Variation in elemental stoichiometry of the marine diatom Thalassiosira weissflogii (Bacillariophyceae) in response to combined nutrient stress and changes in carbonate chemistry

The combined consequences of the multi‐stressors of pH and nutrient availability upon the growth of a marine diatom were investigated. Thalassiosira weissflogii was grown in N‐ or P‐limited batch culture in sealed systems, with pH commencing at 8.2 (“extant” conditions) or 7.6 (“ocean acidification” [OA] conditions), and then pH was allowed to either drift with growth, or was held fixed. Results indicated that within the pH range tested, the stability of environmental pH rather than its value (i.e., OA vs. extant) fundamentally influenced biomass accumul‐ation and C:N:P stoichiometry. Despite large changes in total alkalinity in the fixed pH systems, final biomass production was consistently greater in these systems than that in drifting pH systems. In drift systems, pH increased to exceed pH 9.5, a level of alkalinity that was inhibitory to growth. No statis‐tically significant differences between pH treatments were measured for N:C, P:C or N:P ratios during nutrient‐replete growth, although the diatom expre‐ssed greater plasticity in P:C and N:P ratios than in N:C during this growth phase. During nutrient‐deplete conditions, the capacity for uncoupled carbon fixa‐tion at fixed pH was considerably greater than that measured in drift pH systems, leading to strong contrasts in C:N:P stoichiometry between these treatments. Whether environmental pH was stable or drifted directly influenced the extent of physiological stress. In contrast, few distinctions could be drawn between “extant” versus “OA” conditions for cell physiology.

The combined consequences of the multi-stressors of pH and nutrient availability upon the growth of a marine diatom were investigated. Thalassiosira weissflogii was grown in N-or P-limited batch culture in sealed systems, with pH commencing at 8.2 ("extant" conditions) or 7.6 ("ocean acidification" [OA] conditions), and then pH was allowed to either drift with growth, or was held fixed. Results indicated that within the pH range tested, the stability of environmental pH rather than its value (i.e., OA vs. extant) fundamentally influenced biomass accumulation and C:N:P stoichiometry. Despite large changes in total alkalinity in the fixed pH systems, final biomass production was consistently greater in these systems than that in drifting pH systems. In drift systems, pH increased to exceed pH 9.5, a level of alkalinity that was inhibitory to growth. No statistically significant differences between pH treatments were measured for N:C, P:C or N:P ratios during nutrient-replete growth, although the diatom expressed greater plasticity in P:C and N:P ratios than in N: C during this growth phase. During nutrient-deplete conditions, the capacity for uncoupled carbon fixation at fixed pH was considerably greater than that measured in drift pH systems, leading to strong contrasts in C:N:P stoichiometry between these treatments. Whether environmental pH was stable or drifted directly influenced the extent of physiological stress. In contrast, few distinctions could be drawn between "extant" versus "OA" conditions for cell physiology.
Key index words: acclimation; alkalinity; diatom; growth rate; nutrient; pH Abbreviations: AD, acid drift; AF, acid fixed; ASW, artificial seawater; DIC, dissolved inorganic carbon; ED, extant drift; EF, extant fixed; ES, enrichment solutions; OA, ocean acidification; P/DOM, particu-late/dissolved organic matter; POC, particulate organic carbon; TA, total alkalinity The dissolution of anthropogenically generated carbon dioxide within the surface ocean has led to a decrease in pH since pre-industrial times (Orr et al. 2005), a process termed ocean acidification (OA). How marine phytoplankton communities respond to OA is the subject of intense research activity as phytoplankton form the base of food webs and are a major driver of marine biogeochemical cycles. An area which has received relatively little attention is how OA-related changes in phytoplankton growth are superimposed upon the natural process of seawater alkalization that occurs concurrently with primary production.
During rapid growth, dissolved inorganic carbon (DIC) is consumed by phytoplankton more rapidly than gas exchange occurs between the seawater and atmospheric CO 2 . This imbalance depends on the physical environment (notably wind speed) and the availability of nutrients and light that constrains primary production. As a consequence, during algal bloom development, pH increases and can ultimately be inhibitory to phytoplankton growth (Hinga 2002). Values of pH approaching 10 have been reported in enclosed coastal eutrophic bays and fjords, while in the North Sea, values approaching 9 have been reported (Brussaard et al. 1996, Macedo et al. 2001, Hansen 2002. Changes in nutrient availability and pH during these events are expected to exert a selective pressure (Flynn et al. 2012) and the cells' ability to acclimate using short-term physiological mechanisms will represent a competitive advantage. It is unknown how OA conditions will influence this acclimation process and what the implications will be for cell composition and in turn, trophic interactions and C-export. The important issue here is that the start point of these events will be at a lower pH in an acidified ocean (and indeed is already different from that under historic, pre-industrial conditions ;Flynn et al. 2012), thus scope for biomass acclimation to the upper pH extremes will be extended.
