Carbon content, carbon fixation yield and dissolved organic carbon release from diverse marine nitrifiers

Abstract Nitrifying microorganisms, including ammonia‐oxidizing archaea, ammonia‐oxidizing bacteria, and nitrite‐oxidizing bacteria, are the most abundant chemoautotrophs in the ocean and play an important role in the global carbon cycle by fixing dissolved inorganic carbon (DIC) into biomass. The release of organic compounds by these microbes is not well quantified, but may represent an as‐yet unaccounted source of dissolved organic carbon (DOC) available to marine food webs. Here, we provide measurements of cellular carbon and nitrogen quotas, DIC fixation yields and DOC release of 10 phylogenetically diverse marine nitrifiers. All investigated strains released DOC during growth, representing on average 5–15% of the fixed DIC. Changes in substrate concentration and temperature did not affect the proportion of fixed DIC released as DOC, but release rates varied between closely related species. Our results also indicate previous studies may have underestimated DIC fixation yields of marine nitrite oxidizers due to partial decoupling of nitrite oxidation from CO2 fixation, and due to lower observed yields in artificial compared to natural seawater medium. The results of this study provide critical values for biogeochemical models of the global carbon cycle, and help to further constrain the implications of nitrification‐fueled chemoautotrophy for marine food‐web functioning and the biological sequestration of carbon in the ocean.

A fraction of the carbon dioxide (CO 2 ) that is captured by marine phytoplankton at the surface sinks to depth as dead organic material, supporting a deep ocean food-web of both microbes and higher trophic levels (Hannides et al. 2013;Giering et al. 2014;Choy et al. 2015). Organic matter decomposition in the mesopelagic releases ammonium, a reduced form of nitrogen that can be used as an energy source by chemoautotrophic nitrifying archaea and bacteria to fuel dissolved inorganic carbon (DIC) fixation into biomass (Ward 2011). Chemoautotrophic production provides a new, labile, nonsinking source of particulate organic matter to the deep ocean which is otherwise dominated by refractory organic carbon Middelburg 2011), supporting the heterotrophic microbial community in the mesopelagic (Hansman et al. 2009;Baltar et al. 2010).
The main nitrifiers in the ocean are ammonia-oxidizing archaea, which oxidize ammonia (NH 3 ) to nitrite (NO 2 À ), and nitrite-oxidizing bacteria, which further oxidize NO 2 À to nitrate (NO 3 À ) (Ward 2011). These two steps are assumed to be tightly coupled, as NO 2 À typically does not accumulate in oxic, open ocean waters with measurable nitrification, with the exception of the primary nitrite maximum at the base of the euphotic zone (Lomas and Lipschultz 2006;Santoro et al. 2013). Despite this tight coupling, ammonia-oxidizing archaea are approximately six times more abundant than nitrite-oxidizing Nitrospinae bacteria at a given location and sampling depth throughout the Pacific Ocean (Santoro et al. 2019), possibly owing to their smaller cell size compared to nitrite-oxidizing bacteria (e.g., Watson and Waterbury 1971;Mueller et al. 2021;Qin et al. 2017), or as a result of the higher theoretical energy yield from ammonia compared to nitrite oxidation (Bock and Wagner 2013). Ammoniaoxidizing bacteria are thought to play a minor role in global ocean nitrification due to their overall low abundances (Santoro et al. 2010;Buchwald et al. 2015;Tolar et al. 2016). Despite the known difference in theoretical energy yield, there are many uncertainties regarding the organic carbon *Correspondence: barbara.bayer@univie.ac.at This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Additional Supporting Information may be found in the online version of this article. yield from ammonia vs. nitrite oxidation (hereinafter referred to as DIC fixation yield) and the contribution of both physiological groups to chemoautotrophic DIC fixation in the ocean. Cultures of ammonia-oxidizing archaea have recently been shown to release dissolved organic carbon (DOC) during growth (Bayer et al. 2019a), yet this component of organic carbon is not captured by conventional methods measuring DIC incorporation into biomass. As such, the release of DOC by chemoautotrophs might represent an as-yet unaccounted source of organic material in the deep ocean potentially fueling the microbial loop, with important implications for the marine carbon cycle. However, it remains unclear if DOC release is a phenomenon only observed under specific culture conditions, restricted to select strains of ammonia-oxidizing archaea, or a common feature shared by diverse autotrophic nitrifiers under natural conditions.
Here, we report combined measurements of DIC fixation and DOC release of 10 phylogenetically diverse marine nitrifiers comprising two genera of ammonia-oxidizing archaea (Nitrosopumilus and Ca. Nitrosopelagicus), one genus of ammonia-oxidizing bacteria (Nitrosomonas) and three genera of nitrite-oxidizing bacteria (Nitrospina, Nitrospira, and Nitrococcus). We further explored the effect of different culture conditions, including environmentally relevant conditions of low substrate concentration and temperature, on these measurements using two nitrifier strains isolated from the Pacific Ocean (Nitrosopumilus sp. CCS1 and Nitrospina sp. Nb-3). The results of this study will inform ecological theoretical models to further constrain DIC fixation yields associated with nitrification in order to better understand the dynamics involved in the sequestration of carbon in the ocean.

