Sediment trap experiments were conducted in two areas across the Kuroshio Current along 138°E ( Figure 1 ), designated the slope (34.0°N 138.0°E) and subtropical gyre (28.0–31.5°N 138°E), located on the offshore side of the Kuroshio Current. Experiments were conducted at 10 time points during seven cruises aboard the FRV Soyo-Maru (Japan Fisheries Research and Education Agency) from August 2013 to November 2019 as part of the O-line Ocean monitoring program at 138°E off Japan (Sugisaki et al., 2010) ( Table 2 ). Because the axis of the Kuroshio Current was at different locations on different cruises, the sampling site on the offshore side of the axis was placed depending on the location of the axis position ( Table 2 ). Experiments on the slope were conducted in January, March, August, and November, but no sample was collected in January in the subtropical gyre due to bad weather during the SY1601 and SY1801 cruises.

The sediment traps used in this study were Knauer type, surface-tethered drifting sediment traps (Knauer et al., 1979). The trap array included surface floats to dampen vertical displacement caused by waves and sediment trap cross-pieces holding eight transparent acrylic cylinders at two depths. The collection area and aspect ratio of the cylinder were approximately 0.0037 m² and 8.27 (length, 620 mm; width 75 mm), respectively, except for the SY1308 cruise in which wider cylinders were used (0.0048 m² and 6.18 [length, 618 mm; width 100 mm]). The volume of the collection cup was 295 and 548 mL for the narrow and wide cylinders, respectively. The shallower depth was at Z_eu, and the deeper depth was at 100 m below Z_eu. Z_eu was determined as the depth of 1% photosynthetically active radiation based on direct measurement using the PRR-2600 Profiling Reflectance Radiometer (Biospherical Instruments) during each deployment ( Table 2 ). In several cases, the upper sediment trap was deployed within the surface mixed layer ( Table 2 ). Sediment traps within the surface mixed layer have been reported to under- and over-collect sinking particles (White, 1990; Gardner, 2000; Owens et al., 2013). In those cases, the sinking flux of particles and T₁₀₀ value should be treated with caution.

Before deployment, each cylinder was filled with 0.2-μm-filtered seawater collected from the surface after adding of 5 g L^-1 pre-combusted NaCl (450°C, 3 h). The trap array was deployed for approximately 24 h beginning at 12 PM or 4 PM. After recovery, the sediment traps were left undisturbed onboard for 1–2 h to allow particles to settle to the bottom of the cylinder. The settling time was to allow most sinking particles to settle to the bottom and to minimize biological decomposition of sinking particles especially for small particles with high surface-to-volume ratio after recovery. Before removing the collection cup at the bottom of the cylinder, supernatant seawater in the cylinder was siphoned off. A portion of the slow-sinking-particles, which were resuspended above the collecting cup during the trap recovery, might be removed with the supernatant seawater resulting in underestimation of sinking flux. Sinking particles in the collection cup were collected on pre-combusted (3 h at 450°C) GF/F filters (Whatman). The GF/F filters were wrapped in aluminum foil and stored in a freezer (<-20°C) until analysis of POC and particulate organic nitrogen (PON) in a land-based facility.

Seawater samples were collected using 2.5- or 12- L Niskin bottles (GO-1010X; General Oceanics) mounted on a conductivity–temperature–depth carousel multi-sampling system (SBE 911plus, Sea-bird Scientific). Surface water (0 m) was collected using a plastic bucket. The depth of the surface mixed layer (MLD) was determined based on a density threshold value relative to the near-surface value at 10 m depth (Δσθ = 0.03 kg m⁻³) (de Boyer Montegut et al., 2004).

