Introduction

Vertical mixing is key to supplying nutrients from deeper ocean regions to plankton communities1,2. Among the different nutrient supply mechanisms to the euphotic zone, tidal-driven flows over seamounts are noteworthy because of their enhanced vertical mixing3,4. Moreover, strong currents, such as the Kuroshio Current, also induce disturbances as they traverse topography, thus positively affecting nutrient supplies from greater depths to the surface5,6. However, the effects of topographic mixing on large-scale biological productivity enhancement remain poorly understood due to the sparsity of observations and the difficulty in linking small-scale mixing measurements with larger-scale productivity.

In this study, we introduce a case study to address the above issue by investigating the interaction between a strong current and local topography, revealing that this interplay fosters a vigorous nutrient supply within a confined area at the outlet of the Tsugaru Strait in the North Pacific, Japan (Fig. 1a). The strait features an eastward throughflow, commonly known as the Tsugaru Warm Current, which can reach speeds of 1 m s−1 or more at the fastest points7. The flow path of the current exhibits notable seasonal variations8,9,10,11 (Fig. 1b, c). In winter, the Tsugaru Warm Current flows southwards along the coast of the Shimokita Peninsula as the “coastal mode” (Fig. 1a, b), whereas in summer and autumn, the eastward Tsugaru Warm Current contributes to the formation of a clockwise mesoscale gyre known as the Tsugaru Gyre on the Pacific side8,9 (Fig. 1c–e). The seasonal variations in the horizontal flow path are accompanied by significant changes in the vertical outflow characteristics (Supplementary Fig. 1) over the topographic feature known as Shiriya Spur, which is near 41.4°N–41.7°N, 141.5°E–141.6°E, and extends northwards from Cape Shiriya (Fig. 1e). According to an ocean data assimilation model (called JCOPE-T DA; Downscaled and Assimilated version of the Japan Coastal Ocean Predictability Experiment, Tide enabled12), subsurface upwelling and surface divergence on the eastern side of the topography ( ~ 41.65°N, 141.60°E–141.65°E) can be markedly enhanced during the Tsugaru Gyre development season (Supplementary Fig. 1d and b, respectively). This localized upwelling and expected vertical nutrient supply can sustain surface primary production and the ecosystem of the downstream gyre.

Fig. 1: Observational site.
figure 1

a Location of the Tsugaru Strait. b, c Mean surface velocities and relative vorticity (2014–2021) for December and January–June (b) and July–November (c), respectively, derived from OSCAR surface velocities. d Monthly (2014–2021 average) time series of relative vorticity (normalized to an inertial frequency of 41°N) derived from OSCAR surface velocities in the regions indicated by the yellow rectangles in (b) and (c). The black line and shaded range represent the bootstrap mean and 95% confidence interval, respectively. e Distribution of the pixelwise arithmetic mean surface chlorophyll-a data for August and September from 2014 to 2022, divided by the root of its pixelwise temporal variance relative to the arithmetic mean, defined as an index of Ichl:\({I}_{{chl}}(x,y)={\left\langle {chl}\left(x,y\right)\right\rangle }_{t}/{\left[(\frac{1}{n-1})\sum {\{{chl}(x,y)-{\left\langle {chl}(x,y)\right\rangle }_{t}\}}^{2}\right]}^{1/2}\), where x, y, and n denote longitude, latitude, and the number of data at each pixel, respectively, and t denotes time. The vectors in (e) represent the mean current velocities for August and September from 2014 to 2021, sourced from OSCAR. The magenta points and squares represent stations where the turbulence measurements were deployed and where a carousel water sampler was used, respectively (26th August 2021; Supplementary Table 1). The grey contours denote bathymetry. The black rectangle denotes the area of estimated water column-integrated primary production (40.75°N–41.75°N, 141.75°E–142.75°E; refer to the main text for its estimation details). The onshore circular symbols and labels denote the stations utilized for estimating sunshine duration (refer to Methods for details; Supplementary Table 2). The upward triangles indicate the location of the antenna of the high-frequency radar system. f, The frequency distribution of chlorophyll-a for August and September from 2014 to 2022 is shown in the light green rectangle in (e). Source data are provided as a Source Data file.

Full size image

To quickly test this hypothesis, we normalized the long-term (2014–2022) seasonal (August and September) mean surface chlorophyll-a concentration (mg m−3) by its pixelwise standard deviation (mg m−3) and denoted the normalized data as Ichl (unitless). This index (Ichl) revealed elevated values (>2) within the gyre compared to those outside the gyre (Fig. 1e). Assuming that the chlorophyll-a concentration at each grid is normally distributed during the season, the elevated value of Ichl within the gyre supports the notion of enhanced chlorophyll-a levels with small temporal variation during that season (note that in general, caution should be exercised when applying this model to the distribution of chlorophyll-a, particularly in the case of short-term blooms, although we have assumed a normal distribution model in this study, e.g., Fig. 1f). The Ichl exhibited a horizontal maximum in the region corresponding to the area immediately east of the spur (approximately 41.7°N, 141.6°E–141.7°E; Fig. 1e). This finding suggests the existence of a nutrient source around the spur that supports stable chlorophyll-a levels, as shown by Ichl.

A study by Tanaka et al. 13. reported vertical turbulent nutrient flux over a sill located at approximately 140.35°E (off the Cape Tappi region) in the western part of the strait (Fig. 1e). The reported flux of 1 mmol N m−2 day−1 (below the subsurface chlorophyll-a maximum layer) is significantly greater than the values observed in the open ocean. While this nutrient supply off Cape Tappi is substantial, the sustained productivity in the Tsugaru Gyre suggests the presence of an additional nutrient source, given the distance from Cape Tappi and the spatial distribution of maximum Ichl in the Shiriya Spur (Fig. 1e).

The Tsugaru Warm Current, which passes over the Shiriya Spur, may induce disturbances that more directly impact biological production within the Tsugaru Gyre. However, the nutrient supply capacity at this topographic feature, the Shiriya Spur, has not been assessed, nor has there been a study linking turbulence intensity with large spatiotemporal variations in biological production (over several hours to days). We suggest that this steady supply of localized nutrients at the outlet supports sustained phytoplankton growth in the downstream environment.

To better understand the drivers of sustained phytoplankton productivity downstream of the Tsugaru Warm Current, we conducted drift turbulence measurements on 26th August 2021 following an advected water column in the eastern part of the strait, crossing the Shiriya Spur. In this area, continuous surface current velocity data from a high-frequency radar (HFR) system have been collected since 2014. Here we suggest that nutrient supply at the submesoscale (O(1–10) km) around the Shiriya Spur sustains biological productivity downstream at the mesoscale (O(100) km) gyre, through a combination of turbulence observations, HFR monitoring, and ocean data assimilation model (JCOPE-T DA) outputs.