Nitrifier culture sources
Ammonia-oxidizing archaeal cultures used in this study were three axenic Nitrosopumilus strains and one Nitrosopelagicus enrichment culture. Ca. Nitrosopelagicus brevis U25 originates from a North Pacific Ocean water sample (Santoro and Casciotti 2011;Carini et al. 2018). The level of enrichment during the time of this study was > 90% and co-cultured heterotrophic bacteria belonged to the genera Erythrobacter and Gracilimonas as previously described (Santoro and Casciotti 2011). Nitrosopumilus sp. CCS1 is a novel strain isolated from a seawater sample collected from the California Current system in the North Pacific Ocean (Santoro et al. unpublished). Nitrosopumilus adriaticus NF5 (= JCM 32270 T = NCIMB 15114 T ) and Nitrosopumilus piranensis D3C (= JCM 32271 T = DSM 106147 T = NCIMB 15115 T ) were isolated from the Northern Adriatic Sea and have been described in detail (Bayer et al. 2016(Bayer et al. , 2019c. The four axenic nitrite-oxidizing bacterial strains, Nitrospina gracilis Nb-211, Nitrospina sp. Nb-3, Nitrococcus mobilis Nb-231, and Nitrospira marina Nb-295, were obtained from the culture collection of John B. Waterbury and Frederica Valois at the Woods Hole Oceanographic Institution (WHOI). N. gracilis Nb-211 was isolated from surface waters of the South Atlantic Ocean (Watson and Waterbury 1971), N. mobilis Nb-231 was isolated from a surface water sample obtained from the South Pacific Ocean (Watson and Waterbury 1971) and N. marina Nb-295 was isolated from a water sample collected at a depth of 206 m from the Gulf of Maine in the Atlantic Ocean (Watson et al. 1986). Nitrospina sp. Nb-3 was isolated from the Pacific Ocean off the coast of Peru and has not yet been officially described (Watson and Waterbury, unpublished), however, its genome has recently been sequenced suggesting it is phylogenetically distinct from the species N. gracilis (Bayer et al. 2022).
Ammonia-oxidizing bacteria used in this study, Nitrosomonas marina C-25 and Nitrosomonas sp. C-15 (also referred to as strain Nm51, [Koops et al. 1991]), were both obtained from the culture collection at WHOI and were revived from 60-year-old cryostocks. Strain C-15 was isolated from surface water (1-m depth) of the South Pacific Ocean off the Peruvian continental shelf (Watson and Mandel 1971) and strain C-25 was isolated from surface waters of the South Atlantic Ocean (200 miles off the Amazon River mouth) (Watson and Mandel 1971).

Culture conditions
N. adriaticus NF5, N. piranensis D3C, N. marina C-25, and Nitrosomonas sp. C-15 were grown in HEPES-buffered artificial seawater medium containing 1 mmol L À1 NH 4 Cl, and Ca. Nitrosopelagicus brevis U25 was grown in natural seawater medium containing 50 μmol L À1 NH 4 Cl. N. gracilis Nb-211, N. marina Nb-295, and N. mobilis Nb-231 were grown in artificial seawater medium supplemented with 1 mmol L À1 NaNO 2 . Nitrosopumilus sp. CCS1 and Nitrospina sp. Nb-3 were grown under multiple culture conditions as indicated in the Results and discussion. All strains were routinely grown in 60 mL polycarbonate bottles (Nalgene) containing 50 mL culture medium, and bottles were incubated at 25 C (with the exception of Ca. Nitrosopelagicus brevis, which was always incubated at 22 C) in the dark without agitation.

