As the largest continental shelf in the northwest Pacific Ocean (i.e., east Asian marginal seas), the ECS plays a critical role as a significant carbon reservoir, absorbing atmospheric carbon dioxide at a rate of 2.0–3.0 mol m⁻² yr⁻¹ (Wang et al., 2000; Kim et al., 2013a). It is also one of the most productive fishing grounds for mackerel (Scomber japonicus) and anchovy (Engraulis japonica) and serves as a spawning ground for the Japanese flying squid (Todarodes pacificus) (Rebstock and Kang, 2003; Kim et al., 2013b).

The ECS is a hydrographically dynamic system driven by the Kuroshio current and northeast Asian Monsoon (Fig. S1, Lie and Cho, 2016). Like other large marine ecosystems that receive a tremendous amount of freshwater such as the Amazon Estuary (∼6300 km³ yr⁻¹) and Arctic shelves (∼2000 km³ yr⁻¹), the ECS receives a large amount of freshwater annually (900 km³ yr⁻¹), especially during the summer monsoon period, and exhibits comparable biogeochemical complexity due to intense riverine input, seasonal stratification, and anthropogenic pressures (Milliman and Farnsworth, 2013; Spielhagen and Bauch, 2015). Therefore, the ECS is not only important for its ecological and fisheries productivity but also serves as a large marine ecosystem laboratory for investigating microbial biogeochemical processes in continental shelf systems influenced by enhanced freshwater inputs associated with the increasing trend of extreme precipitation under climate change (Khadgarai et al., 2021; Qiao et al., 2024).

The mixed layer depth (MLD) was defined as the depth immediately above the onset of the thermocline. The euphotic depth was determined based on Lambert–Beer's law described in Poole and Atkins (1929) using a Secchi disk. The Riemann sum method was applied to integrate the parameters within the mixed layer and euphotic depths. Statistical analyses were conducted using IBM SPSS Statistics 27. Detailed descriptions of statistical procedures are presented in Sect. S4. Abbreviations used in the main text and figures are provided in Table S4.

Based on T–S diagrams, the current systems in the study area were defined by 2–4 water masses that vary seasonally (Fig. 2). The two main currents appearing in the nECS are the Kuroshio Source Water and the Shelf Mixed Water (Fig. 2). The Kuroshio Source Water is formed by the convergence of the TWC and the TWW, which are both branches of the main Kuroshio Current in the nECS and are characterized by high temperature and salinity (Fig. S1). In contrast, the Shelf Mixed Water is formed by the mixing of the Kuroshio Source Water with the Chinese Coastal Current. The Chinese Coastal Current, which is mainly observed on the shelf side of the East China Sea, is characterized by cold and low-salinity properties (Li et al., 2006). As a result, the Shelf Mixed Water exhibits relatively lower temperature and salinity due to the influence of the Chinese Coastal Current. During February (winter), May (spring), and November (autumn), the nECS consisted of Kuroshio Source Water and Shelf Mixed Water (Fig. 2a, b, d, e, f, h, i, j and l; Table S1 in the Supplement). However, during August (summer), a more complex water mass distribution was observed due to strong stratification and substantial freshwater input: TWC and YRDW were present near the surface, whereas TWW and Shelf Mixed Water were observed in the mid to lower layers, which differed from other seasons (Fig. 2c, g and k; Table S1). The satellite images of sea surface salinity (Fig. 1), together with the T–S diagrams (Fig. 2), clearly revealed that YRDW (<31 psu, the blue-green colors in Fig. 1c, g and k) originating from the Yangtze River in August expands northeastward to the nECS located approximately 300 km away from the Yangtze River estuary. The expansion of YRDW to the nECS was greatest in August 2020, followed by August 2022 and 2021, respectively (Figs. 1 and 2).

The vertical distributions of DOC reflected seasonal changes in water column structure (Fig. 3). In February and November, DOC concentrations were relatively homogeneous within the MLD (50–70 m), averaging 65±2 µM (53–76 µM) and 64±2 µM (55–83 µM), respectively (Fig. 3c and w). In contrast, during May and August, DOC concentrations increased with decreasing MLD, averaging 72±1 µM (66–74 µM) in May and 86±3 µM (75–130 µM) in August (Fig. 3k and q; Table 1).

