Physiological response of Crocosphaera watsonii to enhanced and fluctuating carbon dioxide conditions.

We investigated the effects of elevated pCO2 on cultures of the unicellular N2-fixing cyanobacterium Crocosphaera watsonii WH8501. Using CO2-enriched air, cultures grown in batch mode under high light intensity were exposed to initial conditions approximating current atmospheric CO2 concentrations (∼ 400 ppm) as well as CO2 levels corresponding to low- and high-end predictions for the year 2100 (∼ 750 and 1000 ppm). Following acclimation to CO2 levels, the concentrations of particulate carbon (PC), particulate nitrogen (PN), and cells were measured over the diurnal cycle for a six-day period spanning exponential and early stationary growth phases. High rates of photosynthesis and respiration resulted in biologically induced pCO2 fluctuations in all treatments. Despite this observed pCO2 variability, and consistent with previous experiments conducted under stable pCO2 conditions, we observed that elevated mean pCO2 enhanced rates of PC production, PN production, and growth. During exponential growth phase, rates of PC and PN production increased by ∼ 1.2- and ∼ 1.5-fold in the mid- and high-CO2 treatments, respectively, when compared to the low-CO2 treatment. Elevated pCO2 also enhanced PC and PN production rates during early stationary growth phase. In all treatments, PC and PN cellular content displayed a strong diurnal rhythm, with particulate C:N molar ratios reaching a high of 22:1 in the light and a low of 5.5:1 in the dark. The pCO2 enhancement of metabolic rates persisted despite pCO2 variability, suggesting a consistent positive response of Crocosphaera to elevated and fluctuating pCO2 conditions.

Introduction

Anthropogenic emissions and land use change are increasing the concentration of carbon dioxide (n class="Chemical">CO2) in the atmosphere and surface ocean waters [1]. The predicted effects of elevated CO2 partial pressure (pCO2) and consequent ocean acidification (OA) on marine ecosystems include a potential upregulation of metabolic processes by CO2-limited phytoplankton species [2]. Because the carboxylating enzyme ribulose-1,5-bisphosphate carboxylase-oxygenase (RuBisCO) is typically not saturated under ambient surface seawater pCO2, many phytoplankton groups invest energy in carbon concentrating mechanisms (CCMs) to increase CO2 concentrations at the catalytic site [3]. Hence, under future OA conditions, phytoplankton with low-affinity RuBisCO could potentially down-regulate CCMs and reallocate energy and elemental resources to allow for increased carbon (C) fixation and growth rates [4]. Indeed, elevated pCO2 has been shown to stimulate C fixation by select monocultures of phytoplankton and natural assemblages (reviewed in [5]).

Elevated pCO2 appears to have a particularly strong metabolic enhancement in n class="Chemical">dinitrogen (N2)-fixing (diazotrophic) cyanobacteria. Initial laboratory experiments using strain IMS101 of Trichodesmium, a group of filamentous, non-heterocystous cyanobacteria, showed that doubling pCO2 increases N2 fixation rates by 35–138% and C fixation rates by 23–40% [4], [6]–[8]. Likewise, the unicellular cyanobacterium Crocosphaera strain WH8501 displays increased rates of C and N2 fixation under elevated pCO2 conditions [9]. Trichodesmium and Crocosphaera are dominant diazotrophic taxa in oligotrophic open-ocean environments [10], where the bioavailable nitrogen (N) from N2 fixation can fuel up to half of production exported from the euphotic zone [11]. In theory, a global pCO2 enhancement of marine N2 fixation could increase oceanic C uptake and export, producing a negative feedback to climate change [6].

In contrast to laboratory findings, recent field experiments using natural diazotrophic assemblages do not reliably show that raising n class="Chemical">pCO2 enhances N- or C-based productivity. Elevating pCO2 in bottle incubations has been shown to increase N2 fixation rates by Trichodesmium colonies isolated from the Subtropical Atlantic and the Gulf of Mexico [12], [13] but not by Trichodesmium colonies isolated from the North Pacific Subtropical Gyre [14]. Furthermore, experiments using whole water diazotrophic assemblages from the North and South Pacific gyres have found no relationship between pCO2 and N2 fixation rates [15], [16]. The inconsistent results from field incubations highlight the importance of assessing how the effect of pCO2 on diazotrophs is influenced by other factors, including community composition [14], [17], physiology, and environmental conditions [18].

