Competition for inorganic and organic forms of nitrogen and phosphorous between phytoplankton and bacteria during an Emiliania huxleyi spring bloom

Abstract. Using 15N and 33P, we measured the turnover of organic and inorganic nitrogen (N) and phosphorus (P) substrates, and the partitioning of N and P from these sources into two size fractions of marine osmotrophs during the course of a phytoplankton bloom in a nutrient manipulated mesocosm. The larger size fraction (>0.8 μm), mainly consisting of the coccolithophorid Emiliania huxleyi, but also including an increasing amount of large particle-associated bacteria as the bloom proceeded, dominated uptake of the inorganic forms NH4+, NO3−, and PO43−. The uptake of N from leucine, and P from ATP and dissolved DNA, was initially dominated by the 0.8–0.2 μm size fraction, but shifted towards dominance by the >0.8 μm size fraction as the system turned to an increasing degree of N-deficiency. Normalizing uptake to biomass of phytoplankton and heterotrophic bacteria revealed that organisms in the 0.8–0.2 μm size fraction had higher specific affinity for leucine-N than those in the >0.8 μm size fraction when N was deficient, whereas the opposite was the case for NH4+. There was no such difference regarding the specific affinity for P substrates. Since heterotrophic bacteria seem to acquire N from organic compounds like leucine more efficiently than phytoplankton, our results suggest different structuring of the microbial food chain in N-limited relative to P-limited environments.

EGU methionine, leucine and tyrosine could be an important source of N for Prochlorococcus, but not for Synechococcus. Apart from this, there has been little work to quantify the importance of DON to the nutrition of these organisms in the natural environment. Laboratory experiments leave no doubt that the majority of aquatic algal species are able to utilize common organic compounds as N sources for growth if sufficient sub-5 strate concentration is provided and enough time is allowed for metabolic adaptation (reviewed by Berman and Bronk, 2003). Most of these studies, however, used axenic laboratory batch cultures growing on high initial substrate concentrations, thus the ability of organisms to exploit the much lower concentrations of these compounds encountered in the environment in situ still remains unclear (Paul, 1983;Berman and Bronk, 2003). This paper focuses on algal and bacterial uptake of different N-and P-substrates during an artificially induced phytoplankton bloom dominated by coccolithophorids, mainly Emiliania huxleyi. The experiment was done in a mesocosm set-up with a time dependent change from presumably C-stressed bacteria and P-stressed phytoplankton, to 15 N-limitation, at least of autotrophic processes. Bacterial and algal uptake of inorganic N (NH + 4 , NO − 3 ) and of leucine as a model substrate for DON, as well as inorganic P (PO 3− 4 ) and the two organic forms ATP and DNA as model substrates for DOP was measured. Analyses of 15 N and 33 P uptake from these sources in algal and bacterial size fractions were performed in order to compare algal vs. bacterial competition for 20 organic and inorganic dissolved N and P by means of their biomass-specific affinities.

Experiment
An outdoor mesocosm experiment was carried out at the Marine Biological Field Station, Espeland, 20 km south of Bergen, western Norway from 4 to 24 May. The study 25 was part of the Pelagic Ecosystem CO 2 Enrichment Study (PeECE) EGU setup and sampling procedures are described elsewhere (Grossart et al., 2006a). The data referred to here were obtained from enclosure number 4 referred to as "present" in the paper of Grossart et al. (2006a) because it represented the present-day level of 370 ppmV atmospheric CO 2 concentration. The enclosure (volume ∼20 m 3 ) was filled with unfiltered, nutrient-poor, post-spring bloom sea water from the fjord and supple-5 mented with 9 µmol L −1 NO − 3 , 0.5 µmol L −1 PO 3− 4 , and 12 µmol L −1 Si(OH) 4 on day 0 of the experiment to induce a phytoplankton bloom.
