The ECS is a typical marginal sea influenced by large rivers and characterized by complex hydrological conditions. The world’s third-largest river, the Changjiang River, carries a large amount of fresh water and nutrients into the ECS annually. Driven by the East Asian monsoon, multiple currents converge in the ECS and interact in a dynamic manner (Guo et al., 2018). Affected by the plume of the Changjiang River, the water column of the ECS becomes stratified during summer, and the surface layer possesses high primary production due to favorable light and temperature and sufficient nutrients (Gong et al., 2003).

Sampling was conducted aboard the R/V Kexue 3 from August 8 to August 20, 2020. Water samples were collected from discrete depths (2, 10, 20, 30, 50 m and bottom layer) at five stations (Figure 2). A Seabird conductivity-temperature-depth (CTD) system incorporated with 12-L Niskin bottles was used for seawater collection and to obtain the in situ salinity and temperature. Water samples for particulate isoGDGTs and particulate organic nitrogen (PON) were filtered through Whatman GF/F membranes (0.7 μm pore size; precombusted at 450°C for 5 h) immediately after collection. The filtrate was collected in duplicate in acid-cleaned, 60-mL high-density polyethylene (HDPE) bottles and stored frozen (-20°C) for further analysis of nutrients. Samples for chlorophyll-a (Chl-a) analysis were collected using cellulose acetate fiber filters (0.45 μm pore size; Whatman) and stored in the dark at -20°C. Water samples for heterotrophic bacterial abundance (HBA) analysis were collected in cryopreservation tubes and immediately fixed with 1 mL of 1% paraformaldehyde. Then, the samples were stored in liquid nitrogen until analysis in the laboratory. The surface sediments (top 0–2 cm) were obtained using a box sampler, and the samples were stored at -20°C.

The temperature and salinity in the study area were in the ranges of 18.9–28.9°C (mean ± SD: 25.0 ± 3.8°C) and 30.10–34.56 (mean ± SD: 34.06 ± 0.87), respectively (Figures 3A,B). The temperature and salinity values in the upper 20 m of the nearshore stations (DH4-0 and DH5-1a) changed dramatically, whereas the salinity was vertically uniform at the offshore stations (DH5-2, DH5-2a, and DH5-3). The pH showed a narrow range (7.98–8.24) and was distributed relatively uniformly throughout the water column (Figure 3C). Concentrations of DO ranged from 125.6 to 238.2 μmol L^–1 and shared a similar distribution pattern with the temperature (Figure 3D). Compared with those of offshore stations (DH5-2, DH5-2a, and DH5-3), the values of temperature, salinity, pH and DO at nearshore stations (DH4-0 and DH5-1a) were relatively low. In contrast, the Chl-a concentrations (0.08–3.63 μg L^–1) were higher at nearshore stations (DH4-0 and DH5-1a; mean ± SD: 0.98 ± 1.09) than at offshore stations (DH5-2, DH5-2a, and DH5-3; mean ± SD: 0.37 ± 0.32) (Figure 3E). A remarkable vertical gradient of Chl-a concentrations was observed at nearshore stations DH4-0 and DH5-1a. The depth profile showed that the HBA values were elevated near the surface (upper 20 m) and decreased with increasing depth (Figure 4F).

The concentrations of NH₄–N, NO₂–N, and NO₃–N ranged from 0.47 to 2.92, 0.01–0.91, and 0.01–12.87 μmol L^–1, respectively (Figures 3F–H). The vertical distributions of NH₄–N and NO₂–N were highly variable. In comparison, the NO₃–N concentration generally exhibited an increasing trend from the surface to the bottom except for the obviously high value at the surface of station DH4-0.

Concentrations of DON ranged from 23.69 to 51.82 μmol L^–1 (average ± SD: 33.91 ± 6.49 μmol L^–1) but displayed no apparent pattern with increasing depth (Figure 5A). The vertical distribution of PON was similar to that of Chl-a and varied between 0.39 and 2.51 μmol L^–1 (Figure 5B).

The concentrations of isoGDGTs were between 3.66 and 52.06 ng L^–1 (average ± SD: 19.63 ± 14.75 ng L^–1) and gradually increased with increasing depth (Figure 4A). The isoGDGT concentrations in this study were comparable to those (8.30-237.10 ng L^–1) reported by Wang et al. (2019) in the same season but lower than those (27.60-6,230 ng L^–1) reported by Lü et al. (2019a). Similar to NO₃–N, relatively high concentrations of isoGDGTs were observed at nearshore stations DH4-0 and DH5-1a. Cren and GDGT-0 accounted for 66.2 ± 6.1 and 15.5 ± 5.2% of the total isoGDGT content, respectively (Figures 4B,C). Cren and GDGT-0 showed similar distribution patterns, and the ratio of GDGT-0/Cren ranged from 0.08 to 0.43 (mean ± SD: 0.24 ± 0.10) among all samples (Figure 4D). The TEX₈₆ ranged from 0.50 to 0.86 and displayed considerable variability in the water column (Figure 4E).