One methodological challenge in pCO2 manipulation studies is producing realistic timescales for n class="Chemical">pCO2 perturbations. In nature, seawater pCO2 varies spatially and temporally: phytoplankton experience pCO2 fluctuations on diurnal, seasonal, episodic, and long-term (e.g. OA-driven) timescales [19]. However, most laboratory OA studies grow cultures under stable pCO2 treatments, allowing cultures to acclimate to a steady pCO2 for multiple generations. These stable pCO2 conditions may affect phytoplankton differently than the dynamic pCO2 experienced in marine ecosystems; for instance, energetic costs associated with resource allocation may be minimized under stable pCO2 regimes [20].

Predicting the future response of diazotrophic assemblages to OA will require an assessment of how n class="Chemical">pCO2 affects marine diazotrophs under variable environmental conditions and physiological states. Here, we present data from experiments tracking the growth, PC and PN production rates of Crocosphaera watsonii WH8501 cultures bubbled with air at three CO2 levels (∼400, 750,1000 ppm), while allowing for biologically induced pCO2 variability in the culture medium resulting from photosynthesis and respiration. This approach contributes to the existing literature on potential responses of marine diazotrophs to future OA, but expands from previous studies by testing the effect of elevated pCO2 under a dynamic pCO2 environment. Our results suggest a consistent response of this organism to elevated pCO2 under variable pCO2 conditions.

Methods

Culture conditions

Unialgal stock cultures of Crocosphaera watsonii strain WH8501 were grown in 0.2 µm-filtered, n class="Chemical">nitrogen-free YBCII medium [21] using 40 µmol L−1 K2HPO4. Cultures were not axenic, but heterotrophic bacterial counts were kept at low levels (1.4–2.1×105 cells mL−1). Light was provided using cool white fluorescent bulbs set on a 12∶12 light/dark cycle. Stock cultures were grown at 24°C and 250 µmol quanta m−2 s−1. For the experiment, cultures were grown at 30°C, a temperature which promotes optimal growth of Crocosphaera in the laboratory [22] and at which high abundances of Crocosphaera cells have been observed at sea [23]. Incoming irradiance for the experiment was 1000 µmol quanta m−2 s−1 as measured by a Biospherical light meter, a level which is saturating but not inhibitory for Crocosphaera WH8501 [24]. Cultures were stirred at least once a day with magnetic stir bars to minimize cells sticking to the glass. The pCO2 was manipulated by gently bubbling cultures with commercially prepared air/CO2 mixtures of ∼400 ppm (‘low-CO2’), ∼750 ppm (‘mid-CO2’), and ∼1000 ppm (‘high-CO2’). Parent cultures were grown under these CO2, light, and temperature conditions for seven days (∼3–4 generations) before productivity rates were measured.

Experimental Design

Crocosphaera cultures were grown under three n class="Chemical">CO2 treatments and monitored over a six-day period (day 0–day 5). Triplicate bottle replicates were used for each CO2 treatment. Preceding the experiment, 2 L glass bottles were filled with 0.2 µm-filtered media that was pre-equilibrated to target pCO2 levels. Initial CO2 equilibration of the media was verified by measuring a stable pCO2 in outflowing gas (>24 h equilibration) using a LI-840 LI-COR gas analyzer (Biosciences). The pH of each replicate was measured and converted through CO2calc (see below) to produce initial pCO2 values of 404±23, 724±51, and 916±34 µatm for low-, mid-, and high-CO2 treatments, respectively. To initiate the experiment, parent cultures in exponential growth phase were diluted into the pre-equilibrated media, producing initial biomass concentrations of 3.9–4.6 µg chlorophyll a (Chl a) L−1 (Table 1). Because parent cultures were shifting media pCO2 through biological processes, the addition of parent cultures to pre-equilibrated media altered the initial pCO2 values to 355±14, 600±21, and 788±11 µatm (Table S1). Bottles were gently bubbled with air/CO2 mixtures at 50 mL min−1 throughout the experiment. Each replicate was sampled once daily after the sixth hour of light (L6) from day 0–day 2, then four times daily after the sixth and twelfth hours of light and darkness (L6, L12, D6, and D12) from day 3–day 5. The pH in each replicate was measured at the time of sampling and samples were preserved for particulate carbon (PC), particulate nitrogen (PN), Chl a, and flow cytometric cell counts (FCM).

