Effects of prey of different nutrient quality on elemental nutrient budgets in Noctiluca scintillans

Noctiluca scintillans (Noctiluca) is a cosmopolitan red tide forming heterotrophic dinoflagellate. In this study, we investigated its ingestion, elemental growth yield and excretion when supplied with different quality food (nutrient-balanced, N-limited and P-limited). Total cellular elemental ratios of Noctiluca were nearly homeostatic, but the ratio of its intracellular NH4 + and PO4 3− was weakly regulated. Noctiluca thus seems able to differentially allocate N and P to organic and inorganic pools to maintain overall homeostasis, and it regulated its internal N more strongly and efficiently than P. The latter was substantiated by its comparatively stable C:N ratio and compensatory feeding on N-limited prey. Using both starvation experiments and mass balance models, it was found that excretion of C, N, and P by Noctiluca is highly affected by prey nutritional quality. However, based on modeling results, nutrients seem efficiently retained in actively feeding Noctiluca for reproduction rather than directly released as was shown experimentally in starved cells. Moreover, actively feeding Noctiluca tend to retain P and preferentially release N, highlighting its susceptible to P-limitation. Recycling of N and P by Noctiluca may supply substantial nutrients for phytoplankton growth, especially following bloom senescence.


Ingestion and growth yield of Noctiluca on different quality food. In incubations of 1 day of
Nocticula and prey of varying nutrient content, Noctiluca consumed nutrient-balanced and P-limited prey with similar, but significantly lower, rates than those of N-limited prey (as clearance or ingestion rates, Fig. 1a,b; ANOVA, Fisher LSD post hoc, p < 0.05). However, regardless of amount ingested, the abundance of Noctiluca did not change significantly after the 1-day incubation period (Table S1; Student's t-test, n = 6, p < 0.05). Neither did the cell diameters of Noctiluca change after feeding, and all had an averaged diameter around 232-240 μm in the different food treatments ( Table 2). The Noctiluca cells collected immediately after feeding contained an average of 11, 18 and 14 T. weissflogii cells per individual Noctiluca in the N-limited, nutrient-balanced and P-limited food treatments, respectively ( Table 2).
Making the assumption that over this 1 day incubation interval, prey items held in Noctiluca remained intact and undigested, their contribution to the elemental balance of C, N and P was small, <5% of the element contents attributed to Noctiluca itself. Nevertheless, the growth yield of Noctiluca in terms C, N, and P varied among food treatments (Fig. 1c). Those Noctiluca cells grown on nutrient-balanced T. weissflogii yielded significantly higher amounts of C, but slightly lower amounts of P (not significant) compared to those cells grown on N-limited food ( Table 3, Fig. 1c; ANOVA, Fisher LSD post hoc, p < 0.05), accounting for 19.43 ± 6.03% of the C content and 95 ± 47.7% of the P content from ingestion, respectively. In contrast, consumption of P-limited prey resulted in a significant loss of elemental contents in Noctiluca cells; about 511 pmol C, 159 pmol N, and 5.39 pmol P were lost per Noctiluca in 1 d (Table 3, Fig. 1c; Student's t-test, n = 6, p < 0.05). The gross growth efficiency based on carbon (GGE C ) of Noctiluca was much lower on nutrient-limited prey, and was even negative on P-limited prey (2.92 ± 2.69% and −30.91 ± 9.2% on N-limited and P-limited prey respectively, Fig. 1d).
