We examined short-term (24–72 h) responses of naturally occurring marine N2 fixing microorganisms (termed diazotrophs) to abrupt increases in the partial pressure of carbon dioxide (pCO2) in seawater during nine incubation experiments conducted between May 2010 and September 2012 at Station ALOHA (A Long-term Oligotrophic Habitat Assessment) (22°45′N, 158°W) in the North Pacific Subtropical Gyre (NPSG). Rates of N2 fixation, nitrogenase (nifH) gene abundances and transcripts of six major groups of cyanobacterial diazotrophs (including both unicellular and filamentous phylotypes), and rates of primary productivity (as measured by 14C-bicarbonate assimilation into plankton biomass) were determined under contemporary (~390 ppm) and elevated pCO2 conditions (~1100 ppm). Quantitative polymerase chain reaction (QPCR) amplification of planktonic nifH genes revealed that unicellular cyanobacteria phylotypes dominated gene abundances during these experiments. In the majority of experiments (seven out of nine), elevated pCO2 did not significantly influence rates of dinitrogen (N2) fixation or primary productivity (two-way analysis of variance (ANOVA), P > 0.05). During two experiments, rates of N2 fixation and primary productivity were significantly lower (by 79 to 82% and 52 to 72%, respectively) in the elevated pCO2 treatments relative to the ambient controls (two-way ANOVA, P < 0.05). QPCR amplification of nifH genes and gene transcripts revealed that diazotroph abundances and nifH gene expression were largely unchanged by the perturbation of the seawater pCO2. Our results suggest that naturally occurring N2 fixing plankton assemblages in the NPSG are relatively resilient to large, short-term increases in pCO2.
Anthropogenic fossil fuel combustion and deforestation continue to alter both atmospheric and oceanic carbon dioxide (CO2) inventories [Meehl et al., 2007]. In large regions of the world's oceans, the partial pressure of carbon dioxide (pCO2) in seawater is increasing at a rate similar to atmospheric pCO2 [Keeling et al., 2004; Santana-Casiano et al., 2007; Dore et al., 2009; Bates et al., 2012]. Changes in the ocean carbonate system have significantly decreased upper ocean pH [Doney et al., 2008; Dore et al., 2009; Feely et al., 2009]. By the year 2150, atmospheric pCO2 is predicted to have tripled relative to contemporary conditions (increasing from ~390 ppm to >1100 ppm), with a concomitant decrease in surface ocean pH of ~0.5 units [Caldeira and Wickett, 2003]. In addition to long-term (decadal-scale) anthropogenic changes in seawater CO2, various naturally occurring physical and biological processes also impart higher frequency variability on the seawater carbonate system. For example, diurnal to seasonal-scale heating and cooling of the upper ocean, or vertical entrainment of dissolved inorganic carbon (DIC)-enriched deep waters through processes such as upwelling or convective mixing can all influence seawater pCO2 and pH [Feely et al., 1999; Dore et al., 2003; Bates, 2007]. Biological processes such as photosynthesis, respiration, and calcification also alter the seawater carbonate system. However, despite considerable research, there remains uncertainty regarding how progressive or abrupt changes in the ocean carbonate system influence marine ecosystems and biogeochemical cycles [Doney et al., 2008; Joint et al., 2011].
Nitrogen (N) is an essential element for biological production in both terrestrial and marine environments [Vitousek and Howarth, 1991]. The most abundant form of N in the biosphere is dinitrogen gas (N2). However, the biological capacity to reduce N2 to ammonia (NH3), a process referred to as N2 fixation, is governed by a restricted group of prokaryotes (termed diazotrophs). In the marine environment, diazotrophs are genetically, physiologically, and ecologically diverse, including free-living and/or symbiotic, filamentous, and unicellular microorganisms [Carpenter and Capone, 2008; Sohm et al., 2011]. Biological N2 fixation represents an important supply of new N to the oligotrophic regions of the open ocean, thereby partly regulating net community production and carbon export [Capone et al., 2005]. In the North Pacific Subtropical Gyre, supply of fixed N by N2 fixation is estimated to be comparable to the delivery of N via mixing, upwelling, and/or advection [Karl et al., 1997; Dore et al., 2002; Johnson et al., 2010]. As a result, oceanic diazotrophs play functionally important roles in marine biogeochemistry.
Most of our understanding of how diazotrophs might respond to future changes in the ocean's carbonate system stems from laboratory studies using cultivated model microorganisms. Such studies have demonstrated that rates of N2 fixation by cyanobacteria belonging to the genera Trichodesmium and Crocosphaera, are sensitive to increases in pCO2, with rates of N2 fixation increasing between 30% to nearly 200% with increases in pCO2 over the range of 750 to 1000 ppm [Barcelos e Ramos et al., 2007; Hutchins et al., 2007; Levitan et al., 2007; Fu et al., 2008]. Similarly, field-based manipulation experiments conducted in both the subtropical Atlantic and the Gulf of Mexico using isolated (“hand-picked”) Trichodesmium colonies from net tows observed 6 to 156% enhancement in rates of N2 fixation during short-term (<24 h) pCO2 enrichment (750 and 800 ppm) experiments [Hutchins et al., 2009; Lomas et al., 2012].