Samples collected for obtaining discrete vertical profiles of the concentrations of summed nitrate and nitrite (N+N), chlorophyll a (Chl a), POC, and PON were collected from the water column within the euphotic zone. Samples for determination of N+N were stored at< −20°C until measurement in a land-based facility. Suspended particles in 200 mL and 10 L seawater were collected on a GF/F filter for determination of the concentrations of Chl a and particulate organic matter, respectively. GF/F filters for determination of the Chl a concentration were immersed in N, N-dimethylformamide (Suzuki and Ishimaru, 1990) immediately after filtration and then stored at< −20°C in the dark until measurement on land. GF/F filters for POC and PON were treated using the same method as for the trap samples.

The particle volume and PSD were measured at depths of 10 and 50 m using an in situ laser diffraction sizer (LISST-100X, Sequoia Scientific), except on the SY1904 cruise. The LISST-100X measures light scattered by particles using 32 concentric, logarithmically spaced ring detectors. The volumetric particle concentration in each size class is estimated by inversion modeling based on Mie theory (Agrawal and Pottsmith, 2000). To minimize overestimation in the upper size bins driven by salinity fluctuation (Styles, 2006) and underestimation in the lower size bins due to low particulate concentrations or high ambient light conditions (Andrews et al., 2011), we used a size class range of 5.2–119 μm (20 LISST bins) (Yamada et al., 2015; Yamada et al., 2021). The LISST-100X was mounted on a winch wire with a protective frame, and the volume of particles at a given depth was measured for 10 min at a 1-s measurement interval (totaling 600 counts).

Total particle volume (TPV, ppm) was calculated as an integral across the whole size window (5.2–119 μm). Volumetric size distribution data were used to calculate the PSD in numerical terms, assuming that particles are spheres. Then, the numerical PSD was determined using the following power-law equation (Bader, 1970; Kostadinov et al., 2009; Buonassissi and Dierssen, 2010; Yamada et al., 2015; Yamada et al., 2021):

where N(D) is the number of particles per volume of seawater normalized to the width of the size bin (particles L^-1 μm^-1). D is the particle diameter (μm). D₀ is the reference diameter (here 1 μm). N₀ is the particle differential number concentration at D₀ (particles L^-1 μm^-1), and ξ is the power-law slope of the PSD. We conducted linear regression of log-transformed data for calculation of ξ and N₀ .

The concentration of N+N was determined using a flow injection analyzer (TrAAcs 2000, Bran+Luebbe) (details in Kodama et al., 2014). The detection limit for determination was 0.1 μM. The reproducibility of seawater measurements had a coefficient of variation of< 3% at an N+N concentration of 15.25 μM (n = 19). The nutrient concentrations were determined using commercially available nutrient standards: the CSK nutrient standard (Wako) and the Reference Material for Nutrients in Seawater (KANSO). Any concentration below the detection limit was set to 0.1 μM for statistical analysis.

The Chl a concentration was measured using a fluorometer (10-AU, Turner Designs) (Welschmeyer, 1994). For measurements of POC and PON, particles retained on GF/F filters were fumed with HCl mist for at least 24 h in a plastic container to remove CaCO₃. The total amounts of organic carbon and nitrogen were determined using an elemental analyzer (Elemental Analyzer-IRMS (FLASH 2000, Thermo Fisher Scientific).

Depth-integrated values of Chl a, POC, PON, and TPV were calculated from discrete vertical profiles using the trapezoidal integration method. The resolution of discrete vertical profile data differed among variables ( Tables S1–S3 ). Depth-weighted mean values of variables within the euphotic zone were calculated via division of the depth-integrated value by the thickness of the euphotic zone.

To examine correlations between variables, Spearman rank order correlation analysis was performed, as the normality assumption was not fulfilled for some variables. For the same reason, the Mann–Whitney rank sum test was used to compare outcomes between two groups. These statistical analyses and linear and non-linear regression analyses were performed using SigmaPlot 14.0 (Systat Software).