Elemental composition and homeostatic regulation of Noctiluca in response to different quality food. Consumption of nutrient-balanced and N-limited prey for 1 day resulted in a slight loss of N content, but higher C and P contents in Noctiluca cells (Table 3). In contrast, ingestion on P-limited prey caused large reductions of all elements in Noctiluca compared to the initial starved condition, differing significantly from those fed on nutrient-balanced and N-limited prey. Generally, Noctiluca fed upon N-or P-limited prey significantly reduced both the amounts and fractions of the intracellular NH 4 + and PO 4 3− compared to those cells fed on nutrient-balanced prey and the initial starved cells (ANOVA, Fisher LSD post hoc, p < 0.05). Consumption of nutrient-balanced prey enhanced the amounts of NH 4 + and PO 4 3− in Noctiluca cells, and also the proportion -N f/2 -P C (pmol cell −1 ) 9.48 ± 0.41 b,c 7.58 ± 0.14 a,c 10.63 ± 0.57 a,b N (pmol cell −1 ) 0. 49 Table 1. Carbon (C), nitrogen (N) and phosphorus (P) contents and molar stoichiometric ratios (N:P, C:N and C:P ratios) of Thalassiosira weissflogii grown in N-limited (-N), nutrient-balanced (f/2) and P-limited (-P) medium. Data are averages with standard deviations. Superscript letters indicate significant differences between nutrient treatments (ANOVA, Fisher LSD post hoc, p < 0.05).
of NH 4 + in its total N pool (as the percentage of total cellular N). The proportion of PO 4 3− (as the percentage of total cellular P) was reduced after feeding, but still comprised the major part of P in Noctiluca, usually >60%, except for those cells grown under P limitation where it comprised ~40% of total cellular P. The molar NH 4 + to PO 4 3− ratio was highest in the Noctiluca cells grown on P-limited prey, while it was lowest in the cells grown on N-limited prey (Table 3).
In all cases, the elemental ratios of Noctiluca showed less variability than those of the algal prey upon which Noctiluca fed (Fig. 2). Based on the categories set by Persson et al. 30 defining conditons of homeostasis 30 , N:P, C:N and C:P ratios of Noctiluca were strictly homeostatic (Table 3). Nevertheless, on the basis of its C:N ratio, Noctiluca was less variable than was the case for its N:P or C:P ratios, as it showed no significant differences after different food treatments (Table 3, Fig. 2a-c; ANOVA, Fisher LSD post hoc, p < 0.05). In addition, stoichiometric coefficients indicated a weak internal stoichiometric regulation of the NH 4 + to PO 4 3− ratio in Noctiluca, with a coefficient value of 0.48 (Table 3, Fig. 2d).
After the 1-day incubation of Nocticula with variable prey, experimental treatments were subjected to a subsequent 6 h starvation period. During this time, the amounts of total cellular C, N and P and the intracellular NH 4 + and PO 4 3− nutrients in the Noctiluca cells previously reared on nutrient-balanced and N-limited prey generally decreased (except the cells previously fed on N-limited prey, of which the amount of NH 4 + increased after starvation), but those previously grown on P-limited prey showed no significant changes in their C and P pools (Table 3; Student's t-test, n = 6, p < 0.05). Both the proportion and molar ratios of NH 4 + and PO 4 3− in Noctiluca cells generally increased after 6 h starvation, except for those cells that had been previously grown on P-limited    Table 3. The amounts and ratios of total cellular elements and intracellular NH 4 + and PO 4 3− of Noctiluca scintillans cells at the initial starved condition, after 1 day incubation with N-limited (-N), nutrient-balanced (f/2) and P-limited (-P) Thalassiosira weissflogii, and after subsequent 6 h starvation of those different grown cells. Superscript letters indicate significant differences between food treatments (ANOVA, Fisher LSD post hoc, p < 0.05), parameters of the Noctiluca from the same food treatment show significant difference after 1 d incubation (before starvation) and after 6 h starvation are labeled in bold. prey, in which case both the total cellular elements and intracellular dissolved inorganic elements had decreased significantly after 1 day feeding (Student's t-test, n = 6, p < 0.05).

