Individual and interactive effects of warming and CO 2 on Pseudo-nitzschia subcurvata and 1 Phaeocystis antarctica , two dominant phytoplankton from the Ross Sea , Antarctica 2

7 Abstract: We investigated the effects of temperature and CO2 variation on the growth and 8 elemental composition of cultures of the diatom Pseudo-nitzschia subcurvata and the 9 prymnesiophyte Phaeocystis antarctica, two ecologically dominant phytoplankton species 10 isolated from the Ross Sea, Antarctica. To obtain thermal functional response curves, cultures 11 were grown across a range of temperatures from 0C to 14C. In addition, a competition 12 experiment examined the relative abundance of both species at 0C and 6C. CO2 functional 13 response curves were conducted from 100 to 1730 ppm at 2C and 8C to test for interactive 14 effects between the two variables. The growth of both phytoplankton was significantly affected 15 by temperature increase, but with different trends. Growth rates of P. subcurvata increased with 16 temperature from 0°C to maximum levels at 8°C, while the growth rates of P. antarctica only 17 increased from 0°C to 2°C. The maximum thermal limits of P. subcurvata and P. antarctica 18 where growth stopped completely were 14°C and 10°C, respectively. Although P. subcurvata 19 outcompeted P. antarctica at both temperatures in the competition experiment, this happened 20 much faster at 6°C than at 0°C. For P. subcurvata, there was a significant interactive effect in 21 which the warmer temperature decreased the CO2 half saturation constant for growth, but this 22 was not the case for P. antarctica. The growth rates of both species increased with CO2 increases 23 up 425 ppm, and in contrast to significant effects of temperature, the effects of CO2 increase on 24 their elemental composition were minimal. Our results suggest that future warming may be more 25 favorable to the diatom than to the prymnesiophyte, while CO2 increases may not be a major 26 factor in future competitive interactions between Pseudo-nitzschia subcurvata and Phaeocystis 27 antarctica in the Ross Sea. 28