Table 2 summarizes sea surface temperature (SST), MLD, Z_eu, and the surface (0 m) concentration of N+N, while the depth profiles of salinity, temperature, and density (σθ) are shown in Figures S1, S2 . SST varied from 15.2°C to 29.0°C and from 19.0°C to 29.6°C in the slope and subtropical gyre areas, respectively, and SST tended to be higher in the subtropical gyre than on the slope during a given cruise. MLD increased with decreasing SST in both areas. N+N in the slope area became high during the mixing period, whereas N+N in the subtropical gyre was below the detection limit except in March 2017. Z_eu was less variable than MLD and showed no apparent coupling with variation in MLD, SST, or N+N. In three cases of trap deployment (01/21/2016, 03/08/2017 and 01/12/2018), the sediment trap at Z_eu was moored in the mixed layer ( Table 2 ).

The depth-integrated N+N within the euphotic zone (diN+N) varied from 100 to 490 mmol m^-2 and 8.7 to 73 mmol m^-2 in the slope and subtropical gyre areas, respectively ( Table 3 ). diN+N and depth-weighted mean N+N (dmN+N) in the slope area were significantly higher than in the subtropical gyre (Mann–Whitney rank sum test, p< 0.05). The diN+N variation in the slope area showed a negative correlation with SST (r_s = − 0.886, p< 0.05, n = 6), but there was no significant correlation between diN+N and SST in the subtropical gyre (p > 0.05, n = 4). Depth-integrated Chl a within the euphotic zone (diChl a) varied from 27.4 to 48.5 mg Chl a m^-2 and from 12.1 to 28.0 mg Chl a m^-2 in the slope and subtropical gyre areas, respectively ( Table 3 ). The variation in diChl a in each area showed no significant correlation (p > 0.05) with SST or N+N, whereas the variation in diChl a in the entire dataset showed a negative correlation with that in SST (r_s = −0.661, p< 0.05, n = 10) and a positive correlation with that in N+N (r_s = 0.806, p< 0.05, n = 10), suggesting that the biomass of phytoplankton along 138°E was regulated by the availability of nutrients transported from the cold deeper layer ( Table S5 ). diChl a was significantly higher on the slope than in the subtropical gyre (Mann–Whitney rank sum test, p< 0.05), and diChl a in the subtropical gyre showed greater variability than on the slope. The depth-weighted mean Chl a concentration within the euphotic zone (dmChl a) ranged from 0.403 to 0.647 μg Chl a L^-1 and from 0.165 to 0.387 μg Chl a L^-1 in the slope and subtropical gyre areas, respectively ( Table 3 ). The variability in dmChl a was similar in magnitude to that in diChl a, and dmChl a was significantly higher on the slope than in the subtropical gyre (Mann–Whitney rank sum test, p< 0.05).

Contrasting Chl a, no significant difference in POC and PON between the slope and subtropical gyre were observed from depth-integrated values (diPOC and diPON, respectively) or depth-weighted mean values (dmPOC and dmPON, respectively) in the euphotic zone (Mann–Whitney rank sum test, p > 0.05; Table 3 ). Moreover, no significant correlations of diChl a with diPOC or diPON were observed within each area or in the entire dataset ( Table S5 ). However, the mean ratios of POC/PON in the slope and subtropical gyre areas were comparable with the Redfield ratio (C:N = 106:16), suggesting that plankton were a major component of the sampled particulate organic matter ( Table 4 ).

PSDs in terms of number and volume within the euphotic zone show temporal variabilities in both the slope and subtropical gyre areas ( Figures 2 , 3 , S3 , S4 ). In both areas, the PSD in number fitted well to Eq. (1), with determination coefficient values of 0.95–1.00 ( Table 5 ). The slope (ξ) of PSD in number varied from −3.5 to −4.1 and from −3.5 and −4.8 in the slope and subtropical gyre areas, respectively, with no significant difference between two areas (Mann–Whitney rank sum test, p > 0.05). The variation in ξ is not reflected in the mean particle diameter based on the PSD in number (mPD_N), but rather in the mean particle diameter based on the volumetric PSD (mPD_V) ( Tables 5 , S5 ). mPD_V shows greater variability than mPD_N ( Table 5 ), ranging from 37.2 to 56.7 μm and from 23.0 to 58.0 μm in the slope and subtropical gyre areas, respectively. Similar to ξ, no significant difference in mPD_V between two areas was observed ( Table 5 ). Depth-integrated TPV (diTPV) varied from 8.75 to 30.1 10^-6 m³ m^-2 and from 7.16 to 21.2 10^-6 m³ m^-2 in the slope and subtropical gyre areas, respectively, showing greater variability than diChl a, diPOC and diPON ( Table 3 ).

