Availability of phosphate for phytoplankton and bacteria and of labile organic carbon for bacteria at different pCO2 levels in a mesocosm study

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. Availability of phosphate for phytoplankton and bacteria and of labile organic carbon for bacteria at different pCO2 levels in a mesocosm study T. Tanaka, T. F. Thingstad, T. Løvdal, H.-P. Grossart, A. Larsen, K. G. Schulz, U. Riebesell

EGU emissions continue to rise on current trends (business as usual), the world oceans will suffer an estimated pH drop of about 0.5 units, which is equivalent to a 3 fold increase in the concentration of hydrogen ions, until the year 2100 (Wolf-Gladrow et al., 1999;Caldeira and Wickett, 2003). While the magnitude of ocean acidification can be predicted with a high level of confidence, its impact on marine organisms, their ecosys- 5 tems, and biogeochemical cycling are largely unknown. Studies dealing with biological responses to increasing CO 2 partial pressure (pCO 2 ) and related changes in carbonate chemistry range from a single-species level in laboratory cultures up to a semi-natural community level in outdoor mesocosms. These studies show that increasing pCO 2 enhances photosynthetic carbon fixation (Zondervan et 10 al., 2001;Leonardos and Geider, 2005) and release of dissolved carbohydrates by phytoplankton (Engel et al., 2004), and also modifies phytoplankton species composition and succession (Tortell et al., 2002). These pCO 2 dependent changes in phytoplankton parameters further enhance growth rate, production as well as α-and β-glucosidase activity of heterotrophic bacteria, especially of attached bacteria (Grossart et al., 2006). 15 On the other hand, no significant increase in primary production (Tortell et al., 2002;Sciandra et al., 2003;, total phytoplankton biomass (Tortell et al., 2002), and total bacterial biomass (Rochelle-Newall et al., 2004, Grossart et al., 2006 have been detected at increasing pCO 2 levels. The elemental composition (e.g. C, N, P) in living organisms is to a certain extent 20 constrained by the necessity to maintain their metabolism (homeostasis) as compared to the rest of the material world (reviewed by Sterner and Elser, 2002). Changes in pCO 2 dependent carbon production by phytoplankton and bacteria (see above) may alter their nutrient demands. On a global scale, these pCO 2 dependent changes will greatly influence carbon and nutrient cycling in the ocean. However, significant differ- 25 ences in consumption ratio of various inorganic nutrients due to increasing pCO 2 levels have been found in one study (Tortell et al., 2002) but not in an other one (Engel et al., 2005). In this context, changes in nutrient availability for phytoplankton and bacteria at different pCO 2 levels seem to be unclear and, thus need to be investigated in greater EGU detail. Nutrient availability (e.g. deficiency, limitation) is not necessarily easy to be examined especially for natural communities of phytoplankton and bacteria. A recent study suggests that the specific affinity for phosphate uptake and the specific alkaline phosphatase activity (APA), which are phosphate uptake rates (inverse of phosphate turnover times) and APA, respectively, normalized to the summed biomass of phytoplankton and bacteria, are useful to examine phosphate availability for the natural phytoplankton and bacteria community in different P starved aquatic systems (Tanaka et al., 2006). Similarly, specific affinity for glucose uptake, which is glucose uptake rates (inverse of glucose turnover times) normalized to bacterial biomass, can be useful to 10 examine the availability of labile organic carbon for bacteria (Tanaka et al., 2007 1 ).
The objective of the present study is to examine how the availability of phosphate for phytoplankton and bacteria and of labile organic carbon for bacteria changes at different pCO 2 levels during a mesocosm experiment. Particulate P concentrations, turnover times of phosphate and glucose, and APA were measured during the experi-15 ment. Combining these results with biomass measurements of phytoplankton and bacteria (Paulino et al., 2007;Schulz et al., 2007