Stormy weather from day 4 eventually led to a collapse of the enclosure on day 6. The enclosure was restored to an upright position, and 705 mg of KH 2 PO 4 was added to the enclosure on day 7 to maintaining the phytoplankton bloom. The addition of 10 KH 2 PO 4 would ideally correspond to a PO 3− 4 enrichment corresponding to a concentration of 0.3 µmol L −1 PO 3− 4 , but was actually slightly higher since the enclosure was not entirely unfolded at the time. From day 10 on, the weather calmed down, and the enclosure was underlayered with an unknown volume of deep, high salinity water (with a soluble reactive phosphorus (SRP) concentration of 0.77 µmol L −1 ) in order to unfold 15 the enclosure after the storm.

Biomass estimations
Free and particle-associated bacteria were enumerated and their volumes calculated 25 as described elsewhere (Grossart et al., 2006a EGU was calculated assuming a constant biomass of 18.5 fg C cell −1 and a C:N:P molar ratio of 50:10:1 for coccoid bacteria (Fagerbakke et al., 1996). Particle-associated bacteria were significantly bigger than free bacteria, and increased from 0.6 to 1.5 µm 3 at the end of the experiment (Grossart et al., 2006a). Because of the large and temporally variable size of the particle-associated bacteria, allometric conversion factors between 5 volume (V ) and C, N, and P content (in femtograms) of C=220×V 0.9 , N=50×V 0.8 , and P=2×V 0.7 was applied. The bacterial volume to C conversion factor is comparable to a factor (C=218×V 0.86 ) for both cultures and bacterioplankton from lakes of different trophic status (Loferer-Krößbacher et al., 1998) and to a factor (C=224×V 0.89 ) for Vibrio splendidus grown along a gradient from C to P limitation (Løvdal et al., 2007a). The volume to N and volume to P conversion factors are adapted from Løvdal et al. (2007a).
Chlorophyll a (Chl a) measurements (Grossart et al., 2006a) were used for estimating phytoplankton C, N, and P biomass. A C:Chl a ratio of 30 (w:w) and a C:N:P molar ratio of 106:16:1 in phytoplankton biomass was assumed. With the conversion factors applied here, we got an estimated "bacterial + algal C-biomass" to Chl a ratio (w:w) of 15 85±22 (mean±SD, n=10), which is close to the reported constant particulate organic C (POC):Chl a ratio (w:w) of 81 (Aure et al., 2000, as cited in Erga et al., 2005 from the waters of the Samnanger Fjord. This site constitutes the fjord system adjacent to the study site. 2.4 Alkaline phosphatase activity (APA) 20 APA was measured fluorometrically using 3-0−methylfluorescein-phosphate (MFP) as substrate (Perry, 1972). Samples were mixed with MFP solution in 0.1 mol L −1 Trizma-HCl pH 8.3 (final concentration 0.1 µmol L −1 ). Fluorescence was measured directly after the addition of the reagent and at two subsequent times according to the expected activity using a Perkin Elmer fluorometer LS50B. After correcting fluorescence values of samples to those of autoclaved samples used as blanks, APA (nmol-P L −1 h −1 ) was calculated using linear regression of fluorescence values versus incubation time. the ambient NH + 4 concentrations were below the measurement limit, unlabelled NH 4 Cl was added to a final concentration of 0.5 µmol L −1 1 to 4 h before tracer addition. The precise portion of tracer added was calculated after determination of the ambient NH + 4 , NO − 3 , or DFAA (Grossart et al., 2006a) concentrations. The bottles were closed gastight and were incubated at in situ temperature and light. Four time points were taken 10 within 3 to 5 h for each measurement. The reaction was terminated by filtering the samples through silver-membrane filters (Osmonics; pre treated 500 • C for 1 h) with pore sizes of 0.2 and 0.8 µm. The filtration volume was 400 to 500 mL. Filters were flushed with 0.2 µm filtered sea water to remove adherent tracer-containing water and stored frozen until analysis. 15 After drying in a drying chamber, the filters were wrapped in silver cups (6×6×12 mm; Elementar Analysensysteme) and formed into pellets with a laboratory press. As the amount of N on the 0.2 µm filters was under the measurement limit, δ 15 N values were measured together with a well defined N isotope standard (peptone; Merck). One µmol L −1 peptone was pipetted onto the 0.2 µm filters before packing and N isotopes 20 were measured by continuous flow isotope ratio mass spectrometry (CF-IRMS) on a Finnigan MAT Delta plus coupled with a Thermo NA 2500 CHN analyser. The N content was calculated using acetanilide (Fisons) as a standard. =0.003676 (Junk and Svec 1958). The uptake rates were estimated from the regression relationship between uptake and time and converted to turnover 5 time (T ; see below).