Once transported into the cell, nutrients are used for the synthesis of structural and functional biomolecules in a way that reflects cellular demand and physiological status. Newly synthesized biomolecules either contribute to cell structure, are retained within the dissolved phase of the cell, or are released as dissolved organic matter. Variability in cellular partitioning of carbon, nitrogen, and phosphorous by phytoplankton is therefore an indication of metabolic flexibility and is reflected in particulate carbon, nitrogen, and phosphorous stoichiometry; C:N:P. Such partitioning, allied to underlying rate processes, has implications beyond cell composition; while such changes may influence C-export flux by modifying cell ballast, the quality of particulate and dissolved organic matter (POM and DOM, respectively) affects heterotrophic activity associated with grazing and nutrient regeneration (Cebrian et al. 1998, Urabe et al. 2003, Mitra and Flynn 2005, Rossoll et al. 2012. Of the phytoplankton, the diatoms are perhaps most synonymous with primary and export production (Mann 1999). Diatoms are also the only major group of autotrophic protists that are not capable of operating mixotrophy through coupled photoautoand phagohetero-trophic activities (Flynn et al. 2013). Although capable like all protists of using dissolved organics, diatoms are more reliant upon the dissolved inorganic nutrient pool to meet cell nutrient requirements, with silicate and/or nitrate limitation of bloom development characteristically cited (Lochte et al. 1993, Sieracki et al. 1993, Leblanc et al. 2009). Developing knowledge of the relationship between fundamental cellular characteristics such as growth rate and nutrient-limited variability in C:N:P stoichiometry is clearly required for model development, including how these may be modified by OA.
The objective of this study was to investigate nutrient-limited growth dynamics of the marine diatom Thalassiosira weissflogii under different pH regimes and to explore how growth conditions influenced cell composition. Sealed culture containers were used into which only acid or base could be added (to maintain fixed pH, as/if required) and from which samples could be withdrawn. The purpose of this approach was to enable a mass balance for C, N, and P. While the addition of acid or base cannot be representative of natural events (not least due to concurrent changes in total alkalinity; TA) in the context of our experimental design, it coincidentally enabled an investigation of the effects of TA on cell physiology.
A total of 12 different pH and nutrient regimes were used for experimentation, with all treatments duplicated (Table 1). The terms N-limited and P-limited are used hereafter to indicate cultures that initially contained nutrient N:P supply ratios which were respectively lower or higher than Redfield ratios and hence were considered conducive to either N or P deprivation. Cylindrical experimental flasks (10 L) were filled with 9 L of culture, of which 3 L was removed for experimental sampling over the time course of the experiment. Culture media was based on artificial seawater (ASW) (Harrison et al. 1980) with a bicarbonate concentration of 2.0 mmol Á L À1 and no organic buffer. The modifications and protocols of Berges et al. (2001) and Andersen et al. (2005) were used. De-ionized and purified water (ELGA Purelab Ultra, VWS UK Ltd., High Wycombe, UK) and chemicals of analytical grade were used for media preparation. Following this protocol, the pH of complete ASW was 8.2.
Under aseptic conditions ASW was filter-sterilized by pumping it into autoclaved experimental flasks through sterile 47 mm diameter, 0.2 lm filters (Durapore; Merck Millipore, Billerica, MA, US). Enrichment Solutions (ES) for vitamins, iron, and trace metals of 10% of the original formula (Andersen et al. 2005) were added to decrease the amount of DOM added as ethylenediaminetetraacetic acid into the medium. Dissolved inorganic nitrogen (as nitrate or ammonium) and phosphate were supplied in different initial concentrations (Table 1). Silicate was added in access at 134 lmol Á L À1 . Starter cultures for experiments were grown with 50 lmol Á L À1 nitrate, 5 lmol Á L À1 phosphate and 134 lmol Á L À1 silicate. As the proportion of macro-nutrients to ES components was lower in the final medium than in the original recipe (Harrison et al. 1980), ES components were inferred to be non-limiting. TABLE 1. Summary of experiment configurations. Initial nutrient concentrations and the intended limiting nutrient are indicated. The asterisk indicates that by the end of the experiment, the nutrient was below the detection limit of nutrient analysis; the incidence of potential secondary N nutrient limitation is therefore indicated. Two-point calibrated pH probes (pH 7.01 and 10.01; HI 1131b; Hanna Instruments, Woonsocket, RI, USA) were surface sterilized with 2 mol Á L À1 HCl for 15 min before being inserted into experimental vessels under aseptic conditions. Experimental vessels were sealed gas-tight with sterilized silicon bungs through which passed the pH probe lead, a sampling tube, two further tubes which facilitated pH correction, and a tube for the addition of N 2 gas to compensate for changes in volume with sample removal.