Introduction
Global temperature is predicted to increase 2.6 to 4.8 • C by 2100 with increasing anthropogenic CO 2 emissions (IPCC, 2014).The temperature of the Southern Ocean has increased even faster than global average temperature (Meredith and King, 2005), and predicted future climate warming may profoundly change the ocean carbon cycle in this region (Sarmiento et al., 1998).The Ross Sea, Antarctica, is one of the most productive area in the ocean, and features annual austral spring and summer algal blooms dominated by Phaeocystis and diatoms that contribute as much as 30 % of total primary production in the Southern Ocean (Arrigo et al., 1999(Arrigo et al., , 2008;;Smith et al., 2000Smith et al., , 2014a)).The responses of phytoplankton in the Ross Sea to future temperature change (Rose et al., 2009;Xu et al., 2014;Zhu et al., 2016) in combination with intensified stratification (Sarmiento et al., 1998) could lead to intensified future diatom blooms (Smith et al., 2014b), and the physiological effects of warming may partially compensate for a lack of iron throughout much of this region (Hutchins and Boyd, 2016).
In the Ross Sea, the colonial prymnesiophyte Phaeocystis antarctica typically blooms in austral spring and early summer, and diatoms including Pseudo-nitzschia subcurvata and Chaetoceros spp.bloom later in the austral summer (Arrigo et al., 1999(Arrigo et al., , 2000;;DiTullio and Smith, 1996;Goffart et al., 2000;Rose et al., 2009).Both diatoms and P. antarctica play an important role in anthropogenic CO 2 drawdown and the global carbon cycle; additionally, they contribute significantly to the global silicon and sulfur cycles, respectively (Arrigo et al., 1999;Tréguer et al., 1995;Schoemann et al., 2005).Furthermore, the N : P and C : P ratios of P. antarc-Published by Copernicus Publications on behalf of the European Geosciences Union.
Z. Zhu et al.: Individual and interactive effects tica are higher than those of diatoms, and thus they contribute unequally to the carbon, nitrogen, and phosphorus cycles (Arrigo et al., 1999(Arrigo et al., , 2000)).Diatoms are preferred by many planktonic herbivores over P. antarctica, and so the two groups also differentially influence the food webs of the Southern Ocean (Knox, 1994;Caron et al., 2000;Haberman et al., 2003).Arrigo et al. (1999) suggested that the spatial and temporal distributions of P. antarctica and diatoms in the Ross Sea are determined by the mixed layer depth, while Liu and Smith (2012) indicated that temperature is more important in shaping the distribution of these two dominant groups of phytoplankton.Smith and Jones (2015) presented evidence for the importance of deep mixing and the critical depth for the timing of transitions from P. antarctica to diatom blooms.Zhu et al. (2016) observed that a 4 • C temperature increase promoted the growth rates of several dominant diatoms isolated from the Ross Sea, including P. subcurvata, Chaetoceros sp., and Fragilariopsis cylindrus, but not the growth rates of P. antarctica.In addition, both field and laboratory research has suggested that temperature increase and iron addition can synergistically promote the growth of Ross Sea diatoms (Rose et al., 2009;Zhu et al., 2016;Hutchins and Boyd, 2016).Thus, it is possible that phytoplankton community structure in this region may change in the future under a global warming scenario.
In addition to temperature increases, ocean uptake of 30 % of total emitted anthropogenic CO 2 has led to a 0.1 pH unit decrease in surface water, corresponding to a 26 % increase in acidity (IPCC, 2014).The global CO 2 concentration is predicted to increase to around 800 ppm by 2100, which will lead to a further decrease in surface seawater pH of 0.3-0.4 units (Orr et al., 2005;IPCC, 2014).CO 2 increases have been found to promote the growth and affect the physiology of many but not all phytoplankton species tested (Fu et al., 2007(Fu et al., , 2008;;King et al., 2011;Xu et al., 2014;Hutchins and Fu 2017).
Research on the effects of CO 2 increases on Phaeocystis antarctica and Antarctic diatoms is still scarce.Xu et al. (2014) suggested that future conditions (higher temperature, CO 2 , and irradiance) may shift phytoplankton community structure towards diatoms and away from P. antarctica in the Ross Sea.Trimborn et al. (2013) discovered that the growth rates of P. antarctica and P. subcurvata were not significantly promoted by high CO 2 relative to ambient CO 2 at 3 • C. In contrast, Wang et al. (2010) observed that the growth rates of the closely related temperate colonial species Phaeocystis globosa increased significantly at 750 ppm CO 2 relative to 380 ppm CO 2 .
Many studies have shown that primary production in various parts of the Southern Ocean is limited by iron supply (Martin et al., 1990;Takeda, 1998;Boyd et al., 2000;Sedwick et al., 2000;Hutchins et al., 2002;Coale et al., 2004), and several have addressed the effects of iron and warming on the growth of phytoplankton from the Ross Sea (Rose et al., 2009;Zhu et al., 2016;Hutchins and Boyd, 2016).Thus, an important goal of phytoplankton research is to also gain an understanding of how global warming together with ocean acidification may shift the phytoplankton community in the Ross Sea (Arrigo et al., 1999;DiTullio et al., 2000).This study aimed to explore the effects of increases in temperature and CO 2 availability, both individually and in combination, on P. antarctica and P. subcurvata isolated from the Ross Sea, Antarctica.These results may shed light on the potential effects of global change on the marine ecosystem and the cycles of carbon and nutrients in the highly productive coastal polynyas of Antarctica.
2 Materials and methods 2.1 Strains and growth conditions P. subcurvata and P. antarctica were isolated from the ice edge in McMurdo Sound (77.62 • S, 165.47 • E) in the Ross Sea, Antarctica, during January 2015; P. antarctica cultures grew as small colonies (∼ 4-12 cells) in all the experiments.All stock cultures were grown in Aquil medium (100 µmol L −1 NO − 3 , 100 µmol L −1 SiO 4− 4 , 10 µmol L −1 PO 3− 4 ) made with 0.2 µM-filtered seawater that was collected from the same Ross Sea locale as the culture isolates (Sunda et al., 2005).Stock and experimental cultures were grown in Fe-replete Aquil medium (0.5 µM).Although phytoplankton in the open Ross Sea polynya are generally proximately iron-limited (Ryan-Keogh et al., 2017), these culture conditions are relevant to the coastal McMurdo Sound ice edge environment in the early spring when Fe is relatively abundant, and typically not limiting.This "winter reserve" iron is then drawn down in this nearshore environment over the course of the seasonal algal bloom to eventually reach limiting levels (Sedwick et al., 2011;Bertrand et al., 2015).Our experiments address warming and acidification responses in P. subcurvata and P. antarctica in the absence of any differential effects of Fe availability; interactive effects of Fe limitation with warming and/or acidification in these two species are presented in Xu et al. (2014) and Zhu et al. (2016).Cultures were maintained at 0 • C in a walk-in incubator under 24 h cold white fluorescence light (80 µmol photons m −2 s −1 ).

Experimental design
For thermal functional response curves, experimental cultures of both phytoplankton were grown in triplicate 500 mL acid-washed polycarbonate bottles and gradually acclimated by a series of step-wise transfers to a range of temperatures, including 0, 2, 4, 6, 8, and 10 • C (P. antarctica died at 10 • C) under the same light cycle as stock cultures.Cultures were diluted semi-continuously following Zhu et al. (2016), allowing them to be maintained in continuous exponential growth and so facilitating comparisons between treatments in the same physiological growth stage.All of the cultures were acclimated to their respective temperatures for 8 weeks before the commencement of the experiment.At this point, after the growth rates were verified to be stable for at least three to five consecutive transfers, the cultures were sampled 48 h after dilution (Zhu et al., 2016).
For CO 2 functional response curves, P. antarctica and P. subcurvata were also grown in triplicate in a series of six CO 2 concentrations from ∼ 100 to ∼ 1730 ppm in triplicate 500 mL acid-washed polycarbonate bottles at both 2 and 8 • C using same dilution technique as above.The CO 2 concentration was achieved by gently bubbling with 0.2 µm filtered air-CO 2 mixture (Gilmore, CA) and carbonate system equilibration was ensured by pH and dissolved inorganic carbon (DIC) measurements (King et al., 2015;see below).
An additional experiment tested whether temperaturerelated trends in growth rates observed in monocultures were maintained when both species were grown together in a simple model community.For this examination of thermal effects on the growth of P. antarctica and P. subcurvata in coculture (pre-acclimated to respective temperatures), the isolates were mixed at equal Chl a (chlorophyll a) concentrations and grown together for 6 days in triplicate bottles at both 0 and 6 • C.These temperatures chosen to span the optimum growth ranges of both species (see Results, below).The relative abundance of each phytoplankton was then calculated based on cell counts taken on days 0, 3, and 6.