The sinking flux of POC at Z_eu ranged from 51.9 to 235 mg m^-2 d^-1 and from 43.7 to 92.0 mg m^-2 d^-1 in the slope and subtropical gyre areas, respectively ( Figure 4 ), with no significant difference between areas ( Table 6 ). At Z_eu, the variation in sinking flux of POC in the entire dataset was closely related to the sinking flux of PON ( Figures 5A , S5 ; r_s = 0.648, p< 0.05, n = 10; Table S5 ) and showed increases during the cold period ( Figure 5B ; r_s = −0.624, p< 0.05, n = 10; Table S5 ) and with increasing diChl a ( Figure 5C ; r_s = 0.648, p< 0.05, n = 10; Table S5 ). However, the three correlations with the sinking flux of POC at Z_eu were not significant (p< 0.05) within each region, except for the correlation between the sinking flux of POC and that of PON in the slope area. The variations in sinking flux of POC at Z_eu and diN+N were not correlated within each area and in the entire dataset ( Table S5 ), despite a significant correlation between the sinking flux of PON at Z_eu and N+N in the entire dataset ( Table S5 ).

The fluctuation in the sinking flux of POC at a depth of 100 m below Z_eu was not similar to that of sinking flux of POC at Z_eu, with T₁₀₀ ranging from 0.307 to 0.646 and from 0.090 to 0.958 in the slope and subtropical gyre areas, respectively. Similar to the areal distribution of sinking flux of POC at Z_eu, the areal distribution of T₁₀₀ showed no significant difference between areas ( Table 6 ). Despite the significant correlation between the sinking flux of POC at Z_eu and SST, no significant correlation between T₁₀₀ and SST was observed within each area or in the entire dataset ( Figure 6A ). In the entire dataset, T₁₀₀ tended to decrease with increasing sinking flux of POC at Z_eu ( Figure 6B ; r_s = −0.661, p< 0.05, n = 10; Table S6 ), while it increased with increasing mPDV ( Figure 6B ; r_s = 0.700, p< 0.05, n = 9; Table S6 ) and ξ ( Figure 6C ; r_s = 0.695, p< 0.05, n = 9; Table S6 ). These three correlations with T₁₀₀ were not significant (p< 0.05) within each region.

The study sites on the slope and in the subtropical gyre were located on different sides of the eastward Kuroshio Current, and their physicochemical environments were affected by the current ( Figure 1 ). The mean surface geostrophic current from 1 August 2013 to 30 November 2019 indicated that Kuroshio water was transported laterally into the surface of the slope area, and that a cyclonic eddy was located on the downstream side of the Kuroshio Current from the sampling sites in the subtropical gyre (28°N–32°N, 133°E–137°E). Because the location of the axis of the Kuroshio Current is unstable ( Table 2 ), it is difficult to generalize the positional relation between sampling sites and the Kuroshio Current or eddies. The seasonal variation of Z_eu in the subtropical gyre is not obvious ( Table 2 ). However, the deeper Z_eu in the subtropical gyre than on the slope is consistent with the difference in dmChl a between the two areas ( Table 3 ). Z_eu in the slope area showed no apparent coupling with the variations in MLD, SST, or N+N at the surface ( Table 2 ), even though the molar ratio of nutrients in the study area indicates that N+N is the potential limiting factor of primary production rather than phosphate or silicic acid (Kodama et al., 2014). The lower variability in diChl a and dmChl a than dmN+N on the slope implies that grazing on phytoplankton or diapycnal mixing with deeper water suppressed the accumulation of diChl a and dmChl a, resulting in low variability in Z_eu (Kaneko et al., 2013; Nagai et al., 2019).