Cellular carbon and nitrogen content measurements
To determine C : N ratios, between 100 and 500 mL of culture was filtered onto combusted (450 C, 4 h) glass fiber filters (Advantec, GF-75, 25 mm; 0.3 μm nominal pore size). Filters were acidified with HCl (10% vol/vol), dried (60 C, 24 h), and packed into tin capsules prior to being analyzed on a CHN elemental analyzer (Exeter Analytical, CEC 440HA). The instrument was calibrated with acetanilide following manufacturer protocols.
Cellular carbon (C) content was calculated using both, CHN elemental analyzer (only for large cells) and 14 C-DIC incorporation measurements (see below), divided by the number of newly produced cells. In addition, C content of the archaeal strain Nitrosopumilus sp. CCS1 was calculated from a dilution series of concentrated cells as described in White et al. (2019). Cells were concentrated using tangential flow filtration (Ultracell 30 kDa membrane; Pellicon; Millipore Sigma) and a dilution series of 1.1-5.6 Â 10 11 cells L À1 was constructed by resuspending cell concentrates in culture medium (Fig. S1). The total organic C content for each vial of the dilution series was directly measured by high temperature combustion using a modified Shimadzu TOC-V as described in Carlson et al. (2010). C content per cell was calculated via linear regression of cell counts and elemental content over the dilution series, where the slope of a Model II least squares regression is considered the elemental content per cell (Fig. S1).

Combined DIC fixation and DOC release measurements
DIC fixation was measured via the incorporation of [ 14 C]bicarbonate as previously described (Herndl et al. 2005) with modifications. [ 14 C]-bicarbonate (specific activity 56 mCi mmol À1 /2.072 Â 10 9 Bq mmol À1 ; PerkinElmer) was added to 5 mL of culture (between 10 and 60 μCi were added depending on the activity of the culture). Different incubation times were tested (see Results section) and all consecutive experiments were performed over the entire length of the growth curve. For every culture condition, at least three replicate live samples and one formaldehyde-fixed blank (3% vol/vol) were incubated in temperature-controlled incubators in the dark. Due to radiation safety procedures which preclude measurements of radioactive samples in general-use equipment, parallel incubations without [ 14 C]-tracer additions were used to determine cell abundance and nitrite concentration (see above). Great care was taken to ensure the exact same culture conditions to reduce potential biological heterogeneity between replicates.
Incubations were terminated by adding formaldehyde (3% vol/vol) to 5 mL of sample. After 30-60 min, every sample was individually filtered onto 25 mm, 0.2 μm pore size polycarbonate filters (Millipore) and rinsed with 0.5 mL of artificial seawater using a glass filtration set (Millipore). The individual filtrates (5.5 mL per sample) were collected and transferred to scintillation vials to determine the fraction of [ 14 C]-DOC. Excess [ 14 C]-bicarbonate from the filters was removed by exposing them to fumes of concentrated HCl (37%) for 24 h. The filters were transferred to scintillation vials and 10 mL of scintillation cocktail (Ultima Gold; PerkinElmer) was added. The filtrates were acidified to pH $ 2 with HCl (25%) as previously described (Marañ on et al. 2004), and filtrates were kept for 24 h in open scintillation vials placed on an orbital shaker before 10 mL scintillation cocktail was added to each vial. Samples were shaken for ca. 30 s and incubated in the dark for at least 24 h prior to counting the disintegrations per minute (DPM) in a scintillation counter (Beckman Coulter LS6500) for 15 min.
Total radioactivity measurements were performed to verify added [ 14 C]-bicarbonate concentrations by pipetting 100 μL of sample into scintillation vials containing 400 μL betaphenylethylamine (to prevent outgassing of 14 CO 2 ). Scintillation cocktail was added, vials were shaken for ca. 30 s and immediately measured in the scintillation counter.
The resulting mean DPM of the samples were corrected for the DPM of the blank, converted into organic carbon fixed over time and corrected for the DIC concentration in the culture media.
DIC fixation rates were calculated using the following formula: where DPM are the disintegrations per minute measured in the scintillation counter, for the sample (s) and the blank (b). DIC w denotes the dissolved inorganic carbon concentration in culture medium and DPM tracer (tr) is the DPM for the [ 14 C]bicarbonate added to the incubations.

