From February to April 2022, a research cruise was conducted aboard the R/V Kexue (voyage NORC2021-09) in the TNWP Warm Pool. During the survey, a total of six sampling stations were set up: EQ-6, E142-3, E142-7, E142-11, E142-13, and E142-19 (Fig. 1).

Seawater samples were collected from 16 depths (5, 50, 100, 150, 200, 300, 400, 500, 600, 800, 1000, 1200, 1400, 1600, 1800, and 2000 m) using 12 L Niskin bottles (KC-Denmark, Denmark) mounted on a shipboard CTD rosette (Conductivity-Temperature-Depth, Sea-bird SBE911, United States). After seawater collection, DO samples were first aliquoted and fixed in 50 mL amber glass bottles. Then, the seawater was filtered through pre-combusted and acid-washed GF/F glass fiber filters (0.7 µm pore size, Whatman; 450 °C for 4 h, 0.5 M HCl for 24 h), followed by the collection of nutrient and DIC samples. Samples for nutrient determination were aliquoted into 250 mL high-density polyethylene bottles, amended with chloroform (approximately 1:500, v/v) to suppress biological activity during delayed analysis, and stored frozen at -20 °C before measurement (Ma et al., 2023). Samples for DIC analysis were collected in 50 mL airtight glass bottles. After overflow, 1 mL of the water sample was withdrawn and the remaining sample was poisoned with saturated mercuric chloride solution to suppress biological activity. The samples were stored at 4 °C until analysis.

Suspended particulate samples for POC analysis were collected by filtering 2–4 L of seawater through pre-combusted and acid-washed GF/F glass fiber filters (0.7 µm pore size, Whatman; 450 °C for 4 h, 0.5 M HCl for 24 h). Samples were stored at -20 °C until measurement. TEP samples were collected by filtering 1–4 L of seawater through polycarbonate filters (0.45 µm pore size, Whatman) under a vacuum of 170 mmHg. The retained particles were stained with 0.5–1 mL of a 400 mg L⁻¹ Alcian Blue solution (Sigma-Aldrich, USA) for 5–10 s, prepared in Milli Q water and acidified to pH 2.5 with glacial acetic acid. The filters were then rinsed with 5 mL of MilliQ water, and stored at -20 °C until measurement (three replicate samples were collected at each depth). Samples for chlorophyll a (Chl a) determination were collected from five depths between 5 and 200 m. First, 2 L of seawater samples were filtered through a 200 µm mesh to remove zooplankton and then filtered onto a 0.7 µm GF/F glass fiber membrane (Whatman; 450 °C for 4 h, 0.5 M HCl for 24 h). The filters were then stored at -80 °C until analysis in the shore-based laboratory. Seawater samples (20 L) for microbial community analysis were similarly pre-filtered through a 200 µm mesh and then filtered through a 0.22 µm mixed cellulose ester (MCE) membrane. The filtered samples were immediately transferred to cryovials, supplemented with RNA Later stabiliser (Qiagen, Germany), and stored at -80 °C.

rPreNO_x and rPrePO₄ were calculated using the modified formulation of Letscher and Villareal (2018), with parameter sensitivity evaluated following Smyth and Letscher (2023). For each observation, 9 calculations were performed using the prescribed DOM and POM parameter combinations. The arithmetic mean was used as the reported diagnostic value, whereas the minimum, maximum, and standard deviation describe sensitivity to the prescribed parameters. Full equations and parameter values are provided in Sect. S2.

Data processing was performed using OriginPro 2021 (v. 9.8.0.200). Depending on whether the variables follow a normal distribution, correlation analyses were performed using either Pearson's or Spearman's rho test (two-tailed, α=0.05), with p<0.05 considered significant. The normality of the data was tested using the Kolmogorov-Smirnov test (two-tailed, α=0.05). The AOU / nutrient ratios were estimated via simple linear regression using the ordinary least squares method, where AOU served as the dependent variable and the nutrient (nitrate or dissolved inorganic phosphate) as the independent variable. The slope of the resulting regression line represents the AOU per mole of nutrient regenerated (Álvarez-Salgado et al., 2014).

Temperature-salinity (T-S) diagram and the vertical distribution of temperature, salinity, and γn indicate pronounced stratification in the water column. Eight distinct water masses were identified and subsequently categorized into upper, intermediate, and deep waters based on their hydrographic properties and vertical positions (Fig. 2a, Table 1) (Sun et al., 2008). γn increases with depth from approximately 21.6 to 27.7 kg m⁻³ (Fig. 2d), reflecting the transition from low-density tropical upper waters to high-density deep waters.