Results

First, we briefly present the seasonal variability characteristics of the Tsugaru Warm Current and Tsugaru Gyre, and the associated topographically driven upwelling and surface velocity divergent field, on the east side of the Tsugaru Strait, using outputs from the JCOPE-T DA (Supplementary Fig. 1, with the 2019 output as an example10; see the Methods section). This analysis was motivated by the challenges in continuously and repeatedly acquiring the spatial structure, including the vertical velocity of the ocean interior. We then introduce results based on field observations using HFR and turbulence measurements. Finally, we discuss the implications for productivity in the Tsugaru Gyre using long-term chlorophyll-a data and empirical estimates of primary productivity.

Seasonal variability characteristics of the topographically relevant upwelling and surface velocity divergence of the Tsugaru Warm Current – implications of numerical modelling

The outputs from the JCOPE-T DA model reasonably reproduce the seasonal variations in the current and the formation of the Tsugaru Gyre8,9,10 (Supplementary Fig. 1a and b). For the vertical east–west section of each season at 41.65°N (Supplementary Fig. 1c and d), in the region where the eastward flow overtakes the Shiriya Spur near 141.5°E, remarkable upwelling (downwelling) is observed on the western side of (above) the summit at 141.4°E–141.5°E (141.5°E–141.6°E) for the January–June and December time averages (Supplementary Fig. 1c), when density stratification is weak. In July–November, when stratification develops, upwelling occurs at depths shallower than 100 m at approximately 141.60°E–141.65°E (Supplementary Fig. 1d), and divergence fields of surface velocities are observed at 41.6°N–41.7°N and 141.5°E–141.7°E (Supplementary Fig. 1b).

Therefore, the surface divergence field that occurs in the stratified season on the east side of the Shiriya Spur summit (approximately 41.6°N–41.7°N, 141.5°E–141.7°E) serves as an indicator of vertical upward flow and nutrient supply from the lower layers. During this season, vertical transport from the lower layers, including turbulent mixing, significantly affects the nutrient supply for phytoplankton in the surface layer, as surface nutrients have been depleted by previous spring blooms14.

Turbulence over the spur and associated diffusive nitrate flux

Then, we focused on the region where the Tsugaru Warm Current intersects the Shiriya Spur. The reason for this is that active mixing is suspected to have occurred at this location, and this mixing may have been accompanied by a vertical supply of nutrients that bolster surface biological production within the Tsugaru Gyre (Fig. 1e). Drift turbulence measurements were conducted near the axis of the Tsugaru Warm Current at a latitude of approximately 41.65°N, allowing us to collect data on changes in nutrient distribution and turbulence intensity as the water column was advected downstream from west to east (Fig. 1e). The vessel’s travel distance of ~14.8 km in approximately 3 h, which was almost entirely due east during the drift observations (Supplementary Table 1), corresponds to an advection velocity of ~1.37 m s−1, which is in good agreement with the actual characteristics of the Tsugaru Warm Current.

The drift turbulence observations from the Shiriya Spur (Fig. 1e) revealed highly elevated turbulent energy dissipation rates, which exceeded 10−6 W kg−1 near the surface and bottom layers from the western region to the top of the spur (west of 141.55°E) (Fig. 2a). Near the spur’s top (approximately 141.55°E), the isopycnal surfaces, such as 24.0 σθ, rapidly deepened (Fig. 2b), and turbulence intensities exceeding 10−8 W kg−1 were observed throughout the entire depth range, reaching 10−5 W kg−1 at certain locations near the bottom (Fig. 2a). The isopycnal surfaces, such as 24.0 σθ, undulated from the top to the dorsal surface on the east side of the summit. The layer with water temperatures between 12 °C and 14 °C (approximately 25.5–26.0 σθ) at approximately 141.6°E was thicker than that on the western side of the spur, with a rising internal surface area of 25.5 σθ, associated with dissipation rates of O(10−5) W kg−1 (Fig. 2a). The decrease and increase in the depth of the isopycnal surface from west to east of the summit (approximately 141.55°E) are analogous to those observed off Cape Tappi13 (Fig. 1e for the location). Although a similar phenomenon was observed for Cape Tappi, notably, the magnitude of mixing was significantly greater in the present study: the vertical diffusivity east of the summit ( ~ 141.60°E) was O(10−1) m2 s−1 (Fig. 2b), which is an order of magnitude greater than that reported for Cape Tappi13. This magnitude of vertical diffusivity was approximately 10,000 times or more significant than the background level in the open ocean15,16,17,18,19,20.

Fig. 2: Vertical transects and longitudinal distribution along the drift.
figure 2

a Turbulent dissipation rate, b vertical diffusivity, c vertical diffusive nitrate flux, d nitrate flux in the longitudinal direction, e nitrate concentration in the isopycnal coordinate system and, zonal velocity obtained from the acoustic Doppler current profiler mounted on the vessel (SADCP). The thin black and grey contours in (a) are the temperature and salinity, respectively. The thick black contours in (b) and (f) denote potential density anomalies. The thick grey and magenta contours in (c) indicate the nitrate concentration and depth of the nitracline (thick) within ±5 m (thin), respectively. The blue line in (f) indicates the layer Froude number, Frl, accompanied by 90% confidence intervals estimated via the bootstrap method (thin lines). The error bars of the flux in (d) indicate the 95% confidence intervals of the means calculated via the bootstrap method. The numbers in (d) denote the numbers corresponding to the literature numbers in the references. T/ST/E, A, AA, Sh/U, and K indicate the tropical/subtropical/equatorial, Arctic, Antarctic, shelf/upwelling, and Kuroshio regions, respectively (Supplementary Table 3). The dashed lines in (d) indicate that the range has been resummarised from the literature values. The circles (crosses) in (d) are the average (median) values with standard deviations (minima and maxima of the literature values). For literature number 24, the circle and bar indicate the mean and 90% confidence interval, respectively. The simple bars without markings in (d) indicate the range from minimum to maximum. Source data are provided as a Source Data file.