A bespoke software-based pH monitoring and control unit was constructed at Plymouth Marine Laboratory (UK) for use with these experiments. Using this system, cultures were independently monitored and, as/if required, the pH controlled. The unit was operated in two modes; a "fixed" mode with pH maintained at a set value throughout the experiment, and a "drift" mode in which pH was monitored as it varied from a set starting pH due to phytoplankton growth (increasing with net C-fixation, and decreasing with net respiration during the dark phase). The values at which pH was either fixed, or from which it drifted, were pH 8.2 representing extant conditions and pH 7.6 representing an acidified ocean. Both temperature (measured in tap water-filled vessels placed adjacent to experimental flasks to avoid the potentially toxic effects of stainless steel probes [HI 7669/2W; Hanna Instruments]) and culture pH were measured at a time step of 70 s.
For cultures at fixed pH, a tolerance of 0.05 pH units was assigned and five consecutive measurements beyond this tolerance resulted in pH correction; metering pumps added either 1 mL of HCl or NaOH to individual cultures. During the growth cycle in these closed system cultures, DIC concentration and thus the buffering capacity of the media decreased. For cultures grown at fixed pH, the concentration of acid and base added to cultures was decreased as culture age increased to minimize the amplitude of pH change during the automated correction procedure. Initial concentrations of 1.2 and 0.25 lmol Á L À1 for acid and base respectively, were used for the acclimation stage and first experimental day, decreasing to 0.12 and 0.125 lmol Á L À1 at day 2 and 0.05 and 0.0625 lmol Á L À1 at day 4. A log of temperature, pH, and acid/base addition for each experimental vessel was recorded by the pH control unit. Changes in TA were calculated from changes in nutrient (including DIC) and pH by reference to the methods described in Flynn et al. (2012).
Four different pH treatments were run simultaneously; these commenced at two different pH values, with the pH then being fixed over the duration of sampling, or allowed to drift. All four pH treatments of one experiment were started with a 5% (v/v) inoculation from the same starter culture resulting in initial cell density of 1,000-3,000 cells Á mL À1 . Inoculated experimental containers were placed on magnetic stirrers within the constant temperature room, connected to the pH control unit, and kept for 14-16 h for temperature and light acclimatization. During this stage, pH was maintained at a constant level of either 8.2 or 7.6 as required. The pH control was then switched to "drifting" or "fixed" modes to initiate experiments. From here on, culture conditions will be referred to using the following nomenclature:extant pH drift -ED; extant pH fixed -EF; acid (i.e., OA) drift -AD; acid fixed -AF.
Sample collection. Samples were collected from vessels on a daily basis for days 0-6, and then at days 8 and 10. Samples were drawn from a polytetrafluoroethylene (PTFE) tube, which was permanently submerged in culture media. The first 10 mL of media drawn was discarded to flush the sampling tubing, and the subsequent 250-450 mL retained. N 2 gas was introduced to the atmosphere above cultures to replace the volume removed by sampling. N 2 was stored in a balloon and passed through a 0.2 lm syringe filter (Minisart plus, 28 mm, hydrophilic, Sartorius, G€ ottingen, Germany) before entering vessels.
Examination of cell numbers and biovolume. Cell numbers and biovolume were measured on sub-samples previously passed through a 200 lm sieve using a FlowCam equipped with 90 lm field of view flow cell and a 910 objective. Image data processing software "Visual Spreadsheet" was used (version 2.1.20 beta, Fluid Imaging Technologies, Yarmouth, ME, USA). Analysis of particles of the size range of 5-30 lm was performed with a picture rate of 6 s À1 in auto-image mode. Typically, 500 images identified as T. weissflogii were counted, although in low density cultures measurements were stopped after 4 min to avoid sedimentation effects during the measurement. Biovolume was estimated using the ABD Flow-Cam algorithm, which equates to a volume estimated from equivalent spherical diameter. Data represent the median of at least three replicate measurements.
Analysis of inorganic nutrients. Under clean conditions (within a filtered air stream in a laminar flow cabinet), subsamples of culture were gravity filtered through pre-combusted (550°C, 20 min) glass fiber filters with a nominal pore size of 0.2 lm (Type A/E, 13 mm, Pall Corporation, Ann Arbor, MI, USA). Filters with retained cells were stored frozen for subsequent elemental and compositional analysis. Filtrates were stored frozen (À20°C) for subsequent inorganic nutrient analysis.