Growth rates
Cell count samples were counted on a Sedgewick Rafter Grid using an Olympus BX51 microscope before and after dilution for each treatment.Samples that could not be counted immediately were preserved with Lugol's iodine solution (final concentration 2 %) and stored at 4 • C until counting.Specific growth rates (d −1 ) were calculated following Eq.(1): where N 0 and N 1 are the cell density at the beginning and end of a dilution period, respectively, and t is the duration of the dilution period (Zhu et al., 2016).The Q 10 of growth rates was calculated following Chaui-Berlinck et al. (2002) as Eq. ( 2): where µ 1 and µ 2 are the specific growth rates of the phytoplankton at temperatures T 1 and T 2 , respectively.The growth rates were fitted to Eq. (3) to estimate the thermal reaction norms of each species: where specific growth rate f depends on temperature (T ) and temperature niche width (w), and other empirical parameters z, a, and b were estimated by maximum likelihood (Thomas et al., 2012;Boyd et al., 2013).Afterwards, the optimum temperature for growth and maximum growth rate were estimated by numerically maximizing the equation (Boyd et al., 2013).The growth rates of all the species at all the CO 2 levels were fitted to Michaelis-Menten equation as Eq. ( 4), to estimate maximum growth rates (µ max ) and halfsaturation constants (K 1/2 ) for CO 2 concentration (S).In the CO 2 curve experiments growth rates for both these autotrophic species were assumed to be zero at 0 ppm CO 2 , and in the thermal curve experiments growth rates were assumed to be zero at −2 • C, approximately the freezing point of seawater.

Elemental and Chl a analysis
Culture samples for particulate organic carbon/nitrogen (POC/PON) and particulate organic phosphorus (POP) analyses were filtered onto pre-combusted (500 • C for 2 h) GF/F filters and dried at 60 • C overnight.A 30 mL aliquot of P. subcurvata culture for each treatment was filtered onto 2 µm polycarbonate filters (GE Healthcare, CA) and dried in an oven at 60 • C overnight for biogenic silica (BSi) analysis.
The analysis method of POC/PON and POP followed Fu et al. (2007), and BSi analysis followed Paasche et al. (1973).An aliquot of 30 to 50 mL from each treatment replicate was filtered onto GF/F filters and extracted with 90 % acetone at −20 • C for 24 h for Chl a analysis.The Chl a concentration was then determined using the non-acidification method on a 10AU TM fluorometer (Turner Design, CA) (Fu et al., 2007).

pH and dissolved inorganic carbon (DIC) measurements
pH was measured using a pH meter (Thermo Scientific, MA), calibrated with pH 7 and 10 buffer solutions.For DIC analyses, an aliquot of 25 mL was preserved with 200 µL 5 % HgCl 2 and stored in the dark at 4 • C until analysis.Total DIC was measured using a CM140 total inorganic carbon analyzer (UIC Inc., IL).An aliquot of 5 mL sample was injected into the sparging column of acidification unit CM5230 (UIC Inc., IL) followed by 2 mL 10 % phosphoric acid.Flow-ratecontrolled pure nitrogen was used as the carrier gas, and the CO 2 released from the DIC pool in the sample was quantified with a CM5015 CO 2 coulometer (UIC Inc., IL) using absolute coulometric titration.The carbonate buffer system was sampled for each of the triplicate bottles in each treatment at the beginning and end of the experiments; reported values are final ones.The pCO 2 in growth media was calculated using CO2SYS (Pierrot et al., 2006).These carbonate system measurements are shown in for each replicate (see above), but for convenience, the CO 2 treatments are referred to in the text using the mean value of all experimental bottles, rounded to the nearest 5 ppm: these values are 100, 205, 260, 425, 755, and 1730 ppm.

Statistical analysis
All statistical analyses and model fitting, including Student's t tests, ANOVA, Tukey's HSD test, two-way ANOVA, and thermal reaction norms estimation, were conducted using the open-source statistical software R version 3.1.2(R Foundation).

Temperature effects on growth rates
Temperature increase significantly affected the growth rates of both P. antarctica and P. subcurvata, but with different trends (p<0.05) (Fig. 1).The specific growth rates of P. subcurvata increased from 0 to 8 • C (p<0.05) and then significantly decreased at 10 • C (p<0.05) (Fig. 1).The growth rates of P. antarctica significantly increased from 0 to 2 • C, plateaued at 4 and 6 • C, and then significantly decreased from 6 to 8 • C (p<0.05) (Fig. 1).P. antarctica and P. subcur- vata stopped growing at 10 and 14 • C, respectively (Fig. 1).The specific growth rates of P. subcurvata were not significantly different from those of P. antarctica at 0, 2, and 4 • C, but they became significantly higher than P. antarctica at 6 • C and remained significantly higher than P. antarctica through 8 and 10 • C (p<0.05) (Fig. 1).The optimum temperatures for growth of P. antarctica and P. subcurvata were 4.85 and 7.36 • C, respectively, both well above the current temperature in the Ross Sea, Antarctica (Table 2).In addition, the estimated temperature niche width of P. subcurvata (−2-12.19• C) is wider than that of P. antarctica (−2.0 to 9.52 • C) (Table 2); calculated minimum temperatures estimated from the thermal niche width equation were less than −2.0 • , the freezing point of seawater, and so growth is assumed to terminate at −2.0 • .The Q 10 value of the growth rate of P. antarctica from 0 to 4 • C is 2.11, which is lower than the Q 10 value of 3.17 for P. subcurvata over the same temperature interval (p<0.05)(Table 2).