Full size image

The spatial distribution of nitrate exhibited significant variation between the western (i.e., upstream) and eastern (i.e., downstream) regions of the topography. Nitrate concentrations were low in the surface layer west of the summit of the Shiriya Spur ( < 141.55°E), with a nitricline observed at approximately 50 m (Fig. 2c). From the summit to the eastern flank of the sill (141.55°E–141.60°E), nitrate concentrations varied along the rippled isopycnal surface. East of 141.60°E, a more concentrated nitrate (~2 mmol N m−3) distribution extended to the near-surface layer (<20 m). This implies a significant nitrate supply to the surface from greater depths during the flow of water through the sill. The highest levels of vertical nitrate flux occurred immediately east of the summit (141.55°E–141.60°E), where the flux reached O(10–102) mmol N m−2 day−1 (Fig. 2c). This value was much greater than the values reported in other open ocean areas2,20,21,22. The magnitude was also approximately 100 times greater than that estimated for another characteristic topography near Cape Tappi in the western part of the strait (Fig. 1e shows the location; i.e., 1 mmol N m−2 day−1)13. This value is also comparable to those reported in areas close to land, such as the western boundary region, where strong currents and topography interact5,6,23,24,25,26,27,28,29,30,31. Fluxes at the nitricline (Fig. 2d) showed large variations but increased from west of the summit (141.55°E), ~0.75 mmol N m−2 day−1, to ~10 mmol N m−2 day−1 east of the summit as the mean. This result suggested that the nutrient content at the surface on the east side of the spur is greatly influenced by vertical nitrate fluxes. The east–west cross-sectional distribution of nitrate concentrations on isopycnal surfaces revealed a tendency for nitrate concentrations to increase within the 24.0–25.5 σθ range at approximately 141.55°E, near the topographic summit (Fig. 2e). A further increase in nitrate concentration was also observed at 141.60°E within a density range of less than 24.5 σθ. The vertical convergence of the fluxes at approximately 24.5–25.0 σθ at 141.58°E–141.60°E (Supplementary Fig. 2) suggests nitrate addition through turbulent vertical mixing on the east side of the summit. This outcome aligns with the process of nitrate addition from lower layers facilitated by turbulent vertical fluxes during the passage of the Tsugaru Warm Current over the topography.

The zonal (east–west) velocities observed along the drift line demonstrated a robust eastward current surpassing 1 m s−1 from 141.50°E to the region just above the sill (Fig. 2f). By employing the method outlined in a previous study13, we calculated the layer Froude number, Frl. The obtained value for Frl was greater than 1 at both 141.56°E and 141.58°E, where intense eastward flow and an undulating isopycnal surface were observed (Fig. 2f). This suggests the presence of an internal hydraulic jump. Specifically, when the background flow velocity approximates the propagation speed of the internal gravity wave, stagnation and the superposition of internal gravity waves near the critical point (Frl = 1) are expected. This condition leads to significant isopycnal displacement, resulting in heightened turbulence due to instability. The enhancement of turbulence by such hydraulic jumps has been investigated in various locations, including the Strait of Gibraltar32. The zonal flow divergence demonstrated a convergent field near the surface from just above the sill to just east (141.55°E–141.60°E), matching the structure of the rippling and deepening density surface (Supplementary Fig. 3). In contrast, the surface layer east of 141.65°E exhibited a divergent field (Supplementary Fig. 3).

Internal hydraulic jumps, upwelling and surface divergence as suggested by JCOPE-T DA data

To demonstrate the close relationship between instability in the ocean interior and the divergence of surface current velocities, we also used JCOPE-T DA outputs in August and September 2021 (Supplementary Fig. 4). We focused on the distributions of vertical flow, the layer Froud number, and the inverse Richardson number (Ri−1, an indicator of shear instability). In August and September 2021, the long-term mean flow axis location of the Tsugaru Warm Current, defined as the latitude where the eastward flow is maximal at each longitude, was distributed at approximately 41.60°N–41.65°N in the region 141.1°E–141.6°E (Supplementary Fig. 4a). The flow velocities showed eastward flow maxima exceeding 1 m s−1 at depths shallower than 50 m west of the Shiriya Spur crest, approximately 141.40°E–141.55°E (Supplementary Fig. 4b). The layer Froude number exceeded 1 at approximately 141.5°E (Supplementary Fig. 4c). Furthermore, the temporal mean of Ri−1 was greater than 1 and closer to its threshold of 4 in the region from 50–100 m and the bottom layer from approximately 141.5°E–141.6°E than in other regions, such as <141.40°E. This increase in Ri−1 indicates a potential field for shear instability there (Supplementary Fig. 4c). The upward flow developed at approximately 50–100 m at 141.6°E, indicating that the area shallower than 50 m has a diverging field of horizontal velocities at approximately 141.6°E (Supplementary Fig. 4d).

A significant positive correlation was observed between the daily averaged Frl in the region 141.45°E–141.60°E, corresponding to the Shiriya Spur, and the daily average Ri−1 in the region 141.55°E–141.65°E east of the summit (averaged at depths greater than 20 m to exclude surface stratification effects) (Supplementary Fig. 4e). A significant positive correlation was also shown between this mean Frl and the divergence field of surface flow velocities averaged over a day in the region 141.60°E–141.65°E (Supplementary Fig. 4f). These modelling results suggest that the enhanced surface divergence field east of the spur is related to the vertical flow and vertical mixing associated with internal hydraulic jumps. Therefore, we investigated the two-dimensional divergence field of the surface velocities obtained by the HFR system.

Intra-seasonal spatial and temporal changes in the surface velocity field around the Shiriya Spur

The seasonal variations in surface divergence observed by the HFR system (Fig. 3a, b) were consistent with the JCOPE-T DA outputs (Supplementary Fig. 1): during July–November, a remarkably divergent region of surface velocity appeared at approximately 41.65°N–41.70°N, 141.60°E–141.70°E (Fig. 3b). The locations of the divergence observed during drift observations at approximately 19:00 on 26th August 2021 (41.65°N–41.70°N, 141.60°E–141.70°E; Supplementary Fig. 5d) were generally consistent with those represented by the July–November observational means (Fig. 3b), although the subdaily variability in velocity was also significant (Supplementary Fig. 5) due to tides around the spur33. Therefore, the directly measured nutrient supply due to turbulent mixing in this study corresponds to the divergence of the surface layer observed from the HFR system. Given the limited opportunities for observing turbulence and the temporal resolution of chlorophyll-a data (1 day or more), we focused on another timescale to further support this idea on the basis of the observed data.

Fig. 3: Characteristics of directly observed surface divergence and impact of the Tsugaru Warm Current path on the divergence.
figure 3

Mean surface current velocity and divergence from December 2020 to June 2021 (a) and July–November 2021 (b), respectively, from the HFR system. The cyan rectangle in (b) represents the area for which the surface divergence (divES) was calculated in (c) and (d). The magenta points in (b) represent the stations where drift observations were made. The grey contours in (a) and (b) denote bathymetry. c Time series of the Tsugaru Warm Current axis latitude at 141.63°E and divES estimated by the HFR system (4-day running mean) for 2021. d Wavelet coherence of the axis latitude and divES for 2021. The solid line represents the 90% significance level. The magenta line represents the August–September period. Source data are provided as a Source Data file.