In contrast to these studies demonstrating increases in N2 fixation, others have reported that diazotroph growth and rates of N2 fixation were either unchanged or inhibited under conditions of elevated pCO2. A laboratory study using the heterocyst-forming diazotroph Nodularia (a bloom-forming diazotroph dwelling in brackish, relatively eutrophic inland aquatic ecosystems) revealed that rates of N2 fixation decreased by approximately 40% with increases in pCO2 (750 ppm [Czerny et al., 2009]). In a series of experiments conducted using whole seawater collected in the subtropical South Pacific Ocean, where the diazotroph assemblages were dominated by a group of uncultivated unicellular cyanobacteria referred to as UCYN-A, rates of N2 fixation were largely unaffected by increased pCO2 (750 ppm [Law et al., 2012]). Similarly, a recent study by Gradoville et al.  demonstrated no enhancement of N2 fixation by Trichodesmium colonies isolated from the North Pacific Subtropical Gyre (NPSG) in response to increased seawater pCO2 (up to ~2000 ppm).
In this study, we sought to quantify the short-term (24 to 72 h) physiological responses of diverse, naturally occurring N2 fixing marine cyanobacteria to abrupt increases in seawater pCO2. We conducted a series of shipboard CO2 enrichment experiments at Station ALOHA (A Long-term Oligotrophic Habitat Assessment), the field site for the Hawaii Ocean Time-series (HOT) program in the NPSG. We experimentally examined responses in rates of N2 fixation and primary productivity, together with changes in nitrogenase gene (nifH) abundances and transcription, to pCO2 enrichments (targeting ~1100 ppm). Overall, we observed no consistent response in diazotroph activities in the pCO2-enriched treatments. Our results provide new information on possible implications of a changing ocean carbon cycle on the dynamics of a functionally important group of marine microorganisms.
2.1 Sample Location, Collection, and Experimental Setup
Carbonate system perturbation experiments were conducted aboard the R/V Ka'imikai-O-Kanaloa and R/V Kilo Moana during several cruises to the field site Station ALOHA between May 2010 and September 2012. Near-surface seawater (~250 L from 5 m depth) was collected prior to midnight using polyvinyl chloride sampling bottles affixed to a conductivity-temperature-depth rosette. Seawater was subsampled under subdued light into acid-washed 20 L polycarbonate carboys fitted with sterile, gas-vented caps. After filling, carboys were placed in outdoor, shaded (~50 % surface irradiance), and surface seawater-cooled incubators. Care was used in setting up and sampling the experiments to minimize trace nutrient contamination; however, none of the experiments were conducted under rigorous trace metal clean conditions. For most experiments, target pCO2 levels (~390 or ~1100 ppm) were achieved by gently bubbling the seawater in the carboys with either air (controls, 390 ppm) or a mixture of air and CO2 (treatments, 1100 ppm) for approximately 6–8 h at 3 L min−1 until the desired, elevated pCO2 was attained. Seawater pH was measured at regular intervals over this 6–8 h period and used to compute the approximate seawater pCO2 using CO2SYS software [Lewis and Wallace, 1998]. When the targeted pCO2 was reached (predawn), the flow rate of the air-CO2 mix was adjusted to maintain near-constant seawater pCO2 for the duration of the experiments (<1.5 L min−1). Subsequently, samples for determinations of total alkalinity (TA), dissolved inorganic carbon (DIC), 14C-based primary productivity, chlorophyll a (Chl a) concentrations, rates of N2 fixation, and diazotroph nifH gene abundances and transcripts were collected from each carboy just before dawn (representing the first sampling time point, T0) as described in the following section 2.2. Typically, two additional time points were sampled during each experiment (hence, the resulting sampling points are reported as T0, T24, T48, or T72). Subsamples were withdrawn from the carboys by blocking the vent on the carboy caps and utilizing the resulting positive pressure to draw seawater through silicone tubing.
We also conducted a single experiment (September 2012) to test whether rates of N2 fixation and C fixation were sensitive to “bubbling” as a means of modifying the carbonate system. For this experiment, bubbled treatments consisted of whole seawater (triplicate 20 L carboys) bubbled with air or mixtures of air and CO2 (as described above) to achieve the targeted pCO2 of ~390 and 1100 ppm. Nonbubbled treatments consisted of whole seawater (triplicate 20 L carboys) where trace metal grade hydrochloric acid (together with sodium bicarbonate) was added to increase pCO2 to ~ 1100 ppm while minimizing changes in seawater TA. Controls (~390 ppm) for the nonbubbled treatments consisted of unperturbed whole seawater in triplicate carboys held in the same seawater incubators. DIC and TA were measured at each time point to examine possible daily scale variations in pCO2.