2.6 33 P-labelling of DNA Radiolabeling of DNA was performed by random oligonucleotide primed synthesis (ROPS) with the DecaLabel DNA labeling kit (Fermentas K0621) in accordance with the manufacturer's instructions. [ 33 P]DNA was prepared for use in uptake studies as described by Løvdal et al. (2007b). This procedure yields labelled DNA products with an average length of 0.5 kilobase pairs (kb) which is at the lower end of the range for naturally occurring dDNA (0.12-35.2 kb) in aquatic environments (DeFlaun et al., 1987). We did not correct for potential shortening of the DNA chain length by radiochemical decay, but the radiolabelled DNA was used well within one half life of the ra- 15 dioactive precursor in order to avoid significant shortening. The final product had a specific activity of approximately 10 8 counts per minute (cpm) µg −1 with 97.0±0.6% (mean ± SD of 3 replicates) incorporation of label and a concentration of 12.6±0.1 ng µL −1 (mean ± SD of 3 replicates) as calculated by the DE-81 filter-binding assay (Sambrook and Russell, 2001 spectively. Incubations were stopped by cold chase (Løvdal et al., 2007b). Bacteria and phytoplankton were separated into different size fractions by filtration onto polycarbonate filters (Poretics) with pore sizes of 0.2 and 0.8 µm according to Løvdal et al. (2007b). Subsamples and filters were radioassayed by liquid scintillation counting (Løvdal et al., 2007b).
where R is the consumed fraction of added label and t is the incubation time. Hence, T represents here the turnover of substrates into particulate matter and does not in- 15 clude the release of hydrolysis products to the water phase. T for 33 P substrates was calculated from one time-point measurements (Thingstad et al., 1993), whereas T for 15 N substrates was calculated from the regression line between four time points.
Biomass-specific affinity for N (α N ; L nmol-N −1 h −1 ) and P (α P ; L nmol-P −1 h −1 ) uptake was estimated according to the procedure proposed by Thingstad and Ras-20 soulzadegan (1999): where f is the fraction of uptake in the respective size fraction and B is the biomass-N (nmol-N L −1 ) or biomass-P (nmol-P L −1 ). The biomass of particle-associated bacteria was contributed to the >0.8 µm size fraction because they could be trapped on a 25 3352 BGD 4,2007 Algal-bacterial competition for N and P EGU 5 µm filter (Grossart et al., 2006a). Hence, due to the particle-associated bacteria, the mechanical separation of bacteria and phytoplankton by size fractionation was not entirely successful. In our estimates of biomass-specific affinity, this was corrected for by assuming that particle-associated bacteria had the same affinity for uptake as free bacteria. From this assumption, the fraction of uptake by particle-associated bacteria 5 was calculated by rearranging Eq.
(1) to inserting the estimated biomass of particle associated bacteria. Corrected estimates corresponding to algae and bacteria was then calculated by subtracting this value from the >0.8 µm size fraction, contributing it to the 0.8-0.2 µm size fraction, respectively.

Statistical analysis
Statistical analysis was performed by student's t-tests according to Sokal and Rohlf (1995). The confidence level of all analyses was set at 95%.