Excretion of Noctiluca from model prediction and starvation experiment. A simple budgetary
model was used herein to describe excretion rates of the biogenic elements (i.e. C, N and P) as a function of C metabolism, growth efficiency, and elemental ratios by assuming that ingested material must be assimilated by the organism before it is used for growth or metabolism 31 . Model simulations showed that excretion rates of C, N and P in Noctiluca were a function of their assimilation efficiencies, and were highly affected by food quality (Fig. 3). Generally, Noctiluca fed on nutrient-limited prey had a higher excretion rate of C and the elements that were in surplus for the growth of its prey. Note that Noctiluca cells grown on nutrient-balanced prey usually released less C and P than when grown on the two nutrient-limited prey conditions, and the cells grown on P-limited prey showed low but constant excretion of P regardless of the assimilation efficiencies. Assuming that the assimilation efficiency of C was 60% for Noctiluca in all food treatments, 80% on the unlimited nutrient, and 100% on the limited nutrient, which are the usual ranges for many feeding micro-and mesozooplankton 14,32,33 , Noctiluca grown on nutrient-balanced prey was estimated to excrete at a rate of 18.04 pmol C Noc −1 h −1 , 3.79 pmol N Noc −1 h −1 and 0.07 pmol P Noc −1 h −1 . Assuming these excretion rates were constant, the C, N and P contents excreted by Noctiluca would account for about 41%, 52% and 21% of those they ingested in 1 day. Elemental excretion rates of Noctiluca on nutrient-balanced prey were much lower than those nutrient-limited prey, especially P-limited prey, of which the elements released were even higher than those from ingestion (Table 4).
In contrast to these mass balance estimates, in the 6 h starvation experiment, Noctiluca previously reared on nutrient-balanced and N-limited prey generally had similar and higher excretion rates than those reared on P-limited prey, as well as those obtained from model predictions ( Table 4). The amount of C, N and P content excreted by the Noctiluca cells previously fed on nutrient-balanced prey during 6 h starvation experiment accounted for an average of 52%, 64% and 67% of those they ingested in 1 day, much higher than those derived from modeling predictions, as well as the amount of elements in food remnants (Tables 1 and 2). Moreover, no significant excretion of the C, N and P contents was observed for the Noctiluca cells previously fed on P-limited prey in the starvation experiment.

Discussion
Stoichiometry of Noctiluca. By examining the balances between the elemental ratios of Noctiluca and those of T. weissflogii, Noctiluca appears to be strictly stoichiometrically regulated. It has a higher degree of stoichiometric regulation than some other phagotrophic flagellates, e.g. Oxyrrhis marina and Gyrodinium dominans 6,34,35 . Such stoichiometric characteristic may provide Noctiluca cells with a competitive fitness advantage when resources are scarce or during periods of starvation 6 . Previous studies have shown that Noctiluca can survive without food for more than 3 weeks and that it requires a threshold of only 15 µg C L −1 T. weissflogii to maintain positive growth 36 . Nevertheless, the C:N ratio of Noctiluca, like other phagotrophic flagellates, is less variable than that of the N:P or C:P ratios 6,15,37 . In addition, Noctiluca appears to engage in compensatory feeding on N-limited prey (Fig. 1a,b), presumably in order to derive enough N to maintain stoichiometric balance, meeting the N requirement for basic metabolism 29 . Noctiluca thus displays stronger regulation of its N content than P content, i.e. relatively greater flexibility in regulating P accumulation. In other words, Noctiluca is more vulnerable to P limitation, which is in consistent with previous findings that dinoflagellates usually have high P demand due to their high DNA content in cell [38][39][40] . This result also indicates that a P limitation signal from the autotrophic phytoplankton would transfer to Noctiluca, but effects would be moderated.
The intracellular NH 4 + and PO 4 3− accounted for substantial amount of the N and P pools of Noctiluca (>32% and >40% respectively, Table 3), but Noctiluca was only able to perform a weak regulation of the molar ratio of dissolved NH 4 + and PO 4 3− over varying resource N:P ratios. It is still unclear what processes and mechanisms are involved in the accumulation of these dissolved inorganic contents in Noctiluca cells, but such processes are important in regulating cell osmosis and buoyancy 41,42 . A weak regulation of NH 4 + and PO 4 3− by Noctiluca indicates a flexible stoichiometry with respect to its unbound nutrients, as well as osmotic and/or buoyancy regulation, and also suggests that Noctiluca could differentially allocate its organic and inorganic N and P pools to maintain overall homeostasis, making the concept of homeostasis regulation even more complex than previously envisioned 2 . Clearly, additional data encompassing a better understanding of NH 4 + and PO 4 3− accumulation processes is necessary before a more complete evaluation of the stoichiometric regulation in Noctiluca can be made.