Temperature effects on elemental composition
The C : N and N : P ratios of P. subcurvata were unaffected by changing temperature (Fig. 2a, b), but the C : P, C : Si, and C : Chl a ratios of this species were significantly affected (p<0.05) (Fig. 2c, d, Fig. 3).The C : P ratios of P. subcurvata were slightly but significantly lower in the middle of the tested temperature range.They were higher at 8 and 10 • C than at 2, 4, and 6 • C (p<0.05) (Fig. 2c), and also significantly higher at 10 • C than at 0 • C (Fig. 2c).The C : Si ratios of P. subcurvata showed a similar pattern of slightly lower values at mid-range temperatures; at 0 and 2 • C they were significantly higher than at 6 and 8 • C (p<0.05) (Fig. 2d), and significantly higher at 2 and 10 • C than at 4 and 8 • C, respectively (Fig. 2d).The C : Chl a ratios of P. subcurvata also showed this trend of somewhat lower values in the middle of the thermal gradient.At 0, 8 and 10 • C, C : Chl a ratios were significantly higher than at 2, 4, and 6 • C (p<0.05), and also significantly higher at 10 • C than at 0 and 8 • C (Fig. 3).
The C : N, N : P, C : P, and C : Chl a ratios of P. antarctica were not significantly different across the temperature range (Figs.2a, b, c, 3).The N : P ratios of P. antarctica were significantly higher than those of P. subcurvata at 2, 6, and 8 • C (p<0.05) (Fig. 2b).Additionally, the C : P ratios of P. antarctica were significantly higher than those of P. subcurvata at 6 and 8 • C (p<0.05) (Fig. 2c), and the C : Chl a ratios of P. antarctica were significantly higher than values of P. subcurvata at all the temperatures tested (p<0.05) (Fig. 3).
Temperature change significantly affected the quotas of cellular carbon (C), cellular nitrogen (N), cellular phospho-rus (P), cellular silica (Si), and cellular Chl a of P. subcurvata (p<0.05)(Table 3).The cellular C and N quotas of P. subcurvata were significantly higher at 8 • C than at 0 • C (p<0.05) (Table 3), the cellular P quotas of P. subcurvata were significantly higher at 4 • C than at 0, 2, and 10 • C (p<0.05) (Table 3), and the cellular Si quotas of P. subcurvata were significantly higher at 8 • C than at 0 and 2 • C. Si quotas were also significantly higher at 4 and 6 • C than at 0 • C (p<0.05) (Table 3).The extreme temperatures significantly decreased the cellular Chl a quotas of P. subcurvata, as the cellular Chl a quotas of this species were significantly higher at 4, 6, and 8 • C than at 0 and 10 • C (p<0.05) (Table 3).
Temperature change significantly affected the cellular P quotas and cellular Chl a quotas of P. antarctica (p<0.05),but not the cellular C and N quotas (p>0.05)(Table 3).The cellular P quotas of P. antarctica were significantly higher at 0 • C than at 8 • C (p<0.05) (Table 3), and the Chl a quotas of the prymnesiophyte were significantly lower at 8 • C than at 0, 2, and 6 • C (p<0.05) (Table 3).

Co-incubation at two temperatures
A warmer temperature favored the dominance of P. subcurvata over P. antarctica in the model community experiment.Although P. subcurvata increased its abundance relative to the prymnesiophyte at both temperatures by day 6, this increase was larger and happened much faster at 6 • C (from 31 to 72%) relative to 0 • C (from 31 to 38%) (p<0.05) (Fig. 4).

CO 2 effects on specific growth rates at two temperatures
The carbonate system was relatively stable across the range of CO 2 levels during the course of the experiment (Table 1).CO 2 concentration significantly affected the growth rates of P. subcurvata at both temperatures (Fig. 5).The growth rates of the diatom at 2 • C increased steadily with CO 2 concentration increase from 205 to 425 ppm (p<0.05),but were saturated at 755 and 1730 ppm (Fig. 5a).Similarly, the growth rates of P. subcurvata at 8 • C increased with CO 2 concentration increase from 205 to 260 ppm (p<0.05), and were saturated at 425, 755 and 1730 ppm (Fig. 5b).The growth rates of the diatom at all CO 2 concentrations tested at 8   the maximum growth rate of P. subcurvata at 8 • C was 0.88 d −1 , significantly higher than the value of 0.60 d −1 at 2 • C (p<0.5) (Table 4).In addition, the pCO 2 half-saturation constant (K 1/2 ) of P. subcurvata at 8 • C was 10.7 ppm, significantly lower than 66.0 ppm at 2 • C (p<0.5) (Table 4).Thus, temperature and CO 2 concentration increase interactively increased the growth rates of P. subcurvata (p<0.05).CO 2 concentration also significantly affected the growth rates of P. antarctica at both 2 and 8 • C. The growth rates of the prymnesiophyte at both 2 and 8 • C increased with CO 2 concentration increase from 100 to 260 ppm (p<0.05), and were saturated at 425 and 755 ppm (Fig. 5c, d).The growth rates of P. antarctica at 2 • C decreased slightly at 1730 ppm relative to 425 and 755 ppm (p<0.05) (Fig. 5c).The maximum growth rate of P. antarctica at 8 • C was 0.43 d −1 , significantly lower than the value of 0.61 d −1 at 2 • C (p<0.05) (Table 4).The pCO 2 half-saturation constants of P. antarctica at 2 and 8 • C were not significantly different (Table 4), and thus no interactive effect of temperature and CO 2 was observed on the growth rate of the prymnesiophyte (p>0.05).