The origins of isoGDGTs can be evaluated by the ratio of GDGT-0/Cren. A low ratio value (<2.0) of GDGT-0/Cren indicates a main source from Thaumarchaeota, while a high ratio value (>2.0) suggests the existence of methanogenic archaea (Blaga et al., 2009). In the present study, GDGT-0/Cren values between 0.08 and 0.43 suggest the dominance of Thaumarchaeota-derived isoGDGTs in the ECS. This is also supported by the archaeal 16S rRNA gene data (Zeng et al., 2007).

The vertical distribution of isoGDGT concentration depicted an increasing trend from the surface to the bottom (Figure 4C). This vertical trend was coupled with the activity of Thaumarchaeota in the water column (Pitcher et al., 2011). Previous studies on the ECS ammonia monooxygenase gene showed that the amoA in bottom water was significantly higher than that in the surface layer (Zhang et al., 2014). This result is consistent with our findings, indicating that Thaumarchaeota prefer to thrive in the deeper layers of the ECS. To further analyze the influence of various physical, chemical and biological parameters on the distribution of isoGDGTs, redundancy analysis (RDA) was performed in this study. The RDA results showed that the first principal component explained 99.09% of the variation in the isoGDGT data (Figure 6). Temperature, DO and pH, as well as nitrogen, presented a strong influence on the distribution of isoGDGTs. The relatively low oxygen and low pH microenvironment of the bottom particles are conducive to the ammonia oxidation process of Thaumarchaeota, which has been confirmed by previous studies (Hanaki et al., 1990; Hsiao et al., 2014). As a reactant in ammonia oxidation, the availability of NH₄–N plays a leading role in the prevalence of Thaumarchaeota (Könneke et al., 2005; Yan et al., 2012; Evans et al., 2018). In summer, stratification is widely developed in the ECS, especially at the nearshore DH4-0 and DH5-1a stations. The pycnocline separates the vertical exchange of nutrients, resulting in two different patterns of nitrogen cycling in the water column. Above the thermocline, the high phytoplankton production, as revealed by the high Chl-a content (Figure 3E), consumed a large amount of NH₄–N. At the same time, the fresh organic matter produced was quickly decomposed by heterotrophic bacteria, which also stimulated heterotrophic bacterial blooms at the surface (Figure 4F). The NH₄–N released by mineralization was utilized by phytoplankton. This rapid NH₄–N cycling made it difficult for nitrogen to accumulate and nitrify at the surface, showing some characteristics oligonitrates (excluding the DH4-0 surface NO₃–N concentration, which was severely affected by terrestrial input). In this pattern, Thaumarchaeota can hardly compete with phytoplankton for NH₄–N (Murray et al., 1999; Galand et al., 2010). Furthermore, the intense radiation in the surface water in summer also exerted a negative effect on the growth of Thaumarchaeota (Merbt et al., 2012). The NO₃–N concentration at the surface of station DH4-0 was remarkably high (Figure 3H), which was attributed to terrestrial input, as revealed by the low salinity (Figure 3B). However, below the pycnocline, the stress from phytoplankton competition gradually decreased due to light limitation. The NH₄–N released by the mineralization of sedimentation particles can be used by Thaumarchaeota to mediate the nitrification process and promote the accumulation of NO₃–N (Hsiao et al., 2014). Even when the concentrations of NH₄–N in the ambient environment are low, archaea can still maintain thriving growth conditions (Martens-Habbena et al., 2009). Although vertical variations in isoGDGTs were observed, the expected increase in NO₂–N with depth was not found, which may be due to its active chemical properties (such as rapid oxidation). However, by excluding the DH4-0 surface data, the strong correlations between isoGDGTs and NO₃–N (R² = 0.75, p < 0.01) and between Cren and NO₃–N (R² = 0.74, p < 0.01) indicate that archaeal ammonia oxidation plays an important role in the ECS nitrogen transformation. Hsiao et al. (2014) found that the nitrification rate was significantly correlated with PON but not with DON in the ECS, implying that PON had a greater potential to provide NH₄–N than DON. The coupling of relatively high PON and isoGDGTs at DH4-0 and DH5-1a (Figures 4A, 5B) and a significant correlation between DO and PON (R² = 0.52, p < 0.01) further indicated that PON as a remineralization substrate was more effective in providing NH₄–N. However, for DON, its vertical distributions were highly variable (Figure 5A), and no significant relationships between DON and DO and isoGDGTs were found (p > 0.05), which may be related to its low bioavailability and diverse sources (Hsiao et al., 2014; Carlson and Hansell, 2015).

As a bioarchive of seawater temperature, TEX₈₆ has been widely used and reported (Pearson and Ingalls, 2013; Schouten et al., 2013; Lü et al., 2015, 2019a). Since TEX₈₆ was proposed, its statistical relationship with surface seawater temperature (SST) has undergone many modifications, but it is not feasible to obtain accurate results in different regions with one formula (Kim et al., 2010; Tierney and Tingley, 2014). Especially in the marginal seas, due to the complex hydrodynamic process and the disturbance of human activities, the records and robustness of TEX₈₆ still need to be further evaluated.