In the upper waters, tropical surface waters and subsurface waters dominate. The North Pacific Tropical Surface Water (NPTSW) and Equatorial Surface Water (ESW) lie at the warmest end of the T-S diagram, with temperatures ranging from approximately 28–30 °C, characteristics of the WPWP. Below the mixed layer, the North Pacific Subsurface Water (NPSSW) and South Pacific Subsurface Water (SPSSW) exhibit pronounced salinity maxima along the 23–25 kg m⁻³ isopycnals. Both water masses have temperatures lower than the overlying surface waters (approximately 19–27 °C). The salinity of the SPSSW is the highest in the upper ocean (mean 35.51), while the NPSSW is slightly fresher (mean 34.80) (Fig. 2a, Table 1). The North Pacific Subtropical Mode Water (NPSTMW) constituted the lower part of the upper water, with stable physicochemical properties. The intermediate waters comprised the North Pacific Intermediate Water (NPIW) and South Pacific Intermediate Water (SPIW). The NPIW is characterized by low salinity (mean 34.12), exhibiting a salinity minimum in the T-S diagram (Fig. 2a). The SPIW has slightly higher temperature and salinity relative to NPIW. Its γn value largely overlaps with those of the NPIW, suggesting substantial mixing between the two within the study area. The deep water was represented by North Pacific Deep Water (NPDW), which had low temperatures, relatively uniform salinity, and high density (Fig. 2b–d, Table 1).

DIN, phosphate, and silicate concentrations were low in the upper waters and increased markedly with depth, reaching their maxima in the deep waters (Fig. 3a–c). The ensemble mean rPreNO_x and rPrePO₄ values were positive in the upper waters and generally increased with depth, reaching their highest values in the deep waters (Fig. 3d, e). Although the absolute magnitudes of the residuals varied among the 9 parameterizations, the main vertical pattern remained consistent across the parameter ensemble. The corresponding ranges and summary statistics are available in the public data set (Tian, 2026).

AOU and DIC showed broadly parallel vertical distributions, with low upper water values and marked increases into the intermediate and deep waters (Fig. 3f, h). In contrast, POC and TEP exhibit a distribution pattern that is nearly opposite to that of nutrients and DIC (Fig. 3g, h, i). TEP concentrations in the upper water ranged from 7.53 to 34.22 µg Xeq L⁻¹ (mean 22.52±8.54 µg Xeq L⁻¹), exceeding those in deeper waters. These concentrations were within the low range reported for open ocean environments (Table S1 in the Supplement). After conversion to carbon equivalents using a factor of 0.51 (Engel and Passow, 2001), the TEP-C / POC ratio ranged from 15.45 % to 71.75 %, with a mean of 43.81±14.62 %. The mean ratio was commonly higher than those observed in eutrophic oceanic regions (Table S1). POC concentrations decrease rapidly in the intermediate and deep waters, while the AOU continuously increased (Fig. 3f, g). In contrast, TEP concentrations decrease more slowly and remain within a measurable range even in the deep waters (Fig. 3i).

Among the five identified dominant genera, Pelagibacter, an oligotrophic bacterial genus, and Alteromonas, an opportunistic heterotrophic bacterial genus, had the highest relative abundances, but displayed different vertical patterns (Fig. 4). In the upper waters, Pelagibacter was the most abundant highlighted genus, accounting for 18.4 % of the community. Alteromonas represented a comparable proportion of 16.6 %. Nitrosopumilus, a chemoautotrophic ammonia-oxidizing archaeal genus, accounted for 4.6 %, whereas Halomonas and Alcanivorax were minor components, accounting for 1.3 % and 1.2 %, respectively. The intermediate waters showed a different community composition, the relative abundance of Pelagibacter decreased to 5.0 %, whereas that of Alteromonas remained relatively high at 16.2 %. The relative abundance of Nitrosopumilus increased to 6.8 %, and those of Halomonas and Alcanivorax increased to 4.2 % and 3.3 %, respectively. In the deep waters, Alteromonas increased to 22.9 % and became the most abundant identified genus, whereas Pelagibacter declined further to 3.0 %. Halomonas reached its highest relative abundance of 6.3 %, while Nitrosopumilus and Alcanivorax accounted for 5.0 % and 2.5 %, respectively.