Table 1

Time series biomass measurements for cultures of Crocosphaera watsonii WH8501 grown under three CO2 treatments.

	Low-CO2 treatment				Mid-CO2 treatment				High-CO2 treatment
Time point	PC	PN	Cells	Chl a	PC	PN	Cells	Chl a	PC	PN	Cells	Chl a
Day 0 L6	124 (12)	8.8 (1)	-	4.3 (0.6)	120 (4)	8.5 (0.7)	-	4.6 (0.3)	118 (2)	7.6 (0.1)	-	3.9 (0.4)
Day 1 L6	166 (24)	14.2 (2)	5.6E+05 (7.4E+04)	7.3 (1.6)	192 (1)	16.2 (0.2)	8.0E+05 (6.3E+04)	9 (0.4)	172 (21)	13.2 (2)	7.5E+05 (1.9E+04)	7.9 (0.4)
Day 2 L6	374 (145)	22 (4)	7.4E+05 (1.2E+05)	11.6 (1.5)	424 (73)	28.2 (3)	1.2E+06 (1.1E+05)	16.3 (3)	547 (63)	28.3 (2)	1.2E+06 (9.7E+03)	15.1 (1)
Day 3 L6	542 (65)	42.5 (6)	1.4E+06 (2.5 E+05)	24.8 (3)	706 (88)	55.4 (6)	2.4E+06 (1.9E+05)	39.6 (6)	740 (44)	55.9 (4)	2.6E+06 (1.3E+05)	35.3 (6)
Day 4 L6	833 (121)	66 (7)	2.3E+06 (3.1E+05)	31.6 (2.6)	1144 (174)	91.7 (10)	3.6E+06 (1.5E+06)	50.6 (9)	1308 (121)	97.4 (7)	4.8E+06 (2.1E+05)	49.1 (9)
Day 5 L6	1037 (58)	84.2 (3)	3.3E+06 (2.9E+05)	55.4 (2.1)	1589 (186)	130.2 (12)	6.7E+06 (4.4E+05)	82.5 (8)	1852 (65)	155.1 (2)	8.1E+06 (3.8E+05)	101 (2)

Concentrations of particulate carbon (µmol L−1; PC), particulate nitrogen (µmol L−1; PN), cells (# mL−1), and chlorophyll a (Chl a; µg L−1) are provided for L6 time points. Data are mean values from three replicate bottles; standard deviations are presented in parentheses.

Analytical Measurements

For PC/n class="Chemical">PN and Chl a measurements, three subsamples of 5–50 mL (depending on cell density) were withdrawn from each replicate and filtered onto glass fiber filters (GF/F, Whatman), using pre-combusted GF/F filters for PC/PN. Samples were immediately frozen at −80°C (PC/PN) or −20°C (Chl a). PC/PN samples were dried at 60°C overnight, packaged into silver and tin capsules, and analyzed using a Carlo Erba elemental analyzer. Acetanilide (71.09% C and 10.36% N by weight) served as a standard, and filter blanks were <10% of total C and N content. Chl a was extracted in 90% acetone at −20°C for 48 hours and analyzed with a Turner Model 10-AU fluorometer using the acidification method of Strickland and Parsons [25]. On day 0 and day 5 L6 time points, 25 mL samples were withdrawn from GF/F filtrate and immediately frozen for soluble reactive phosphorus (assumed to be equivalent to PO4) and NH4 analyses. NH4 concentrations were measured with a Technicon AutoAnalyzer II, using a modified indophenol blue method [26] and PO4 via the standard ascorbic acid-molybdate method [25].