A segmented flow auto-analyzer (model AA3; Seal Analytical Ltd., Fareham, UK) was used for inorganic nutrient analysis, which consisted of a random access XY-2 auto-sampler, a high precision pump, chemistry module trays, and a dualbeam high-resolution digital colorimeter operating within the range 340-900 nm. Nitrate and nitrite were determined by the cadmium reduction method and sudan-1 synthesis, which was measured at 550 nm. The Berthelot reaction was used for the determination of ammonium, with indophenol measured at 660 nm. The limit of detection for inorganic nitrogen was 0.5 lmol Á L À1 . The analysis of inorganic and total phosphorous was based on the method of Murphy and Riley (1962), with the blue phosphor-molybdenum complex measured at 880 nm. The limit of detection for inorganic phosphate was 0.25 lmol Á L À1 .The determination of DIC was based on an inverse chemistry with the use of buffered phenolphthalein indicator, with measurement at 550 nm; this method was an adaptation of that presented by Stoll et al. (2001), with a detection limit of 25 lmol Á L À1 .
Analysis of particulate C, N, and P. Particulate C and N content was analyzed in duplicate for each growth cycle against isoleucine standards using an elemental analyzer and mass spectrometer (ANCO PDZ-Europa and 20-20 IRMS PDZ-Europa; SerCon Ltd Crewe, UK) combined with Callisto 530 software (SerCon). Both samples and standards were wrapped in tin foil. Helium gas served as system carrier and auto-sampler purge. The combustion column (packed with quartz wool/chromium oxide/silver wool) operated at 1,000°C, with sample combustion taking place in the presence of an O 2 pulse to improve combustion efficiency. Combustion gases were reduced in a reduction column composed of silica wool/copper wire at 600°C. Sample-derived CO 2 and N 2 gases were separated by a gas chromatography column at 70°C and quantified by mass spectrometry.
Particulate P content was derived in duplicate for each growth cycle by alkaline persulphate oxidation followed by analysis of the solubilized phosphate. Algal samples retained on glass fiber filters were heated for 20 min at 121°C (autoclave) in sealed glass ampoules containing 2 mL of a solution containing potassium peroxodisulphate (50 g Á L À1 ), boric acid (30 g Á L À1 ), and sodium hydroxide (15 g Á L À1 ). After 642 DARREN R. CLARK ET AL.
shaking, 1.5 mL of the supernatant was withdrawn into a micro-centrifuge tube, the filter remains removed by centrifugation (20,000 g for 5 min) and 1 mL of the supernatant then stored frozen (À20°C) until analysis. The inorganic P thus produced was analyzed as phosphate by segmented flow auto-analyzer after dilution using purified water (ELGA Purelab ultra, VWS UK Ltd.).

RESULTS
An overview of the experimental design for this study is presented in Table 1. Figure 1 presents an example of the temporal changes in culture pH, cell density, and nutrient concentration, in this instance for nitrate-limited T. weissflogii. For fixed pH cultures, automated pH control maintained values (AE1 SD) of 7.60 AE 0.05 for simulations of OA (AF) and 8.19 AE 0.04 for simulations of extant conditions (EF). For drift cultures in which pH changes due to phytoplankton growth were allowed to take place, pH increased from initial values of 7.6 (AD) and 8.2 (ED) to final values of~9.5. A comparable value was reached for all 12 cultures (i.e. those limited by nitrate, ammonium, and phosphate) in which pH was allowed to drift. However, since OA treatments started from a lower pH value, the magnitude, and coincidentally the rate of pH increase was greatest in these compared to extant pH drift treatments. Nitrate-limited cultures (Fig. 1) became nitrate deplete within 3-4 d (noting that the cells had been acclimated to growth conditions for 14-16 h prior to commencement of sampling to minimize lag phase duration) and cells entered stationary phase within 7 d. Associated changes in carbonate chemistry are presented in Figure 2 and demonstrate strong contrasts between fixed pH and drift pH systems. While both fixed pH and drift pH cultures were subject to changes in TA linked to nutrient utilization, changes in TA were far greater for fixed pH cultures (Fig. 2). It may therefore be concluded that pH (i.e., H + concentration) and not TA was the major factor affecting growth, other than nutrient exhaustion.
Culture biovolume was correlated with particulate organic carbon (POC) concentration (r 2 = 0.82 for data derived from 24 batch culture cycles i.e., 12 conditions duplicated; Fig. 3) and used to derive carbon-specific growth rate (Cl). The ratio between POC and culture biovolume (AE1 SD) was similar in all treatments during nutrient-replete growth at 349 AE 109 g C Á L À1 , but there was increasing scatter in nutrient-deplete cultures, with the ratio decreasing to 319 AE 86 g C Á L À1 ; a value of 327 AE 93 g C Á L À1 was derived for all data. Most of the increase in culture biovolume (and carbon) occurred following inorganic nutrient depletion, with higher biovolumes consistently achieved under both conditions of fixed pH (AF and EF).