Elevated DOC concentrations were particularly evident in August within the low-salinity surface layer (upper 10 m), where values averaged 95±19 µM (76–130 µM; Fig. 3p and q). Among the summer observations, August 2020 exhibited the highest DOC concentrations (104±5 µM), which were significantly higher than those in August 2021 and 2022 (77±1 µM and 80±3 µM, respectively; p=0.002), coinciding with the strongest influence of YRDW observed during the study period (Figs. 1c and 4a). Chl a concentrations within the MLD followed a similar pattern: homogeneously low in February (0.47±0.02 µg L⁻¹; 0.42–0.49 µg L⁻¹), increasing in May (1.05±0.24 µg L⁻¹; 0.77–1.22 µg L⁻¹) with the development of thermal stratification, remaining elevated in August (avg. 0.74±0.22 µg L⁻¹; 0.58–0.99 µg L⁻¹) under strong haline stratification associated with YRDW intrusion, and decreased in November (avg. 0.53±0.02 µg L⁻¹; 0.50–0.55 µg L⁻¹) (Fig. 3i, p, l and r).

PP, integrated over the euphotic depth, followed a similar seasonal pattern, increasing from May to August and peaking at 336±55 mg C m⁻² d⁻¹ (August 2021) and 263±73 mg C m⁻² d⁻¹ (August 2022) (Fig. 4d). Notably, PP in August 2020 (182±77 mg C m⁻² d⁻¹) was significantly lower than other years (p=0.028) (Fig. 4d), despite high YRDW input.

Mean PA within the euphotic depth was lowest and vertically homogeneous in February (0.46±0.03×109 cells per L; 0.35–0.59×109 cells per L). In contrast, PA reached its highest values in May (1.28±0.09×109 cells per L; 0.48–1.88×109 cells per L), before declining in August (0.99±0.09×109 cells per L; 0.37–2.36×109 cells per L) and November (0.66±0.08×109 cells per L; 0.28–1.29×109 cells per L) (Fig. 3; Table 1). PA correlated positively with DOC (ρ=0.255, p<0.001) and Chl a (ρ=0.281, p<0.001; Table S2).

HPP within the euphotic depth followed a seasonal cycle: lowest in February (0.33±0.12 µg C L⁻¹ d⁻¹; 0.20–0.58 µg C L⁻¹ d⁻¹), rising in May (3.73±1.37 µg C L⁻¹ d⁻¹; 1.76–5.86 µg C L⁻¹ d⁻¹), reaching a consistent summer peak in August (5.51±4.08 µg C L⁻¹ d⁻¹; 0.69–17.43 µg C L⁻¹ d⁻¹), and then declining in November (0.83±0.26 µg C L⁻¹ d⁻¹; 0.43–1.24 µg C L⁻¹ d⁻¹) (Fig. 3). HPP correlated positively with temperature (ρ=0.525, p<0.001), DOC (ρ=0.457, p<0.001) and Chl a (ρ=0.398, p<0.001; Table S2) consistent with the observed summer peak.

In the present study, one of the most prominent features observed in oceanographic and microbiological properties was the consistent summer peak in HPP (Fig. 4e). Considering the typically enhanced HPP coupled with phytoplankton bloom in spring in the middle latitudes of the northern hemisphere (Amon and Benner, 1998; Lemée et al., 2002; Gomes et al., 2015), our results showing consistently high summer peaks in microbiological parameters are noteworthy. Plankton activities are positively correlated with temperature in many marine systems, a pattern that has been widely reported in previous studies (White et al., 1991; Robinson and Williams, 1993), suggesting that elevated summer temperatures may partially contribute to high HPP. However, assuming a Q10 of 2, a 10° temperature increase from 15 to 25° would yield an approximately twofold increase in HPP. In contrast, estimates from the observed seasonal temperature–HPP relationship suggest a fivefold increase from 28 to 140 mg C m⁻² d⁻¹ (Fig. S3), exceeding that expectation. In addition to temperature, positive correlations between HPP and DOC (ρ=0.457, p<0.001) and between HPP and Chl a (ρ=0.398, p<0.001) indicate that these factors acted as co-drivers of the consistently elevated HPP observed in summer, coinciding with the expansion of YRDW (Fig. 1; Tables 1 and S2).