Results
3.1 Nutrients, alkaline phosphatase activity (APA), and bloom development 15 Dissolved inorganic N (DIN), represented as the sum of NH + 4 +NO x was dominated by NO x concentrations ranging from 8 µmol L −1 on day 1, to 4 µmol L −1 on day 7, before it rapidly declined to under the detection limit (<0.01 µmol L −1 ) on day 16, whereas NH + 4 concentrations were low (<0.02 µmol L −1 ) the whole study period. SRP concentrations declined from approximately 0.4 µmol L −1 on day 1 to ∼0.04 µmol L −1 on day 20. Thus, 20 the DIN:SRP ratio was well above the Redfield ratio of 16 beyond the second addition of PO 3− 4 on day 7 before it rapidly declined ( Fig. 1 EGU between 0.4 and 1.6 µmol L −1 , with the lowest concentrations after the peak of the bloom (Grossart et al., 2006a).
Alkaline phosphatase activity (APA) ranged from 1 nmol-P L −1 h −1 on day 2 of the experiment, to 5 nmol-P L −1 h −1 on day 20. Specific values for APA (S-APA) and for phosphate affinity (S-α PO 4 ) normalized for the summed P-biomass of phytoplankton 5 and bacteria are shown in Fig. 2. Both values increased after day 8, but remained below 0.04 h −1 and 0.008 L nmol-P −1 h −1 , respectively. According to Tanaka et al. (2006), S-APA above 0.2 h −1 or S-α PO 4 above 0.02 L nmol-P −1 h −1 , is indicative of systemic P-limitation. According to these criteria, our data on S-APA and S-α PO 4 (Fig. 2) do not indicate severe P-limitation during any parts of the study period. 10 The algal bloom, almost exclusively dominated by coccolithophorids (mainly Emiliania huxleyi), was initiated by the addition of inorganic nutrients on day 0 and reached its maximum on day 12 (Grossart et al., 2006a) where the DIN:SRP ratio was ∼5 (NO x :SRP ∼2). Highest numbers of total as well as free bacteria were recorded on day 8 (Grossart et al., 2006a) when the DIN:SRP ratio was ∼11 and NO x accounted 15 for 99% of DIN. Particle-associated bacteria contributed to 40-80% of the total bacterial volume (Grossart et al., 2006a). From our estimates, the contribution of particleassociated bacteria to the total C-biomass of algae and bacteria increased from ∼30% on day 0 to ∼65% on day 20. 20 During the first week of the experiment, T of all substrates were in general long and T of the organic N and P substrates measured were shorter than T of the respective inorganic substrates (Fig. 3). This may indicate C-limitation of heterotrophic processes in this phase (see discussion). The peak of the phytoplankton bloom on day 12 coincided with the shortest T NH4 (4.8 h) and T NO3 (

EGU
whereas T Leucine decreased from >100 h to 34 h at the same time (Fig. 3a). Despite high DIN:SRP ratios (Fig. 1), which could have been interpreted to indicate phosphate limitation, T PO4 was longer than T ATP at the beginning of the experiment (Fig. 3b). Because P limitation was not expected immediately after the addition of PO 3− 4 on day 7, T for P substrates was not measured between day 6 and day 12, the 5 peak of the phytoplankton bloom. On day 12, T PO4 (5.8 h) was shorter than T ATP and was comparable to T NH4 and T NO3 , before it reached a minimum of 1.4 h at day 16 (Fig. 3b). T ATP and T dDNA declined from 24 and 346 h, respectively, on day 12 before it stabilized on 6±2 h (mean ± SD, n=) and 84±8 h (mean ± SD, n=4), respectively, for the rest of the study period (Fig. 3b). The distribution of added activity taken up by the two size fractions is shown in Fig. 4. The 0.8-0.2 µm size fraction took up most of the organic substrates during the first part of the study period, whereas the >0.8 µm size fraction dominated uptake of the inorganic substrates. In our set of data, both T NH4 and T PO4 dropped below 10 h at the 15 peak of the phytoplankton bloom (Fig. 3). There are strong indications for a