The AOU of the ocean water column is an important indicator of the intensity of organic matter remineralization and oxygen consumption (Sulpis et al., 2023). In the upper waters of the study area, AOU is slightly negative and close to zero, reflecting photosynthetic oxygen production and bubble injection. From the upper to the intermediate waters, AOU increases rapidly, indicating intensified organic-matter remineralization dominated by heterotrophic respiration (Fig. 3f). In deep waters, AOU increases to a maximum of 227.80±8.69 µmolL-1 and then remains relatively stable, suggesting a dynamic balance between oxygen consumption and supply (Fig. 3f). In this study area, the AOU / DIN and AOU / DIP of the entire water column are 6.28 and 86.79, respectively, much lower than the ideal Redfield model (i.e., AOU / DIN = 8.6, AOU / DIP = 138) (Redfield, 1960) and also lower than the reported values from the Atlantic and Indian Oceans (AOU / DIN = 10.6, AOU / DIP = 170) (Anderson and Sarmiento, 1994). Although previous studies have shown that in strongly mixed ocean areas, enhanced water aeration can systematically suppress linear regression slopes (AOU / DIN = 3.1, AOU / DIP = 57.6) (Zhu et al., 2025), our study area is located in the WPWP, where strong stratification inhibits vertical mixing and exchange (Hu et al., 2015). Hence, physical “noise” is weak, and the low ratios observed here are more likely to originate from biogeochemical processes operating within the waters.

To examine how these biogeochemical processes vary across water column, we calculated AOU / nutrient regression slopes separately for the upper, intermediate, and deep waters (Fig. 5). The AOU / DIN value is lowest in the upper waters (6.87), increases in the intermediate waters (7.22), and becomes statistically insignificant in the deep waters (p>0.05) (Fig. 5a). The AOU / DIP slope is higher in the upper waters (116.04), decreasing in the intermediate waters (103.17), and non-significant in the deep waters (Fig. 5b). The low AOU / DIN and AOU / DIP slopes in the upper and intermediate waters indicate greater nutrient accumulation per unit increase in AOU than expected from the classical reference ratios. Limited vertical nutrient supply maintains persistently oligotrophic conditions in the upper water column (Chl a in the deep chlorophyll maximum: 0.19–0.31 µgL-1). Under nutrient limitation, phytoplankton may allocate a relatively large proportion of photosynthetically fixed carbon to extracellular polysaccharides and TEP with high C:N ratios rather than to nutrient-rich cellular biomass (Smyth and Letscher, 2023; Curran et al., 2025). The observed TEP-C / POC ratios in the upper waters (mean of 43.81±14.62 %) indicate that estimated TEP-associated carbon constituted a substantial component of the local POC pool (Fig. 3i, Table S1). Positive ensemble average rPreNO_x and rPrePO₄ residuals occurred together with elevated TEP concentrations in the upper water column (Fig. 3d, e, and i). This indicates that measured nutrient concentrations were higher than predicted from AOU under the prescribed DOM and POM remineralization parameters, resulting in a lower AOU / nutrient ratio in the upper waters. Although standing TEP concentrations do not directly quantify production, the substantial contribution of TEP-associated carbon to the local POC pool and its co-occurrence with low AOU / nutrient ratios support carbon allocation to TEP as one plausible contributor to the observed deviations. Future studies should examine whether similarly low AOU / nutrient ratios occur in other ocean regions with comparably high TEP-C / POC ratios to assess the spatial generality of this association. Furthermore, evidence indicates that heterotrophic microorganisms preferentially consume components with low C:N, such as proteins, in organic matter during aerobic respiration (Hwang et al., 2006; Tian et al., 2025). Therefore, in the intermediate waters, preferential degradation of relatively nutrient rich and labile organic components may enhance nutrient regeneration relative to bulk organic carbon oxidation, thereby contributing to the low AOU / nutrient slopes (Fig. 5a, b). In the deep waters, the absence of significant AOU / nutrients relationships precludes the calculation of meaningful remineralization ratios. In this depth horizon, AOU remains relatively stable whereas nutrient concentrations continue to increase, suggesting that their distributions are governed primarily by the cumulative effects of water-mass aging, lateral transport, and diverse biogeochemical transformations operating on different timescales. Among these, nitrification mediated by ammonia-oxidizing archaea such as Nitrosopumilus may consume oxygen without concomitant phosphorus regeneration (Martens-Habbena et al., 2009), while DOP hydrolysis associated with Alteromonas may mobilize phosphate independently of bulk organic matter oxidation (Saavedra et al., 2025). Collectively, these processes likely contribute to the observed decoupling between AOU and nutrients in the deep ocean.