2.2 Total Alkalinity, Dissolved Inorganic Carbon, and pH
Seawater samples for determinations of TA, DIC, and pH were collected and analyzed following HOT program procedures and methodologies [Winn et al., 1998; Dore et al., 2009]. DIC and TA samples were collected into 300 mL, precombusted borosilicate bottles and immediately preserved with 100 μL of saturated mercuric chloride (HgCl2) for subsequent analysis in the laboratory. TA was determined using an automated, closed-cell potentiometric titration and DIC was measured coulometrically using a Versatile INstrument for the Determination of Total inorganic carbon and titration Alkalinity 3S system [Dickson et al., 2007]. Accuracy and precision of the DIC and TA measurements were validated through analysis of certified seawater CO2 reference material [Dickson, 2001], with a resulting accuracy of approximately 1 µmol L−1 and 3 µmol L−1, respectively. Salinity-normalized values for DIC and TA were calculated as follows: nDIC = (35 × DIC)/(measured salinity) and nTA = (35 × TA)/(measured salinity). Seawater pH (measured at 25°C) was analyzed by the spectrophotometric detection of m-cresol purple with a precision of 0.001 [Clayton and Byrne, 1993]. To examine seasonal variability in the carbonate system at Station ALOHA, long-term (1988–2011) trends in seawater pCO2, pH, nDIC, and nTA (determined by least squares linear regression analyses) were subtracted from the time series data.
2.3 14C-based Primary Productivity (C Fixation) and Concentrations of Chlorophyll a
Samples for analyses of inorganic nutrient concentrations were subsampled from each carboy into clean, acid-washed 500 mL polyethylene bottles and immediately frozen upright at −20°C until analyses. At the shore-based laboratory, samples were thawed and split for measurements of nitrate plus nitrite (NO3− + NO2−), phosphate (PO43−), and silicic acid (Si(OH)4). High-sensitivity analytical techniques were chosen for [PO43−] and [NO3− + NO2−] because concentrations of these nutrients at Station ALOHA are at or below the detection limits of standard autoanalyzer methods [Karl et al., 2001]. Samples for the determination of [NO3− + NO2−] were analyzed using the chemiluminiscent method described by Garside  as modified by Dore and Karl . [PO43−] was determined using the MAGnesium Induced Coprecipitation assay [Karl and Tien, 1992]. [Si(OH)4] was quantified following the procedures described by Hansen and Koroleff .
2.5 N2 Fixation Rates, Gene Abundances, and Transcripts of Diazotrophs
Rates of N2 fixation were assessed by the 15N2 isotopic tracer technique [Montoya et al., 1996]. For experiments conducted between May 2010 and April 2011, whole seawater samples were subsampled from each carboy into clean, acid-washed 4.5 L polycarbonate bottles without introducing air bubbles. Bottles were sealed with caps fitted with silicone septa. After capping, 3 mL of 15N2 gas (98%; Isotech Laboratories®) was injected into each polycarbonate bottle through the silicone septum using a gas-tight syringe and a stainless steel needle. However, a previous study [Mohr et al., 2010] reported underestimation of rates of N2 fixation by these standard procedures, a finding these authors attributed to the time required for the 15N2 gas bubble to equilibrate with the seawater. Hence, for the experiment carried out in September 2012, the 15N2 gas was first dissolved into seawater (prepared as described in Wilson et al. ) and the resulting 15N2-enriched water was added to 4.5 L polycarbonate sampling bottles. Regardless of the method used to estimate rates of N2 fixation (adding 15N2 gas in gaseous or dissolved form) in the present study, rates of N2 fixation in the pCO2 perturbed treatments were compared to controls, which were measured by identical procedures during each individual experiment. All 15N2-enriched bottles were incubated for 24 h in the same surface seawater shaded incubator previously described. After the incubation period, the entire volume was filtered onto a precombusted 25 mm diameter, GF/F filter and subsequently stored frozen at −20°C. In the shore-based laboratory, these filters were dried for 24 h at 60°C and pelleted, and 15N assimilation on each filter was analyzed with an elemental analyzer-isotope ratio mass spectrometer (Carlo-Ebra EA NC2500®). Rates of N2 fixation were calculated as described in Capone and Montoya .
In addition, whole seawater samples (4 L) were collected from each carboy into clean, acid-washed 4 L polycarbonate bottles for subsequent determination of cyanobacterial nifH genes and transcripts. Approximately 2 L each was filtered through 25 mm diameter 0.22 µm Durapore filters (Millipore®) to concentrate cells for subsequent extraction of DNA and RNA, respectively; nucleic acid filtrations were typically completed within 45 min. Filters for subsequent RNA analysis were stored in microcentrifuge tubes containing 500 μL of a buffer containing 10 μL ß-mercaptoethanol to 1 mL Buffer RLT (Qiagen®). Filters for subsequent DNA analysis were kept in 500 μL buffer containing 20 mM Tris-HCl, pH 8.0; 2 mM EDTA, pH 8.0; 1.2% Triton X and 20 mg mL−1 lysozyme. Microcentrifuge tubes containing the filters and buffers were flash frozen in liquid nitrogen and kept at −80°C until extraction. In the shore-based laboratory, DNA and RNA were extracted from filters using the DNeasy kit (Qiagen®) or RNeasy Mini Kit (Qiagen®), respectively, following the manufacturer's protocols.
Determinations of diazotroph nifH gene abundances (gene copies L−1) and nifH gene transcripts (gene transcripts L−1) were assessed by quantitative polymerase chain reaction (QPCR). Six different nifH phylotypes were targeted by QPCR, including UCYN-A, Crocosphaera spp., Trichodesmium spp., and three types of heterocyst-forming cyanobacteria (termed Het1, Het2, and Het3) identified as deriving from Richelia spp. (Het1 and Het2) and Calothrix spp. (Het3 [Foster and Zehr, 2006]). These phylotypes are commonly retrieved from nifH clone libraries at Station ALOHA [e.g., Church et al., 2009] and were amplified using group specific primers and probes as described in Church et al. [2005a, 2005b]. Both QPCR (for gene abundances) and reverse transcriptase (RT)-QPCR (for gene transcripts) reactions were performed in duplicate. For RT-QPCR, extracted mRNA was reverse transcribed (SuperScript III RT cDNA synthesis Kit, Invitrogen®) as described in Church et al. [2005a, 2005b], and the resulting cDNA was amplified in the QPCR reactions.