Crocosphaera and heterotrophic bacterial cell densities were measured using n class="Chemical">FCM. Two 3-mL subsamples were withdrawn from each replicate, pipetted into 4 mL cryovials, and fixed with paraformaldehyde at a final concentration of 1% (volume volume−1). Samples were inverted and allowed to sit in the dark for ∼10 minutes before being frozen at −80°C. For analysis of Crocosphaera cell densities, samples were thawed on ice in the dark then spiked with a known number of 3 µm Polysciences Fluoresbrite yellow-green beads and run on a Becton-Dickinson FASCaliber flow cytometer with a 488 nm laser. Crocosphaera cells and beads were distinguished from other particulate matter by their side light scatter and fluorescence in orange wavelengths. The bead count determined the volume of sample run, and thus the concentration of Crocosphaera cells. A similar method was used to enumerate the background heterotrophic bacteria in these cultures. The samples were spiked with Fluoresbrite 1 µm beads, stained with SYBR Green I according to the method of Marie et al. [27], and differentiated by their side light scatter and green fluorescence.

The pH of each replicate was measured directly using a VWR sympHony electrode calibrated with VWR buffers (n class="Disease">NBS scale). pH values were converted to pCO2 by assuming a constant total alkalinity (TA) for the YBCII medium (2500 µM). This TA value was determined by analyzing DIC and pCO2 of a separate batch of YBCII medium according to the methods of Bandstra et al. [28], then the program CO2calc [29] was used to convert DIC and pCO2 to TA (with CO2 constants from Merbach et al. [30] refit by Dickson and Millero [31], and a correction to the NBS scale by the CO2calc program). Finally, CO2calc was used to calculate pCO2 from our measured pH data and the constant TA.

Our assumption of constant TA of the YBCII medium throughout the experiment is based on the observation that bubbling with air/n class="Chemical">CO2 mixtures perturbs DIC but not TA [32]; thus, any change in TA through the experiment was due to biological activity. While the process of N2 fixation does not affect TA, photosynthesis can have a small effect due to hydrogen ion uptake to balance anionic nutrient (N, P, and S) acquisition [33]. Assuming no inorganic N uptake (as cultures were grown in N-free media) and a 2.4∶1 S:P uptake ratio (as in [33]), we estimate that the average PO4 drawdown of ∼9 µM observed in our experiment (see ) increased TA by an average of ∼52 µM by day 5, generating a maximum pCO2 error of ∼2%. In addition, we assume that calcium carbonate (CaCO3) minerals were not precipitated in our experiment, as the presence of PO4 has been shown to inhibit CaCO3 precipitation, even at high CaCO3 saturation states [34], [35].

Rate calculations

Specific growth rates were calculated for each of the biomass parameters measured: cell density, PC, n class="Chemical">PN, and Chl a (Table 2). Growth rates (μ) were determined using Eq. 1,where N is the biomass at day 3, N is the biomass at day 1, and ΔT is the time interval in days. The day 1–day 3 time interval was chosen for growth rate calculations because this was the phase of exponential growth (Fig. 1A).

Table 2

Biomass-specific growth rates of Crocosphaera watsonii WH8501 cultures grown under three CO2 treatments.

	Specific Growth rate µ (d−1)			Tukey HSD p-value
	Low-CO2	Mid-CO2	High-CO2	Low-CO2 vs. high-CO2	Low-CO2 vs. mid-CO2
Cell density	0.45 (0.02)	0.54 (0.02)	0.60 (0.02)	<0.001	<0.01
Particulate nitrogen	0.54 (0.02)	0.60 (0.05)	0.71 (0.04)	<0.01	0.2
Particulate carbon	0.58 (0.02)	0.63 (0.06)	0.71 (0.06)	<0.05	0.44
Chlorophyll a	0.60 (0.06)	0.72 (0.05)	0.72 (0.07)	0.12	0.14

Rates were calculated from L6 time points between day 1 and day 3 (exponential growth phase). Tukey HSD p-values are provided for comparisons among CO2 treatments. Standard deviations of growth rates from three replicate bottles are presented in parentheses.