DIC concentration measurements
Total alkalinity (TA) of unfixed natural and artificial seawater medium was measured via an open-cell endpoint titration using a Mettler-Toledo T5 autotitrator, and pH was measured spectrophotometrically using a Shimadzu UV-1280 UV-VIS spectrophotometer as described previously (Dickson et al. 2007;Hoshijima and Hofmann 2019). DIC concentrations were calculated from TA and pH using the CO2SYS software (Pelletier 2007). To calculate DIC concentrations of HEPES-buffered media, TA values were taken from unbuffered artificial seawater medium and the pH was remeasured after adding HEPES.

Calculations of Gibbs free energy (ΔG)
The effective Gibbs free energy (ΔG) for ammonia and nitrite oxidation was calculated for the culture conditions in this study using the following formula: where R is the ideal gas constant (8.314 J mol À1 K), Q is the reaction quotient, and T is the temperature in K. ΔG 0 values were obtained from Amend and Shock (2001).
Q was calculated based on the following measurements and estimates: NO 2 À concentrations were measured directly (see  (2019). Calculations can be found in the Supporting Information (Table S1).

Statistical analyses
Pairwise comparisons were performed with a two-sided Mann-Whitney U test (pairwise.wilcox.test) using the R software environment (R Core Team 2013). p values were adjusted for multiple comparisons using the Benjamini-Hochberg correction (p.adjust.method = "fdr") (Benjamini and Hochberg 1995).

Elemental composition of cultured nitrifiers
We determined the cellular carbon (C) content of cultured isolates of ammonia-oxidizing archaea, ammonia-oxidizing bacteria, and nitrite-oxidizing bacteria belonging to six different genera. The cellular C contents of ammonia-oxidizing archaea were $ 11-17 fg C cell À1 (Table 1), which is slightly higher than values reported for natural populations in the deep Atlantic Ocean ($ 8.39 fg cell À1 , Herndl et al. 2005) and an enrichment culture from the Baltic Sea (9 fg cell À1 , Berg et al. 2014), but much lower than values reported for ammonia-oxidizing archaea from hypoxic shelf waters of the Gulf of Mexico (50 AE 16 fg cell À1 , Kitzinger et al. 2020). All investigated marine ammonia-and nitrite-oxidizing bacteria had higher cellular C quotas compared to archaeal nitrifiers (Table 1), with Nitrospina exhibiting the lowest ($ 28-55 fg C cell À1 ) and Nitrococcus the highest ($ 272-1207 fg C cell À1 ) values (Table 1). The C content of ammonia-oxidizing archaea cells remained fairly constant during different growth phases, while C contents of all investigated nitrite oxidizer strains drastically decreased ($ 40-70%) from early exponential growth to stationary phase, which was supported by the observation of smaller cells in stationary compared to exponentially growing cultures (data not shown). Cell sizes of natural populations of Nitrospinae bacteria have been reported to be 4-fold (Kitzinger et al. 2020) and 50-fold (Pachiadaki et al. 2017) larger than cells of ammonia-oxidizing archaea, potentially reflecting these variations in cell size and C content during different growth phases.
Due to the small cell sizes of ammonia-oxidizing archaea and the potential to lose some cells via filtration through 0.2 μm pore size filters, we compared the cellular C content of Nitrosopumilus sp. CCS1 to values obtained from a filtrationindependent method using a dilution series of concentrated cells (see Methods section). There was no difference between the cellular C content of exponentially growing cells of strain CCS1 when using either method (12.0 AE 2.0 vs. 12.5 fg cell À1 , see Table 1). In addition, we quantified the amount of strain CCS1 cells passing through 0.2 μm filter sizes via flow cytometry and found that 0.2 μm polycarbonate filters had very high retention efficiency with only 0.14 AE 0.03% of cells not being recovered, further confirming that our results are not biased by the applied filtration procedure.
The molar C : N ratios of all investigated nitrifiers were in the range of 3.4-4.6 : 1 (Table 1), with the exception of previously published values of Nitrosopumilus maritimus NAOA6 (Meador et al. 2020) and two ammonia-oxidizing bacteria strains (Glover 1985). The values observed are lower than average values of heterotrophic marine bacteria cultures ($ 5 : 1) including Pelagibacter ubique ($ 4.6 : 1) (White et al. 2019, and references therein), with Nitrospina cells exhibiting the lowest average C : N ratio ($ 3.4) of all cultured nitrifiers in our study (Table 1). These low cellular C : N ratios are surprising considering the observation of glycogen storage deposits in cells of N. gracilis, N. mobilis, and N. marina (Watson and Waterbury 1971;Watson et al. 1986), as well as polyhydroxbutyrate storage in N. mobilis (Watson and Waterbury 1971). Cellular N contents in our study might be underestimated as measurements were performed on acidified samples, which could lead to partial hydrolyzation of proteins and amino acids. However, C : N ratios obtained from non-acidified cell pellets of N. adriaticus and N. piranensis (Bayer et al. 2019c) are comparable to those obtained for Nitrosopumilus sp. CCS1 in this study (Table 1), suggesting a negligible bias for at least some of the studied species.