CO 2 effects on elemental composition at two temperatures
CO 2 concentration variation did not affect the C : N, N : P, or C : P ratios of P. subcurvata at either 2 or 8 • C. The C : Si ratios of P. subcurvata were significantly higher at 1730 ppm relative to lower pCO 2 levels, except at 755 ppm at 8 • C (p<0.05) (Table 5).The N : P ratios of P. subcurvata at 8 • C were significantly higher than at 2 • C at all the CO 2 levels tested except 100 ppm (p<0.05)(Table 5).The C : P ratios of P. subcurvata at 8 • C were significantly higher than at 2 • C at all the CO 2 levels tested (p<0.05)(Table 5).The C : Si ratios of P. subcurvata at CO 2 levels lower than 755 ppm at 8 • C were significantly lower than at 2 • C (p<0.05) (Table 5).The higher temperature also significantly increased the C : Chl a ratios of P. subcurvata at all the CO 2 levels tested (p<0.05)(Table 5).Additionally, the temperature increase and CO 2 concentration increase interactively decreased the C : Chl a ratios of P. subcurvata (p<0.05)(Table 5).
The CO 2 concentration increase did not affect the C : N, N : P, and C : P ratios of P. antarctica at either 2 • C or 8 • C. The carbon to Chl a ratios of P. antarctica were significantly higher at 1730 ppm than at all lower CO 2 concentrations at 2 • C. Similarly, at 8 • C the carbon to Chl a ratios of this species were also significantly higher at 425, 755, and 1730 ppm than at lower CO 2 concentrations (p<0.05)(Table 5), and significantly higher at 1730 ppm than at 425 and 755 ppm (p<0.05)(Table 5).
The warmer temperature significantly decreased the C : N ratios of P. antarctica at 260 and 755 ppm CO 2 (p<0.05)(Table 5), and C : P ratios also decreased at 100 and 205 ppm (p<0.05)(Table 5).The C : Chl a ratios of P. antarctica at CO 2 levels higher than 205 ppm were significantly higher at 8 • C relative to 2 • C (p<0.05) (Table 5).Temperature and CO 2 concentration increase interactively increased the C : Chl a ratios of P. antarctica (p<0.05)(Table 5).
The CO 2 concentration increase did not affect the cellular C, N, P, or Si quotas of P. subcurvata at 2 • C, nor the C quotas and N quotas at 8 • C. The Si quotas of P. subcurvata were significantly lower at 1730 ppm CO 2 than at 100 and 205 ppm at 8 • C (p<0.05) (Table 6).The cellular Chl a quotas of P. subcurvata were significantly lower at 8 • C relative to 2 • C at CO 2 higher than 205 ppm (p<0.05)(Table 6).The temperature increase significantly increased the celluwww.biogeosciences.net/14/5281/2017/Biogeosciences, 14, 5281-5295, 2017 Table 3.The effects of temperature on the C quota (pmol cell −1 ), N quota (pmol cell −1 ), P quota (pmol cell −1 ), Si quota (pmol cell −1 ), and Chl a per cell (pg cell  The C, N, and P quotas of P. antarctica were not affected by CO 2 increase at 2 • C, and N and P quotas were not affected by CO 2 increase at 8 • C either.However, the C quota of P. antarctica at 1730 ppm CO 2 was significantly higher than at CO 2 levels lower than 755 ppm at 8 • C (p<0.05) (Table 6).The Chl a per cell of P. antarctica at 1730 ppm CO 2 was significantly less than at lower CO 2 levels at both 2 and 8 • C (p<0.05) (Table 6).For P. antarctica, the values of Chl a per cell at 100, 205, and 755 ppm CO 2 at 8 • C were significantly lower relative to 2 • C (p<0.05) (Table 6).Temperature increase and CO 2 concentration increase interactively increased the C and N quotas of P. antarctica (p<0.05)(Table 6).