Full size image

Owing to the unique characteristics of the Tsugaru Strait, this current–divergence mechanism can change drastically on timescales ranging from a few days to approximately 10 days (i.e., subinertial frequency range10; Fig. 3c) as well as on a seasonal timescale. This is due to the half-open channel on the northern side (41.8°N, 141.20°E–141.50°E; e.g., Fig. 1e), the protruding spur being in the middle of the strait ( ~ 41.7°N at 200 m; e.g., Fig. 1e), and the disturbing excitation in the subinertial frequency range by wind and other factors10. In the 2021 example, a decrease in the surface velocity divergence east of the Shiriya Spur summit (41.65°N–41.70°N, 141.60°E–141.65°E; hereafter divES) was noticeable during July–November, coinciding with the northward shift of the current axis at 141.63°E (Fig. 3c). The wavelet coherence of the spatially averaged divES and the axis latitude at 141.63°E revealed that there was an increase in the signal in the subinertial frequency range around August and September (Fig. 3d). Readers can also grasp the correspondence between the north–south shift of the current axis and the intensity changes in the layer Froude number, subsurface upwelling, and surface divergence on the east face of the Shiriya Spur reproduced by the JCOPE-T DA output (Supplementary Fig. 6a–f).

On the basis of these findings, we contend that the divergence of the surface flow field on the east side to the sill observed by the HFR system might act as a proxy of the vertical nutrient supply to the surface layer via significant turbulent vertical mixing and the associated nutrient supply occurring east of the sill to a certain extent. We used a daily average of the surface velocity divergence field for the subsequent analysis, accounting for the shortest timescale in the GlobColour chlorophyll-a data34,35. Focusing on the daily mean divES for July–November 2021, we calculated the distribution of the ratio of the 1-day lag daily surface chlorophyll-a concentration when divES was positive to negative. The daily lag was assumed here to be the advection time, as discussed below. Consequently, the distribution of ratios greater than one was consistent with the anticyclonic Tsugaru Gyre distribution (Fig. 4). These results suggest that divES is closely linked to the chlorophyll-a concentration in the broader area. Then, we computed and mapped the correlation between the broader chlorophyll-a distribution east of the sill and the surface divergence east of the summit of the spur during August–September utilizing data from 2014–2020 (excluding 2015 due to the data defects for the HFR system). The results are presented in the next section.

Fig. 4: Impact of directly observed surface divergence on downstream chlorophyll-a.
figure 4

Ratios of the daily surface chlorophyll-a concentration from July to November 2021: the concentration a day after when the daily divES is positive to that in a day after when the daily divES is negative (divES: the surface velocity divergence observed by the HFR system east of the Shiriya Spur summit, 41.65°N–41.70°N, 141.60°E–141.65°E is denoted by the cyan rectangle). The grey contours denote bathymetry. Source data are provided as a Source Data file.

Full size image

Divergence east of the spur and chlorophyll-a variation in the summertime Tsugaru Gyre

The HFR system data revealed a significant positive 1-day lag correlation on the basis of the Pearson linear correlation coefficient between the daily and spatial averages of divES and the surface chlorophyll-a concentrations in the downstream area between August and September of 2014 and 2022 (except for the 2015 data; Fig. 5a), which overlaps well with the distribution area of the Tsugaru Gyre (indicated by the contours and vectors in Fig. 5a).

Fig. 5: Correlation maps between the surface chlorophyll-a concentration in the Tsugaru Gyre development season and the divergence of the surface velocity.
figure 5

The area for divergence estimation is delineated by the cyan rectangle, which is situated on the eastern (western) flank of the Shiriya Spur, ranging from 41.65°N–41.70°N, 141.60°E–141.65°E (41.65°N–41.70°N, 141.50°E–141.55°E) in (a) and (b) (c and d). a, c Cases with a 1-day lag, whereas b and d represent estimations with a 7-day lag. The grey dots indicate the grids in which the p value is greater than 0.05. The grey contours denote bathymetry. The black contours and vectors are the standardized mean relative vorticity by the inertial frequency at 41°N and the velocity at the surface estimated from OSCAR data between August and September for the years 2014–2021, respectively. Source data are provided as a Source Data file.

Full size image

To confirm the above lag-1-day correlation with respect to the timescale of advection, we performed offline particle tracking using surface velocities from JCOPE-T DA in late August 2021 (Fig. 6). This tracking experiment effectively reproduced a pronounced clockwise circulation at approximately 41°N–42°N, 141.5°E–143.0°E (Fig. 6). It was also suggested that the particles released from the eastern side of the Shiriya Spur generally travel approximately half a gyre in approximately one day. This indicates that intense surface flows, which are not detectable in long-term time averages (e.g., Fig. 1e), may be responsible for active horizontal transport. Therefore, the broad-scale positive correlation observed on lag-1 day in Fig. 5a is not incomprehensible or widely improbable, given the advection timescale. The 2021 case study also revealed that significant wavelet coherence between divES and chlorophyll-a dominated in the Area 1 and Area 2 regions, as shown in Fig. 6, at subinertial frequencies with small phase differences (Fig. 7a and b, respectively). This suggests that variations in divES intensity associated with the north–south shift of the flow axis impact the surface chlorophyll-a concentration in these areas (see also Fig. 4).

Fig. 6: Example of particle tracking providing advection timescale information.
figure 6

Example of offline particle tracking for 7 days from 30th August 2021 at 0:00 using JCOPE-T DA surface velocities. The colours represent each day. The black dots are the initial points. The grey contours represent the seafloor topography. The areas of the magenta rectangles correspond to the areas shown in Fig. 7. Source data are provided as a Source Data file.

Full size image
Fig. 7: Wavelet coherence of the surface divergence east of the Shiriya Spur and chlorophyll-a in the area along the advection pathway.
figure 7

The spatially averaged surface chlorophyll-a and divES wavelet coherence (2021) values in each area are shown in Fig. 6. a Area 1 (41.60°N–41.75°N, 141.75°E–142.25°E), b Area 2 (41.60°N–41.75°N, 142.25°E–142.75°E), c Area 3 (41.50°N–41.65°N, 142.75°E–143.25°E), d Area 4 (41.35°N–41.50°N, 143.25°E–143.75°E), e Area 5 (40.75°N–40.90°N, 144.00°E–144.50°E), and f Area 6 (40.85°N–41.00°N, 142.25°E–142.75°E). The black line indicates the 95% significance level. The arrows indicate the phase difference. The horizontal direction of the arrows indicates a phase difference of 0 or π, the vertical direction indicates π/2 or − π/2, and the right up and left down directions indicate that divES precedes surface chlorophyll-a. The magenta lines represent the August–September duration. Source data are provided as a Source Data file.