Particulate P
Particulate P samples were size-fractionated in triplicates on polycarbonate filters (47 mm diameter) with 10, 5, 1, and 0.2 µm pore sizes, respectively. Filters were transferred to polycarbonate test tubes with 5 ml of Milli-Q water and oxidized by acid persulphate at 121 • C. Liberated P was spectrophotometrically measured (Koroleff, 1983 Uptake rate of orthophosphate was measured using 33 P-orthophosphate (Thingstad et al., 1993). Carrier-free 33 P-orthophosphate (Amersham, 370 MBq ml −1 ) was added to 12 ml samples in 15 ml sterile Nunc test tubes to give a final radioactivity of ∼10 5 counts per minute ml −1 . Samples for subtraction of background and abiotic adsorption 5 were fixed with 100% TCA before isotope addition. Samples were incubated under subdued (laboratory) illumination at in situ temperature. Incubation time was short enough to assure a linear relationship between the fraction of isotope adsorbed vs. the incubation time but it was long enough to reliably detect isotope uptake above background levels. Incubation was stopped by a cold chase of 100 mmol L −1 KH 2 PO 4 (final conc. 1 mmol L −1 ). Subsamples were filtered in parallel on 25 mm polycarbonate filters with 10, 5, 1, and 0.2 µm pore sizes, which were placed on a Millipore 12 place manifold and supported on Whatman (GF/C) glass fiber filters saturated with 100 mmol L −1 KH 2 PO 4 . After filtration, filters were placed in polyethylene scintillation vials with Ultima Gold (Packard), and radio-assayed. After the radioactivities of the filter were corrected

EGU
Days 0-3 due to a technical problem.

Calculation of specific affinity for phosphate and glucose uptake
Under the assumption that a community of phosphate users (i.e. bacteria and phytoplankton) consists of a set of n groups with biomass B i (nmol-P L −1 : i =1, . . . , n), and all of which take up PO 4 proportional to phosphate concentration (S [PO4] : nmol-P 5 L −1 ) and to each biomass, one has i is the specific affinity for PO 4 uptake (L nmol-P −1 h −1 ) by osmotroph group i and f i is the fraction (no dimension) of total uptake V [PO4] that goes into osmotroph group i (Thingstad and Rassoulzadegan, 1999). Since the sum of PO 4 uptake equals to V [P O4] (sum of f i =1) and PO 4 turnover time is the ratio between the phosphate concentration and the P flux (1) Then, the total specific uptake rate (h −1 ) (Eq. 1) can be normalized to total biomass of bacteria and phytoplankton (B), and is defined as specific affinity for PO 4 uptake by the osmotroph community (α e [PO4] : hereafter, specific PO 4 affinity): Specific affinity for glucose (α e [Glu] : L nmol-P −1 h −1 ) by bacteria is defined as

EGU
where T [Glu] is the function of glucose concentration (S [Glu] : nmol-glucose L −1 ) and its flux (V [Glu] : nmol-glucose L −1 h −1 ) through this pool (T [Glu] =S [Glu] /V [Glu] ) and B BAC is bacterial biomass (nmol-P L −1 ). Since all mesocosms were P-depleted rather than labile organic carbon depleted (see Results) and we wanted to unify the unit of specific affinity, specific glucose affinity was expressed in the unit of P. 5 To calculate P-biomass, data on bacterial abundance and Chla were used (Paulino et al., 2007;Schulz et al., 2007 2 ). Bacterial abundance and Chla were converted to C-biomass under the assumption that bacterial carbon content is 20 fg C cell −1 (Lee and Fuhrman, 1987) and C: Chla (w:w) is 30 (w:w), respectively. Fixed C:P molar ratios of 106 for phytoplankton (Redfield et al., 1963) and 50 for bacteria (Fagerbakke 10 et al., 1996) were used to calculate P biomass of bacteria and phytoplankton, respectively. Although C:P ratios are variable for both phytoplankton and bacteria (e.g. Fagerbakke et al., 1996;Geider and La Roche, 2002), we applied the average C:P ratios for phytoplankton and bacteria. This is because direct measurement of P biomass of osmotrophs was not done in this study (see Results and Discussion for potential biases 15 by these fixed ratios).
2.5 Alkaline phosphatase activity APA was measured fluorometrically using 3-0-methylfluorescein-phosphate as substrate (Perry, 1972). Samples were mixed with the substrate solution in 0.1 mol L −1 Trizma-HCl pH 8.3 (final concentration 0.1 µmol L −1 ). Fluorescence was measured 20 just after the addition of the substrate solution and at two or three subsequent times according to the fluorescence increase. After correcting fluorescence values of samples to those of autoclaved samples, APA (nmol-P L −1 h −1 ) was calculated by using a linear regression of fluorescence values versus incubation time. The APA measurement was done only for the period with low concentrations of soluble reactive phosphorus (SRP) 25 (Days 10- 25: Schulz et al., 2007 2 ). As APA is derepressed when cellular P falls below a certain threshold level (e.g. Fitzgerald and Nelson, 1966;Rhee, 1973 and Sakshaug, 1983) and biomass of phytoplankton and bacteria were variable with time and between mesocosms, biomass-specific APA (h −1 ) was calculated by dividing APA by estimated P biomass of osmotrophs (nmol-P L −1 ). Measurements of APA and phytoplankton biomass were not always synchronized due to logistic constraint. To calculate specific APA, Chla values on Days 19, 21, and 23 (Data from Schulz et al., 5 2007 2 ) and APA values on Day 24 were linearly interpolated.