2.6 Statistical Analysis
Two-way analysis of variance (two-way ANOVA) was utilized to identify statistical patterns, using treatments (ambient versus CO2-elevated) and incubation time as factors, checking data for normality and heterogeneity of variance [Zar, 1998]. Standard deviations of the replicate treatments (n = 3) are reported whenever applicable.
3.1 Habitat Characteristics and Diazotroph Community Structure at Station ALOHA
Short-term CO2 perturbation experiments were conducted on nine cruises to Station ALOHA between May 2010 and September 2012 with the majority of experiments occurring during the spring and summer months (Table 1). Near-monthly HOT program sampling of this site since October 1988 provides robust temporal context in which to place our carbonate system perturbation experiments. The CO2 perturbation experiments conducted for this study were initiated under conditions spanning the full annual range of surface ocean carbonate system properties observed at Station ALOHA (Figure 1). During our study period, seawater pCO2 in the near-surface ocean (5 m) fluctuated between 339 and 404 µatm (Figure 1a). Detrended seawater pCO2 revealed a seasonal pattern of elevated pCO2 (>365 µatm) during the late summer and fall (August– October) when seawater temperatures peak (>25°C; Table 1), with lower pCO2 during the winter months. Near-surface ocean pH varied by ~ 0.03 units during our sampling (ranging from 8.055 to 8.081), typical for the region. Removal of the long-term downward trend in seawater pH revealed a seasonal pattern in pH out of phase with pCO2, with lowest pH observed in the summer when pCO2 was maximal (Figure 1b). Detrended nDIC fluctuated ~10 µmol C L−1 over the study, varying between 1964 and 1974 µmol C L−1, with elevated concentrations in the late winter and spring (Figure 1c). Consistent with the long-term HOT measurements, nTA varied little during our sampling, ranging between 2302 and 2315 µmol L−1 with no apparent seasonality (Figure 1d). In addition, the near-surface waters at Station ALOHA were persistently oligotrophic throughout our study, as reflected by low concentrations of inorganic nutrients (i.e., N + N was consistently <5 nmol N L−1, PO43− was <75 nmol P L−1), and chlorophyll a (<0.15 µg L−1; Figures 2a–2c and Table 1). Rates of 14C-based carbon fixation in the near-surface ocean were typical for the site, ranging between 0.3 and 0.7 µmol C L−1 d−1 (Figure 2d, Table 1).
Table 1. Near-Surface (5 m) Seawater Characteristics and Sampling Time Points for Each of the CO2 Perturbation Experiments Between May 2010 and April 2011a
Time points (hours)
PO43− (nmol P L−1)
NO3− + NO2− (nmol N L−1)
Chl a (µg L−1)
14C-PP (µmol C L−1 d−1)
nDIC (µmol C L−1)
nTA (µmol L−1)
Temperature (T), phosphate (PO43−), nitrate + nitrite (NO3− + NO2−), chlorophyll a (Chl a), C fixation (14C-PP), salinity-normalized dissolved inorganic carbon (nDIC), salinity-normalized total alkalinity (nTA), and partial pressure of carbon dioxide (pCO2). ND = not determined.
0, 24, 48
0, 24, 48
0, 48, 72
0, 24, 72
0, 24, 72
3.2 Stability and Variability of pCO2 During Experiments
We measured DIC and TA at all sampling time points during the CO2 perturbation experiments and used those two variables (as recommended by Dickson ) to compute seawater pCO2 in the control and CO2 perturbed treatments (Table 2) using CO2SYS software. The pCO2 in the control carboys varied between 336 ± 4 and 415 ± 7 µatm, but was relatively constant (varying <5%) over the course of any one experiment (Table 2). The pCO2 attained in the perturbed treatments ranged from 920 ± 17 to 1144 ± 30 µatm, and temporal variations within each experiment were comparable to the control carboys. During the experiment conducted in September 2012, we perturbed the CO2 system by adding hydrochloric acid (HCl) and sodium bicarbonate (NaHCO3). However, the pCO2 was not maintained throughout the remaining duration of the experiment (as compared to the experiments which were continuously aerated with mixtures of CO2 and air), resulting in a decline (by ~19%) in pCO2 toward the end of the experiment (72 h) in the pCO2-elevated treatment (Table 2).
Table 2. Mean (±Standard Deviation of Triplicate Carboys When Available) Partial Pressure of Carbon Dioxide (pCO2) in the Control and CO2 Perturbed Treatments at the Various Time Points of Each Experiment
Control pCO2 (µatm)
Elevated pCO2 (µatm)
Time Points (h)
First Time Point
Second Time Point
Third Time Point
First Time Point
Second Time Point
Third Time Point
CO2 system was modified through bubbling mixtures of CO2 and air.
CO2 system was modified through the addition of hydrochloric acid (HCl) and sodium bicarbonate (NaHCO3).