DIC fixation yields of marine nitrifiers
We conducted combined measurements of DIC fixation, DOC release and ammonia/nitrite oxidation rates of 10 nitrifier cultures. The biological variability of NO 2 À and DIC fixation measurements between replicate bottles across all nitrifier strains and culture conditions was 5.2 AE 3.7% and 4.8 AE 3.2%, respectively. Here, we use the term "DIC fixation yield" to describe the number of moles of inorganic carbon (CO 2 or Bayer et al. C fixation and release by nitrifiers HCO 3 À ) that are fixed for every mole of N (NH 3 or NO 2 À ) oxidized, including the proportion that is released/lost as DOC.
DIC fixation yields of marine nitrite-oxidizing bacteria (Nitrospina/Nitrospira: mean AE SD = 0.036 AE 0.005, n = 47; Nitrococcus: mean AE SD = 0.018 AE 0.002, n = 11) were lower compared to those of ammonia oxidizers (Fig. 1). N. mobilis, which uses the CBB cycle for DIC fixation (Füssel et al. 2017) had $ 2 times lower DIC fixation yields compared to Nitrospina and Nitrospira which use a O 2 -tolerant version of the reverse TCA cycle (Lücker et al. 2010(Lücker et al. , 2013. Zhang et al. (2020) measured $ 1.7-times lower DIC fixation yields of N. gracilis 3/211 and a terrestrial Nitrospira isolate compared to values in our study. We observed that radiotracer incubations conducted over the entire length of the growth curve (until early stationary phase, see Fig. S2) resulted in $ 1.4-1.7-times higher DIC fixation yields of nitrite oxidizers compared to incubations conducted until late exponential growth (when NO 2 À was completely oxidized) (Fig. 2), suggesting that, in contrast to Nitrosopumilus sp. CCS1 where ammonia oxidation and DIC fixation were tightly coupled, nitrite oxidation might be partly decoupled from DIC fixation in nitrite-oxidizing bacteria. This observation was also supported by an increase of cell numbers after NO 2 À was completely depleted (Fig. S2).
While incubation times < 72 h are typically favored over longer times for environmental measurements to avoid crossfeeding of reaction products, our results indicate that DIC fixation yields of nitrite oxidizers might be underestimated using these established protocols (Fig. 2). We further explored the effect of multiple culture conditions, including environmentally relevant conditions of low substrate concentrations (1 μmol L À1 ) and low temperature (15 C), on DIC fixation yields of Nitrosopumilus sp. CCS1 and Nitrospina sp. Nb-3. We observed that Nitrospina sp. Nb-3 was $ 1.4 times more efficient in converting energy to growth when grown in natural seawater compared to artificial seawater medium, which was not observed for Nitrosopumilus sp. CCS1 (Fig. 3). We hypothesize that reduced N compounds present in natural seawater (ammonium and/or organic N compounds) might be responsible for the observed differences due to the metabolic costs of six reduced ferredoxins associated with assimilatory NO 2 À reduction in Nitrospina and Nitrospira (Lücker et al. 2013;Bayer et al. 2021). Those reduced ferredoxins could instead be used as electron donors for DIC fixation via the rTCA cycle (Berg 2011). Additions of ammonium or tryptone to artificial seawater medium likewise resulted in significantly higher DIC fixation yields (Figs. 3, S3), corroborating this hypothesis. Environmental populations of Nitrospinae have previously been shown to favor ammonium and the organic N sources urea and cyanate over nitrite (Kitzinger et al. 2020). Our data suggest that in addition to urea and cyanate, marine nitrite-oxidizing bacteria can assimilate more complex organic N sources such as peptides and/or amino acids thereby saving energy that can instead be invested in C assimilation. We cannot exclude the possibility that some of the organic C present in natural seawater or added via tryptone might also be assimilated, however, the observed higher net DIC fixation yields suggest that organic C assimilation plays-if at all-a very minor role in marine  Table S2. nitrite oxidizers. Furthermore, we observed slightly higher DIC fixation yields of Nitrosopumilus sp. CCS1 in HEPESbuffered artificial seawater compared to unbuffered culture medium (Fig. 3), which coincided with higher cellular C quota (Table 1). While we cannot explain these observations, the differences in DIC fixation yield did not seem to be caused by variations in pH, which remained constant in unbuffered culture medium containing low substrate concentrations (1 μmol L À1 NH 4 + ).
The theoretical Gibbs free energy release (ΔG) estimated for conditions in our study was 3.6 times higher for ammonia compared to nitrite oxidation (Table 2), yet DIC fixation yields of Nitrosopumilus sp. CCS1 and Ca. Nitrosopelagicus brevis U25 (Table 2) were only 2-2.6 times higher compared to Nitrospina sp. Nb-3. Similar observations were made by Kitzinger et al. (2020) who reported that Nitrospinae bacteria in low O 2 waters of the Gulf of Mexico are more efficient in translating energy gained from nitrite oxidation to C assimilation than ammonia-oxidizing archaea are in translating energy gained from ammonia oxidation. Thermodynamic properties and the efficiency of the DIC fixation pathway itself can contribute to realized energy yields. Additional factors include the requirement of four out of six generated electrons by ammonia monooxygenase to reduce molecular oxygen in ammonia oxidizers (Stahl and de la Torre 2012;Caranto and Lancaster 2018). When considering that a maximum of 53.8% of the energy released from catabolism are available to ammonia oxidizers for growth (Gonz alez-Cabaleiro et al. 2019), ammonia-oxidizing archaea are estimated to have slightly higher DIC fixation efficiencies compared to nitrite-oxidizing bacteria encoding the rTCA cycle (Table 2). While oxygen protection likely increases the energy demands of the rTCA cycle (Berg 2011), our results indicate that the cycle might also be highly efficient under oxic conditions that are found in most regions of the global ocean.
Multiple studies have used estimates of DIC fixation yields to infer DIC fixation rates associated with nitrification in diverse marine and estuarine environments (e.g., Dore and Karl 1996;Lam et al. 2004;Lee et al. 2015), and a value of 0.1 for archaeal ammonia oxidation has widely been used in the literature (Wuchter et al. 2006;Reinthaler et al. 2010;Middelburg 2011) without direct experimental evidence. Previous measurements of DIC fixation yields were mainly derived from cultures of ammonia and nitrite oxidizers that are not representative for the majority of nitrifiers found in marine environments and were highly variable (ammonia-oxidizing n.d.  Table S3. Representative growth curves can be found in the Supporting Information (Fig. S2).
bacteria: 0.033-0.130; nitrite-oxidizing bacteria: 0.013-0.031; Prosser 1990, and references therein). The variations in DIC fixation yields we observe for marine nitrifiers across different species and culture conditions are comparably low within ammonia-oxidizing archaea (mean AE SD = 0.091 AE 0.012; n = 47) and Nitrospina/Nitrospira (mean AE SD = 0.036 AE 0.005; n = 56), suggesting that these values are more constrained than previous estimates and particularly useful for modeling approaches in marine systems.