Furthermore, Noctiluca cells grown on nutrient-balanced or N-limited prey had a significant yield in terms of C and P contents ( Table 3, Fig. 1c), indicating accumulation of these two elements during feeding. This has implications for the cells also in terms of reproduction. Noctiluca seems to have high requirement of C and P for cell growth. In contrast, Noctiluca grown on P-limited prey not only resulted in considerable loss of cell elements ( Table 3, Fig. 1c), but even cell death 29 . It is known that P is the structural element in the RNA and DNA skeleton, and is also required in the ATP-ADP system that are all directly related to cell division and growth 43,44 . Therefore, inefficient housekeeping of cellular elements under stronger P deficiency might imply that P limitation disturbed the normal cell functions of Noctiluca metabolism and growth, making it vulnerable to direct P limitation.

Mode predation
Experimental excretion rates (6 h starvation)  Table 4. Computational and experimental excretion rates of C, N and P (R C , E N and E P , pmol Noc −1 h −1 ) and percentages of the excretion to ingestion (%) of Noctiluca scintillans grown on N-limited (-N), nutrientbalanced (f/2) and P-limited (-P) Thalassiosira weissflogii. Note: In the starvation experiment, percentages of the excretion to ingestion (%) of Noctiluca were calculated by dividing the elemental excretion in 6 h to 1-day ingestion and assuming that Noctiluca's assimilation efficiency was 60% for C content in all food treatments, 80% for the excessive nutrient and 100% for the limited nutrient. Dashes mean no significance difference of the element content of Noctiluca before and after starvation.
ScienTific RepoRts | 7: 7622 | DOI:10.1038/s41598-017-05991-w Excretion of Noctiluca from model prediction and starvation experiment. The measurement of elemental excretion by a feeding zooplankton in natural conditions is a technically difficult task 45,46 . The primary difficulties come first, from the fact that co-existing phytoplankton can rapidly remove released nutrients from the dissolved pool 46,47 . Second, Noctiluca, as is the case with many flagellates, is fragile and easily damaged during manipulation. The few attempts to study nutrient excretion rates of Noctiluca have focused on the inorganic nutrients NH 4 + and PO 4 3− , and rates were measured without food 23,48 . However, this sort of measurement may only determine the basal metabolic release of a starving grazer. Such measurements are also not necessarily representative of the actual excretion rate(s) from actively feeding organisms, as nutrients may also be released from the metabolic processes that are associated with feeding itself besides basal metabolism, or nutrients may be more efficiently retained in cells for actively feeding organism for reproduction rather than directly released 2, 45 . Moreover, the elements released by Noctiluca after metabolism are not only in oxidized (CO 2 , PO 4 3− ) or reduced (NH 4 + ) inorganic forms; other possible dissolved organic forms, e.g. free amino acids 49 , phosphonates and P-esters (DOP) 50 and carbohydrates or dissolved organic C 51 , associated with food vacuole egestion and mucus formation, may also contributed substantially to C, N and P release by Noctiluca 26 . As it is difficult to directly measure all these nutrient components, an elemental budget model can help to overcome these difficulties 2, 31 and provide a general insight to the magnitude of these fluxes.