Figure 1

Growth of C. watsonii WH8501 batch cultures over a 6-day period under three CO2 treatments.

Shown are concentrations of PN (a), PC (b), and cells (c), molar C:N ratios (d), and pCO2 in µatm (e) within each treatment. For (a–c), the concentrations for each time point (Table S1) were first normalized to the concentration at the day 1 L6 time point, then ln-transformed. The derived slopes between day 1 L6 and day 3 L6 time points correspond to the exponential growth rates (µ) as shown in Table 2. The lines in (a) represent linear regressions through the day 1, day 2, and day 3 L6 time points for high-CO2 (dashed line) and low-CO2 (dotted line) treatments. The regression lines have been extended to the full time period (day 0–5) for visualization of exponential growth (day 0–3 L6 time points) transitioning to early stationary growth (L6 time points after day 3). The dotted line in (d) represents the 6.6 C:N ratio expected from Redfield stoichiometry. Shaded areas represent the dark periods. Error bars represent standard deviations from three replicates.

Shown are concentrations of PN (a), n class="Chemical">PC (b), and cells (c), molar C:N ratios (d), and pCO2 in µatm (e) within each treatment. For (a–c), the concentrations for each time point (Table S1) were first normalized to the concentration at the day 1 L6 time point, then ln-transformed. The derived slopes between day 1 L6 and day 3 L6 time points correspond to the exponential growth rates (µ) as shown in Table 2. The lines in (a) represent linear regressions through the day 1, day 2, and day 3 L6 time points for high-CO2 (dashed line) and low-CO2 (dotted line) treatments. The regression lines have been extended to the full time period (day 0–5) for visualization of exponential growth (day 0–3 L6 time points) transitioning to early stationary growth (L6 time points after day 3). The dotted line in (d) represents the 6.6 C:N ratio expected from Redfield stoichiometry. Shaded areas represent the dark periods. Error bars represent standard deviations from three replicates.

Carbon-normalized PC and PN production rates (∼net C and N2 fixation rates) were calculated using Eq. 2,where N is the biomass (PC or PN) at the final time point, N is the biomass at the initial time point, PC is the initial PC concentration, and ΔT is the time interval in days. Production rates were calculated for both exponential (day 1–day 3) and early stationary (day 3–day 5) growth phases.

Growth rates, PC and PN production rates were all calculated using data from L6 time points. Day 0 was excluded from these analyses due to missing FCM samples on this day.

Statistics

The effects of n class="Chemical">pCO2 on growth rates, PC production, PN production, and molar C:N ratios were assessed using the one-way ANOVA. Differences between CO2 treatments were determined using the Tukey Honest Significance Difference (HSD) test of multiple comparisons. All data reported in this study are averages from triplicate bottles. Statistical tests were run using the program R (http://www.r-project.org/).

Results and Discussion

Our study tested how enhanced pCO2 afn class="Chemical">fects the growth, PC and PN production rates of high-density Crocosphaera cultures. In agreement with a previous study [9], we found that PC production, PN production, and growth rates were all positively correlated with pCO2 (Table 2, Fig. 2). This pCO2 enhancement was observed for Crocosphaera cultures during both exponential and early stationary growth phases (Fig. 2). The high growth rates and cell densities observed in our study produced a strong diurnal rhythm of C and N metabolism in Crocosphaera cultures as well as daily pCO2 variability (Fig. 1).

Figure 2

Carbon-normalized PN (a) and PC production rates (b) of Crocosphaera WH8501 cultures grown under three CO2 treatments during periods of exponential (day 1–day 3) and early stationary (day 3–day 5) growth phases.

Production rates are calculated as increases in PC and PN concentrations (data provided in Table 1) per time normalized to initial PC concentrations within the time interval. Error bars represent standard deviations from three replicates.

Production rates are calculated as increases in PC and n class="Chemical">PN concentrations (data provided in Table 1) per time normalized to initial PC concentrations within the time interval. Error bars represent standard deviations from three replicates.