transition from a non-N-limited to an N-limited system at this point (see Discussion). Hence, we used T NH4 as an indicator to split the dataset into non-N-limitation (T NH4 >10 h) and Nlimitation (T NH4 <10 h) which also represents the periods before and after the peak of the phytoplankton bloom, respectively. After the peak of the phytoplankton bloom, the 20 >0.8 µm size fraction dominated uptake of all substrates, and NH + 4 uptake was almost exclusively by the >0.8 µm size fraction. There were small changes in the distribution of inorganic N and P compared to before the phytoplankton peak, with the >0.8 µm size fraction taking up slightly more after this peak (Fig. 4). Table 1 is a compilation of data presented in Fig. 4 showing uptake distributions to 25 the >0.8 µm size fraction grouped for the two periods. NO EGU was a shift towards DON and DOP uptake dominated by the >0.8 µm size fraction during T NH4 <10 h (Table 1). There were no drastic shifts in the distribution of inorganic substrates, with the >0.8 µm size fraction dominating uptake in both periods, but more so during T NH4 <10 h (Table 1). S-α NO 3 and S-α NH 4 , and S-α PO 4 are specific affinities normalized for the summed N 5 and P biomass, respectively, of algae and bacteria, and are shown in Fig. 2. Affinities for all substrates were low during the first half of the study period. Maximum values of S-α NH 4 paralleled the peak of the phytoplankton bloom, whereas S-α PO 4 peaked on day 16. Comparing competitive ability of osmotrophs in the 0.8-0.2 µm and the >0.8 µm size 10 fraction by means of biomass-specific affinity for N-uptake in the two size fractions, focusing on the N-limited phase (T NH4 <10 h), reveals a significant shift depending on whether N is present in organic or inorganic form (Fig. 5). The >0.8 µm size fraction had a significantly higher biomass-specific affinity for NH + 4 than the 0.8-0.2 µm size fraction (p=0.005), and a significantly lower biomass-specific affinity for leucine-N than 15 the 0.8-0.2 µm size fraction (p=0.003) (Figs. 5a-b). No such shift between the competition for DIP and DOP were observed; no significant differences were found between biomass-specific affinity in the two size fractions for PO 3− 4 (p=0.449), ATP-P (p=0.372), or dDNA-P (p=0.494) (Figs. 5c-e).
From Eq.
(2), it was estimated that much of the uptake in the >0.8 µm size frac-20 tion was by particle-associated bacteria. Particle-associated bacteria took up larger shares of DON compared to DIN (Table 2). In fact, it was estimated that the majority (73±6%, n=5) of total DON uptake was by particle-associated bacteria during Nlimitation (T NH4 <10 h), leaving phytoplankton with <2% (Table 2), and thus free-living bacteria with approximately 25% of total DON uptake. Hence, from the corrected es- 25 timates, phytoplankton had a significantly lower biomass-specific affinity for DON than heterotrophic bacteria (p<0.0001) (Fig. 5b) but a significantly higher biomass-specific affinity for NH + 4 (p<0.0001) (Fig. 5a). No shift between the competition for DIP and DOP were observed; no significant differences were found between biomass-specific affinity

Discussion
It has been hypothesized that phosphate limitation, classically indicated by DIN:DIP ratios >16, is one of the critical factors allowing the coccolithophore Emiliania huxleyi 5 to bloom (reviewed by Lessard et al., 2005). This hypothesis is based on physiological studies showing that E. huxleyi has an exceptionally high affinity for PO 3− 4 and is able to use DOP (Kuenzler and Perras, 1965;Riegman et al., 2000). High NO − 3 :PO 3− 4 ratios, however, appear to be the exception rather than the rule in E. huxleyi blooms. In fact, in most blooms studied to date, NO − 3 :PO 3− 4 ratios were low, and nitrate was