The most important implication of the elemental budget model is that the excretion rate of an element in Noctiluca, analogous to other zooplankton, depends largely on its physiological conditions, as well as the chemical nature of its prey 11,31 . Noctiluca has nearly Redfield N:P and C:P ratios, and lower ratios than those of T. wessfogii under nutrient-balanced condition (Tables 1 and 3). In the framework of stoichiometry, the grazer tends to retain P and preferentially recycle N, a phenomenon that was shown in both the growth yield in the 1-day incubation and in the elemental excretion in the starvation experiment in this study. In addition, the released elements, in particular C and N, usually accounted for a significant fraction of the total metabolic budget of Noctiluca (41-91% and 52-107% for C and N, respectively), indicating that the cost of the growth of an actively feeding Noctiluca, in terms of C and N, is considerable. Obviously, even though applied here in the budget model, it is imprecise to assume an assimilation efficiency (AE) for C of 60%, and of the non-limiting nutrient of 80% and of the limiting nutrient of 100%, even though these are the typical values for many feeding micro-and mesozooplankton 14, 32, 33 . Herbivores, for example, may adjust their AE of each element, by decreasing the efficiency with which they assimilate C-rich compounds during digestion 52,53 . Therefore, it is important to consider the different assimilation efficiencies of each element in Noctiluca with respect to its nutritional status in future energetic studies.
In the starvation experiment, the Noctiluca cells that previously fed on nutrient-balanced and N-limited prey exhibited similar excretion rates of C, N and P, rates substantially higher than the model predictions (Table 4). A mismatch of the excretion rates obtained from the model prediction and the starvation experiment suggest that growth (from model prediction) and maintenance (from starvation experiment) of Noctiluca may have a different set of elemental demands, with maintenance seeming to have higher requirements for all biogenic elements 2, 34 . Results of the starvation experiment also further suggest that P limitation might have disturbed Noctiluca basal metabolism, as there was no elemental excretion by the Noctiluca cells previously grown on P-limited prey.
Both the computational and experimental excretion rates of N and P reported herein are in the ranges of those reported in previous studies when NH 4 + and PO 4 3− excretion rates were determined before and after starvation (Supplementary Table S3). For example, Drits et al. 48 showed that the averaged excretion rates of Noctiluca in a day were 0.74 ± 0.04 pmol N cell −1 h −1 and 0.34 ± 0.02 pmol P cell −1 h −1 . Ara et al. 23 found that the excretion rates of Noctiluca decreased rapidly with time, but the highest rates found in the first 1 h were 243 pmol N cell −1 h −1 and 24 pmol P cell −1 h −1 , which they thought were less influenced by starvation and closer to the actual excretion rate. Variation in the excretion rates obtained in the present and previous studies may be due to the differences in the cell size and physiological condition of the Noctiluca cells tested (Table 2, Supplementary Table S2). Besides, higher rates for the specimens in a short period may be also because of the increased activity during manipulation, while lower rates may result from prolonged starvation 23 . Furthermore, our study reveals that the amount of C, N or P reduced in Noctiluca cells after 6 h starvation was much higher than those retained in the food remnant, and the proportion of NH 4 + and PO 4 3− in Noctiluca' N and P pools generally increased with starvation. Therefore, the elements excreted in the starvation experiment were not only ascribed to the digestion of food remains, but also to the metabolism of its cellular organic matter, and the intracellular NH 4 + and PO 4 3− nutrients were possibly metabolic products 54 . Measurement of elemental excretion rates by determining the nutrients excreted in the experimental bottle, and culturing the grazer in the starved condition, clearly cannot reflect the actual excretion rates of an actively feeding organism 45 .