Since the expansion of YRDW in summer was the most notable observation in the nECS (Fig. 2), we further analyzed the impact of YRDW in regulating the HPP to better interpret the cause-effect relationship on the high HPP in summer. Firstly, the correlation analysis between salinity and DOC for three years (2020–2022) consistently revealed a significant negative relationship (Fig. 5a–c). Especially higher concentrations of DOC were observed in August with low salinity conditions (<31 psu), which indicated a substantial input of DOC into the nECS via the inflow of YRDW. Secondly, the HPP showed a significantly positive relationship with DOC in the summer of all three years, which suggested that the high HPP in summer is greatly stimulated by the DOC supply (Fig. 5d–f). Moreover, the three-year pooled summer Chl a data also showed a significant correlation with HPP, indicating the combined effect of DOC from terrestrial and phytoplankton sources (Fig. 6a). Consequently, these correlation analyses between salinity-DOC-HPP, along with the tight coupling between Chl a and HPP, clearly demonstrated that the high HPP in summer is intrinsically stimulated by the input of DOC-rich YRDW and enhanced summer phytoplankton bloom driven by nutrient input (Figs. 5 and 6a). Similar findings have been reported in the southern ECS (25.4–31.6° N, 120.5–127.0° E), where HPP in the euphotic depth was highest during the summer months, coinciding with the wider distribution of YRDW in summer periods (Shiah et al., 2003; Chen et al., 2009, 2021). Our results, combining extensive satellite images analyses, variations of hydrographic conditions, concentrations of DOC and Chl a, and HPP variations with season, provide comprehensive and direct evidence on the impact of YRDW in stimulating consistently high HPP in summer.

Recent climate modeling studies suggest that the intensity of precipitation in East Asia during summer monsoon period will be increasing with global warming (Li et al., 2019b, Khadgarai et al., 2021; Qiao et al., 2024). In this regard, the high HPP in summer in the nECS has a significant ecological implication to the microbial oceanography of the adjacent seas (i.e., Yellow Sea and East Sea) where high HPP still prevails in spring. To investigate further the potential impact of YRDW on HPP in summer in the adjacent seas, we assembled available HPP data: (1) from the East Sea that is affected by the intrusion of YRDW in summer, and (2) from the Yellow Sea, relatively less affected by the YRDW (Figs. S1 and 5). Indeed, the average HPP in summer (62±72 mg C m⁻² d⁻¹) of the East Sea was not significantly different from that measured in spring (77±94 mg C m⁻² d⁻¹) (Fig. S5b, p=0.518). This relationship suggests that the YRDW directly flowing into the East Sea via nECS (Chang and Isobe, 2003) stimulates the HPP in summer. In contrast, the average HPP in summer (22±9 mg C m⁻² d⁻¹) of the Yellow Sea remained significantly lower than that measured in spring (75±58 mg C m⁻² d⁻¹) (Fig. S5c, p<0.001). Our findings of elevated HPP in summer of the nECS and the East Sea highlight the importance of sustained ecological monitoring to better understand the extensive influence of climate change-induced precipitation on the HPP in the nECS and its adjacent seas (Chen et al., 2021).

Despite the extraordinarily enhanced DOC input in August 2020 (Figs. 4a, 5a and S4a) resulting from the unusually heavy rainfall compared to that measured in August 2021 and 2022 (Figs. 1c, 2c and 4a), the HPP value in 2020 remained similar to that of 2021 and 2022 (Figs. 4e and S4b). This result suggests that a substantial fraction of the DOC transported to the nECS via YRDW in summer 2020 contained much more recalcitrant dissolved organic matter (DOM) for HP metabolism (Guo et al., 2014; Li et al., 2020; Chen et al., 2024). Indeed, the fluorescence intensity of the FDOM_H (i.e., humic-like FDOM, Coble, 1996) was highest in the high DOC range (>90 µM) (Fig. 7a) within the YRDW fraction (<28 psu) (Fig. 7b) in August 2020. Consistent with this, the HIX, a proxy for the refractory nature of DOM (Hansen et al., 2016; Li et al., 2019a), also reached its highest level (1.79±0.36) in August 2020 (Fig. 6c). HIX reflects the relative contribution of humic-like FDOM in the DOM pool, which is generally characterized by high aromaticity, molecular weight, and structural complexity (Hansen et al., 2016). Such humic-like FDOM are known to exhibit low biogeochemical reactivity and resistance to microbial degradation, resulting in their persistence in aquatic environments (Yamashita and Tanoue, 2008; Cao et al., 2019). Under these conditions, HP tend to utilize organic matter primarily for cell maintenance (i.e., for energy-yielding respiration) rather than for biomass synthesis (i.e., production) (Carlson et al., 2007; Fasching et al., 2014).