Discussion
As has been documented in previous work, the diatom P. subcurvata and the prymnesiophyte P. antarctica responded differently to warming (Xu et al., 2014;Zhu et al., 2016).In the Ross Sea as elsewhere, temperature determines both phytoplankton maximum growth rates (Bissinger et al., 2008) and the upper limit of growth (Smith and Sakshaug, 1990) in a species-specific manner.Thermal functional response curves of phytoplankton typically increase in a normally distributed pattern, with growth rates increasing up to the optimum temperature range, and then declining when temperature reaches inhibitory levels (Boyd et al., 2013;Fu et al., 2014;Xu et al., 2014;Hutchins and Fu, 2017).Specific growth rates of P. subcurvata reached optimal levels at 8 • C, demonstrating that this species grows fastest at temperatures substantially above any temperatures found in the present-day Ross Sea.In contrast, growth rates of P. antarctica saturated at 2 • C.This suggests that P. subcurvata may be a superior competitor over P. antarctica in any realistically foreseeable warming scenario.Zhu et al. (2016) found that 4 • C warming significantly promoted the growth rates of P. subcurvata but not P. antarc- tica.Xu et al. (2014) found that the growth rates of another strain of P. antarctica (CCMP3314) decreased in a multi-variable "year 2100 cluster" condition (6 • C, 81 Pa CO 2 , 150 µmol photons m −2 s −1 ) relative to the "current condition" (2 • C, 39 Pa CO 2 , and 50 µmol photons m −2 s −1 ) and the "year 2060 condition" (4 • C, 61 Pa CO 2 , and 100 µmol photons m −2 s −1 ).In our study, the Q 10 value of P. subcurvata from 0 to 4 • C was 3.11, nearly 50 % higher than the Q 10 value of P. antarctica across the same temperature range (2.17), and similar to the Q 10 values ob-served for different strains of these two species in Zhu et al. (2016).Such Q 10 values that substantially exceed the canonical value of 2 are often observed in polar marine organisms (Clarke et al., 1983;Hutchins and Boyd, 2016).Our results showed that the maximal thermal limit of P. antarctica was reached at 10 • C, as was also observed by Buma et al. (1991), while P. subcurvata did not cease to grow until 14 • C. Clearly, P. subcurvata has a superior tolerance to higher temperature compared to P. antarctica.
Table 6.The effects of CO 2 on the C quota (pmol cell −1 ), N quota (pmol cell −1 ), P quota (pmol cell −1 ), Si quota (pmol cell −1 ), and Chl a per cell (pg cell −1 ) of P. subcurvata and P. antarctica at 2 and 8 • C. Values represent the means and errors are the standard deviations of triplicate bottles.
Besides temperature, mixed layer depth and irradiance also likely play a role in the competition between diatoms and P. antarctica (Arrigo et al., 1999(Arrigo et al., , 2010;;Smith and Jones, 2015).Arrigo et al. (1999) observed that P. antarctica dominated the southern Ross Sea region with deeper mixed layers, while diatom dominated the regions with shallower mixed layer depths.The niches of these two groups of phytoplankton are difficult to define by either light or by temperature, since shallow surface stratification tends to promote both solar heating and high irradiance, while deep mixing often lowers both light and temperatures.It is worth considering whether these two phytoplankton groups are each best adapted to a different environmental matrix of both variables.This concept of different light/temperature niches for Ross Sea diatoms and P. antarctica is worthy of further investigation.
Our experiments used nutrient-replete conditions, which are relevant to most of the Ross Sea HNLC region throughout most of the growing season.However, major nutrients sometimes become depleted late in the season on McMurdo Sound, the origin of our culture isolates, as Fe inputs are somewhat higher in these nearshore waters (Bertrand et al., 2015).Experiments using nutrient-limited phytoplankton frequently find differing responses to CO 2 and temperature compared to those of nutrient-replete cells, including sometimes enhanced effects of these global change factors on elemental ratios (Taucher et al., 2015;Sala et al., 2016).Our experiments under high-nutrient, Fe-replete conditions thus are likely to best predict possible biological effects of future high CO 2 and temperature during the first half or more of the Ross Sea growing season.
Temperature change affected the C : P, N : P, and C : Si ratios of P. subcurvata, due to the combined effects of the different responses of cellular C, P, and Si quotas.The C : P and N : P ratios of P. subcurvata increased at the two highest temperatures tested.This might be due to an increase in protein translation efficiency and a corresponding decrease in phosphate-rich ribosomes with warming, which can result in a decreased cellular P requirement per unit of carbon in marine phytoplankton (Toseland et al., 2013).Similarly lowered P quotas at higher temperatures have been documented in other studies as well (Xu et al., 2014;Boyd et al., 2015;Hutchins and Boyd, 2016).This result suggests that the amount of carbon exported per unit phosphorus by P. subcurvata (and perhaps other diatoms) in the Ross Sea may increase as temperature increases in the future (Toseland et al., 2013).
In contrast, the decreasing trend of C : Si ratios in P. subcurvata appears to be largely due to higher cellular Si quotas at temperatures of 4 • C and above.Although the physiological reason(s) for increased silicification with warming are currently not understood, this trend also may have biogeochemical consequences.This decrease in cellular C : Si ratios at higher temperature may tend to enhance Si export, with the qualification that biogenic Si remineralization rates also increase with temperature (Ragueneau et al., 2000) and thus could potentially offset this trend.
Previous studies have shown that nutrient drawdown by diatoms and P. antarctica is different, due to differing elemental ratios of these two groups (Arrigo et al., 1999;Smith et al., 2014a;Xu et al., 2014).Our results generally corresponded to this trend, as the N : P ratios of P. antarctica were higher than P. subcurvata at 2, 6, and 8 • C and C : P ratios of P. antarctica were higher than P. subcurvata at 6 and 8 (p<0.05) (Fig. 2).Although elemental ratios of the prymnesiophyte were largely unaffected by temperature, a predicted increase in diatom and decrease in P. antarctica contributions to phytoplankton production caused by warming will likely change nutrient export ratios (Smith et al., 2014a, b).It is possible that N and C export per unit P may decrease with a phytoplankton community shift from P. antarctica dominance to diatom dominance (Arrigo et al., 1999;Smith et al., 2014a, b;Xu et al., 2014).However, food web effects may compensate for the effects of temperature on biogeochemical cycles, as diatoms are a preferred food source for zooplankton grazers, compared to Phaeocystis (Knox, 1994;Caron et al., 2000;Haberman et al., 2003).
