Introduction

The constant elemental stoichiometry of phytoplankton (C:N:Si:P = 106:16:16:1) is known as the Redfield ratio. It is considered as a foundation of ocean biogeochemistry, and also reflects the utilization of dissolved nutrients by marine phytoplankton (Falkowski 2000; Redfield 1958). Growing evidence show, however, that dissolved nutrient ratios vary widely in different oceanographic settings (e.g., Deutsch and Weber 2012). Along with the availability of nitrogen (N) and phosphorus (P), as well as other nutrients, such as silicon (Si) and iron, the nutrient ratios largely determine the phytoplankton community structure in the ocean (Deutsch and Weber 2012).

Non-Redfield ratios have been observed in various oceanic settings: coastal upwelling systems (Bruland et al. 2005; Hutchins and Bruland 1998), mesoscale eddies (Li and Hansell 2008), micro-mesocosm and iron fertilization sites (DeBaar et al. 1997). The underlying mechanism for such variation, however, remains debatable and/or uncertain. Martiny et al. (2013) demonstrated a global pattern of N/P ratios in particulate form in the oceans, of which elevated ratios were observed in nutrient depleted subtropical gyres (e.g., particulate N/P = 37) and reduced ratios were found in nutrient enriched, high latitude regions (e.g., particulate N/P = 11). Such variable particulate stoichiometry is mainly driven by the variable uptake ratio and/or growth rate among different dominant phytoplankton groups for resource competition (Klausmeier et al. 2004). For example, picoplankton tends to dominate in oligotrophic regions and assimilate nutrients with a higher N/P ratio (relative to the Redfield ratio of 16), while large diatom cells are dominant in high nutrient regions with a lower N/P uptake ratio (Arrigo et al. 1999; DeVries 2018). On the other hand, high dissolved N/P ratios were observed in the subsurface of BATS station (Bermuda Atlantic Time-series Study, 31o40’N, 64o10’W) and the cause for such high N/P ratios was attributed to N addition from remineralization of N2-fixation (Deutsch and Webber 2012). Overall, nutrient stoichiometry and the composition of the phytoplankton assemblage show reciprocal causation in the illuminated ocean.

The ubiquitous existence of mesoscale eddies may further complicate the nutrient dynamics in the oligotrophic oceans (Benitez-Nelson and McGillicuddy 2008). In the subtropical ocean, new nutrients could be brought up into the sunlit zone by eddies to change the community composition and fuel the growth of phytoplankton (McGillicuddy 2016; **. Their distributions presented an irregular “circular cone-like” shape within the eddies. The eddy-influenced area of enhanced nutrient levels is usually smaller at the surface, but larger at the subsurface. Clearly, nutrient concentrations were elevated in eddy cores relative to their surrounding waters at the same depth (Fig. 2).

Fig. 2
figure 2

Sectional distributions of N + N, SRP, Si(OH)4, dissolved N:P and Si:N ratios in the upper 300 m in decay eddy (left panel) and mature eddy (right panel) (see Fig. 1 for the locations of the transects). Core and edge stations are denoted at the top. The isopycnal lines of σθ = 25 are also highlighted in black

Full size image

Nutrient depth profiles showed substantial vertical displacement across the eddy core section (Fig. 2). Nutrient-enriched subsurface water with concentrations of 13.3 μmol L−1 N + N, 0.86 μmol L−1 SRP, and 12.8 μmol L−1 Si(OH)4 in decay eddy rose to ~ 70–80 m along the specific density, σθ = 25 isopycnal surface from ~ 150 m (the depth where a similar nutrient level was observed outside the eddy). Such water displacement was even more pronounced in mature eddy, where nutrient-enriched subsurface water concentrations of 16.3 μmol L−1 N + N, 1.08 μmol L−1 SRP, and 16.4 μmol L−1 Si(OH)4 could be brought from ~ 160 m up to 50 m along the σθ = 25 isopycnal surface. The difference in upward water displacement between the two eddies also affected the nutrient inventory of 0–100 m depth interval (Table 1).

Table 1 Nutrient concentrations and ratios (mean ± standard deviation) in both dissolved and particulate phases, and biomass and phytoplankton community composition in three water types: eddy C2 (mature stage), eddy C1 (decay stage) and non-eddy water (SouthEast Asian Time-series Study (116oE, 18oN))
Full size table

For both eddies, the depth of the upper nutriclines coincided with the depth of SCM. The average of depth integrated nutrient concentration (0–100 m) in mature eddy was significantly elevated relative to non-eddy references (P < 0.05, t test). In detail, N + N, SRP and Si(OH)4 concentration was, respectively, 378%, 338% and 209% of those in non-eddy references. Similarly, the nutrient concentrations in decay eddy were higher than the references (P < 0.05, t test), but with a less degree as compared to mature eddy. The N + N, SRP and Si(OH)4 concentration was 290%, 269% and 183% of non-eddy references, respectively. Compared to N + N and SRP, the difference of Si(OH)4 concentrations between two eddies was minor, indicating a stronger uptake by diatoms in mature eddy. Nutrient concentrations at the depth of 150–300 m, on the other hand, showed similar levels among three water types. The ratios of N + N, SRP and Si(OH)4 in both mature and decay eddies were all close to 1 relative to non-eddy waters.