low or 10 undetectable (Lessard et al., 2005). In the current experiment, the addition of NO − 3 and PO 3− 4 in a molar ratio of 18 initiated a bloom of coccolithophorids, dominated by E. huxleyi, reaching its maximum after 12 days when the DIN:SRP ratio was ∼5. The first half of the period prior to the peak of the bloom was characterized by high DIN:SRP ratios (Fig. 1), thus the phytoplankton growth may be suspected to have been restricted by the availability of inorganic P. Low S-APA and low S-α PO 4 , S-α NO 3 and S-α NH 4 values (Fig. 2), however, does not indicate severe mineral nutrient limitation, and T of the organic N and P substrates measured where shorter than T of the respective inorganic substrates (Fig. 3), possibly indicating C-limitation of heterotrophic processes in this phase. Bacteria hydrolyze specific DON and DOP-compounds in the presence of inor-20 ganic mineral nutrients, presumably to obtain other associated mineral nutrients and C (Jørgensen et al., 1993;Benitez-Nelson and Buesseler, 1999). DIN:SRP ratios rapidly declined from day 7 (Fig. 1) due to the PO 3− 4 enrichment. However, the additional PO 3− 4 was rapidly consumed followed by increasing values of S-APA and S-α PO 4 (Fig. 2). Sα NO 3 and S-α NH 4 values rapidly increased from day 8 with S-α NH 4 peaking at day 12 25 (Fig. 2). Additionally, T NO3 and T NH4 got shorter as the experiment proceeded, with 3357 BGD 4, 2007 Algal-bacterial competition for N and P EGU T NH4 reaching minimum values at day 10 before stabilizing, and T Leucine declining from day 10 to the end (Fig. 3). Conclusively, the splitting of the investigation period into a non-N-limited and an N-limited phase can be justified. From our data, there is evidence for a shift in terms of algal -bacterial competition when N is available in the form of leucine contrary to NH + 4 (Fig. 5), with organisms in the 5 >0.8 µm size fraction having significantly lower biomass-specific affinity for leucine-N and significantly higher biomass-specific affinity for NH + 4 compared to smaller organisms (0.8-0.2 µm). Correcting for the contribution of particle-associated bacteria in the >0.8 µm size fraction by assuming they have similar affinity per biomass as free-living bacteria, probably represents a best estimate for the competition between algae and heterotrophic bacteria. However, since specific aminopeptidase (Karner and Herndl, 1992;Middelboe et al., 1995), protease (Becquevort et al., 1998) and phosphatase (Smith et al., 1992;Simon et al., 2002) activities in particle-associated bacteria is often higher than those characteristic for free-living bacteria, our affinity values for heterotrophic bacteria may have been underestimated, and the values for phytoplankton 15 correspondingly overestimated.
In a previous investigation in P-limited estuarine mesocosms, Løvdal et al. (2007b) found no significant shift in algal -bacterial competition for P from ATP and dDNA relative to PO 3− 4 , in terms of biomass-specific affinity. In the current experiment, using comparable methodology, a shift from bacteria dominating the competition for DON 20 towards algae dominating the competition for DIN was evident (Fig. 5). This does indeed indicate that the structure of the microbial food web in N-limited environments may be different from that in P-limited environments. If amino acids are utilized almost exclusively by bacteria, this changes the food web structure relative to P-limited environments, since there will be a pool of dissolved N for which bacteria do not experience 25 significant competition from phytoplankton to achieve. One effect of this is, theoretically, that the probability for C-limitation of bacteria could be larger in N-deficient regions than in P-deficient regions (Thingstad, 2000).