Rates of N2 fixation in experiments conducted between May 2010 and April 2011 in both the controls (0.6 ± 0.2 and 6.7 ± 3.5 nmol N L−1 d−1) and CO2 treatments (0.3 ± 0.1 and 6.8 nmol N L−1 d−1; Table 3) were in the range of previously reported data for the region [Church et al., 2009]. The only exception when N2 fixation rates were substantially higher occurred during the experiment conducted during September 2012, where we utilized a revised 15N2 tracer technique; the 15N2 tracer was dissolved in seawater prior to adding to the experimental incubations instead of adding the 15N2 as a gaseous bubble [Mohr et al., 2010; see section 2.5]. As has been previously observed [Wilson et al., 2012; Großkopf et al., 2012], N2 fixation rates in both the controls and CO2 perturbed treatments during this experiment were substantially greater than in experiments where 15N2 was added as a bubble with rates ranging between 19.4 ± 2.8 and 32.2 ± 2.2 in the controls and from 16.8 ± 2.5 to 31.5 ± 0.9 nmol N L−1 d−1 in the pCO2-elevated samples (Table 3). Rates among the replicate carboys varied by 3–71% (= coefficient of variance), with variability among replicates similar in both the controls and CO2 perturbed treatments (Table 3). In most cases (seven of the nine experiments), rates of N2 fixation in the CO2 perturbed treatments were not significantly different from the controls (two-way ANOVA, P > 0.05; Table 3). In the remaining two experiments, rates of N2 fixation in the CO2 treatments were significantly lower (two-way ANOVA, P < 0.05) than the controls (Table 3 and Figure 3a).
Table 3. Mean Rates (±Standard Deviation of Triplicate Carboys) of N2 Fixation and Primary Productivity in Control and CO2-Elevated Treatments at Different Time Points of the Experimentsa
Sampling Time (h)
Control (nmol N L−1 d−1)
CO2 Treatment (nmol N L−1 d−1)
P Level (Treatment)
P Level (Time)
Control (µmol C L−1 d−1)
CO2 Treatment (µmol C L−1 d−1)
P Level (Treatment)
P Level (Time)
Also depicted are statistical results from two-way analyses of variance. Significance level (p): ns = p > 0.05,
= p <0.05,
= p <0.01, and
= p <0.001.
CO2 system was modified through bubbling mixtures of CO2 and air.
CO2 system was manipulated by the addition of hydrochloric acid (HCl) and sodium bicarbonate (NaHCO3).
N2 fixation rates were estimated by injecting 15N2 in gaseous form.
N2 fixation rates were assessed by adding 15N2 dissolved in seawater.
We also investigated the response in rates of 14C-based primary productivity to CO2 enrichment. Rates of primary productivity measured in the controls ranged from 0.3 ± 0.03 to 1.2 ± 0.3 µmol C L−1 d−1, while rates in the CO2 treatments ranged between 0.2 ± 0.03 and 0.8 ± 0.2 µmol C L−1 d−1. As observed with N2 fixation, rates of productivity did not demonstrate consistent changes in response to CO2 enrichment (Table 4). In seven of the nine experiments, primary productivity in the CO2 treatments was not significantly different from the controls (two-way ANOVA, P > 0.05), and in the remaining two experiments, rates were significantly lower (by 52 to 72%) than the controls (two-way ANOVA, P < 0.05; Figure 3b).
Table 4. Summary of the Effects of Ocean Acidification on Processes Associated With Marine Diazotrophsa
Laboratory experiments targeted monospecific cultures, whereas field-based experiments were conducted with isolated field populations or whole, mixed assemblages. + = increase in measured property under elevated relative to contemporary pCO2 conditions, − = decrease in measured property under elevated pCO2 conditions, and o = no difference between elevated and contemporary pCO2 conditions. Gulf of Mexico (GOM), Subtropical North Atlantic (SNA), North Pacific Subtropical Gyre (NPSG), South Pacific Subtropical Gyre (SPSG).
During one of our field-based experiments (September 2012), we simultaneously estimated rates of N2 fixation and primary productivity in treatments that were aerated at the targeted pCO2 (“bubbled”) or altered by the addition of HCl and NaHCO3 (“nonbubbled”; Figure 4). Comparison of the resulting rates of N2 fixation and 14C-bicarbonate fixation revealed generally no significant differences among these methods of perturbing the seawater carbonate system (one-way ANOVA, P > 0.05; Figure 4).
3.4 Response of nifH Gene Abundance and Transcripts to Increasing Seawater pCO2
We examined diazotroph population dynamics in our experiments based on QPCR and RT-QPCR amplification of nifH genes and gene transcripts. In five of eight experiments where nifH genes were quantified, the UCYN-A phylotype accounted for between 40 and 98% of the gene abundances at the beginning of the experiments (Figure 5). In a single experiment (August 2010), gene abundances of Crocosphaera rivaled those of UCYN-A, and in the remaining two experiments (September and October 2010), Crocosphaera dominated gene abundances, contributing 80% and 95%, respectively, of the quantified nifH genes (Figure 5). Relative to unicellular cyanobacteria, the abundances of filamentous diazotrophs (including Trichodesmium and the three phylotypes of heterocystous cyanobacteria) were low in all of the experiments. The nifH gene abundances of the two unicellular phylotypes (UCYN-A and Crocosphaera) were often at least 2 orders of magnitude greater than abundances of these filamentous phylotypes (Figure 6). Overall, there were no significant differences in nifH gene abundances for any of the six phylotypes between the control and CO2-enriched treatments (Figure 6). In addition, gene abundances did not vary significantly in time over the course of the various experiments (two-way ANOVA, p > 0.05), suggesting the incubation conditions (confinement of the organisms) did not stimulate or inhibit the growth of these organisms.
