because of teeth recrystallization and exchange with pore fluid Sr²⁺ during burial and diagenesis (Martin and Scher, 2004). In this process, the interference of the Sr isotope from different sources with the original Sr isotope in fish teeth cannot be ignored.

The average ⁸⁷Sr/⁸⁶Sr in fossil fish teeth from the CIOB is 0.70942 (from 0.70808 to 0.71075), significantly more radiogenic than the ancient and modern seawater (increasing from 0.70770 at about 40 Ma to 0.70920 at present) (McArthur et al., 2001). Previous studies have indicated that the Sr isotope may be recrystallized and exchanged with Sr in pore water after deposition under the influence of diagenesis, which affected the original Sr isotope record (Martin and Scher, 2004). The ⁸⁷Sr/⁸⁶Sr of the fossil fish teeth in the 2-m cores from the CIOB was higher than the fossil fish teeth (⁸⁷Sr/⁸⁶Sr=0.70853) and foraminifera (⁸⁷Sr/⁸⁶Sr=0.70849) buried deep in the Indian Ocean but was very close to the ⁸⁷Sr/⁸⁶Sr of the silicate components (⁸⁷Sr/⁸⁶Sr=0.70944) in the surface sediments of the CIR and has a strong terrigenous property ( Table 7 ). The results of chemical leaching experiments from Western Pacific sediments show that although the ⁸⁷Sr/⁸⁶Sr of phosphate components (⁸⁷Sr/⁸⁶Sr=0.70830–0.70917) is between the range of seawater since 25 Ma, it does not decrease with the increase of depth, which also indicates a strong terrigenous attribution (Ren et al., 2022).

It is interesting that the corresponding lower ϵNd under the influence of terrigenous debris was not observed in the fossil fish teeth, and, on the contrary, the ϵNd was more radioactive with an average of -4.7 (from -8.1– -2.2) ( Table 7 ). It was also more radioactive than the Antarctic circumpolar current (ACC) and the Indian Ocean mean water [ϵNd=-7–9, Albarède et al. (1997)] recorded by iron manganese oxides in the seabed, and appears to be very close to the Pacific Ocean [ϵNd=-5–4, Ling et al. (2005)] and the silicate components in the surface sediments of the SWIR [ϵNd=-4.6, Subha Anand et al. (2019)], indicating the uncertainty of the paleoceanographic information record in the samples of this study.

Recent studies have found that REEs in pore water near the sediment-seawater interface may be an important source of rare earth elements in rare earth-rich sediments (Liao et al., 2019b; Deng et al., 2022). Bioapatite can obtain REEs from terrigenous detrital deposits through pore water (Elderfield and Pagett, 1986) and the Nd isotope of biological apatite in volcanic detrital deposits has obvious volcanic origin characteristics (Zhao et al., 2013). A study found that ϵNd in fossil fish teeth was unevenly distributed and varied widely, gradually increasing from the root to the tip of fish teeth and was significantly different from the seawater (Deng et al., 2022). An inference from this is that the original ϵNd values of bioapatite in deep-sea sediment may be altered by pore water with high Nd concentrations and low ϵNd values during diagenesis. The difference was also found in this study, but the trend was reversed so that the ϵNd of the fossil fish teeth was more radioactive than the Indian Ocean water. This may be related to the same high Nd concentrations but different ϵNd of pore water under the influence of sediment components.

For the sediment input of the CIOB, a higher ⁸⁷Sr/⁸⁶Sr ratio and lower ϵNd near Antarctica indicate old granitic rocks as the source, and the opposite is observed in sediments from the Central and the Southwest Indian Ridges, indicating seafloor spreading and mid‐ocean ridge hydrothermal activity as the source (Subha Anand et al., 2019). It is noteworthy that the ⁸⁷Sr/⁸⁶Sr ratio increases along with decreasing Sr content in the fossil fish teeth from GC03 (0.70924) and GC04 (0.70940) to GC11 (0.70964). It is possible that the GC11 core, at a lower latitude than the GC03 and GC04 cores, was more influenced by continental detritus from Antarctica sources with high ⁸⁷Sr/⁸⁶Sr. Furthermore, high field strength elements Hf and Zr, large ion lithophile element Sr, and rare earth elements exhibited a decreasing trend in the fossil fish teeth with GC03>GC04>GC11 ( Table 2 ), which indicates a weakened terrigenous signal from GC11 and GC04 to GC03. It is reasonable to speculate that fine terrestrial debris with high ⁸⁷Sr/⁸⁶Sr may enter the fine pores of fish teeth or through exchange of Sr isotopes at the microscopic level after deposition but with minor impact on the Sr content. If so, fossil fish teeth may preserve the influence from enhanced continental weathering signal and prove the long diagenesis influence after deposition (Martin and Scher, 2004).

As for the significantly more radioactive ϵNd, it may be closely related to the release of Nd from volcanic hydrothermal components in sediments into pore water. While cores GC03 and GC04 were less affected by Antarctic continental debris, they received more from seafloor spreading and mid-ocean ridge hydrothermal activity (Subha Anand et al., 2019), and the radioactive ϵNd of these materials may have been absorbed by fish teeth through pore water after deposition. This means that fossil fish teeth may also preserve the influence from volcanic hydrothermal weathering signal and the influence is multi-factor controlled. In general, the suspected sources or continuous absorption of Sr and REEs from pore water could occur, which may cast doubt on the reliability of fish teeth fossils as valid records of deep-sea environments in Earth’s history over millions of years. The fractionation of Sr-Nd isotopes in fish teeth and the related diagenetic change after burial are worthy of further study.