Unlike DIN and DIP, SiO₃-Si and AOU do not exhibit a clear linear relationship in the entire water column, but instead show distinct vertical stratification. In the upper waters, intense silicate uptake by phytoplankton such as diatoms drives concentrations to near-zero values (Fig. 5c) (Brzezinski et al., 2024). In the intermediate waters, the variation in silicate concentration is mainly influenced by diatom deposition and remineralization. As AOU increases, silicates are rapidly released, resulting in an AOU / SiO₃-Si ratio (1.47) lower in these waters than in the upper waters (3.70) (Fig. 5c). In the deep waters, AOU tends to remain nearly constant, whereas SiO₃-Si concentration continued to increase owing to the delayed dissolution of siliceous biogenic debris (Yu et al., 2022). It is estimated that about two-thirds of biogenic silica dissolves at depths shallower than 2000 m, with the remainder released more slowly at greater depths (Tréguer and De La Rocha, 2013). Consequently, SiO₃-Si levels at depth are largely decoupled from organic matter supply and exhibit no significant linear relationship with AOU (Fig. 5c). A similar decoupling between biological consumption and accumulation is observed in the distribution of DIC, which exhibits a characteristic three-stage pattern from upper to deep waters (Fig. 5d). In upper waters, photosynthesis consumes DIC; however, this signal is masked by intense air-sea CO2 exchange, which leads to a net increase in DIC and consequently a low AOU / DIC ratio of 0.37 (Ito and Reinhard, 2025). In intermediate waters, organic matter decomposition dominates, and DIC increases rapidly with increasing AOU (AOU / DIC = 0.86, Fig. 5d) (Hu et al., 2016). In deep waters, AOU remains stable while DIC accelerates again, due to the accumulation of respired CO2 from above and enhanced calcium carbonate dissolution with depth (Chen et al., 2022; Feely et al., 2004).

Furthermore, DIN and DIP in each layers of the study area show a strong linear correlation, and the slope decreases progressively from 17.89 in the upper waters to 11.08 in the intermediate waters and further to 7.79 in the deep waters, with an overall slope of 14.01, which is lower than the Redfield ratio (Fig. 5e). The overall low N:P ratio (14.01) and its vertical decline reflect that phosphorus is regenerated more efficiently than nitrogen in this region. Denitrification, which removes fixed nitrogen, can lower the N:P ratio. However, classical water-column denitrification typically requires dissolved oxygen concentrations below 5 µmolL-1 (Deutsch et al., 2011). In the present study, the minimum observed oxygen concentration exceeds 55 µmolL-1, making extensive pelagic denitrification unlikely. Denitrification may nevertheless occur within particles suspended in oxygenated water when respiratory oxygen demand exceeds oxygen supply, particularly in large particles with high respiratory activity (Ploug, 2001; Ploug et al., 2008). The importance of this process depends on particle abundance and size distribution (Bianchi et al., 2018), and has been shown to increase with particle size and concentration in oxygenated coastal waters with high particle loads (Wan et al., 2023). In our study, POC and TEP concentrations were within the lower reported ranges (Table S1), indicating relatively low particle loads. Although these bulk measurements cannot exclude denitrification within individual particles, they suggest that this process was unlikely to dominate the observed DIN : DIP pattern. Alternatively, enhanced phosphate release from dissolved organic phosphorus (DOP) via alkaline phosphatase (APase) activity represents one potential mechanism. Although direct APase measurements were not made in this study, previous work has demonstrated that APase produced by Alteromonas can maintain high expression and abundance under phosphorus-limited conditions and even in deeper waters where inorganic phosphorus is not scarce (Saavedra et al., 2025). Given that Alteromonas accounts for a high proportion in all waters (Fig. 4), it is plausible that this genus contributes to DOP hydrolysis and thereby to the observed low N:P ratio.