N/P and Si/N in dissolved and particulate form in two eddies

The dissolved N/P and Si/N nutrient ratios in three water types (i.e., non-eddy, mature eddy and decay eddy) showed similar vertical patterns with variable values, exhibiting a wider range in the upper 100 m but relatively stable below 100 m (Fig. 2). In the upper 100 m, both N/P < 16 and > 16 co-existed in three water types (Fig. 3a, e, i). The high N/P of > 16 was usually observed in the eddy center, where the upward water displacement was most significant, while low N/P was mostly seen at edges. Noted that ammonia was not included for N/P ratio calculation. Zhu et al. (2021) reported an ammonia concentration range of 10–20 nmol L−1 in the SCS basin, which was overall < 5% of total dissolved N, and was thus excluded in the calculation of N/P ratios.

Fig. 3
figure 3

Vertical profiles of N/P ratios (a, e, i), N/P ratios vs. N + N concentrations (b, f, j), bSiO2/PN (c, g, k) and the diatom pigment fucoxanthin (d, h, l), in three different water types: non-eddy water (blue color), mature eddy (green color) and decay eddy (red color). The lines of N/P = 16 and Si/N = 1 are also shown as black solid lines. The dashed black lines are the average lines for those N/P < 16 and N/P > 16. The profiles of fluorescence were also shown with N/P ratios and diatom pigment. Note that the fluorescence scale for mature eddy is set differently from decay eddy and non-eddy water. For c, g and k, the left and right whisker represent the 10th and 90th percentile of bSiO2/PN data from the same depth of all stations, respectively. The left and right boundaries of the box represent 25th and 75th data percentiles, respectively, with the central line representing the data median

Full size image

It is noteworthy that N/P ratios in the upper 100 m was ~ 1.4-fold higher in mature eddy relative to the other two water types (P < 0.05). The mean N/P ratio of 0–100 m for mature eddy was up to 19.1, which is higher than that observed in subsurface water (~ 14) and the Redfield ratio of 16. Likewise, the Si/N ratios in the upper 100 m for the three water types were also quite variable, being 0.9–3.1 (mature eddy), 0.8–18.3 (decay eddy) and 0.9–6.0 (non-eddy), comparing with values of 1.0–1.6, 1.1–1.6, 1.1–2.1 and 1.0–1.8, respectively, for 150 to 300 m depth interval. The mean Si/N ratio for upper 100 m in decay eddy (1.1 ± 0.55 as shown in Table 1) was lower than those (1.8–1.9) in the other two water types with 90% confidence interval (P = 0.07). Overall, most of those high N/P ratios were observed near the SCM layers regardless the water type and the mature stage hold the highest N/P when the N + N concentration was lower than 10 μmol L−1 (Fig. 3b). Similarly, high dissolved Si/N values appeared near the SCM relative to the subsurface water (150–300 m) regardless water type, while the mature stage held lower Si/N values in the upper water column among three water types.

As for the particulate phase, significantly higher bSiO2/PN appeared at the mature stage. For instance, the bSiO2/PN ratios in the upper 100 m was 2.8-fold and 3.5-fold higher compared to those in decay eddy and non-eddy reference, respectively (Table 1 and Fig. 3). The enhancement of bSiO2/PN in mature eddy is consistent with the growth of diatoms. The concentration of Fucoxanthin (average = 16.1 ± 22.4 ng L−1) in the upper 150 m of mature eddy was 1.5-fold and 3.0-fold higher than the values in the decay eddy (average = 10.6 ± 10.6 ng L−1) and non-eddy references (average = 5.4 ± 5.0 ng L−1).

Model-derived N/P and Si/N

In the model, the depth profiles of dissolved N/P for both mature and decay eddies have been simulated under different scenarios of the uptake ratio by diatoms. If the uptake ratio was set as 16 (Redfield ratio), the dissolved N/P in the upper 100 m would be less than 16 for both eddies (orange lines in Fig. 4). Once the uptake ratio was reduced to 10, the dissolved N/P presented elevated values of > 16 in the SCM layer, where most of the diatom resided (blue lines in Fig. 4). Such increase of N/P is more obvious in mature eddy compared to decay eddy. For example, N/P reached to ~ 35 in mature eddy, while it was < 25 in decay eddy. The depth profiles of dissolved Si/N are also presented in Fig. 4. Similar with N/P, Si/N also showed higher values (> 1) in the SCM layer when Si/N uptake ratio by diatoms is assumed to be 1–2. Possibly due to the faster Si(OH)4 uptake by diatoms, such Si/N elevation in the subsurface is less obvious in the mature eddy.