Changes in particulate elemental stoichiometry with carbon-specific growth rate (Cl) are presented in Figure 4. Throughout the 24 culture experiments in this study, the N:C, P:C, and N:P ratios of exponentially growing cells were, as means with their coefficients of variation, 0.1043 AE 9.2%, 0.0089 AE 34.3%, and 12.32 AE 24.9%, respectively (Fig. 5). This indicates that T. weissflogiii maintained a very tight control of the relative proportions of cellular N to C between pH treatments during nutrient-replete growth, but expressed higher variability in P:C and N: P ratios. N:C ratio variability increased and was comparable to that of P:C and N:P ratios during nutrient deficient growth (Fig. 4), with values of 0.0701 AE 46.2%, 0.0066 AE 24.0%, and 9.35 AE 37.6% respectively.
A decrease in particulate N:C mass ratio with decreased Cl may not be expected for P-limited cultures. However, secondary N nutrient depletion took place (Table 1). For those P-limited cultures that also exhausted nutrient-N, N:C decreased at fixed pH, but increased under conditions of drifting pH. These results suggested that T. weissflogii had a significant capacity for uncoupled carbon uptake, which was most readily expressed under conditions of fixed pH, with elevated pH being detrimental to growth.
General trends in particulate P:C ratios with Cl were evident and comparable to N:C results, although less well defined. Thus, for N-limited and P-limited cultures under conditions of drifting pH, particulate P:C ratios were little changed from exponential growth values as cultures entered nutrient depletion. In contrast, P:C ratios decreased under nutrient depletion at fixed pH for N-limited and P-limited cultures. The N:P ratio decreased as cultures entered N-limited growth, but was relatively invariant during P-limitation (Fig. 4).
Changes in cell volume in relation to Cl or N:C are presented in Figures 6 and 7, respectively. As a general feature, cell volumes (AE1 SD) were at their greatest (1428 AE 285 lm 3 ) during nutrient sufficient exponential growth. Nutrient limitation was associated with both a decrease in Cl and cell volume. The associated changes in N:C with cell volume of N-limited cultures indicated a clear distinction between fixed pH and drift pH systems (Fig. 7); as N:C decreased in response to N-depletion, cell volumes decreased to a greater extent under the extreme pH conditions reached in drift pH systems (values of pH 9.5). For fixed pH systems, the decrease in cell volume was less significant while the greater capacity for uncoupled C-uptake resulted in very low N:C ratios. Consequently, N-limited diatoms grown under fixed pH conditions were distinguished from those grown under drift pH conditions on the basis of both particulate N:C ratio and cell volume. This implied, especially since drift pH cultures did not attain the same biomass levels, that high pH (i.e., low [H + ]) imparted a physiological stress associated with an  Mehrbach et al. (1973), as refitted by Dickson and Millero (1987), in combination with known values of DIC, pH, temperature, salinity, and inorganic nutrient concentrations. Data represent the average of duplicate growth cycles. TA, total alkalinity. Culture pH conditions are indicated thus: EF, extant fixed; ED, extant drift; AF, acidic fixed; AD, acidic drift; DIC, dissolved inorganic carbon. increase in surface area:volume ratio. This general trend for N-limited cells appeared unrelated to the oxidation state of the N-source (ammonium vs. nitrate). Trends in particulate P:C and N:P ratios with cell volume for N-limited cultures were comparable to N:C data, although less distinct (results not shown).
For P-limited cultures, the decrease in cell volume and Cl following phosphate depletion (Fig. 6) was also associated with pH treatment-specific changes in N:C (Fig. 7), which may have been related to the depletion of N as previously noted. Changes in P:C and N:P ratio with cell volume were not clearly distinguished (results not shown).

DISCUSSION
Overview. There are two substantive conclusions to be drawn from the results presented in this work. First, drifting pH commencing at extant or OA conditions appeared to be far more detrimental to diatom growth than fixed pH commencing at the same pH values. Thus, biomass accumulation was always greatest at fixed pH (EF or AF), even if OA conditions were simulated, and in consequence, these cultures also attained higher C:(N:P) ratios. Secondly, and of value from a pragmatic point of view for experiments, changes in TA over a very wide range (>1,500 lmol Á Kg À1 ) did not appear to be detrimental to growth. Although the experiments were not explicitly designed to test for this eventuality, the fixed pH systems in which the diatom grew extremely well experienced a near exhaustion of DIC, associated with a more than halving of TA (Fig. 2).
DIC versus pH stress. Using a carbon draw-down technique, Clark and Flynn (2000) demonstrated that a range of phytoplankton species (including the diatom strain used here) were capable of maintaining high growth rates across a broad DIC range at fixed pH, suggesting that phytoplankton were relatively insensitive to changes in DIC (or pCO 2 ) and that pH rather than carbonate chemistry may have a more significant role to play in moderating growth. Results presented here develop this earlier work, suggesting that initial pH per se, within realistic bounds, was not the most important factor influencing growth, nutrient acquisition or particulate stoichiometry.