Diurnal rhythm in growth and pCO2

Diazotrophic cyanobacteria employ various mechanisms to separate the n class="Chemical">oxygen (O2) evolved through photosynthesis from the enzyme nitrogenase, which catalyzes biological N2 fixation and is irreversibly inactivated by O2 [36]. Crocosphaera circumvents this problem by restricting N2 fixation to the nighttime, when O2 is not being produced. The energy needed to fix N2 is generated photosynthetically in the light and stored primarily as carbohydrate granules [37]; respiration of these organic C reserves fuels N2 fixation in the dark. The temporal separation and energetic linkage of photosynthesis and N2 fixation in Crocosphaera produces a daily pattern in the timing and magnitude of PC and PN production and loss. In our study, cultures were grown under high light conditions, producing high growth rates and an especially pronounced diurnal rhythm.

We observed a strong daily pattern in PC production, n class="Chemical">PN production, and cell division by Crocosphaera in all CO2 treatments (Fig. 1). The PC concentration of Crocosphaera cultures fluctuated widely between the light and dark periods: PC increased 48–216% in the light (∼C fixation) and decreased 17–79% in the dark (∼respiration) (Fig. 1B). This substantial dark PC loss is consistent with previous studies of Crocosphaera and reflects the respiration of carbohydrate reserves to fuel N2 fixation [37]–[39]. A fraction of PC may have also been exuded from cells, as it has been shown that Crocosphaera WH8501 can release ∼10% of total C content daily as extracellular polymeric substances [38]; however, dissolved organic C was not measured in our study. The sharp increase in cell concentration following the D12 measurement indicates that cells divided in the first half of the light period (Fig. 1C).

PN production was restricted to the dark period, when n class="Chemical">PN concentrations increased between 48 and 93% (Fig. 1A). Rates of PN increase approximate net N2 fixation rates, though we cannot rule out NO3 or NH4 utilization as driving some small fraction of PN production. While cultures were grown in N-free media [21], the initial dilution of the parent cultures into fresh media resulted in NH4 concentrations of 0.3±0.1 µM on day 0. By day 5, NH4 had increased to 1.1±0.3 µM, supporting the interpretation that the rate of accumulation of inorganic plus particulate N in our cultures was fueled by N2 fixation. The accumulated NH4 had presumably been fixed by Crocosphaera and released from cells; Crocosphaera have been previously observed to release 23–67% of recently fixed N [38]. PN decreased slightly (1–8%) during the light period of the experiment (Fig 1A), which may indicate daily NH4 release. The total NH4 accumulated through the experiment (0.8 µM) represents 0.5–1% of total PN production (day 5–day 0, Table 1).

Together, the high rates of PN production, n class="Chemical">PC production in the day (∼C fixation) and PC loss in the night (∼C respiration) led to large fluctuations in the particulate C:N ratio over the daily cycle: molar C:N ratios in our study ranged from 5.5–22.1 (Fig. 1D). The C:N ratios at D12 time points had a relatively consistent range (∼5.5–8) encompassing the 6.6 ratio predicted from Redfield stoichiometry [40]. Elevated C:N ratios at L6, L12, and D6 time points were driven by the accumulation of organic C reserves to fuel dark N2 fixation and other cellular processes. These daily C:N deviations were independent of pCO2 treatment. Previous studies have reported less dramatic stoichiometric fluctuations in Crocosphaera, with daily C:N content ranging from 6.5–8.5 [39] and 5.0–8.8 [38]. The strong daily C:N deviations observed in our study are consistent with metabolic rates at high growth rates (∼0.5 d−1) of cultures grown at optimum temperature (30°C, [20]) and saturating incoming irradiance (1000 µmol quanta m−2 s−1, [24]). Our experiment spanned both exponential and early stationary growth phases. Cultures grew exponentially from day 0 to day 3, at which point growth rates began to decline (Fig. 1A). The shift in growth phase is evident from non-linearity of natural log-normalized PN growth curves at L6 time points (Fig. 1A) and from decreased PC and PN production rates from day 3–day 5 (Fig. 2). Declines in the magnitude of daily PC fluctuations indicate decreased rates of photosynthesis (∼positive derivative, in light) and respiration (∼negative derivative, in dark) after day 3 (Fig. 1B). Crocosphaera cultures were grown in an artificial medium initially containing an ample supply of macro and micronutrients [16], and PO4 remained replete throughout the experiment (PO4 decreased from 36±1.8 µM on day 0 to 27±1.8 µM on day 5, data not shown). Thus, although we cannot exclude the possibility of limitation by a micronutrient, we hypothesize that the shift to early stationary growth phase resulted from self-shading or a decreased RuBisCO carboxylation efficiency during the second half of the photoperiod, either through direct CO2 limitation or through competitive inhibition of carboxylation from photorespiration at high O2:CO2 ratios [41].