Full size image

The particle tracking example shows that surface circulation of the Tsugaru Gyre (41°N–42°N, 141°E–143°E) can complete a full cycle in approximately two or three days (Fig. 6). The fact that the positive correlation region becomes even larger at a lag of 3 days (Supplementary Fig. 7b) than at 0 lag (Supplementary Fig. 7a) and at a lag of 1 day (Fig. 5a) suggests the effect of this recirculation. At a lag of 5 days, the area of regions with significant positive correlations within the Tsugaru Gyre decreased (Supplementary Fig. 7c).

In the 7-day lag case, a notable (p < 0.05) positive correlation zone was identified at approximately 41.0°N–41.25°N and 142.5°E–143.0°E, which lies distant from the region where the divergence was calculated (Fig. 5b). This is partly an effect of the second round of recirculation. In addition, there may be a converging effect of phytoplankton advected by another larger circulation (39.5°N–41.5°N, 142°E–144°E), as expressed in Fig. 6. This larger anticyclonic gyre is likely caused by the influence of a mesoscale eddy, which also appeared in the long-term average field in August (Supplementary Fig. 8). The wavelet coherence analysis shown in Fig. 7 with divES and chlorophyll-a revealed subinertial frequency variability along this large circulation, with a tendency for the phase shift to increase along the path from Area 2 to Area 5, as shown in Fig. 6 (Fig. 7b–e). In the particle tracking example, the timescale for the advection of particles along this large circulation into the positively correlated area (41.0°N–41.25°N and 142.5°E–143.0°E, around Area 6) was approximately 7 days. Chlorophyll-a anomalies, excited by subinertial frequency fluctuations and advected along this large circulation, likely contributed to the strengthening of lag correlations of approximately 7 days in Area 6 (Fig. 7f; e.g., September).

In contrast, the divergence west (i.e., upstream) of the summit (41.65°N–41.70°N, 141.50°E–141.55°E) did not exhibit a significant positive correlation within the region where the gyre was expected (Fig. 5c and d). The results support the idea that the areas where mixing and nutrient additions influencing broad-scale production occur are concentrated in the eastern part of the Shiriya Spur. We also conducted a correlation analysis using high-passed time series data using a 3-day moving average to eliminate the recirculation resonance effect (Fig. 6). As the area deviated from approximately 41.60°N–41.70°N in latitude and moved away from the Shiriya Spur, the positive correlation area to the east tended to diminish (Supplementary Fig. 9). These findings suggest that, primarily, the nutrient supply area is geographically quite confined and can be characterized as submesoscale, likely dictated by the width of the Tsugaru Warm Current. The results of this long-term correlation analysis (2014–2022) also suggest that physical processes that significantly control phytoplankton production within the broad Tsugaru Gyre, such as the nutrient supply associated with upwelling (e.g., Supplementary Figs. 1d and 4d) and vertical mixing (Fig. 2a–d), are restricted to the eastward (leeward) side of the spur.

Discussion

The positive correlation of divES with chlorophyll-a within a small lag ( < 5 days) over the region 41°N–42°N, 141.5°E–143.5°E (Fig. 5a, and Supplementary Figs. 7a, b, and 8) suggests that nutrient supply into the surface layer via turbulent vertical mixing east of the Shiriya Spur can promote phytoplankton proliferation in the Tsugaru Gyre through horizontal advection and the recirculation of nutrient-enriched surface water (Fig. 6). The even longer positive lag correlations ( > 5 days) found in the southeastern region of the Tsugaru Gyre (39.75°N–41.25°N, 142.25°E–142.75°E; Fig. 5b) may be due to the second-round recirculation of the Tsugaru Gyre and/or phytoplankton advection from the larger clockwise circulation (39.5°N–41.5°N, 142.5°E–144.5°E; Fig. 6 and Supplementary Fig. 8). Nonetheless, the presence of significant correlations at such long lags can be useful for predicting productivity fluctuations and, ultimately, for fisheries forecasting in the area.

The present investigation revealed that intense turbulent vertical mixing occurred on the adjacent east side of the summit of the Shiriya Spur, when the throughflow surmounted the spur (Fig. 2a and b). This mixing can increase the nutrient supply from the lower layers to the surface, reaching O(10–102) mmol N m−2 day−1, which is at least 10 times greater than that in the open ocean (Fig. 2c and d). We determined the 95% confidence interval for the vertical turbulent nitrate flux into the euphotic layer, spanning 141.60°E–141.65°E, to be between 1.0 and 49 mmol N m−2 day−1, employing the bootstrap method, using the fluxes at the nitricline (Fig. 2d), whereas the vertical average of the flux for all layers observed ranged from 10 to 48 mmol N m−2 day−1. The magnitude of O(10) mmol N m−2 day−1 exceeded the nitrate uptake rate within the euphotic layer observed in August 2012 in the region south of the Tsugaru Gyre slightly downstream of the Tsugaru Warm Current (approximately 39.25°N, 142.5°E–143.0°E)36, which was 0.73 mmol N m−2 day−1. The depth of the euphotic layer was estimated to be approximately 60 m prior to the start of the drift observation (see the Methods section), so it is not unrealistic to assume that the nitrate flux at the euphotic later depth above the spur was on the order of O(10) mmol N m−2 day−1 (Fig. 2c).

Furthermore, the monthly mean primary production (PP) within the Tsugaru Gyre (40.75°N–41.75°N, 141.75°E–142.75°E, the mean area indicated in Fig. 1e as the black rectangle) was estimated in two ways. First, we utilized the chlorophyll-based empirical model (CBEM in this paper) previously suggested37,38 to link primary production and the surface chlorophyll-a concentration, expressed as log10(PP) = a + b log10(Chl). In this equation, PP and Chl represent the primary production and chlorophyll-a concentration at the sea surface, respectively, whereas the empirically given constants a and b have fixed values of 2.793 and 0.559, respectively37,38. The estimates based on the CBEM were approximately 36 and 34 mmol C m−2 day−1 for August and September, respectively. Moreover, according to the vertical generalized production model (VGPM39), which considers additional parameters, such as local sea surface water temperature, photosynthetically available radiation (PAR), euphotic layer depth, and photoperiod, the estimated values of PP ranged from 37 to 40 mmol C m−2 day −1 for August and September (Supplementary Fig. 10; refer to the Methods section for detailed calculation methods). The above PP estimates exceed the reported values for summer in the western North Pacific subtropical gyre (at 16°N–24°N in the central subtropical gyre at 155°E, the values are 18.2–22.8 mmol C m−2 day−1 40).