Our results showed that the growth rates of both P. subcurvata and P. antarctica exhibited moderate limitation by CO 2 levels lower than ∼ 425 ppm at both 2 and 8 • C; this observation is significant, since during the intense Ross Sea summertime phytoplankton bloom pCO 2 can sometimes drop to very low levels (Tagliabue and Arrigo, 2016).However, at CO 2 concentrations beyond current atmospheric levels of ∼ 400 ppm, growth rates of P. subcurvata or P. antarctica were CO 2 -saturated.Although a general model prediction suggests that an atmospheric CO 2 increase from current levels to 700 ppm could increase the growth of marine phytoplankton by 40 % (Schippers et al., 2004), our results instead correspond to previous studies which showed negligible effects of elevated CO 2 on various groups of phytoplankton (Goldman, 1999;Fu et al., 2007;Hutchins and Fu 2017).In particular, Trimborn et al. (2013) found that increasing CO 2 had no effect on growth rates of Southern Ocean isolates of P. subcurvata and P. antarctica.The minimal effects of changing CO 2 levels on many phytoplankton groups have been suggested to be due to efficient carwww.biogeosciences.net/14/5281/2017/Biogeosciences, 14, 5281-5295, 2017 bon concentrating mechanisms (CCMs) that allow them to avoid CO 2 limitation at low pCO 2 levels (Burkhardt et al., 2001;Fu et al., 2007;Tortell et al., 2008).For instance, both P. subcurvata and P. antarctica have been shown to strongly downregulate activity of the important CCM enzyme carbonic anhydrase as CO 2 increases (Trimborn et al., 2013).Clearly, though, for our two species their CCM activity was not sufficient to completely compensate for carbon limitation at low pCO 2 levels.Although speculative, it is possible that P. antarctica could have an ability to subsidize growth at very low CO 2 levels through oxidation of organic carbon from the colony mucilage.Our results also showed that very high CO 2 (1730 ppm) significantly reduced the growth rate of P. antarctica relative to 425 and 755 ppm at 2 • C.This inhibitory effect might be due to the significantly lower pH at 1730 ppm (∼ 7.4), which could entail expenditures of additional energy to maintain pH homeostasis within cells.Similar negative effects of high CO 2 have been observed on P. antarctica in natural communities (Hancock et al., 2017), as well as on a mixed Antarctic microbial assemblage (Davidson et al., 2016).
Warming from 2 to 8 • C had a significant interactive effect with CO 2 concentration in P. subcurvata, as maximum growth rates were higher and the half-saturation constant (K 1/2 ) for growth was much lower at the warmer temperature.In contrast, warming decreased the maximal growth rates of P. antarctica over the range of CO 2 concentrations tested, and failed to change its K 1/2 for growth.The decreased CO 2 K 1/2 of P. subcurvata at high temperature might confer a future additional competitive advantage over P. antarctica in the late growing season, when pCO 2 can be low (Tagliabue and Arrigo, 2016) and temperatures higher, although temperatures are generally never as high as 8 • C in the current Ross Sea (Liu and Smith, 2012).The interactive effects of temperature and CO 2 on P. subcurvata might be due to elevated enzyme and protein translation efficiencies at higher temperature, which may decrease the CO 2 requirement of the Calvin cycle and facilitate allocation of fixed carbon to growth (Toseland et al., 2013;Hutchins and Boyd, 2016).On the other hand, 8 • C is clearly close to the upper thermal limit of P. antarctica, suggesting that its biochemical efficiencies decline rapidly above this temperature.The K 1/2 of P. antarctica for CO 2 was, however, significantly lower than that of P. subcurvata at 2 • C, which may be advantageous to the prymnesiophyte when water temperatures are low in the spring.
The effects of pCO 2 variation on the elemental ratios of P. subcurvata and P. antarctica were minimal relative to those of temperature increase.Previous research on the effects of CO 2 on the elemental ratios of phytoplankton has shown that the elemental composition of phytoplankton may change with CO 2 availability (Burkhardt et al., 1999;Fu et al., 2007Fu et al., , 2008;;Tew et al., 2014;reviewed in Hutchins et al., 2009).Hoogstraten et al. (2012) found that CO 2 concentration change did not change the cellular POC, PON, C : N ratios, or POC to Chl a ratios of the temperate species Phaeocystis globosa.In contrast, Reinfelder (2014) observed that the N and P quotas of several diatoms decreased with increasing CO 2 and led to increased C : N, N : P, and C : P ratios.King et al. (2015) found that high CO 2 could increase, decrease or not affect the C : P and N : P ratios of several different phytoplankton species.Our results resemble those of studies with other phytoplankton that found that the effects of CO 2 concentration can be negligible on C : N, N : P, or C : P ratios (Fu et al., 2007;Hutchins et al., 2009;Hoogstraten et al., 2012;King et al., 2015), including incubation studies with Antarctic communities (Deppeler et al., 2017).It is possible that such contrasting effects of CO 2 concentration on the elemental ratios of phytoplankton are due to speciesspecific differences in biochemical composition (e.g., proteins are enriched with N and membranes with P relative to other cellular components) or to differences in experimental design, which can make intercomparisons problematic (Hutchins and Fu, 2017).
In contrast to C : N : P ratios, we observed that the C : Si ratios of P. subcurvata were significantly higher at 1730 ppm compared to almost all of the lower CO 2 levels.This increase in C : Si ratios was due to a decrease in cellular Si quotas at 1730 ppm CO 2 .Milligan et al. (2004) observed that the silica dissolution rates of a temperate diatom increased significantly in high CO 2 relative to in low-CO 2 cultures.Tatters et al. (2012) found a similar trend in the temperate toxic diatom Pseudo-nitzschia fraudulenta, in which cellular C : Si ratios were higher at 765 ppm than at 200 ppm CO 2 .This suggests that future increases in diatom silicification at elevated temperature could partially or wholly offset the decreased silicification and higher dissolution rates of silica observed high CO 2 (above); to fully predict net trends, further interactive experiments focusing on silicification as a function across a range of both temperature and pCO 2 are needed.
In conclusion, our results indicate that P. subcurvata from the Ross Sea is better adapted to higher temperature than is P. antarctica.Diatoms are a diverse group, but if their general thermal response is similar to that of this Pseudonitzschia species, they may thrive under future global warming scenarios, while the relative dominance of P. antarctica in this region may wane.In contrast, another recent study has suggested that warming might indirectly favor P. antarctica springtime dominance by leading to large areas of open water at a time when incident light penetration is low and mixed layers are still relatively deep (Ryan-Keogh et al., 2017).Because of the differences in elemental ratios in the two groups, ecological shifts that favor diatoms may significantly increase the export of phosphorus and silicon relative to carbon and nitrogen, while increased P. antarctica dominance will increase carbon export relative to nutrient fluxes, as well as enhancing the organic sulfur cycle.Our conclusions must be qualified as they were obtained using Fe-replete culture conditions, similar to conditions often found early in the growing season in McMurdo Sound.However, Fe limitation generally prevails later in the season here, and elsewhere in the offshore Ross Sea.Irradiance is an additional key environmental factor to consider in both the present and future in this region (Smith and Jones, 2015).Thus, in addition to warming and CO 2 increases, the interactive effects of light and Fe with these two factors should also be considered (Xu et al., 2014;Boyd et al., 2015;Hutchins and Boyd, 2016;Hutchins and Fu, 2017).Considering the differences between the responses of the diatom and P. antarctica to warming and ocean acidification seen here, as well to warming and Fe in previous work (Zhu et al., 2016), models attempting to predict future changes in community structure and primary production in the Ross Sea polynya may need to realistically incorporate a complex network of interacting global change variables.