Statistical analysis
Analysis of covariance (ANCOVA) was done using the R software (http://www.r-project. org/) with log-transformation of all data. Experimental day was used as the covariate and pCO 2 was used as the nominal predictor. Simple linear regression (Model I) was 10 used to describe the relationship between specific glucose affinity and dissolved organic carbon (DOC: Data from Schulz et al., 2007 2 ). As the temporal variations of these parameters were similar between the three mesocosms (see Results and Discussion), all data except on Day 4 were pooled for the correlation analysis. Statistical significance of linear regression was tested by F -test.

Results and discussion
In this study, samples were taken only from one mesocosm of each pCO 2 level (M2, M5, and M8) because of logistic constraints. Since no significant differences between samples of each pCO 2 level were found for temporal changes in inorganic and organic nutrients, biomass and production of phytoplankton and bacteria (ANCOVA test,  Paulino et al., 2007;Schulz et al., 2007 2 ), we assume that the three mesocosms selected in this study were representative for each pCO 2 level. The increase of total particulate P (>0.2 µm) between Days 0-10 was accompanied by an increase of particulate P in the >10 µm fraction in all three mesocosms (Fig. 1). 5 While particulate P concentrations in the 0.2-1 µm fraction varied little except on Day 6 in 3×CO 2 during this period, the proportion of the 0.2-1 µm fraction to total particulate P decreased. The particulate P concentration in the >10 µm fraction peaked on Day 10 in all mesocosms, while it was significantly higher in 3×CO 2 (0.61 µmol L −1 ) than in 1×CO 2 (0.44 µmol L −1 ) on Day 10 (t-test, P<0.05). While the highest proportion of 10 the >10 µm fraction in 2×CO 2 and 1×CO 2 was observed on Day 6 (60% and 71%, respectively), the particulate P concentration in the >10 µm fraction on Day 6 was significantly higher in 3×CO 2 (0.46 µmol L −1 ) than in 2×CO 2 (0.29 µmol L −1 ) (t-test, P<0.005). However, the statistical difference of the >10 µm fraction between 3×CO 2 and 1×CO 2 (0.38 µmol L −1 ) was marginal on Day 6 (t-test, P=0.053). After the peak 15 on Day 10, the proportion of the >10 µm fraction decreased to 33-43%, while that of the 1-10 µm fraction increased to 30-45% in all three mesocosms. Temporal variations in substrate turnover time reflect either those of substrate concentration or those of substrate flux through this pool or both (see Materials and Methods).

EGU
From Day 16 onward, turnover time oscillated (0.4-21 h) and tended to be shorter in 3×CO 2 than in 2×CO 2 and 1×CO 2 . When phosphate turnover time was shorter, more phosphate was taken up by the smaller fraction and vice versa. During the initial period (Days 0-6) with turnover time >100 h, 33 PO 4 uptake by the >10 µm fraction increased up to 70% in 3×CO 2 , while it was up to 50% in 2×CO 2 and 1×CO 2 . The mean uptake 5 was highest (47-53%) by the 0.2-1 µm fraction and smallest (8-11%) by the 5-10 µm fraction during the experiment. APA ranged from 1.3 to 24.6 nmol-P L −1 (Fig. 3). After SRP depletion around Day 10 (Schulz et al., 2007 2 ), APA increased toward Days 13-15, and the fastest and highest increase in APA was observed in 3×CO 2 . After this peak, APA decreased in all three mesocosms. Glucose turnover time was long (>100 h) between Days 4-6, and rapidly decreased to ca. 16 h on Day 14 (Fig. 4). Thereafter, it rapidly increased toward Day 18 (74-91 h) and fluctuated between 39 and 73 h onward. No significant differences in any of the parameters presented in Figs. 1-4 were detected between the different pCO 2 treatments during the experimental period (ANCOVA, P>0.05). 15 Particulate organic matter (POM) and phytoplankton biomass increased from the start of the experiment, peaked around Day 10, and decreased onward, while Si, SRP, and nitrate concentrations changed from replete to deplete during the experiment (Schulz et al., 2007 2 ). This nutrient dynamics can be summarized as follows: (1) no obvious nutrient depletions between Days 0-6, (2) only Si depleted between 20 Days 7-9, (3) Si and phosphate depleted between Days 10-12, (4) Si, phosphate, and nitrate depleted from Day 13 onward. Interestingly, this approximately corresponded to the dynamics of phosphate turnover time: (1) the longest turnover time (>100 h) between Days 0-6, (2) decreasing turnover time between Days 7-11, (3) the shortest turnover time (<1 h) between Days 12-16, (4) increasing turnover time between Days 17-20, and (5) the short-term oscillation between Days 21-24 (Fig. 2