The spatial mean primary productivity for August and September in the region bounded by 40.75°N–41.75°N and 141.75°E–142.75°E (Supplementary Fig. 10), as estimated from the VGPM in the aforementioned findings corresponds to values of 5.6 and 6.0 and mmol N m−2 day−1, assuming a Redfield ratio41 (C:N = 106:16) and conversion to the nitrogen content. The nitrate fluxes on the eastern side of the Shiriya Spur, conservatively estimated at approximately O(1–10) mmol N m−2 day−1 (Fig. 2c and d), are comparable to the nitrogen uptake rate resulting from biological production derived from the aforementioned primary productions. Assuming that the area of nitrate supplied by turbulent fluxes (10–48 mmol N m−2 day−1) covers approximately 0.1° latitude ( ~ 11 km) and 0.1° longitude ( ~ 8.3 km), the nitrate vertical transport rate in this area is estimated to be approximately 9.2–44 × 105 mol N day − 1. Conversely, assuming that this water mass influences the production of a region covering 0.2° latitude ( ~ 22 km) and 1° longitude ( ~ 83 km) per day (e.g., Fig. 6) and that the productive capacity of this region is approximately 5.6 mmol N m − 2 day − 1, the regionally integrated basic production is approximately 1.0 × 107 mol N day − 1. The ratio of the former to the latter is therefore approximately 0.09–0.43. The maximum value (0.43) exceeds the f-ratio of 0.10–0.22 reported for the subtropical region (15–34°N, 155°E)36,40, suggesting that turbulent vertical mixing in a narrow area can have a significant quantitative impact on Tsugaru Gyre production.

More detailed investigations into spatiotemporal variations in nitrate fluxes over the eastern flank area of the spur, including its areal extent, and other nitrogen sources, such as regenerated production and nitrogen fixation, are warranted in the future. For example, a more pronounced divergent field at 41.65°N–41.79°N, 141.60°E–141.70°E was suggested, particularly from 15:00 to 16:00 on 26th August 2021 (Supplementary Fig. 5a and b), than at approximately 19:00 (the time of deployment of our microstructure profiler there; Supplementary Fig. 5d). This stronger surface velocity divergence may indicate increased vertical upwelling and enhanced vertical mixing than those observed in the present study (see Supplementary Fig. 4e and f for the impact of temporal variation).

The timescale for the ultimate consumption of the nutrients supplied at the east side of the Shiriya Spur should also be further investigated in relation to advection for the Tsugaru Warm Current. Nonetheless, the results of this study (e.g., Figs. 4, 5, and Supplementary Fig. 9) suggested that the submesoscale key area east of the summit may control the mesoscale chlorophyll-a distribution in the mesoscale Tsugaru Gyre. The intrinsic north–south displacement within the subinertial frequencies of the Tsugaru Warm Current10 presents an optimal setting for conducting a field experiment to explore the impact of vertical nutrient injection, particularly east of the Shiriya Spur, through situational changes (Figs. 3c and 4).

The intensity of the Tsugaru Warm Current has also been proposed to be linked to decadal-scale fluctuations in the North Pacific, such as the North Pacific Index42. Moreover, the Tsugaru Gyre is recognized as a productive fishing ground43,44,45. Therefore, ongoing surveillance of the pivotal area delineated by this investigation will also yield significant insights into long-term shifts in this isolated system, the Tsugaru Gyre, which encompasses both the ecosystem and fisheries. The notion of a submesoscale impact mechanism could extend to regions manifesting analogous characteristics, such as the circulation patterns observed in the Alboran Sea46.

Methods

Shipboard observations were conducted over the Shiriya Spur on 26th August 2021 with the research vessel (R/V) Wakataka-Maru, which is owned by the Japan Fisheries Research and Education Agency. The microstructure measurements were repeated 24 times via a free-fall turbulence ocean microstructure acquisition profiler (TurboMAP; JFE Advantech Co., Ltd.; Nishinomiya, Hyogo, Japan) that was loosely tethered during drift observation and equipped with two shear probes (Supplementary Table 1). The calculation method for the estimation of the dissipation rate, ε, was consistent with that used in a previous study conducted on the western Tsugaru Strait13, where shear probe data were sampled at 512 Hz and processed via standard procedures47. For each 8-metre segment of each profile, the turbulent kinetic energy dissipation rate, ε, was estimated by integrating the shear spectrum from 1 cycle per metre to the Kolmogorov wavenumber, employing the Nasmyth universal spectrum48. Segments with spectral shape deviations from the Nasmyth universal spectrum due to various factors, such as rapid changes in the fall rate, were excluded from the ε evaluation. The turbulent vertical eddy diffusivity, Kρ, was calculated as Kρ = Γ × ε/N2, assuming that the mixing coefficient of Γ was constant at 0.249. The buoyancy frequency, N, was derived from the temperature and conductivity sensors attached to the TurboMAP.

Simultaneously, with the turbulence observations, nitrate profiling was carried out via Deep SUNA v2 (Sea-Bird Scientific, Inc.) attached to TurboMAP. During the drift observations, surface water samples were taken at approximately every other point of TurboMAP deployment (Supplementary Table 1) for correction of offset values at the surface nutrient concentrations by the nitrate sensor. The profile offset was adjusted with reference to the surface water sample data. Unfortunately, SUNA’s batteries ran out during the drift observations, so data for deployments east of ~141.65°E could not be obtained (Supplementary Table 1). To estimate the vertical turbulent nitrate fluxes, we adopted an approach from a previous study13,50:

$${F}_{{{NO}}_{3}}=-{K}_{\rho }\frac{\partial {{NO}}_{3}}{\partial z}=\frac{\partial {{NO}}_{3}}{\partial \rho }\frac{\Gamma \varepsilon \rho }{g},$$
(1)