Figure 1 .
Figure 1.Thermal functional response curves showing specific growth rates of Pseudo-nitzschia subcurvata and Phaeocystis antarctica across a range of temperatures from 0 to 14 • C. Values represent the means and error bars represents the standard deviations of triplicate samples.

Figure 2 .
Figure 2. The C : N ratios (a), N : P ratios (b), and C : P ratios (c) of Pseudo-nitzschia subcurvata and Phaeocystis antarctica and (d) the C : Si ratios of Pseudo-nitzschia subcurvata from the thermal response curves shown in Fig. 1 for a range of temperatures from 0 to 10 • C. Values represent the means and error bars represents the standard deviations of triplicate samples.

Figure 3 .
Figure 3.The C : Chl a ratios of Pseudo-nitzschia subcurvata and Phaeocystis antarctica from the thermal response curves shown in Fig. 1 for a range of temperatures from 0 to 10 • C. Values represent the means and error bars represents the standard deviations of triplicate samples.

Figure 4 .
Figure 4.The relative abundance of Pseudo-nitzschia subcurvata in a 6-day competition experiment with Phaeocystis antarctica at 0 and 6 • C. The competition experiments were started with equal Chl a concentrations for both species, and the relative abundance was calculated based on cell counts.Values represent the means and error bars represents the standard deviations of triplicate samples.

Figure 5 .
Figure 5. CO 2 functional response curves showing specific growth rates (and fitted curves) across a range of CO 2 concentrations from ∼ 100 to ∼ 1730 ppm at 2 and at 8 • C. Pseudo-nitzschia subcurvata at 2 • C (a) and 8 • C (b) and Phaeocystis antarctica at 2 (c) and 8 • C (d).Values represent the means and error bars represents the standard deviations of triplicate samples.

Table 1 .
The measured pH and dissolved inorganic carbon (DIC) as well as calculated pCO 2 of P. subcurvata and P. antarctica at 2 and 8 • C in each treatment.Values represent the means and errors are the standard deviations of triplicate bottles.

Table 2 .
Statistical comparison of the results for each of the three thermal traits: optimum temperature ( • C), maximum growth rate (d −1 ), and temperature niche width (W) * of P. subcurvata and P. antarctica.The statistical results for the lower bound of temperate niche width in both species were lower than −2.0 • C, the freezing point of seawater. * −1 ) of P. subcurvata and P. antarctica.Values represent the means and errors are the standard deviations of triplicate bottles.

Table 4 .
Comparison of the curve fitting results for maximum growth rate (d −1 ) and half-saturation constants (K 1/2 ), calculated from the CO 2 functional response curves of P. subcurvata and P. antarctica at 2 and 8 • C. Values represent the means and errors are the standard errors from fitting.

Table 5 .
The effects of CO 2 on the C : N, N : P, C : P, C : Si, and C : Chl a ratios of P. subcurvata and P. antarctica at 2 and 8 • C. Values represent the means and errors are the standard deviations of triplicate bottles.