At the level of the whole cell, relatively few distinctions could be drawn between acid and extant conditions when pH was fixed or when it was drifting, FIG. 4. Changes in particulate carbon, nitrogen, and phosphorous stoichiometry (mass ratios) with carbon-specific growth rate (Cl) during nutrient-replete (filled symbols) or -deplete (open symbols) conditions. Only primary (not secondary) nutrient depletion is considered. Dashed lines for N:C and N:P indicate Redfield mass ratio values. All P:C data were considerably below the N:P Redfield mass ratio value of 0.0243 and is therefore not shown. Culture pH conditions are indicated thus: EF, extant fixed; ED, extant drift; AF, acidic fixed; AD, acidic drift.

NUTRIENT AND PH STRESS IN A DIATOM
implying that any cellular acclimations in response to OA conditions, such as subtle shifts in biochemical composition, were sufficient to mitigate any potential impacts of [H + ]. In contrast, strong distinctions could be drawn between cells grown at fixed versus drifting pH. Consequently, T. weisfloggii could mitigate against OA-induced impacts upon growth but that changes in pH that take place during natural blooms may present a greater challenge. Indeed, since elevated pH developing in dense blooms can exert selective pressures and cause bloom termination (Hinga 2002), it could be argued that the more important difference between OA and extant conditions is the shift in the pH envelope experienced during bloom progression (i.e., starting at a lower pH under OA conditions). This event also interacts with nutrient loading on two counts: (i) the amount of nutrient available affects the biomass development and hence pH increase and (ii) we speculate that the nutrient status of the cells likely affects their ability to survive adverse pH conditions. From an historic perspective and particularly relevant to studies of phytoplankton that are adversely affected by aeration (such as dinoflagellates), reports of physiological responses in the literature likely have invoked some level of pH drift and its associated stresses. Experiments under drift pH conditions, commencing at different pH values, are thus of interest, and arguably, at least as important as experiments run under fixed pH for our understanding of the implications of OA.
Experimental approach within the context of previous OA studies. The dissolution of anthropogenic CO 2 in the surface ocean modifies both the speciation of DIC and seawater pH; these aspects are related and inseparable. This complexity challenges the development of robust experimental approaches to OA simulations, with generic examples of clear physiological responses in phytoplankton remaining elusive (Hutchins et al. 2009, Liu et al. 2010. In their review of experimental techniques, Hurd et al. (2009) contrasted the use of CO 2 bubbling with HCl/NaOH additions (the latter adopted here), concluding that there is a need to undertake experiments using both approaches in order to account for differences due to the method of carbonate system manipulation. This duel approach was adopted by Piontek et al. (2010) who reported that microbial polysaccharide degradation in natural cultures was not influenced by the method of carbonate system manipulation. Results between laboratory experiments that use different methods of carbonate system manipulation may therefore be broadly comparable. The more important issue is that of operating fixed or drift pH systems; the former may be considered (in relative terms) more analogous to oceanic or shelf sea systems during periods of low productivity, which do not drive a significant change in pH, while drift pH systems may be considered analogous to shelf sea and coastal/eutrophic systems during periods of high productivity. We view these treatments as independently representing the opposing ends of a dynamic continuum in terms of seawater pH; the rate and extent of phytoplankton biomass accumulation (amongst other factors) dictating the extent to which seawater pH changes subsequently exert a selective pressure.
A feature of our approach was the use of closed systems. This was originally intended to allow system C, N, and P mass balance, but unfortunately, this was not achievable because cumulative errors in analysis (notably of DOM; results not presented) generated excessive variability in the data. Consequently, we were unable to assess the fractionation of carbon, nitrogen, and phosphorous between dissolved inorganic, dissolved organic and particulate organic phases as a consequence of diatom growth, including any assessment of CO 2 outgassing to the culture headspace. However, the unanticipated consequence was the demonstration that T. weissflogii was capable of maintaining growth over a very broad TA range. It was anticipated that in closed systems with limiting amounts of other nutrients, the diatom would draw DIC down as previously observed (Clark 2000), but that this would not result in C-limitation of growth (Clark and Flynn 2000). CO 2 removal during diatom growth would result in a re-equilibration of DIC species as pH changed, with a decrease in the proportion of HCO 3 À , and an increase in CO 3 2À . This process drove pH higher in drift pH cultures (an upper limit of~pH 9.5 was attained), but would be countered by the addition of HCl in fixed pH cultures. Consequently, while all cultures started and subsequently decreased from the same initial [DIC] (and [CO 2 (aq) ]) values, as culture growth progressed the fraction of DIC available as CO 2 would diminish further as pH increased in drift pH cultures. In fixed pH cultures, this fraction remained constant, although the actual concentration of forms of DIC available for the support of photosynthesis (i.e., CO 2 (aq) and HCO 3 À ) fell in all instances. Figure 2 presents derived changes in carbonate system parameters in cultures of T. weissflogii under contrasting pH conditions, demonstrating that the diatom was capable of biomass accumulation over a wide range of TA and CO 2(aq) availability. This range was greater than reported to occur in natural systems (Hinga 2002 and references therein), where pH and significant [DIC] drawdown have been observed in dense blooms under calm conditions (where C-fixation would greatly exceed gas exchange between the sea and atmosphere), and in estuarine systems of decreased salinity (and buffering potential). These results are consistent with previous laboratory investigations which indicate that phytoplankton are FIG. 7. Changes in cell volume in relation to particulate N:C stoichiometry (mass ratios) during nutrient-replete (filled symbols) or -deplete (open symbols) conditions. Only primary (not secondary) nutrient depletion is considered. Dashed lines indicate Redfield mass ratio values. Culture pH conditions are indicated thus: EF, extant fixed; ED, extant drift; AF, acidic fixed; AD, acidic drift.