The high Crocosphaera growth rates and cell densities afn class="Chemical">fected the stability of C chemistry within CO2 treatments: photosynthesis and respiration produced pCO2 variability despite continuously bubbling cultures with air/CO2 mixtures (Fig. 1E). The magnitude of pCO2 variability increased throughout the experiment, and by day 5, the high-CO2 treatment had fluctuated between 1216 and 96 µatm (Fig. 1E). In addition, by day 4, all cultures, independent of pCO2 treatment, reached consistent minimum pCO2 values close to 100 µatm in measurements taken at the end of the light cycle (L12); hence, ∼100 µatm appears to represent a physiological limit for the uptake of inorganic C by this organism. Despite the extreme pCO2 fluctuations within treatments, mid- and high-CO2 treatments always had higher pCO2 values than the low-CO2 treatment (which fluctuated between 376 and 74 µatm), with the exception of the final time point. To separate the effect of inflowing pCO2 from the possible confounding effect of cell densities, we present growth rates from day 1 to day 3 when pCO2 fluctuations were less extreme and treatments did not overlap in pCO2 range (Table 2); the mean pCO2 levels measured in this time interval were 252, 477, and 665 µatm for the low-, mid-, and high-CO2 treatments, respectively. The response of Crocosphaera to elevated pCO2 combined with daily, biologically induced pCO2 fluctuations may have ecological implications for bloom scenarios (discussed below).

pCO2 enhancement of growth, PC and PN production

Consistent with previous studies of Crocosphaera WH8501 [7] and n class="Species">Trichodesmium IMS101 [4], [6]–[8], we found that elevating pCO2 increased Crocosphaera growth, PC and PN production rates (Table 2, Fig. 2). Growth in the high-CO2 treatment was significantly higher than in the low-CO2 treatment for growth rates specific to PC, PN, and cell density (Table 2). Cell-specific growth rates were also significantly higher in the mid-CO2 treatment than the low-CO2 treatment (Table 2). Chl a growth rates were not significantly different between CO2 treatments, possibly because of the large coefficient of variation between replicates (7–10%) (Table 2). Biomass-normalized PC and PN production rates were significantly enhanced in mid-CO2 and high-CO2 treatments compared to the low-CO2 treatment (Fig. 2). The pCO2 enhancement of PC and PN production rates was observed both during exponential (day 1–day 3) and early stationary (day 3–day 5) growth phases (Fig. 2).

The only previous CO2 manipulation study using n class="Species">Crocosphaera strain WH8501 tested a pCO2 range of 190–50 µatm and observed that raising ambient pCO2 (380 µatm) to 750 µatm produced 1.2- and 1.4-fold higher rates of C and N2 fixation, respectively [9]. In our study, PC and PN production rates during exponential growth were both ∼1.2-fold higher in the mid-CO2 treatment than the low-CO2 (Fig. 2). The magnitude of pCO2 enhancement we observed is strikingly similar to those reported by Fu et al. [9], especially considering that the environmental conditions utilized in our study differed from the low light (80 µmol quanta m−2 s−1), steady pCO2 conditions of Fu et al. [9]. In our study, including a higher pCO2 treatment displayed even larger enhancements: both PC and PN production rates were ∼1.5-fold higher in the high-CO2 treatment than the low-CO2 treatment. Higher pCO2 treatments would need to be included to determine the threshold pCO2 condition that saturates C and N2 fixation rates of Crocosphaera WH8501.