We also quantified nifH gene transcription by these same six cyanobacterial phylotypes (Figure 6). In all of our experiments, gene transcripts of Trichodesmium spp. were 1–2 orders of magnitude greater than nifH gene transcripts derived from the unicellular or the heterocystous phylotypes quantified. The dominance of Trichodesmium nifH transcripts likely reflects the time of day the samples were collected (predawn), which is consistent with previously reported diel changes in Trichodesmium nifH expression at this location [Church et al., 2005a]. As observed with rates of N2 fixation and primary productivity, increasing pCO2 had little influence on nifH gene transcription (Figure 6) with the exception of Crocosphaera nifH gene transcripts (during the experiment conducted in October 2010), which were nearly tenfold greater in the pCO2 treatment than the control after 72 h of incubation (1.9 × 104 ± 9.8 × 103 versus 2.1 × 105 ± 1.4 × 105 gene transcripts L−1, respectively).
We sought to examine whether projected future changes in the ocean carbonate system might impact diazotroph activities and abundances in the NPSG, one of the largest marine ecosystems on the planet. Although our goal was to gain insight into potential responses by contemporary marine diazotroph assemblages to shifts in the seawater carbonate system projected for the latter part of this century, we acknowledge that results from short-term incubation experiments do not provide information on shifts in the microbial species or gene evolution resulting from selective pressure due to long-term environmental changes. In this context, our experiments were designed to examine the physiological responses of naturally occurring contemporary diazotroph assemblages to short-term (48 to 72 h) abrupt perturbations in a single global change variable, namely seawater pCO2.
In the majority of our experiments, we observed no statistically significant responses (positive or negative relative to the controls) in rates of N2 fixation, nifH gene abundances, or nifH gene transcription in response to increases in seawater pCO2. During two experiments, rates of N2 fixation were significantly lower in the pCO2-enriched treatments relative to the experimental controls (Table 3). Similarly, we did not observe consistent responses in rates of primary productivity (Table 3) or concentrations of Chl a to elevated seawater pCO2. In most experiments, rates of primary productivity in the CO2 perturbed treatments were statistically indistinguishable from the controls. During two experiments, rates of productivity were significantly lower in the elevated CO2 treatments than in the controls. Such findings suggest that rates of both N2 fixation and primary productivity in the NPSG are relatively insensitive to large, abrupt changes in seawater pCO2.
Findings from our field-based experiments using naturally occurring N2-fixing cyanobacteria, including those for which there are currently no cultivated representatives, contrast with results from laboratory-based studies in which significant stimulation of both C and N2 fixation by cultivated representatives of the marine cyanobacteria Trichodesmium and Crocosphaera has been observed under enhanced pCO2 conditions similar to those tested in the present study (Table 4). Such laboratory studies have highlighted the possibility that long-term increases in seawater pCO2 could promote the growth and activities of marine diazotrophs. However, there have been comparatively few field-based assessments on the responses of diazotrophs to changes in seawater CO2 (Table 4). In two field experiments conducted using naturally occurring Trichodesmium colonies from the Gulf of Mexico and the Sargasso Sea, rates of N2 fixation (as measured by acetylene reduction or 15N2 assimilation) increased between 6 and 156% in pCO2-enriched treatments (where pCO2 was ~750 ppm) relative to unperturbed controls [Hutchins et al., 2009; Lomas et al., 2012]. However, a similar study conducted in the NPSG detected no enhancement of N2 fixation by isolated Trichodesmium colonies to elevated pCO2 [Gradoville et al., 2014].
In the South Pacific Subtropical Gyre (SPSG), Law et al.  conducted a series of pCO2 perturbation experiments using whole seawater, and similar to the results of the present study, these authors found no consistent response in rates of N2 fixation to abrupt increases (750 ppm) in seawater pCO2. Taken together, results from the Law et al.  and Gradoville et al.  studies and the experiments conducted as part of the present work suggest that CO2 is not a limiting factor controlling the growth and activities of natural assemblages of ocean diazotrophs in the large regions of the Pacific Ocean (NPSG and SPSG). Moreover, these results suggest that the activities of contemporary diazotroph assemblages in these habitats are relatively insensitive to large, abrupt changes in the seawater carbonate system. Such findings have potentially important implications for our capacity to predict future changes to plankton biogeochemistry and net ecosystem productivity in some of the largest ecosystems on the planet. Specifically, based on results from laboratory studies, marine diazotrophs are often predicted to become increasingly important to ocean carbon cycling with progressive increases to seawater pCO2. However, our field-based experiments, together with those of Law et al. , suggest that enrichment of the upper ocean with CO2 may not stimulate the activities or net growth (as reflected by no changes in nifH abundances in our experiments) of naturally occurring N2 fixing microorganisms in the Pacific Ocean.