E. huxleyi has been shown to be a moderate competitor for nitrate and a good com- EGU petitor for phosphate compared to other algal species (Riegman et al., 1992(Riegman et al., , 2000. During natural E. huxleyi blooms, E. huxleyi primarily use NH + 4 and urea (Kristiansen et al., 1994;Fernández et al., 1996;Rees et al., 2002). The high P acquisition capacity and ability to use non-nitrate N has been suggested to (Lessard et al., 2005), at least in part, explain the success of E. huxleyi in nutrient-depleted waters. This is supported 5 by our data because the biomass-specific affinity for all P-substrates in E. huxleyi was comparable to that in heterotrophic bacteria, whereas E. huxleyi, although they seemed to have low amino acid-N acquisition capacity, dominated the competition for NH + 4 . Our results also agree with previous findings that DFAA are an important N source to marine bacteria (Hollibaugh and Azam, 1983;Jørgensen et al., 1993). 10 The biomass-specific affinity values estimated in the present study can be compared with those given by diffusion limitation of substrate transport to the cell surface -that is, the theoretical maximum. Assuming that the cell is diffusion-limited, i.e., that the cell's uptake system is so efficient (and the bulk nutrient concentration so low) that all substrate molecules hitting the cell surface are captured, it is possible to derive a 15 theoretical expression for maximum specific affinity (α max ) for a spherical cell ( Fig. 6; see also Thingstad and Rassoulzadegan, 1999): where D is the diffusion constant for the substrate molecules and σ is the volumespecific content of the element in question. Assuming free-living and particle-20 associated bacteria to have volumes of 0.2 and 1.5 µm 3 , respectively, and particleassociated bacteria to make up 30% of the total bacterial community (cf. Grossart et al., 2006a), their mean equivalent r can be calculated to be ∼0.5 µm. Assuming the phytoplankton community is dominated by N-limited E. huxleyi with volumes of ∼40 µm 3 (cf. Riegman et al., 2000), their mean r may be set to 2.1 µm. One should 25 however keep in mind that genetic and physiological intraspecific diversity in E. huxleyi has been documented, and multiple strains can be present within an environment (Paasche, 2002), complicating comparison between laboratory and field studies. The E. huxleyi (L-strain) used as a model organism by Riegman et al. (2000)  EGU in continuous cultures and was more than three times smaller and had lower internal N and P cell quotas than E. huxleyi reported elsewhere (Ho et al., 2003). Thus, the results of their study cannot be applied directly to the natural environment. Hence, the cell radius applied here and the lines representing the theoretical maximum for biomass-specific affinity in E. huxleyi (Fig. 6) is to be understood as extremes.

5
The mean biomass-specific affinity for NH + 4 uptake in phytoplankton was higher than that for bacteria, approaching that predicted by the diffusion model (Fig. 6a), contrary to that estimated for leucine (Fig. 6b). Although E. huxleyi is reported to grow fairly on leucine as the sole N-source (Ietswaart et al., 1994), our data suggest that the natural amount of leucine does not serve as an important N-source for E. huxleyi in our study.
The reason for the low biomass-specific affinity for N-substrates estimated for bacteria (Figs. 6a-b) is unknown. One possibility, contradictory to our previous conclusion, could be that the growth rate of heterotrophic bacteria was limited by C or other factors, rather than N. Another possibility could be artefacts in our estimation procedure. However, the main reason that biomass-specific affinities for N-compounds, both in bacteria 15 and algae, seem unrealistically low for a presumably N-stressed system, compared to that of P-compounds, stems probably from a methodological limitation. Although the continuous flow isotope ratio mass spectrometry (CF-IRMS) is very sensitive, it requires a certain threshold level of particulate material for the analysis of microbial N uptake. Hence, following long filtration time due to large sample volumes, this tech-20 nique only allows for analysis of relatively high levels of isotope enrichment to avoid exhaustion of the added tracer. Thus, it is not optimal in environmental labelling studies when the ambient turnover time is extremely short. In our experiment, effort was done to add below 10% of ambient concentrations of tracer 15 N isotope. In the case of NH + 4 uptake studies, this was occasionally achieved only by adding additional un-25 labelled NH 4 Cl to the incubation bottles. During the N-limited phase, this enrichment could increase the DIN:SRP ratio in the incubation bottles by a factor of up to 10 compared to the ambient concentrations (Fig. 1), but the DIN:SRP ratio in the incubation bottles was still below 16. We acknowledge that this treatment presumably relaxed EGU the N-limitation in these bottles, and led to significant overestimates of the ambient NH + 4 turnover times (Fig. 3a). Additions of 15 N-leucine were based on measurements of ambient DFAA concentrations. Most likely, the DFAA pool measured chemically is larger than that utilized by osmotrophs since both bacteria and phytoplankton prefer certain amino acids (including the neutral leucine) over others (Ietswaart et al., 1994).