Roles of Noctiluca in nutrient recycling. Based on our model predictions, N and P excretion rates of actively feeding Noctiluca are lower than those reported for micro-and mesozooplankton with the same C biomass or dry weight, assuming C weight to be 43.9% of dry weight for Noctiluca (Table 4, Supplementary  Table S2) 55,56 . Therefore, the more significant role for Noctiluca as a nutrient regenerator and supplier ascribes to its extremely high concentration of NH 4 + and PO 4 3− in its cell, which herein accounted for 32-63% and 40-68% of its total N and P contents, depending on its physiological condition. These dissolved inorganic nutrients have been shown to contribute considerably to the N and P pools in natural assemblages [24][25][26]57 . For example, Pithakpol et al. 58 showed that in Seto Inland Sea, Japan, NH 4 + and PO 4 3− contained in Noctiluca cells contributed up to 119% of the N pool and 80% of the P in the water column (0-35 m depth). This is especially true when Noctiluca blooms are formed, as the cells at this stage stop feeding as they go into stationary growth, becoming starved and/ or nutrient limited, and their mortality rate increases. Liberation of NH 4 + and PO 4 3− and other nutrients of these cells would stimulate the growth of phytoplankton species living near the red tide patches and improve the food quality for Noctiluca again 19 , Noctiluca also contains high amounts of organic substances in cell 26,27 . Decaying Noctiluca cells, therefore, would contribute organic matter for bacteriovorous protozooplankters and could supply food to actively feeding Noctiluca 26,27 . In addition, Noctiluca excretes mucus to trap food items, and the decomposition of this organic matter by heterotrophic marine bacteria would fuel the microbial loop and result in an increase of recycled nutrients 26,59 . Therefore, the lysis of Noctiluca cells and subsequent mineralization activity could be a significant source of inorganic and organic nutrients fueling further phytoplankton production, prolonging the bloom duration and existence of Noctiluca 19,20,26,58 . For example, Schaumann et al. 26 found that the marked release of nutrients, especially NH 4 + and PO 4 3− by Noctiluca contributed to autochthonous eutrophication in the German Bight, initiated by diatoms bloom, e.g. Rhizosolenia sbrubsolei, R. setigera and Guinardia flaccida that are then fed upon by Noctiluca. Harrison et al. 19 conceptualized Noctiluca as an offshore manifestation of eutrophication since it feeds on the phytoplankton bloom caused by anthropogenic nutrients.
The interaction between Noctiluca and phytoplankton is not a simple mutually supportive relationship 20,23,26,58 . Rather, Noctiluca accumulates and releases each element with different efficiencies according to its physiological status (the present study), which would intensify the effect of nutritional imbalances in primary producers and then strengthen the trophic feedback to Noctiluca. As stated above, Noctiluca weakly regulates its intracellular NH 4 + to PO 4 3− ratio, and it is an efficient recycler of N, but recycles P with a low efficiency, thus it would usually suffer P limitation. This would intensify P-limitation from phytoplankton to Noctiluca, causing a negative impact on it. Our findings may provide an important clue for the population dynamics of Noctiluca in the field. For instance, the occurrence of Noctiluca in Hong Kong waters has a strong seasonality, with high abundance in winter-spring, while almost no occurrence in summer-fall 60 . The Hong Kong coastal waters, especially the eastern waters, could potentially experience N-limitation in winter and spring 61 . A possible explanation for its high abundance and long residence, besides the suitable temperature during winter-spring, is that Noctiluca is less susceptible to N-limitation and able to efficiently recycle N. In contrast, in summer and fall, the diatom-dominated phytoplankton assemblages in Hong Kong waters become proportionately more P limited 60,62,63 , and the reduction of the P content in phytoplankton could be intensified through Noctiluca' feeding activity, which would be detrimental to Noctiluca, leading to its disappearance in the water column 29 .
In conclusion, Noctiluca is nearly homeostastic, but it regulates its internal N more strongly and efficiently than its P content, thus is more vulnerable to P-limitation. It is able to differentially allocate N and P to organic and inorganic pools to maintain overall homeostasis. Excretion of C, N, and P by Noctiluca is coupled through processes that help to maintain its elemental composition, and also depends highly on resource nutritional quality. Noctiluca seems to have a shifting role in planktonic food web, from mainly exerting top-down control during most of its pelagic life to fuelling bottom-up processes due to the liberation of intracellular nutrients during senescence.