EGU
Phase II, Prymnesiophytes and Prasinophytes in Phase III, Prasinophytes, Dinoflagellates, and Diatoms in Phase IV, and Dinoflagellates and Cyanobacteria in Phase V. These results suggest that temporal changes in availability of the inorganic nutrients influenced those of phytoplankton biomass and succession during the experiment. Specific phosphate affinity, which was calculated for the period of low SRP concen-5 trations (Days 10-24: Schulz et al., 2007 2 ), ranged from 4.0×10 −4 to 1.3×10 −2 L nmol-P −1 h −1 in the three mesocosms (Fig. 5). As the biomass of phytoplankton and bacteria was similar in all three mesocosms (Paulino et al., 2007;Schulz et al., 2007 2 ), temporal variation of specific phosphate affinity mirrored that of phosphate turnover time. After SRP depletion, specific phosphate affinity increased toward Days 12-16. However, 10 it decreased between Days 16-20, while, between Days 20-24, it showed short-term oscillations and tended to be higher in 3×CO 2 than in 2×CO 2 and 1×CO 2 . Specific APA ranged from 0.003 to 0.141 h −1 and peaked 4-6 days later (Day 19) than the APA (Figs. 3 and 5). This was because the temporal changes of the summed biomass of phytoplankton and bacteria were greater than those of the APA between Days 13-19 15 (Paulino et al., 2007;Schulz et al., 2007 2 ). However, no significant differences in specific phosphate affinity and specific APA between M2, M5, and M8 occurred between Days 10-24 or 25 (ANCOVA, P>0.05). Tanaka et al. (2006) have recently proposed that either specific APA >0.2 h −1 and/or specific phosphate affinity >0.02 L nmol-P −1 h −1 indicates P limitation, i.e. the growth 20 rate of the existing organisms is reduced due to the P availability, in different aquatic systems. They have also proposed that either specific APA in the order of 0.01 h −1 and/or specific phosphate affinity in the order of 0.001 L nmol-P −1 h −1 indicate a situation less strict than limitation, i.e. P deficiency or suboptimal nutrient supply for phytoplankton and bacteria community. In this context, specific phosphate affinity and 25 specific APA suggested P-deficiency of the phytoplankton and bacteria community in all three mesocosms between Days 11-24, except for 2×CO 2 and 1×CO 2 on Days 20 and 24 (Fig. 5). We acknowledge that the estimate of P biomass, which was used to determine specific phosphate affinity and specific APA, includes elements of uncertainty EGU in this study (see Materials and Methods). However, the estimated P biomass never exceeded the chemically measured particulate P (>0.2 µm) (range: 14-60%, n=36). The ratios of POC to particulate P were similar for all mesocosms and slightly higher (mean±SD: 129±28, n=38) than the Redfield ratio of 106 (see Schulz et al., 2007 2 for POC data). If the specific phosphate affinity is recalculated by correcting the assumed 5 C: P ratios by the POC to particulate P ratios, only for one occasion (3×CO 2 on Day 23) we found P-limitation whereas all other data points indicated P-deficiency. Thus, we believe that the uncertainties in P biomass estimation would not significantly change our conclusion of phosphate availability for the phytoplankton and bacteria community. During the P deficient period, viral abundance was high in all three mesocosms , suggesting that DOM production increased because of viral lysis of bacterial and phytoplankton cells. However, DOP concentrations increased gradually and slightly throughout the experiment (Schulz et al., 2007 2 ). This can be explained by the fact that DOP produced via viral lysis is rather labile, and thus rapidly degraded by DOP hydrolyzing enzymes such as APA (Berman, 1969) and 5'-nucleotidase (Am-15 merman and Azam, 1985). Both enzymes are essential for PO 4 uptake by osmotrophs from organic compounds when the phosphorus demand is in excess compared to the available phosphate pool. Specific glucose affinity between the three mesocosms varied similarly with a range of 5.8×10 −5 to 5.4×10 −4 L nmol-P −1 h −1 (Fig. 6). It increased toward Day 9 and gradu-20 ally decreased onward. The differences in temporal variation between turnover time and specific affinity were due to the large temporal variations in bacterial biomass ( Fig. 4; Paulino et al., 2007). Linear regression analysis showed a significant negative