NUTRIENT AND PH STRESS IN A DIATOM capable of growth over a wide pH and CO 2 range (Tortell et al. 1997, 2000, Ishida et al. 2000, Hinga 2002, Herv e et al. 2012. Consequently, it may not be surprising that many studies of marine diatoms have failed to demonstrate an OA-related impact upon growth rate when pH is fixed (Crawfurd et al. 2011 and references therein).
The greatest impact of pH in our study was related to its stability, or specifically to whether significant alkalinization occurred. Biomass accumulation continued beyond the point at which N or P was exhausted from the medium as N-limited and P-limited cultures, respectively, entered uncoupled growth. However, the extent to which uncoupled growth continued was significantly greater in fixed pH cultures. Under drift pH conditions, commencing from extant or acidic pH, stationary phase [DIC] (n = 18 for each pH treatment AE1 SD) was 889 AE 149 and 915 AE 135 lmol Á L À1 respectively. In contrast, at fixed extant or acidic pH, stationary phase [DIC] was 113 AE 76 and 77 AE 66 lmol Á L À1 , respectively, below the half saturation constant for carbon-specific growth of 135-258 lmol Á L À1 (Clark and Flynn 2000). Results from fixed pH experiments demonstrated that T. weissflogii had the potential to decrease [DIC] further under conditions of fixed pH than drift pH; the implication is that processes influenced by extracellular [H + ] affected cell physiology and constrained nutrient utilization. The nature of environmental pH fundamentally influenced the cells' ability to accumulate biomass.
The activity of extracellular enzymes and transmembrane transport processes are influenced by pH and contribute to the differences observed between drift pH and fixed pH systems. Herv e et al. (2012) demonstrated that changes in external pH affected the metabolic state of T. weisfloggii cells due to intracellular acid-base balance regulation; an increase in extracellular pH leads to increased intracellular pH with implications for fundamental processes such as cell cycle progression.
Adding complexity to the interpretation of the results is that phytoplankton experience an environmental pH, which deviates from that of the bulk water (Flynn et al. 2012). The pH of the boundary layer surrounding phytoplankton cells is modified in a way that relates to cell size (or colony aggregate size) and the organisms metabolic activity, with changes being greater for larger cells (poorer diffusion rates), greater in faster growing cells (dis-equilbria between fluxes) and greater when commencing from a lower pH (due to a lower buffering capacity of seawater under OA conditions). The combination of bulk water and boundary layer pH changes may have conspired to affect cell physiology over lightdark cycles during these experiments. Consequently, the diatom's tolerance of broad changes in carbonate system parameters may represent a competitive and selective advantage under conditions of prolific growth, although alkalization of the growth medium ultimately appeared to be a self-limiting process, which may constrain bloom development, and contribute to its collapse. The strong contrast in the diatoms capacity for uncoupled growth between drifting and fixed pH conditions also implicates changes in environmental pH with those of cellular C:N:P stoichiometry.
Diatoms in the natural world and their role in CNP biogeochemistry. Diatoms play a key role in global biogeochemical cycles. They frequently dominate the early stages of seasonal primary production, characteristically achieving high growth rates in nutrient-replete environments (Martin et al. 2011, Boyd et al. 2012). Nutrient limitation constrains diatom productivity and the sedimentation of diatom cells, enhanced by relatively dense siliceous frustules, which continue to build following non-Si limited nutrient exhaustion (Flynn and Martin-J ez equel 2000), represents an important mechanism for carbon, nitrogen, and phosphorous cycling and export (Takahashi et al. 2000, Klass andArcher 2002). Estimations of carbon export flux can be informed by consideration of particulate C:N:P ratios.