There are a number of reasons diazotroph responses in our field-based experiments might differ from those observed in laboratory-based studies. The lack of response, or in some cases, detrimental impacts of increased pCO2 on rates of N2 fixation and primary productivity observed in the present study presumably reflect differences in the physiological responses of diverse, naturally occurring planktonic assemblages in the NPSG relative to the marine diazotroph isolates examined to date. The resulting responses to CO2 in these field experiments reflect the aggregate behavior of a mixed assemblage of plankton, including diazotrophs, which possess diverse metabolic capabilities and presumably differing physiological sensitivities to variations in CO2. Similar to the Law et al.  study, in the majority of our experiments, the diazotroph assemblage (as assessed based on nifH gene quantifications) was dominated by UCYN-A (comprising between ~40 and 98% of the quantified nifH genes; Figure 5). Time series measurements of N2 fixation rates at Station ALOHA suggest that small, unicellular diazotrophs (<10 µm) consistently dominate rates of N2 fixation at this site [Church et al., 2009]. Although UCYN-A remains uncultivated, gene-based surveys and isotope-labeling studies have provided insights into the ecology and metabolic potential of these microorganisms. The recent work of Thompson et al.  for instance discovered that this organism can exist as an epiphyte associated with a unicellular picoplanktonic prymnesiophyte. Metagenomic analyses of cell sorts enriched in UCYN-A revealed the absence of suites of genes necessary for oxygenic photosynthesis (including photosystem II) and fixation of inorganic carbon (via the Calvin-Benson-Basham cycle); however, these organisms appear to have retained key genes encoding components of photosystem I [Zehr et al., 2008; Tripp et al., 2010]. Such results hint at a presumed photoheterotrophic metabolism for UCYN-A [Zehr et al., 2008]. A number of studies have hypothesized that photoautotrophs relying on carbon-concentrating mechanisms (CCM) to concentrate CO2 at the active site of the RuBisCO enzyme may benefit from increases in pCO2 through reduction in energetic demands of the CCM components [Kaplan et al., 1991; Price et al., 2008; Raven, 2003]. There is some evidence supporting this hypothesis in laboratory experiments with Trichodesmium [Levitan et al., 2007; Kranz et al., 2009]; in these studies, pCO2 was increased (up to 900 ppm) and Trichodesmium was observed to downregulate CCM activity and allocate greater energy reserves toward fueling N2 fixation. The absence of RuBisCO and CCMs in the UCYN-A genome suggests that these organisms may not experience the same increase in growth efficiency under conditions of elevated pCO2. The dominance of UCYN-A (and relatively low abundance of Trichodesmium) and the inability of UCYN-A to fix CO2 may underlie the lack of response in N2 fixation to increases in CO2 observed in our experiments.
In an attempt to assess the potential impact of the mode of manipulating the CO2 system in field-based experiments, we simultaneously estimated rates of N2 fixation (and primary productivity) in gently aerated versus carboys where the carbonate system was perturbed through the addition of HCl and NaHCO3 (Figure 4). We observed no significant difference in rates of N2 fixation between the controls and CO2 perturbed treatments in this experiment. Moreover, we also observed no significant differences in rates of N2 fixation and 14C-carbon fixation between the gently aerated and the HCl/NaHCO3 perturbed carboys. These results suggest that the gentle aeration procedures recommended for pCO2 perturbation experiments [Riebesell et al., 2011] and utilized for most of our experiments were not detrimental to N2 fixation.
On three occasions (August, September, and October 2010), we initiated experiments in which the unicellular cyanobacterium Crocosphaera contributed significantly to the total nifH gene abundances (47–95%; Figure 5). Crocosphaera can be a significant contributor to oceanic N2 fixation [e.g., Montoya et al., 2004; Moisander et al., 2010] and cultured isolates of Crocosphaera (strain WH8501) have been shown to increase rates of N2 fixation in iron (Fe)-replete conditions after brief acclimation to elevated CO2 concentrations [Fu et al., 2008]. In one of our experiments (October 2010), we did observe an increase in Crocosphaera nifH expression under elevated pCO2 conditions (Figure 6), but even when Crocosphaera was a dominant component of diazotroph abundance (based on quantification of nifH genes), rates of N2 fixation in our experiments remained either unchanged or decreased under enhanced pCO2 conditions. To date, it remains unclear how variability in gene transcript abundances relate to rates of N2 fixation; however, the lack of a measurable response in N2 fixation to elevated pCO2, despite significant increases in transcriptional activity could reflect posttranscriptional and/or posttranslational modification of nifH gene transcripts by Crocosphaera. In a study examining genetic variability among strains of Crocosphaera (including several from the North Pacific), Webb et al.  found that despite low apparent genetic variability, these strains exhibited notable differences in their physiologies. Hence, there may be “ecotype”-dependent differences in the response of Crocopshaera to changes in seawater pCO2; unfortunately, the QPCR nifH gene assays utilized in the present study do not allow us to distinguish potential ecotype-specific responses by Crocosphaera. In addition, a recent study by Hutchins et al.  shows considerable taxon-specific differences in the CO2 sensitivities of rates of N2 fixation. For example, the CO2 half-saturation constant for Crocosphaera watsonii (WH8501) was estimated at 162 ppm, suggesting N2 fixation by this strain might be nearly saturated under present CO2 conditions.