Materials and Methods
Preparation of experimental organisms. An initial inoculum of cells of Noctiluca was gently collected from Port Shelter in eastern Hong Kong in October 2011 using a plankton net of 120 µm mesh size. Cells were isolated and maintained in culture as described by Zhang et al. 29  P), respectively. The seawater used for preparing the media was the surface water collected from Port Shelter, and aged before usage (the concentrations of N and P in the seawater were negligible compared to the limiting nutrients in the media). Algal cultures were maintained at the same conditions as described above with illumination of 80 μmol photons m −2 s −1 . Experiments were started once the cultures reached stationary phase. All cultures were sampled for cell counts and for cellular C, N and P analyses before the experiments were started. Experimental manipulations. Experiments were designed to determine the stoichiometric regulation and element budgets of Noctiluca when facing different quality food. To avoid any potential effects of food carryover, Noctiluca used in this study were gently washed and resuspended with 0.2 µm-filtered autoclaved seawater 24 h prior to the experiment to void the food vacuoles. Feeding experiments involved 3 different T. weissflogii cultures (cultures grown in N-limited, nutrient-balanced and P-limited media) that were diluted to 1.5 × 10 4 cells mL −1 with appropriate amount of autoclaved filtered seawater. Noctiluca cells were inoculated into these food suspensions (2 L) in triplicate with a final concentration of ~13 Noctiluca cells mL −1 . Duplicate bottles with each type of prey were used as the controls. All bottles were incubated at dim light (~10 μmol photons m −2 s −1 ) under the same conditions described above. To avoid cell aggregation or settlement, cultures were gently agitated manually 2-3 times a day. Preliminary trials using a plankton wheel (at various rotation speeds) significantly reduced the rate of growth of Noctiluca (data not shown), and thus the best growth of predator and prey was achieved through regular manual agitation.
After 1 d incubation, duplicate 20 mL aliquots were withdrawn from the bottles with 25 mL plastic long pipettes for determining Noctiluca abundance, and 1 mL aliquots were collected with pipette (Eppendorf) for ultimately determining prey abundance. Both aliquots were preserved with acid Lugol's solution (final conc. 2%). The preserved cells of Noctiluca were also measured to determine cell size (first 100 cells encountered) using the SPOT image program (Version 3.5.0). All measured cells were examined for the presence of prey.
The remaining Noctiluca cells were collected on a mesh with a size of 100 μm, and gently washed and resuspended with autoclaved seawater to initiate excretion experiments. The resuspended Noctiluca cells of each food treatment (N-limited, nutrient balanced and P-limited) were immediately transferred to 3 (triplicate), 150 mL polycarbonate bottles, each yielding a concentration ~ 50 cells mL −1 , and the bottles were filled with autoclaved filtered seawater and incubated under the same condition as in the feeding experiment for 6 h. Subsamples for ScienTific RepoRts | 7: 7622 | DOI:10.1038/s41598-017-05991-w total cellular C, N and P contents, as well as the intracellular dissolved nutrient contents NH 4 + and PO 4 3− in Noctiluca cells were collected and determined as stated below before and after 6 h incubation. The 6 h incubation period was based on previous data showing that Noctiluca can void its food vacuoles in 4.5 to 7 h after feeding for 1 d 65-67 . Analytical measurements. Before the grazing experiment started, samples for analyses of cellular C, N, and P of differently grown T. weissflogii were taken from the respective culture bottles by filtering 15 to 25 mL cultures onto pre-combusted (550 °C, 4 to 5 h) GF/C glass-fiber filters. Samples for determining total elemental composition (C, N and P) and intracellular nutrients (NH 4 + and PO 4 3− ) of Noctiluca were collected at the beginning and end of the feeding experiment, and also after 6 h starvation. Usually more than 1200 Noctiluca cells were filtered on pre-combusted 25 µm GF/C filters for cellular C, N and P analyses. A similar amount of Noctiluca cells were filtered onto a 20-μm PC membrane and the filters were immediately submerged into 10 mL MilliQ, sonicated (50/60 hz, 20 min), and then filtered through a 0.2 µm disk filter. The filtrate was used to determine the amount of dissolved inorganic NH 4 + and PO 4 3− in N. scintllans. Subsamples for measuring these elements were in triplicate.