The role of stoichiometric relationships in constraining nutrient cycles is well established and related to structural and functional requirements for carbon, nitrogen, and phosphorous (Vrede et al. 2004). However, plasticity in the C:N:P ratio has implications for biogeochemical cycles and trophic dynamics. For example, increases in carbon relative to nitrogen content of N-limited cells may be viewed as a potential enhancement of carbon export via the biological pump, but also equates to a deterioration in food quality for heterotrophic grazers, lowering their growth efficiency, grazing pressure, secondary productivity, and egg production (Cebrian et al. 1998, Jones et al. 2002, Urabe et al. 2003, Mitra and Flynn 2005, Mitra et al. 2007, Rossoll et al. 2012. Furthermore, the rate of detrital decomposition has been linked to the particulate C:N:P ratio (Enriquez et al. 1993), with implications for particulate matter remineralization as a function of depth. Consequently, it is recognized that the use of fixed Redfield stoichiometry in biogeochemical models (Moore and Doney 2007), although convenient and computationally cheap, is insufficient (Flynn 2010a); mechanisms which influence diatom growth rate, biomass accumulation, and particulate C:N:P variability need to be understood.
Elemental requirements for cellular machinery impose constraints on composition and functional relationships to fundamental processes such as growth rate (Vrede et al. 2004, Flynn 2010b), although support for generalized trends is sparse, especially between species (Flynn 2010a). Using t-tests or Mann-Whitney Rank Sum Tests in cases were Shapiro-Wilk normality tests failed, no statistically significant differences were found between the mean N:C, P:C, or N:P ratios of nutrient-replete cells grown under the various pH treatments (Fig. 5,   648 noting that this figure contains additional data to those presented in Fig. 3 because a growth rate for day 0 cannot be derived). However, considerably more plasticity in P:C and N:P ratios is evident than for N:C, suggesting that tight control of cellular carbon and nitrogen content was required to achieve maximum Cl. This is consistent with the form of the relationship between N:C and growth rate (linear) and that for P:C and growth rate (distinctly curvilinear ;Flynn 2008a,b).
Developing from this point, we aimed to investigate how the relationship between C:N:P and growth rate (Cl) changed with nutrient limitation and under differing pH conditions. This necessarily required that physiological stress was induced by our experimental approach and that its extent could be assessed; cell quotas were used for this purpose (Fig. 8). Although N and P became depleted from growth media in systems intended to be N-limited and P-limited, respectively, analysis of cellular N:C and P:C quotas in comparison to the form of the respective quota curves suggested that in fact effective P-stress never developed, while N-stress typically did develop. This state of affairs is a further reflection in the form of the relationship between N:C and Cl versus P:C and Cl (Flynn 2010a,b). In consequence of the pH-limited growth of drift cultures, these never developed such extreme stress (such extreme N:C) as did the fixed pH cultures. This was true for both the ED and AD (vs. EF and AF) series. The fact that the cells growing in the drift pH systems displayed physiological stress by being smaller is thus all the more interesting (Figs. 6 and 7).
Cell size may impact many aspects of a cells physiology including nutrient transport and accumulation capacity, buoyancy, and grazing susceptibility (Legendre and Rivkin 2002, L opez-Urrutia et al. 2006. As a fundamental physiological parameter, cell size links biomass formation with elemental transfer to the deep-ocean or higher trophic levels and may moderate global biogeochemical cycles. Quantitative relationships between cell size and physiological/ecological function have been developed for model applications . In marine diatoms, cell size for individual cell-lines is often expected to progressively decrease during asexual reproduction, until sexual reproduction restores full-size cell dimensions. Results demonstrated that cell size varied over a relatively broad range during nutrient sufficient growth, while near maximum Cl values were maintained. However, with N and P depletion, both cell size and Cl decreased, although, as noted above, this was likely to reflect N, or N secondary-nutrient limitation and associated physiological stress. By decreasing cell size in response to nutrient limitation, diatom cells NUTRIENT AND PH STRESS IN A DIATOM increased surface area:volume ratio enhancing nutrient transport capacity under low nutrient conditions. Furthermore, the range in boundary layer pH changes associated with photosynthetic carbon uptake, respiration, and nutrient transport over light-dark cycles will decrease with decreased cell size (Flynn et al. 2012). This modification to cell physiology will effectively increase the stability of environmental pH experienced by the cell, an acclimation that is conducive to enhanced nutrient utilization and biomass accumulation, as demonstrated here.
Concluding remarks. Within the constraints of the pH conditions tested, pH stability or the extent of alkalization fundamentally influenced diatom cell physiology and C:N:P stoichiometry. The implications of simulated OA versus extant conditions (i.e., AD vs. ED, or AF vs. EF) for cell physiology could not be distinguished. Although the ability to tolerate changes in pH (and TA) may be a competitive and selective advantage during bloom development, the alkalization of seawater is likely to have a direct role in competitive interactions between phytoplankton species and growth termination. The role of pH stability or alkalization upon cell physiology and particulate stoichiometry may need to be considered when interpreting community level responses to simulated OA conditions, notably for mesocosm studies in which alkalization has been reported (Bellerby et al. 2008, Schulz et al. 2013). This work was supported by the UK Natural Environment Research Council (award NE/F002564/1) to KJF and DRC.