Fig. 4
figure 4

Model-derived vertical N/P ratios (a, b) and Si/N ratios (c, d) in decay eddy and mature eddy by assuming a diatom uptake N/P ratio of 16 (orange line) and 10 (blue line) and Si/N ratio of 2 (blue line) and 1 (green line), respectively. Here, the uptake N/P ratios for non-diatoms are arbitrarily set as 16. The green dots are the in situ N/P and Si/N ratios in both C1 and C2 cores; the black line shows the N/P and Si/N profiles controlled only by physical upwelling and mixing

Full size image

Discussion

In our case, the dissolved N/P ratios of subsurface nutrient source (below 150 m) were 14–15, lower than the classical Redfield ratio. Under the scenario of Redfield uptake, the N/P ratios of remaining nutrient in the euphotic zone should be less than the Redfield ratio. The consecutive changes in dissolved N/P ratio and its elevation during the mature stage of eddy in this study might thus be caused by several reasons: (1) preferential utilization of P relative to N; (2) N addition through N2-fixation and/or N deposition; (3) preferential P loss relative to N by particle sinking and/or active migration; and/or (4) stronger N than P supply from the remineralization of dissolved organic matter (DOM).

The phytoplankton community has been investigated for both eddies (Wang et al. 1). It can thus be concluded that N2-fixation may play only a minor role in explaining the N/P elevation during the eddy’s mature phase. More importantly, our results are consistent with the concept proposed by Mills and Arrigo (2010) that the enhanced subsurface diatom growth by eddy pum** may filter P preferentially to hamper the P supply, thus, the niche development of N2-fixers in surface water. On the other hand, the N deposition was equally important relative to N2-fixation in the SCS (Yang et al. 2014). In the present study, however, its contribution to the elevated N/P ratio could be excluded, since both decay and mature eddy were located in the same region receiving similar N deposition.

We noted that low N/P ratios (< 16:1) occurred above the SCM for both eddies, which could be a result of a high N/P uptake ratio by other phytoplankton taxa such as Prochlorococcus and Synechococcus (Singh et al. 2017) as well as preferential release of P relative to N by particle remineralization (Clark et al. 1998; Monteiro and Follows 2012). The preferential release of P could be the case particularly in decay eddy, where shallow remineralization was observed as indicated by excess 234Th (Zhou et al 2010). To examine the temporal evolution of nutrient dynamics, the ideal strategy is to observe on single eddy over its entire lifespan. However, it is extremely difficult for shipboard measurements that include nutrient concentrations and dynamics due to the need of temporally and spatially resolved sampling. This study happened to encounter two eddies at different evolution stages that could mimic the eddy evolution as an alternatively best approach. Moreover, the two eddies were formed in the similar region by the same physical forcing of the coastal jet separation and are believed to experience the similar temporal evolution (Zhou et al. 2020).

Fig. 5
figure 5

Conceptual model of nutrient dynamics within eddies at different developmental stages in the cyclonic eddies of western South China Sea modified from Zhou et al. (2020). The size of the circles denotes the relative nutrient concentration; nutrient ratios are shown in the circles

Full size image

However, the high dissolved N/P ratios had not been reported in the SCS basin and North Pacific Subtropical gyre in previous studies (Du et al. 2013; Wong et al. 2007; Deutsch and Weber 2012) indicating that such an eddy-induced high N/P ratio phenomenon was transient and was likely missed by routine field observations. Our results further emphasize that cautions should be made using subsurface nutrient ratios to derive N*(= N-16P) for N2-fixation estimate in oligotrophic oceans (Deutsch et al. 2001; Hansell et al. 2004; Singh et al. 2013; Wong et al. 2007).

Conclusions

Nutrient ratios were measured to be higher than the canonical Redfield ratio in a mature cyclonic eddy in the western SCS, while they were lower than Redfield ratio in another eddy under the decay stage. We found that such variability in the nutrient ratios within eddies is tightly linked to the temporal evolution in the abundance and nutrient uptake of large-cell diatoms, mediated by the physical dynamics of eddies. It implies that the regional nutrient addition by N2-fixation could be overestimated if such a high N/P ratio was used as a tracer for N2-fixation rate. The time series of nutrient ratios needs to be fully investigated in the eddies to understand the nutrient dynamics in the ocean.