Conclusions and ecological implications

We observed that elevated pCO2 conditions significantly enhanced n class="Chemical">PC production, PN production, and growth rates of Crocosphaera strain WH8501. This pCO2 enhancement persisted despite biologically induced pCO2 variability in all treatments. By allowing photosynthesis and respiration to drive pCO2 deviations from target values, our methods contrast with those of many previous OA studies, which often keep cultures optically dilute and/or do not report the measured pCO2 time course for each replicate. Though pCO2 in our study varied within treatments, the mid- and high-CO2 treatments had higher pCO2 values than the low-CO2 treatment for nearly all time points (Fig. 1E). Thus, the higher rates of growth, PC and PN production observed in mid- and high-CO2 treatments can be attributed to the elevated pCO2. Furthermore, the differences between treatments observed in the early stationary growth phase reflect low-end estimates of potential pCO2 enhancements: cell densities were highest in the elevated CO2 treatments, possibly leading to more severe growth limitations and dampening evidence for CO2 enhancement.

Our study allows for new insights into the response of Crocosphaera to enhanced n class="Chemical">pCO2 under a variable pCO2 environment. Elevated mean pCO2 enhanced the growth of Crocosphaera cultures despite large pCO2 fluctuations on timescales of less than a generation time, showing that this organism does not need to be acclimated to a stable pCO2 regime to benefit from elevated pCO2. Testing the response of microbes to CO2 perturbations on multiple timescales is ecologically relevant, because net community metabolism, temperature and salinity effects on CO2 solubility, and advective processes cause pCO2 fluctuations on episodic, diurnal, and seasonal timescales. Phytoplankton will experience the long-term OA pCO2 signal superimposed onto this existing pCO2 variability. Furthermore, future OA will increase the DIC:TA ratio of surface waters, leading to a reduced capacity to buffer processes like photosynthesis and respiration, ultimately increasing the magnitude of pCO2 fluctuations [42].

The pCO2 fluctuations observed in our study are probably more extreme than the natural variability experienced by n class="Species">Crocosphaera populations in open-ocean habitats. In the North Pacific Subtropical Gyre, surface pCO2 varies by ∼20–50 µatm seasonally [43]; mesoscale features in this region may cause biologically induced pCO2 swings of ∼150 µatm [44]. However, aggregated cells may experience larger pCO2 swings; for example, Crocosphaera nifH genes have been observed associated with Trichodesmium colonies [14] and could thus experience more extreme pCO2 fluctuations during blooms and subsequent crashes in these concentrated-biomass microhabitats [45]. Regardless of the large magnitude of pCO2 fluctuations employed in our study, our results suggest that elevated mean pCO2 impacts the growth response of Crocosphaera despite short-term variability.

Overall, our study contributes to the growing literature on the response of marine diazotrophs to elevated n class="Chemical">pCO2. We observed that growth, PC and PN production rates of Crocosphaera WH8501 were enhanced under elevated and variable pCO2 in both exponential and early stationary growth phases. It should be noted that a recent study by Garcia et al. [46] found that Crocosphaera strains WH0401 and WH0402 appear to be fully saturated under present day pCO2 conditions (∼400 µatm); thus, elevated pCO2 seems to have strain-specific effects within Crocosphaera. Further research investigating how community composition and environmental conditions regulate the response of marine diazotrophs to elevated pCO2 will be key to predicting whether global rates of N2 fixation will increase under future OA scenarios.

Time series measurements for cultures of Crocosphaera watsonii WH8501 grown under three n class="Chemical">pCO2 treatments. Measured pH, calculated pCO2 (µatm, see Methods) and concentrations of particulate carbon (µmol L−1; PC), particulate nitrogen (µmol L−1; PN), cells (# mL−1), and chlorophyll a (Chl a; µg L−1) are provided for every time point available. Data are mean values from three replicate bottles; standard deviations are presented in parentheses. Dashes indicate no data available.

(DOCX)

Click here for additional data file.