5
Hence, in the N-limited phase, 15 N enrichment presumably exceeded that typically regarded as true tracer levels, leading to an overestimate of turnover times, and hence underestimates of biomass-specific affinities for N-compounds (Figs. 5a-b). The liquid scintillation counting technique on the other hand, applied for 33 P uptake studies, is not hampered with this problem because the high specific activity of 33 P allows for 10 true tracer level enrichment, e.g. picomolar concentrations. Although this problem with the CF-IRMS technique can be expected to have given significant underestimates of biomass-specific affinity, we believe that it did not affect the relative distribution of Nsubstrates in the two size fractions considerably. Thus, qualitative differences between the 0.8-0.2 µm and the >0.8 µm size fractions were attributed to the differences be-15 tween bacterial and algal metabolism. DNA and ATP contain 16% and 14% N, and 10% and 18% P, respectively. Hence, they contain as much N per dry weight as amino acids of algal proteins (Laws, 1991). Therefore these substrates may serve as potential N and P sources at the same time. The high biomass-specific affinities for dDNA and ATP, considering their diffusion con-20 stants, compared to PO 3− 4 (Figs. 6c-e), may indicate that these substrates where hydrolyzed for their N content rather than their P content. dDNA is mainly utilized as a P source by marine bacteria, and accounts <10 of their N and C requirements (Jørgensen et al., 1993;Jørgensen and Jacobsen, 1996). The possibility then exists, that organic compounds, such as dDNA, are more bioavailable to phytoplankton than 25 DON-compounds, such as amino acids, and are thus utilized as alternative sources for N by phytoplankton when the ambient DIN-concentrations are too low to support their growth. The potential for dDNA and ATP to also have rapid turnover times in non-Plimited environments, and to substantially support algal N-demand, indicates that phy- EGU toplankton may rely on these compounds to support their N-demand, rather than those traditionally looked upon as important DON-sources. Yet, isotope P-labelled substrates may yield limited information about N uptake. Therefore, an N-isotope is preferable in assessing phytoplankton dDNA and ATP utilization. The main conclusion of our study is that the competitive fitness of bacteria and al-5 gae for nutrients (and particularly N) vary dramatically depending on whether the substrates are organic or inorganic. Whereas bacteria appear to be superior competitors for organic forms of N, phytoplankton appear to be superior competitors for inorganic N supplied as NH + 4 . This observation has major implications for our understanding of marine food webs and effects of nutrient status on community structure. EGU Kristiansen, S., Thingstad, T. F., van der Wal, P., Farbrot, T., and Skjoldal, E. F.: An Emiliania huxleyi dominated subsurface bloom in Samnangerfjorden, western Norway. Importance of hydrography and nutrients, Sarsia, 79, 357-368, 1994. Kuenzler, E. J. and Perras, J. P.: Phosphatases of marine algae, Biol. Bull., 128, 271-284, 1965. 5 Laws, E. A.: Photosynthetic quotients, new production and net community production in the open ocean, Deep-Sea Res., 38, 143-167, 1991. Lessard, E. J., Merico, A., and Tyrrell, T.: Nitrate: phosphate ratios and Emiliania huxleyi blooms, Limnol. Oceanogr., 50, 1020-1024, 2005. Loferer-Krößbacher, M., Klima, J., and Psenner, R.: Determination of bacterial cell dry mass by transmission electron microscopy and densitometric image analysis, Appl. Environ. Microbiol., 64, 688-694, 1998. Løvdal, T., Skjoldal, E. F., Heldal, M., Norland, S., and Thingstad, T. F Table 2. Uptake of labelled substrates by particle-associated bacteria and phytoplankton given as percent of total uptake. Estimates from Eq.

BGD BGD
(2) assuming similar affinity in particle-associated bacteria as in free-living bacteria (see text). Pooled data for inorganic and organic N and P substrates, grouped in samples with T NH4 longer than and shorter than 10 h. Means with standard deviation.