Differences in the areal distributions of particle properties between the two areas were not statistically significant, except for Chl a/TPV ( Table 4 ). Therefore, subsequent analyses of particle properties, sinking POC flux, and T₁₀₀ were conducted using the entire dataset, followed by the evaluation of trends within each area. No significant correlation of diTPV with diChl a, diPOC, or diPON was observed ( Table S5 ), implying that the chemical properties of particles (Chl a/TPV, POC/TPV, and PON/TPV) exhibited short-term fluctuations. Only Chl a/TPV showed a regional difference between the slope and subtropical gyre ( Table 4 ). The variations in Chl a/TPV had significant positive correlations with Chl a (r_s = 0.783, p< 0.05, n = 9; Table S6 ), indicating that the volumetric contribution of non-phytoplankton particles to total particles increases in an oligotrophic environment. Furthermore, POC/TPV was negatively correlated with dmTPV (r_s = −0.667, p< 0.05, n = 9; Table S6 ), whereas the mean ratios of POC/PON in the slope and subtropical gyre areas suggest that plankton were a major component of the particulate organic matter ( Table 4 ). These results imply that the change in the POC/TPV was the results in the difference in porous structures, which made a smaller contribution to the total organic matter of each particle than did planktonic organisms. This is consistent with the accumulation of larger particles detectable by LISST-100X in the presence of a high concentration of TEP, which has a porous, gel-like structure, in the subtropical northwest Pacific along the Kuroshio Current (Yamada et al., 2021). Therefore, porous gel-like materials may be a determinant of particle dynamics in subtropical environments.

In marine environments, the sinking flux of POC shows large regional and temporal variability (Buesseler and Boyd, 2009; Cavan et al., 2019). The sinking POC at Z_eu in the entire dataset showed temporal variability correlated with SST and diChl a ( Figures 5B, C ). However, seasonal variability in the subtropical gyre was modest ( Figure 5B ). The range of the sinking flux of POC at Z_eu in the subtropical gyre exceeds values in the subtropical North Pacific Ocean near Hawaii (station ALOHA) and is comparable with the range in the Gulf of Mexico ( Table S7 ), but the range of of the sinking flux of POC at Z_eu on the slope, where NPP is greater than in the subtropical gyre (Yokouchi et al., 2006), is comparable with those in mid-latitude areas (Iberian basin, Spanish coast, and Gulf of Mexico) and in high-latitude areas ( Table S7 ). It is worth comparing the sinking flux of POC and T₁₀₀ in the Kuroshio region to those in the Gulf of Mexico. In the Gulf of Mexico, mesoscale activities driven by the Loop Current and the eddies affects hydrography and biogeochemistry in the oligotrophic environments (Green et al., 2014; Stukel et al., 2021; Lee-Sánchez et al., 2022). Similarly, diverse routes of nutrient supply along the Kuroshio Current, including diapycnal turbulent mixing, wind- or tide-induced near-inertial internal waves, and double-diffusive convection, support higher primary productivity in the Kuroshio Current area relative to the subtropical gyre (Nagai et al., 2019).

T₁₀₀ in marine environments varies and may exceed 1 in areas where particle production or lateral transport of particles occurs below the euphotic zone ( Table S7 ), but it shows no marked regional trend. The range of T₁₀₀ in this study was within the range of previous studies shown in Table S7 , except for the lowest value (0.090). The areal averages of T₁₀₀ on the slope (0.437 ± 0.130) and in the subtropical gyre (0.504 ± 0.0380) are comparable with the range in the Gulf of Mexico (0.437–0.657; Table S7 ). However, no seasonal trend was observed ( Figure 6A ). In the following section, we test the hypothesis that particle size affects the attenuation of sinking POC by altering sinking speed in slope and subtropical gyre areas along the Kuroshio Current.