Our results may also differ from findings with monospecific cultures due to the types of resources limiting diazotroph growth in the ocean relative to conditions examined in the laboratory. Besides temperature and light, various essential nutrients have been hypothesized as possible controls on diazotroph growth in the open sea, including phosphorus, molybdenum, iron, and nickel [e.g., Sañudo-Wilhelmy et al., 2001; Mills et al., 2004; Ho, 2013]. Several laboratory studies have examined whether nutrient limitation plays an interactive role in CO2 stimulation of diazotroph growth. In a series of laboratory-based experiments with a strain of Crocosphaera, Fu et al.  measured elevated rates of N2 fixation in response to increased CO2 in Fe-replete cultures, but no CO2-dependent response was observed under Fe-limited growth conditions. Shi et al.  observed a pH-dependent decrease in rates of N2 fixation in iron-limited cultures of Trichodesmium, a finding the authors attribute to changes in iron chemistry of the seawater medium rather than physiological alteration of Trichodesmium growth. A recent field-based study conducted by Gradoville et al.  found that increasing pCO2 did not significantly enhance N2 fixation by Trichodesmium colonies under Fe-enriched conditions (100 nmol Fe L−1). We did not employ rigorous trace metal clean sampling techniques for the experiments described in the present study and did not measure concentrations of iron in our incubations; hence, we are unable to determine the extent to which we may have inadvertently introduced trace nutrients to these incubations. Iron concentrations in the near-surface ocean at Station ALOHA have been reported to range between 0.2 and 0.7 nmol L−1 [Boyle et al., 2005] and these concentrations have been considered nonlimiting [Dutkiewicz et al., 2012]. Thus, we suspect that the elevated Fe concentrations typical of ALOHA likely resulted in Fe-replete conditions during our experiments, making it unlikely that Fe limitation resulted in the lack of response in our CO2 perturbation experiments.
In contrast, the availability of phosphorus (P) in the upper ocean at Station ALOHA, where concentrations are generally <0.1 µmol L−1 [Björkman and Karl, 2003], has been shown to restrict the metabolic activity of natural populations of diazotrophs [e.g., Grabowski et al., 2008; Sohm et al., 2008; Watkins-Brandt et al., 2011]. Hutchins et al.  demonstrated that rates of N2 fixation by Trichodesmium strains GBR (isolated from the Pacific) and IMS (isolated from the Atlantic) grown either under P-replete (20 µmol P L−1) or P-limiting (0.2 µmol P L−1) conditions were enhanced with increasing CO2. Thus, even if the diazotrophs sampled as part of these experiments were P limited, we still should have anticipated stimulation in rates of N2 fixation in response to increases in CO2. The lack of response in our experiments suggests that CO2 is not limiting diazotroph growth in the NPSG; rather, we suspect the persistently low concentrations of other important growth-requiring nutrients (e.g., vitamins and nickel) observed in the surface ocean of this ecosystem might be more likely to restrict diazotroph growth [e.g., Barada et al., 2013; Ho, 2013; Sañudo-Wilhelmy et al., 2014] and obscure potentially stimulatory influences of CO2 on these microorganisms.
The inconsistencies between field-based CO2 manipulation experiments conducted with diverse natural assemblages of marine diazotrophs and laboratory experiments conducted using monospecific isolates highlights the utility of both cultivation-dependent and independent approaches for informing our understanding of diazotroph physiology, ecology, and biogeochemistry. While field-based experiments such as those conducted as part of this study provide insight into plankton ecology and biogeochemistry in a more realistic setting that includes interactions among organisms, laboratory experiments provide important clues into potential physiological and genetic changes by isolated microorganisms. Moreover, the relatively slow growth rates of diazotrophs (~0.15 day−1 for Trichodesmium spp. or ~ 0.35 for Crocosphaera spp. [LaRoche and Breitbarth, 2005; Webb et al., 2009]) make laboratory experiments well suited for examining hypotheses focused on acclimatization and adaptability by diazotrophs to slow and progressive ecosystem changes, including those accompanying future changes to seawater CO2. Further laboratory-based and field-based experiments will hopefully provide additional information on how rising CO2 in combination with other changes in climate-sensitive processes that were not explored in the present study (e.g., warming) may influence the physiology of marine N2 fixing cyanobacteria, and the ocean C and N cycle.
We thank the scientists and staff of the Hawaii Ocean Time-series (HOT) for their support of this work. We particularly acknowledge the efforts of D. Sadler for his assistance in setting up experiments and overseeing the carbonate system measurements in this work, and B. Updyke for his assistance with analyses and sampling. We would also like to thank J. Dore (MSU) for providing seawater carbonate determinations from Station ALOHA. In addition, B. Wai, A. Mine, S. Thomas, and C. Johnson aided in sampling and laboratory analyses. We also thank the captains and crew of the R/V Kilo Moana and R/V Ka'imikai-O-Kanaloa. This project was supported by a grant from the National Science Foundation (OCE-08-50827) to M.J.C. and R.M.L. The comments of two anonymous reviewers significantly improved the manuscript. Additional support was derived through the Center for Microbial Oceanography: Research and Education (C-MORE; NSF grant EF04-24599), HOT (NSF grant OCE-09-26766), and the Gordon and Betty Moore Foundation (to D.M.K.).