A plausible explanation for maintaining a low bioavailable DOC pool in August 2020 is the reduced PP observed during that period (Fig. 4d). The Yangtze River carries a large amount of sediment into the estuary (ca. 4.86×108 t yr⁻¹), making it one of the most turbid estuaries in the world (Guo et al., 2014). This high turbidity can result in limited photodegradation of humic substances in the Yangtze River Estuary (Li et al., 2015). The YRDW, which contains abundant terrestrial humic substances, can reach the nECS after passing through this highly disturbed area, driven by southeasterly winds (Fig. S1). The strong signals of FDOM_H in August 2020, when YRDW input was greatest, suggest both elevated terrestrial input and limited photodegradation (Fig. 7b). This implies that a more turbid water plume reached the nECS in 2020 than in 2021 and 2022. Indeed, the average euphotic depth in August 2020 was only 26±4 m, compared to 33±6 m and 34±11 m in 2021 and 2022, respectively. This higher turbidity likely limited light penetration, resulting in reduced PP and a diminished supply of labile DOC (Huguet et al., 2009; Jiang et al., 2016). As a result, the DOM pool in August 2020 was characterized by a higher relative contribution of refractory DOC, as reflected in the elevated HIX values (Fig. 7c).

Consequently, although the intrusion of YRDW supplying a substantial amount of bioavailable DOC in summer significantly stimulates HPP (Fig. 5), exceptionally great YRDW input, as it was observed in August 2020 (Fig. 5a), likely altered DOC quality by oversupplying terrestrial recalcitrant DOM and reduced PP by limiting light availability in turbid water column, thereby limiting the additional stimulation of HPP (Fig. 4e).

The Yangtze River has been reported to change the nutrient balance in the ECS by excessive DIN influx and relatively low DIP influx (Chen et al., 2020). In the turbid estuary, the DIP is generally removed by suspended particles (Lopez et al., 1996), which consequently triggers a nutrient imbalance (i.e., high N : P ratio) in the Yangtze River Estuary. In general, DOC is recognized as the major limiting substrate for HP growth in the ocean. However, since pico-sized prokaryotes have a high surface area-to-volume ratio that requires more phospholipids to make up cell membranes and contain higher nucleic acids (P-rich) fractions than phytoplankton on a per-cell basis, they have relatively high P demand for growth (Goldman et al., 1987; Fagerbakke et al., 1996). This frequently results in a P-limited growth of HP in various marine environments (Thingstad et al., 2005; Yuan et al., 2011).

To directly demonstrate the impact of YRDW in regulating HP growth, we conducted three incubation experiments under different DOC and P conditions (Fig. 8; Table 2). The first experiment was carried out in August 2020 at the nECS site (Fig. 8a; Exp. 1 in Table 2) where the hydrographic condition was characterized by the expansion of the YRDW with high N : P ratio (34) and exceptionally high DOC containing high humic-like FDOM (Fig. 7). The second experiment was conducted during the same summer as the first experiment in the East Sea where the water mass was dominated by the high T–S TWW with N : P ratio of 15.09 (Fig. 8b; Exp. 2 in Table 2). The HPP was stimulated in P-amended samples exclusively at the YRDW-prevailing nECS site (Fig. 8a), while HPP was stimulated in glucose amended samples at TWW-prevailing East Sea site (Fig. 8b). Considering the highest DOC concentration (91 µM) and an abnormally high N : P ratio (34) (Table 2), the results indicated that intrusion of P-depleted YRDW was primarily limiting the HPP in the nECS in summer (Fig. 8a; Table 2). Based on the two experiments, we further hypothesized that if YRDW is not prevailing in the nECS, the HPP is not likely to be limited by the P deficiency. To confirm the hypothesis, we conducted the third experiment at the nECS site in April 2021 when the water mass was dominated by the Kuroshio source current (TWW and TWC) with high salinity (Fig. 8c; Exp. 3 in Table 2). Indeed, the HPP was stimulated only in the samples amended by glucose (+C, +CN and +CP). Combining these three growth-limiting resource experiments conducted at different water regimes in time and space demonstrates that the HPP in the nECS is closely regulated by the hydrographic conditions varying with season. It is primarily limited by P availability in summer when YRDW prevails, whereas organic carbon limitation is more significant in other seasons when Kuroshio source current (TWC and TWW) dominates (Fig. S1).