The low AOU / DIN (6.28) and AOU / DIP (86.79) ratios observed in the TNWP result from the metabolic strategies of the resident microbial community. Within the study area, Pelagibacter and Alteromonas emerge as the two most abundant bacterial genera, their complementary metabolic capabilities directly modulate the coupling between organic carbon oxidation and nutrient regeneration, thereby driving the observed deviations from Redfield stoichiometry. Pelagibacter exhibits a significant positive correlation with organic matter and a negative correlation with inorganic nutrients (Fig. 6), consistent with its streamlined genome and oligotrophic adaptive strategy (Giovannoni, 2017; Carini et al., 2013). This bacterium lacks the ability to synthesize several essential metabolites and relies on exogenous supply, but its cell surface is enriched with high-affinity transport proteins that enable efficient uptake of small-molecule dissolved organic matter (Giovannoni, 2017). Critically, Pelagibacter selectively removes nitrogen-rich organic substrates such as amino acids and proteins, while leaving behind carbon-rich polymers (Malmstrom et al., 2005).

In contrast, Alteromonas displays the opposite environmental correlation pattern: showing a significantly positive correlation with inorganic nutrients and a negative correlation with POC (Fig. 6), reflecting its role as a particle-associated opportunist with a versatile genome encoding diverse polysaccharide-degrading enzymes (Koch et al., 2019; Lopez-Perez et al., 2012). This genus is a major contributor to alkaline phosphatase gene expression (particularly phoD and phoX) and enzymatic activity (Saavedra et al., 2025). Under phosphorus-limited conditions, these enzymes hydrolyze dissolved organic phosphorus (DOP) to release bioavailable phosphate (Saavedra et al., 2025). Alteromonas was abundant in all waters (Fig. 4), resulting in faster phosphorus regeneration than nitrogen regeneration in the study area, leading to an observed DIN : DIP ratio of approximately 14:1, lower than the Redfield value of 16:1 (Fig. 5e). Beyond these two dominant genera, the ammonia-oxidizing archaeon Nitrosopumilus, which is abundant in the intermediate waters (Fig. 4), can mediate high-affinity nitrification and consume oxygen even at extremely low ammonium concentrations (<10 nmol L⁻¹) (Martens-Habbena et al., 2009).

The ecological division of labor between these functional groups creates a vertically structured processing network: Pelagibacter sustains the microbial loop and small-molecule cycling in oligotrophic upper waters through high-affinity uptake, while Alteromonas dominates the degradation of settling TEP and other complex polymers, releasing inorganic nutrients in deeper waters (Robertson et al., 2024). This functional differentiation enables the preferential degradation of low C:N substrates, coupled with active phosphorus scavenging from organic pools, driving the system toward enhanced nutrient regeneration relative to organic carbon oxidation, thereby maintaining a dynamic equilibrium in its biogeochemical cycles and overall ecosystem function.

The strong vertical stratification of the TNWP creates a multi-water-mass structure. Nutrients are primarily retained in deep reservoirs and remain chronically depleted in upper waters. Nutrient regeneration in intermediate waters relies mainly on the release during organic matter degradation (Hirose and Kamiya, 2003). In this nutrient-poor environment, the synthesis of TEP with a high C:N ratio creates an imbalance between oxygen production and nitrate consumption (Fig. 3i), leading to positive anomalies in the rPrePO₄ and rPreNO_x in surface waters (Fig. 3d, e) (Curran et al., 2025). POM is gradually degraded during settling, while intermediate and deep waters are active areas for heterotrophic respiration. It is estimated that approximately 90 % of the organic matter exported to this area is remineralized and releases CO2 (Robinson et al., 2010). This process consumes a large amount of DO, and the strong stratification hinders the timely replenishment of DO in the intermediate and deep waters, thereby forming low oxygen zones in these waters (Fig. 7).

Simultaneously, nutrient release also mainly occurs in the intermediate and deep waters, exhibiting a clear regeneration gradient from shallow to deep (Fig. 3a–c) (Hirose and Kamiya, 2003). The distribution patterns and functional traits of key bacteria in the water column are closely linked to their distinct ecological niches. Pelagibacter is abundant in oligotrophic upper waters (Fig. 4), and its streamlined genome and metabolic mechanisms enable it to drive the remineralization of small-molecule organic carbon and nutrient recycling through slow but continuous growth (Sun et al., 2011). Alteromonas relies on a diverse enzyme system to efficiently respond to organic substrates, thereby utilizing complex polymers (Koch et al., 2019). This functional differentiation within the microbial community maintains the slow cycling of dissolved organic matter under oligotrophic conditions and ensures the efficient degradation of complex polymeric organic matter when available, thereby regulating both the pathways and the intensity of carbon, nitrogen, and phosphorus cycling in the water column.