where g is the gravitational acceleration and where ρ is the potential density. In this study, we estimated the nitrate gradient on the density surface for each profile via a fitting function. To obtain the vertical distribution of the background nutrient distribution, eliminate contamination due to spike errors in SUNA and exclude fluctuations due to short-term disturbances, we utilized a multiple regression model with density as an explanatory variable, ranging from the first to the tenth order, and log-transformed nitrate concentration. The model with the smallest Akaike information criterion was chosen. From this fitting function for density and nitrate concentration, we estimated the nitrate gradient in the density plane (Supplementary Fig. 11). We also processed each vertical nitrate flux profile to address the near-surface extrapolation effect resulting from the aforementioned fitting process. Specifically, we ensured that the flux was zero at depths where the nitrate concentration was determined to be zero. The nitrate profiles for this terminal processing underwent offset correction through surface samples during the drift observations following the application of a 30-metre moving average due to spike errors in SUNA. Even with the moving average scale of 10 metres, there was no notable disparity observed in the flux magnitude estimates depicted in Fig. 2d. In the present study, the depth of the nitricline was defined as the depth at which the vertical gradient was greatest for each 30-metre filtered nitrate vertical profile. This depth corresponded relatively well with the 2 mmol N m − 3 nitrate concentration, except for 141.58°E–141.60°E immediately east of the summit of the Shiriya Spur (141.55°E) (Fig. 2c). Fluxes were averaged over a range of ±5 m at this depth and were defined as fluxes in the nitricline (Fig. 2d). Even at the 10-metre scale of the moving average for the nitrate profiles, there was no noticeable difference in the estimates of the magnitude of the fluxes shown in Fig. 2d.

Other vertical water sampling independent of the turbulence observations was also performed via a carousel water sampler (SBE 32; Sea-Bird Scientific, Inc.; Bellevue, Washington, USA) at two stations, i.e., at the start of the drift observation (41.60°N, 141.41°E) and at the end of the drift observation (41.69°N, 141.72°E). The samples were collected at 10, 20, 30, 50, 60, 80, 100, 125, 150, 200, and 300 dbar in addition to the surface. However, the starting point of the drift observation was located slightly further from the initial deployment point of water sampling (41.60°N, 141.41°E) because the latitude at which the drift observation started was fine-tuned to account for the position of the flow axis of the Tsugaru Warm Current (approximately 41.68°E). The end point (41.69°N, 141.72°E) was also far from the correction station because the SUNA battery ran out during the drift observations. Therefore, the nitrate distribution from this water sampler was used to visually confirm the consistency of the vertical structure of nitrate and was not used to correct the nitrate profile during drift (Supplementary Fig. 12).

We utilized velocity data obtained from an acoustic Doppler current profiler mounted on the R/V Wakataka-Maru (SADCP, Ocean Surveyor, Teledyne RD instruments, 38 kHz, Poway, California) in our study. The data were corrected for misalignment angle error due to the SADCP installation51, and data with a data flag value generally called “percent good” (indicating the fraction of available pings for a given ensemble) of 80% or lower were excluded. We used short-period average data (STA) at 1-minute intervals (including 6 pings per 1 ensemble) obtained while crossing the Shiriya Spur. The data were spatially averaged in the longitudinal direction to match the horizontal resolution of the HFR of 3 km and the grid. In the vertical direction, the first layer is 25 m long with a vertical resolution of 8 m and a maximum of 60 layers (497 m at the maximum depth).

The HFR monitoring system employed in this study was a SeaSonde (13.9 MHz, CODAR, Mountain View, USA). The HFR data provided a spatial resolution of ~3 km covering an area of ~3 to 60 km from each antenna (Fig. 1e)9,10,11. Approximately every 30 min, the system calculates the distribution of surface currents, which are uploaded to the Mutsu Institute for Oceanography (MIO) Ocean Radar data site for the eastern Tsugaru Strait (MORSETS; http://www.godac.jamstec.go.jp/morsets/e/top/). This HFR system has been operational in the eastern Tsugaru Strait since 2014. The basic processing procedure followed that of prior studies52. A raw Doppler spectrum was acquired at 2 Hz for 512 sweeps, approximately every 4.5 min. Four raw Doppler spectra were averaged (i.e., a 15-minute average) to estimate the Doppler spectrum for determining the radial velocity via the multiple signal classification (MUSIC) algorithm53, which was output every 10 min. Additionally, seven 15-minute averages were also utilized to estimate the final radial velocity. Therefore, each 30-minute output contained approximately 75 min of time information. We excluded grids with data acquisition rates below 90% in 2021 for Supplementary Fig. 5 and 95% for 2014 and 2016–2022 for the correlation analysis with reference to the previous method10. We excluded data for 2015 from the correlation analysis because of an observed data malfunction. Additionally, we verified the consistency of the HFR velocity with that obtained from the SADCP data along the drift trajectory (Supplementary Fig. 13).

We analysed the velocity fields for a region larger than the coverage of the HFR using surface velocity data from the Ocean Surface Current Analysis Real-time (OSCAR) dataset, which provides a third-degree resolution of ocean surface currents (Ver. 1. Physical Oceanography Distributed Active Archive Center, National Aeronautics and Space Administration, California, USA.). The data covering the period from 2014 to 2021 were accessed on 9th June 2022 at https://doi.org/10.5067/OSCAR-03D0154. To delineate the extent of the Tsugaru Gyre, the relative vorticity was calculated via the following definition: ∂V/∂x −∂U/∂y, where U and V represent the east‒west and north‒south currents of the OSCAR, respectively, and where x and y denote the east–west and north–south directions, respectively.

The surface distributions of chlorophyll-a were analysed via data from GlobColour (http://globcolour.info), which was developed, validated, and distributed by ACRI-ST, France55,56. We selected data products with 4-km resolution and 1-day resolution via the level-three Garver-Siegel-Maritoren (GSM01) merged model55,56 for correlation analysis (Fig. 5, Supplementary Figs. 7–9). The merged data encompassed information acquired from three satellite sensors: SeaWiFS (Sea-Viewing Wide Field-of-View Sensor) on GeoEye’s Orbview−2 mission, MODIS (Moderate Resolution Imaging Spectroradiometer) on NASA’s (National Aeronautics and Space Administration) Aqua mission, and MERIS (Medium Resolution Imaging Spectrometer) on ESA’s ENVISAT (Environmental Satellite) mission. The period of study extended from 2014 to 2022.

We used the output of the ocean data assimilation model (JCOPE-T DA12) to describe the physical processes linking the observed turbulence and the surface divergence field. The system provides hourly outputs of variables, such as temperature, salinity, and ocean current velocities, with a horizontal resolution of 1/36°. It incorporates tidal fluxes and sea level anomalies for 11 constituents at the lateral boundaries and surface pressure gradients because of the 25 potential tidal constituents. The atmospheric forcing data were distributed from the Global Forecast System of the National Centers for Environmental Prediction. Using multiscale three-dimensional variational data assimilation, the JCOPE-T DA system assimilates observational data, including satellite sea surface height anomalies, satellite sea surface temperatures, and in situ temperature–salinity profiles. Further details on the JCOPE-T DA configuration are provided in a previous study12. The correspondence between the model and observations of seasonal variations around the Tsugaru Strait was documented in our previous study10. The data selection year includes outputs from 2019, which are part of the results from the previous study, and 2021, when turbulence observations were made. The vertical layers were output every 5 m from 0 m to 350 m.