Cellular C and N were analyzed with a CHN elemental analyzer (Perkin-Elmer) and cellular P was analyzed as orthophosphate after acidic oxidative hydrolysis with 1% HCl 68 . Analyses of NH 4 + and PO 4 3− were conducted manually according to Strickland and Parsons (1972) 69 , and the detection limit was 0.5 µmol L −1 for PO 4 3− and 0.5 µmol L −1 for NH 4 + . Rate process and elemental composition calculations. Clearance (F, µL Noc −1 d −1 ) and ingestion (I, cells Noc −1 d −1 ) rates were calculated according to Harris et al. 70 and Frost 71 , respectively: ) where C′ t and C t (cells mL −1 ) are the prey concentrations at the end of the incubation in control and experimental bottles, respectively; V is the volume of the culture (mL), t (d) is the incubation period and n is the number of Noctiluca used; [C] is the prey concentration in the experimental bottle averaged over the incubation period. The homeostasis coefficient, Η, was calculated as: where x is the resource nutrient stoichiometry, y is the organism's nutrient stoichiometry and c is a constant 2 . Therefore, 1/H is the slope of the regression between log(x) and log(y) which is based on values between zero and one. According to Persson et al. 30 , if the regression relationship is non-significant (p > 0.1), the organism is considered 'strictly homeostatic' and an organism with 1/H = 1 is not considered to be homeostatic. Homeostatic plots with significant regressions and 0 < 1/H < 1 are classified as: 0 < 1/H < 0.25 'homeostatic' , 0.25 < 1/H < 0.5 'weakly homeostatic' , 0.5 < 1/H < 0.75 'weakly plastic' , 1/H > 0.75 'plastic' . In the present study, estimations of the homeostatic coefficient H for Noctiluca were only considered for Noctiluca cells collected immediately after 1 d incubation.
Determination of the stoichiometric composition and elemental growth yield of Noctiluca in term of C, N and P were corrected by subtracting the amount of elements of the food remnant in Noctiluca (assuming those prey items were intact and undigested, and an assimilation efficiency for C of 60%, and that of the non-limiting nutrient was 80% and that of the limiting nutrient was 100% for the remaining prey items). Elemental growth yield was calculated by comparing the elements per individual cell before and after 1 d incubation, and gross growth efficiency in terms of C (GGE C ) was calculated by dividing the growth yield by the ingestion per Noctiluca in terms of C. Based on the assumption that ingested material must be assimilated by organism before it is used for growth or metabolism, the expression for N and P excretion rates as a function of C metabolism, C-based growth yield, and C to nutrients ratios can be derived from elemental budget models modified from Landry 31 : where AE C , AE N and AE P is the assimilation efficiency of C, N and P (as a percentage); I C (pmol C Noc −1 h −1 ) is the ingestion of C; R C (pmol C Noc −1 h −1 ) indicates a combination of C respiration and other forms of C excretion, E N (pmol N Noc −1 h −1 ) and E P (pmol P Noc −1 h −1 ) are the excretion of N and P related nutrients; G C is C-based growth yield (pmol C Noc −1 h −1 ); C:N prey and C:P prey are C:N and C:P of prey; C:N pred and C:P pred are C:N and C:P of predator.
Rates of nutrients excretion were estimated by comparing the elemental contents per individual cell before (right after 1 d feeding experiment) and after 6 h starvation, and only those that had significant differences were reported.
Statistical analysis. Distributions of the data (log transformed before analysis as necessary) were evaluated by Shapiro-Wilk test before analysis of variance (ANOVA) and post hoc comparisons. Data that were normally distributed were analyzed using standard one-way ANOVA and Fisher LSD's post hoc comparisons with significance levels of p < 0.05. Comparisons between two groups were conducted using Student t-test (2-tailed) with significance levels of p < 0.05. All analyses were conducted using Sigma Plot 11.0 (Systat Software Inc., San Jose, CA). The models were simulated using R software v. 3.0.2 (R Development Core Team 2013).