relationship between specific glucose affinity and DOC concentrations (r 2 =0.390, P<0.0006, n=26). This suggests that the temporal changes in DOC concentration of 25 this study largely related to those of the labile DOC fraction (Fig. 7). This may be explained by DOC production through DIC overconsumption by phytoplankton (Riebesell et al., 2007). In order to examine substrate availability for osmotrophs, the experimentally determined specific affinity (α e ) can be compared with the theoretical maximum EGU (α max ). A theoretical expression for the maximum specific affinity for a spherical cell of radius r can be derived (Thingstad and Lignell, 1997): where D is the diffusion constant for the substrate molecules, and σ is the volume specific constant of the element in question. If we assume that bacterial cells have 5 a density of 1.2 g cm −3 , 50% dry weight of wet weight, 50% carbon of dry weight, a C:P ratio of 50 (Fagerbakke et al., 1996), and a cell radius of 0.25 µm and that the diffusion constant for glucose is 6×10 −6 cm 2 s −1 (e.g. Koch, 1971), the maximum specific affinity given by diffusion limitation of substrate transport is calculated to be 0.207 L nmol-P −1 h −1 . The theoretical maximum is 3-4 orders of magnitude higher 10 than the values calculated in this study (in the order of 10 −5 to 10 −4 L nmol-P −1 h −1 , Fig. 6). This suggests that bacterial growth was not limited by the availability of labile DOC. Riebesell et al. (2007) report that enhanced DIC overconsumption was observed in higher pCO 2 levels toward the peak of phytoplankton bloom (Day 12), but the ratio of 15 POC to PON remained very close to the Redfield ratio at the three pCO 2 levels. While this deviation between DIC/NO 3 drawdown and POC/PON build-up was attributed to enhanced DOC production in higher pCO 2 levels, higher loss of organic C from the upper mixed layer in higher pCO 2 levels suggests enhanced export of POC including transparent exopolymer particles (TEP) originated from DOC (Riebesell et al., 2007). 20 A difference in cumulative primary production between the three pCO 2 levels became evident after the peak of phytoplankton bloom or during the nutrient depleted period (Egge et al., 2007 4 ). Whereas contribution of picophytoplankton to total phytoplankton biomass was relatively small in all mesocosms (Riebesell et al., 2007;Paulino et al., 2007), the difference was mainly due to enhanced organic 14 C production in the 25 0.2-1 µm fraction in higher pCO 2 levels, a fraction overlapping TEP size (Egge et al., 2007 4   . This size shift of the phytoplankton community should have reduced vertical sinking flux, but the loss of organic C from the upper mixed layer continued toward the end of the experiment and was consistently higher at higher pCO 2 levels (Riebesell et al., 2007). The results thus indicate that nutrient depletion (e.g. P deficient, Fig. 5) reduced the bacterial capacity to degrade organic C despite increased labile DOC at 5 higher pCO 2 levels, and may suggest that DOC or small C-rich organic particles, such as TEP, developed into sinking particles.
In summary, all parameters measured in the current study did not show statistically significant differences between the three mesocosms, and were thus apparently independent of different pCO 2 levels. In the accompanying studies, effects of increased 10 pCO 2 level were, if found, relatively small and appeared toward the end of the experiment (Allgaier et al., 2007 3 ;Egge et al., 2007 4 ;Paulino et al., 2007;Schulz et al., 2007 2 ). Although phosphate availability tended to be lower in the 3×CO 2 mesocosm compared to the mesocosms toward the end of the experiment, no statistical differences were detected. This may be partly due to the lack of parallel measure-15 ments (one mesocosm for each pCO 2 level). However, such small differences could be explained by variance between "parallel treatments" rather than "different treatments" (Martínez-Martínez et al., 2006). These results suggest that availability of glucose and of phosphate availability was similar between three pCO 2 levels. This implies that the plankton food webs can buffer to a large extent the effects of increased pCO 2 and the 20 related changes in carbon chemistry changes during a short time frame (ca. 1 month).