This P deficiency may have also influenced phytoplankton composition and their production of bioavailable organic matter, ultimately affecting HPP. In August 2020, despite the tight coupling between HPP and DOC (p<0.001, Fig. 5d), there was no significant correlation between HPP and Chl a (p=0.405, Fig. 6b), unlike the patterns observed in other years (Fig. 6c and d). This uncoupling between HPP and phytoplankton-derived organic matter suggests a possible shift toward smaller phytoplankton communities under P limitation (Fig. 6b). Previous studies in the ECS have reported that elevated N : P ratios associated with YRDW inputs can drive shifts in phytoplankton community structure, transitioning from larger diatom dominance to smaller phytoplankton such as cyanobacteria, small dinoflagellates, and cryptophytes that prefer environments with low DIP availability (Gomes et al., 2018; Park et al., 2022). Such DIP limitation on phytoplankton growth may be further intensified by competition with HP that can be more efficient in acquiring phosphorus under high-DOC and low-P conditions driven by YRDW (Drakare, 2002). Moreover, exudates released from P-depleted phytoplankton tend to be less bioavailable for HP growth by supplying more recalcitrant organic carbon that is poorly suited for bacterial uptake (Kragh and Søndergaard, 2009; Puddu et al., 2003). Therefore, severe P deficiency induced by the excessive input of YRDW results in uncoupling between HPP and phytoplankton-derived organic matter (i.e. P-depleted exudates), leading HPP in August 2020 to rely primarily on terrestrial DOC (Fig. 5d).

Together with the refractory properties of the DOC and the reduced PP driven by high turbidity (Fig. 7), this P-limited HPP (Fig. 8a) and its uncoupling with phytoplankton (Fig. 6b) in August 2020 provide further evidence explaining why HPP was not so great even in summer 2020 with exceptionally high DOC compared to that measured in summer 2021 and 2022 (Fig. 4e). In a P-starved condition, HP is unable to synthesize nucleic acids, protein, and ATP, which limits their ability to replicate cells and metabolize (i.e., degrade and uptake) organic carbon (Obernosterer et al., 2003; Kritzberg et al., 2010).

Given that the YRDW delivers high DOC but low phosphorus to the continental shelves of the ECS, it may alter substrate availability and affect the structure of both heterotrophic and autotrophic communities (Gomes et al., 2018; Kim et al., 2022), potentially leading to shifts in the relative contributions of basal producers (i.e., HP and phytoplankton) to food web dynamics and ecosystem carbon cycling. To evaluate the carbon flow through the microbial food web, we calculated the ratio of HPP to PP, representing the proportion of carbon flowing through HPP relative to PP (Fig. 9). The HPP : PP ratios observed in February and November fell within the global average 0.15 (Fig. 9a, Ducklow, 2000). However, in May, HPP : PP ranged from 0.25 to 0.53, and notably, it consistently exceeded 0.5 in August (Fig. 9a). This result indicates that the microbial food web is significantly enhanced in the nECS, especially in summer supporting a comparable amount of carbon flow through the microbial food web (Rowe et al., 2025). Notably, in August 2020, when the influence of YRDW was particularly strong, the high HPP : PP ratio (up to 0.84) is likely due to the stimulated HPP and reduced PP, resulted from increased DOC input and high turbidity, respectively (Fig. 4d and e). This suggests that climate change-induced precipitation may increasingly deliver turbid, DOC-rich, but nutrient-imbalanced freshwater to the ECS, ultimately resulting in basal production being more strongly sustained by HP than by phytoplankton, thereby driving the ECS toward a more microbially dominated food web. Such a shift could reduce the efficiency of energy transfer to higher trophic levels, potentially affecting fishery resources (Berglund et al., 2007; Rowe et al., 2025). Supporting this trend, compiled historical summer data over the past two decades revealed a gradual increase in HPP : PP ratios in the ECS (Fig. 9b). This finding suggests ongoing environmental changes, including warming and intensified precipitation events (Belkin, 2009; Khadgarai et al., 2021; Qiao et al., 2024), may increasingly shift carbon flow toward the microbial food web rather than the classical grazing food chain. Therefore, our results highlight that YRDW runoff driven by climate change can significantly alter the trophic structure and energy transfer efficiency in the ECS, ultimately affecting fisheries by favoring smaller-sized consumers. To better understand these dynamics, further microbial oceanographic studies are needed to elucidate how YRDW influences interactions between HP parameters (i.e., respiration and carbon demand) and phytoplankton parameters (i.e., biomass and production). Such efforts are essential for assessing current ecosystem functioning and predicting future shifts in microbially mediated carbon cycling and metabolic balance (i.e., heterotrophy vs. autotrophy) in the nECS.