DOC release by chemolithoautotophs
We measured DOC release rates of 10 nitrifier cultures and tested how different culture conditions affected the amount of DOC released in proportion to the amount of fixed DIC. All Table 2. Thermodynamic considerations and comparison of DIC fixation efficiencies and biomass yields of marine ammonia-oxidizing archaea and nitrite-oxidizing bacteria grown under environmentally relevant conditions (substrate concentration: 1 μmol L À1 ; temperature: 15 C) in artificial and natural seawater medium. Gibbs free energy calculations for NH 3 oxidation and NO 2 À oxidation can be found in Table S1.   Table S3.
investigated strains released DOC during exponential growth, and DOC release ceased when cultures reached stationary phase (as determined by comparing the total amount of released DOC until late exponential vs stationary phase, see Fig. S4), suggesting that DOC release is a feature of metabolically active nitrifiers. This is in agreement with earlier observations of amino acid release by exponentially growing Nitrosopumilus cells (Bayer et al. 2019a). The amount of chemoautotrophically fixed DIC that was released as DOC by nitrifiers made up on average $ 5-15% (Fig. 4a). This is within the range observed for phytoplankton, which released 2-10% and 4-42% of their photosynthetically fixed DIC in culture and environmental studies, respectively (Carlson 2002, and references therein). To assess the potential stimulation of DOC release during the 30-60 min of formaldehyde fixation (see Methods section), we compared the fraction of fixed DIC released as DOC between 24 h and long-term (7-and 10-day long) incubations ( Fig. S5). If DOC release was occurring during the fixation period, it would make up a larger fraction of the total DOC release in 24 h compared to long-term incubations. However, we did not observe any significant differences between incubation times (Fig. S5), suggesting that formaldehyde fixation did not bias our results. DOC release varied between closely related species (Fig. 4a). N. piranensis released more DOC compared to the two other investigated Nitrosopumilus species, which is in agreement with Bayer et al. (2019b) who reported higher amino acid release rates of N. piranensis compared to N. adriaticus. Differences in the amount of released DOC have also been recently reported between the closely related aquarium strain N. maritimus SCM1 (9-19% of fixed DIC) and the environmental strain N. maritimus NAOA6 (5% of fixed DIC) (Meador et al. 2020). Within nitrite oxidizers, Nitrospina sp. Nb-3 consistently released less DOC compared to N. gracilis Nb-211 and the two phylogenetically more distantly related species N. marina and N. mobilis.
The fraction of released DOC remained constant across different culture conditions including environmentally relevant conditions of low substrate concentration (1 μmol L À1 ) and at low temperature (15 C) in natural seawater (Fig. 4b). This suggests that DOC release is not an artifact of unrealistic culture conditions but likely a feature exhibited by nitrifier  Table S3.
populations in the environment. However, given the differences in DOC release between closely related cultured species (Fig. 4a) and the greater diversity of nitrifiers observed in the ocean, it is possible that DOC release might differ in more complex natural environments, particularly of marine nitrite oxidizers for which environmentally relevant clades escaped cultivation thus far (Pachiadaki et al. 2017). In addition, in situ pressure conditions could further affect DOC release by nitrifiers in nature. While the composition of DOM released by bacterial nitrifiers is currently unknown, a fraction of the DOM released by ammonia-oxidizing archaea has been shown to consist of labile compounds, such as amino acids, thymidine, and B vitamins, that can be limiting for heterotrophic microbes in open ocean waters (Bayer et al. 2019a).