To obtain information on the spatiotemporal scale of advection of potentially eutrophicated surface water near the Shiriya Spur, we performed an offline particle tracking experiment using surface velocity data from JCOPE-T DA. Using hourly velocity data, we investigated the particle transport pathway from 00:00 on 30th August 2021 for 7 days as an example (Fig. 6). The discharge locations were three north–south grids centred at 41.65°N and 141.65°E. For each time step, data from the nearest grid to the expected transport location latitude and longitude were used. The effect of horizontal dispersion was added via random walk steps57, and 10 sequences were performed per emission grid. Here, the effect of horizontal diffusion is expressed by the following equation:

$$\left[\begin{array}{c}{l}_{x}\\ {l}_{y}\end{array}\right]=\sqrt{2{D}_{H}\varDelta t}\left[\begin{array}{c}{R}_{x}\\ {R}_{y}\end{array}\right],$$
(2)

where R is a random number ranging from − 1 to 1. The subscripts x and y represent the longitude and latitude, respectively. Here, Δt was set to 3600 sec. In the present study, a horizontal diffusivity coefficient, DH, of 102 m2 s − 1 was employed.

We computed the divergence field of surface velocity within the latitudinal band of 41.65°N–41.70°N, encompassing the latitude of the drift turbulence observations, within the 141.60°E–141.65°E area. This calculation utilized the 1-day average of the divergence field of surface current velocities (〈∂u/∂x + ∂v/∂y〉t, where u and v represent the east‒west and north‒south currents observed by the HFR, respectively, and the subscript t indicates the time average). We defined this divergence east of the summit of the Shiriya Spur as divES. Subsequently, employing 1-day resolution data of surface chlorophyll-a in the Tsugaru Gyre, we conducted a lag correlation analysis between the divergence and chlorophyll-a, utilizing data from August and September spanning the years 2014 to 2022, excluding 2015, due to an observed data malfunction. The Pearson linear correlation coefficients were examined.

Following the methodology presented in a previous study in the western Tsugaru Strait13, we determined the layer Froude number via the following formula:

$${{Fr}}_{l}=\frac{{U}_{l}}{{N}_{l}{h}_{l}},$$
(3)

where Ul, Nl, and hl represent the zonal velocity, vertically averaged buoyancy frequency, and thickness of the layer between the 24.0 σθ isopycnal surface and the bottom layer to which the free-fall measurement instrument (TurboMAP) descended (for N calculation), respectively58. We also calculate the layer Froude number via the JOCPE-T DA outputs.

Furthermore, we determine the gradient Richardson number (Ri) via JCOPE-T DA outputs via the following definition to examine the potential shear instability field:

$${Ri}=\frac{{N}^{2}}{{(\partial u/\partial z)}^{2}},$$
(4)

where ∂u/∂z is the vertical shear of the zonal velocity. Here, we assumed predominance of the eastward flow over the meridional velocity. We used the inverse of the Richardson number, Ri − 1, as an indicator of instability. In this case, the criterion suggesting shear instability is Ri − 1 = 4. However, we primarily used the time average of the logarithm, which understates the impact of instantaneous instability. We also calculated the divergence from the horizontal velocities of the JCOPE-T DA outputs.

To estimate primary production within the Tsugaru Gyre, we employed two models that use surface chlorophyll-a concentrations. First, we utilized the CBEM previously suggested37,38, as mentioned in the main text. The CBEM estimation equation is expressed as log10(PP) = a + b log10(Chl). In this equation, PP and Chl represent the primary production and chlorophyll-a concentration at the sea surface, respectively, whereas the empirically given constants a and b have fixed values of 2.793 and 0.559, respectively37,38. We also used VGPM39 to estimate primary production, accounting for the effects of the local sea surface water temperature and light environment on primary production. Here, vertically integrated primary production over the euphotic layer was calculated via the following equation:

$${PP}={P}_{{opt}}^{B}{Chl}\frac{0.66125{E}_{0}}{{E}_{0}+4.1}{Z}_{{eu}}{D}_{{irr}}[{{{\rm{mg\; C}}}}{{{{\rm{m}}}}}^{{-}2}{{{{\rm{day}}}}}^{{-}1}],$$
(5)

where \({P}_{{opt}}^{B}\) is the maximum carbon fixation rate within the water column and is calculated via the sea surface temperature, T, as follows:

$${P}_{{opt}}^{B}= -\!3.27\times {10}^{-8}\times {T}^{7}+3.4132\times {10}^{-6}\times {T}^{6}-1.348\times {10}^{-4}\times {T}^{5} \\ \quad+2.462\times {10}^{-3}\times {T}^{4} - 0.0205\times {T}^{3}+0.0617\times {T}^{2}+0.2749\times T\\ \quad+1.2956$$
(6)

The variables E0, Zeu, and Dirr are the sea surface daily photosynthetically available radiation (PAR), euphotic layer depth, and photoperiod, respectively. In the present study, the monthly average (August and September of 2014 to 2022) sea surface temperature and PAR were employed; the former was calculated from the ocean data assimilation reanalysis products (JCOPE2M59,60), and the latter was the result of Aqua-MODIS (Moderate-Resolution Imaging Spectroradiometer; level 3, 4 km) distributed from the NASA Ocean Biology Distributed Active Archive Center (OB.DAAC). The photoperiod was subsequently calculated from the monthly mean sunshine duration (defined as the period during which direct solar irradiance exceeds 120 W m − 2) for the onshore observation points surrounding the Tsugaru Gyre (Fig. 1e and Supplementary Table 2) provided by the Japan Meteorological Agency for 1991–2020. The depth of the euphotic layer was estimated to be 60 m from the observed PAR at 41.60°N, 141.41°E (the first station for the carousel water sampler, Sea-Bird Scientific), and information from a previous study36.

We utilized ETOPO161 for our topographic data.

Ethics and inclusion statement

The collaborators involved in this study meet the authorship criteria stipulated by the Nature Portfolio, as their contributions were integral to the study’s design and implementation. Prior to commencement, roles and responsibilities were delineated among all coauthors. This study is aimed at fostering the capacity and enabling the involvement of local researchers. It encompasses data pertaining to local attributes. Pertinent local and regional studies are referenced throughout the text. This research is not bound by severe restrictions or prohibitions on researcher access, nor does it entail any risk of stigmatization, incrimination, discrimination, or personal harm to participants.