Using numerical models, Cram et al. (2018) and Omand et al. (2020) demonstrated that increasing the sinking speed of particles enhances the transfer efficiency of sinking organic matter to deeper layers by reducing residence time within the water column. The positive correlation between T₁₀₀ and mPD_V ( Figure 6B ) and the positive correlation between T₁₀₀ and ξ (i.e., lower ξ indicates a greater contribution of smaller particles) ( Figure 6C ) are consistent with the results obtained by Cram et al. (2018) and Omand et al. (2020). However, T₁₀₀ decreased during the period of high sinking flux of POC at Z_eu, suggesting that active biological decomposition enhanced vertical attenuation of sinking flux of POC ( Figure 6A ). Honda et al. (2016) reported close coupling between the attenuation of POC flux and zooplankton carbon demand. Active decomposition of zooplankton might alter the shape of the PSD via size-selective consumption of particles (Fuchs and Franks, 2010; Stamieszkin et al., 2017).

To conduct a more process-oriented comparison, we assumed that attenuation of POC flux can be expressed using first-order kinetics from the flux and time, t:

where F_z and F₀ are the sinking fluxes of POC at the depth of z and the reference depth z₀ , respectively (Suess, 1980; Armstrong et al., 2002; Lutz et al., 2002). k is the rate of decomposition including bacterial decomposition and zooplankton grazing. Time can be reformulated based on the sinking rate, w, and the distance between the reference depth and z:

k/w in Eq. (3) corresponds to z* in previous studies (Boyd and Trull, 2007; Marsay et al., 2015). For spherical particles, Stokes law predicts that

where D is the particle diameter, Δρ is the density difference between the particle and seawater; μ is seawater viscosity, and g is gravitational acceleration. For marine particles, Δρ tends to decrease with increasing D, as w ~ D ^γ with γ< 2 (Alldredge and Gotschalk, 1988). If z₀ is Z_eu, T₁₀₀ can be simplified to

where a is the constant defining the empirical power-law relationship between w and D. Thus, the logarithmic value of T₁₀₀ (−ln T₁₀₀ ) is proportional to k/aD^γ . Figure 7 shows the relationship between −ln T₁₀₀ and mPD_V . The exponent of the non-linear regression is −1.74 (95% confidence interval: −2.57 to −0.89) and −1.98 (95% confidence interval: −3.43 to −0.53) for the entire dataset and in cases in which the sediment trap at Z_eu was deployed below MLD, respectively. The exponent of the non-linear regression is consistent with γ< 2, suggesting that T₁₀₀ is controlled primarily by w in the study area.

Figures 6C and 7 show that T₁₀₀ tended to be greater in the subtropical gyre than in the slope area based on its relationship with mPD_V. Among particle properties (POC/PON, Chl a/TPV, POC/TPV, and PON/TPV) within the euphotic layer, only Chl a/TPV showed a significant difference between the two areas ( Table 4 ). Yamada et al. (2021) reported that the ratio of TEP concentration, which has a porous gel like structure, to Chl a concentration was elevated in oligotrophic areas along the Kuroshio Current. Low Chl a/TPV in the subtropical gyre may indicate a smaller contribution of phytoplankton to TPV and lower food quality relative to the slope. In addition, the species composition of planktonic copepods differs between the slope and the subtropical gyre along 138°E with large taxa abundant on the slope (Hirai et al., 2015 and Hirai et al., 2021). Large copepod species (Calanidae and Eucalanidae) were abundant in the copepod community of the slope water and may actively consume sinking particles in the area. These differences in both particle properties and species composition of copepods may affect the dynamics of T₁₀₀ .