Conclusions
Our results suggest that DIC fixation yields of marine nitrite oxidizers might be underestimated by conventional < 72 hlong tracer incubations, due to a partial decoupling between NO 2 À oxidation and C assimilation over short timescales. In addition, DIC fixation yields of Nitrospina were positively affected by the presence of ammonium or complex organic N compounds, which might influence metabolic interactions with ammonia oxidizers and/or heterotrophic prokaryotes in the environment, suggesting a potentially underappreciated role for competition in the N cycle (Santoro 2016). DIC fixation yields of marine nitrifiers obtained in our study will help to further constrain the relationship between C and N fluxes in the nitrification process and inform theoretical models about how to connect observations at microscale to regional and global scales. Using a mean global value of organic C export from the euphotic zone of $ 6 Pg C yr À1 (Siegel et al. 2014) and a mean C : N ratio of sinking marine particles (at the surface) of $ 7.1 (Schneider et al. 2003), we estimate that the resulting global ocean organic N export of $ 0.85 Pg yr À1 could fuel up to 0.13 Pg C y À1 of chemoautotrophic DIC fixation (0.094 Pg C yr À1 by ammonia-oxidizing archaea and 0.037 Pg C yr À1 by nitrite oxidizers) in the dark ocean, which is in the lower range of previous estimates (0.15-1.4 Pg C yr À1 , see Table S4, and references therein). Furthermore, we show that nitrifiers release significant amounts of DOC under environmentally relevant conditions, equating to fluxes of 0.006-0.02 Pg C yr À1 of fixed DIC released as DOC. Elucidating the lability and fate of the DOM released by nitrifiers will be crucial to understand its implications for the marine carbon cycle.

Data availability statement
Data and metadata are available at the Biological and Chemical Oceanography Data Management office (BCO-DMO) under project 806565.