The reconstruction of paleo-SST by TEX₈₆ is based on the assumption that sedimentary isoGDGTs originated from surface archaea. However, our study found that the isoGDGT concentration increased with increasing depth in the water column. This indicated that, at least in summer, the isoGDGTs in ECS surface sediments are not derived from the surface and are mainly controlled by bottom archaeal production. The depth profile including the surface sediments showed that the surface sediments and the bottom layer exhibit similar isoGDGT concentrations (Figure 7), further supporting this hypothesis. From this perspective, the sedimentary TEX₈₆ cannot accurately retrieve the surface water temperature, and it is more likely to record the bottom temperature signals. This also explains why TEX₈₆ in surface sediments usually does not accurately represent surface temperature in summer (Duan et al., 2019). To further verify the suitability of sedimentary TEX₈₆ as a bottom temperature proxy, we collected and reanalyzed the previously published isoGDGT data of surface sediments in the ECS (Lü et al., 2014). We found that the sedimentary TEX₈₆ had a stronger relationship with the annual mean bottom seawater temperature (BST) (R² = 0.82, p < 0.01; Figure 8B) than with the annual mean SST (R² = 0.22, p < 0.01; Figure 8A), indicating that the sedimentary TEX₈₆ in the ECS is more suitable as a proxy for BST. The optimal relationship between sedimentary TEX₈₆ and annual average BST was also found by Wang et al. (2019) in the eastern China marginal sea, further confirming that sedimentary TEX₈₆ is a proxy for ECS bottom temperature. In this scenario, in coastal settings, sedimentary isoGDGTs could be mainly derived from bottom archaeal production, and sedimentary TEX₈₆ best reflects BST. However, in the open sea, the sedimentary isoGDGTs mainly originate from the subsurface water, resulting in sedimentary TEX₈₆ reflecting the subsurface temperature (Huguet et al., 2007; Hernández-Sánchez et al., 2014).

In addition to the temperature records by TEX₈₆, whether TEX₈₆ only depends on temperature is also debatable. In this study, the relationship between TEX₈₆ and temperature presented two scenarios (Figure 9). When the temperature was greater than 25.0°C, TEX₈₆ was significantly related to temperature (R² = 0.58, p < 0.01), while this relationship disappeared when the temperature was below 25.0°C. Previous studies have found that DO and pH could also affect TEX₈₆, which was confirmed in pure archaeal culture experiments (Oger and Cario, 2013; Elling et al., 2015; Qin et al., 2015). Significant correlations were found between TEX₈₆ and DO (R² = 0.14, p < 0.01) and pH (R² = 0.36, p < 0.01), indicating that TEX₈₆ does not only serve temperature in the ECS during summer. Samples with temperatures below 25.0°C have relatively low DO content and pH values driven by stratification, and TEX₈₆ increased compared to the relational equation for samples with a temperature greater than 25.0°C. This phenomenon was consistent with previous findings, indicating that a decrease in DO or pH could lead to an increase in TEX₈₆ (Elling et al., 2015; Zhang et al., 2016; Lü et al., 2019a; Cao et al., 2020). For samples with a temperature below 25.0°C, a significant correlation was found between pH and TEX₈₆ (R² = 0.33, p < 0.05), but no significant correlation was found between DO and TEX₈₆, indicating that the effect of pH on TEX₈₆ was greater than that of DO. However, more data are needed to confirm this hypothesis. Another argument for TEX₈₆ bias is the supply of NH₄–N. Recent studies have shown that the low ammonia oxidation rate and energy limitation caused by a low supply of NH₄–N resulted in a positive deviation of TEX₈₆ (Hurley et al., 2016; Evans et al., 2018). Lü et al. (2019a) investigated isoGDGT concentrations in the water column of the ECS and found that the deviation between the reconstructed temperature based on isoGDGTs and the in situ temperature gradually increased with increasing water depth in summer. They attributed this phenomenon to the reduction in the ammonia oxidation rate at the bottom layer. However, in this study, no significant relationship was found between NH₄–N and TEX₈₆, indicating that NH₄–N had a limited effect on TEX₈₆. Alternatively, the rapid turnover of NH₄–N in the water column could have caused the statistical relationship between NH₄–N and TEX₈₆ to be lost. Furthermore, the growth phase of archaea also has a profound effect on the composition of membrane lipids, and the bias of TEX₈₆-derived temperature between the growth and stationary phases can be up to 9°C (Elling et al., 2014). However, it is difficult to distinguish the contribution of various influencing factors in this study. Overall, the coupling of multiple factors leads to the divergence of TEX₈₆ to the summer temperature in the ECS. More seasonal data are needed to further understand this issue.