Another factor that may affect attenuation of sinking POC flux is water temperature. Marsay et al. (2015) showed that the geographical pattern of flux attenuation is correlated with water temperature, suggesting that the temperature dependence of the biological consumption rate of sinking POC controls flux attenuation. In this study, no significant correlation was found between T₁₀₀ and mean temperature between Z_eu and 100 m below Z_eu, which varied from 14.2°C to 22.5°C ( Table 6 ). The Q₁₀ coefficient enables quantification of the temperature sensitivity of bacterial decomposition and zooplankton grazing (Ducklow et al., 2010; Ikeda, 2014). The rate of decomposition, k, at water temperature T can be expressed using Q₁₀ :

where k₀ is the rate of decomposition at the reference water temperature, T₀ . Typical Q₁₀ values of respiration are 2 and 1.9 for marine bacteria and meso- and microzooplankton, respectively (Ducklow et al., 2010; Ikeda, 2014). If the Q₁₀ temperature coefficient of k is assumed to be 2, k will increase at 1.78 times at the observed temperature increase from 14.2°C to 22.5°C. Thus, the observed range of −ln T₁₀₀ (0.0429–2.401) was greater than the expected variation in k with increasing temperature.

The LISST instrument observes particles only within a limited size range of 5.2 to 119 μm, which excludes large sinking particles such as marine snow (> 0.5 mm). Furthermore, particles detected using LISST within the euphotic zone could include suspended particles that do not sink. Thus, PSD measured by LISST is not an accurate representation of that of sinking particles. Cael and White (2020) reported that the PSD measured using LISST for particles collected in sediment traps showed a flatter shape than the PSD for suspended particles within the euphotic zone at station ALOHA, indicating that the mean size of sinking particles is greater than that of suspended particles in the euphotic zone. However, Puigcorbé et al. (2015) presented compiled data from the literature showing that flux attenuation tends to be greater for small particles (< 53 μm in diameter) than for large particles (> 53 μm in diameter) in low and mid-latitude areas where pico- and nano-phytoplankton are dominant. Along the Kuroshio Current, pico- and nano-phytoplankton account for a greater proportion than that of diatoms near the axis and on the offshore side of the axis (Endo and Suzuki, 2019). Furthermore, the particle dynamics of the observed size range may propagate to larger particles through aggregation and disaggregation processes (Li et al., 2004; Burd and Jackson, 2009). The relatively rapid attenuation of sinking POC in small particles (< 53 μm diameter) and dynamic changes in particle size may tighten the coupling of PSD measured using LISST with the variability in T₁₀₀ .

However, the mechanism that controls PSD in the slope and subtropical gyre region along the Kuroshio Current remains unclear. No significant correlations were found between either mPD_V or ξ and the tested variables ( Tables S5 , S6 ). Buonassissi and Dierssen (2010) reported that ξ determined using LISST varied from −4.1 to −2.9, exhibiting a gradual decrease from the shelf, across the Gulf Stream, and into the Sargasso Sea, and had a positive correlation with the Chl a concentration. Yamada et al. (2021) reported a similar range of ξ obtained using LISST (−4.2 to −3.2) along the Kuroshio Current but found a negative correlation between ξ and the Chl a concentration. Those authors also found that the PSD became flatter in the oligotrophic area, with a positive correlation between ξ and the TEP concentration (i.e., at higher TEP concentrations, larger particles make a greater contribution). In this study, ξ showed temporal variability with similar ranges on the onshore and offshore sides of the current along a N+N concentration gradient, although the correlations between Chl a/POC and Chl a concentration and between POC/TPV and TPV ( Table S6 ) suggest that the contribution of phytoplankton to particle volume is lower in the oligotrophic environment. Further research is needed to elucidate PSD dynamics, particle properties (density, stickiness, carbon content and biological availability), and the mechanisms regulating the sinking flux of materials into the deep ocean.