Aerobic media were prepared with an artificial seawater base containing 20 g L^-1 sodium chloride (Fisher Scientific), 3 g L^-1 magnesium chloride hexahydrate (Fisher Scientific), 0.15 g L^-1 calcium chloride (Fisher Scientific), 0.5 g L^-1 potassium chloride (Fisher Scientific) in ultrapure water produced by Milli-Q^® EQ 7000 Ultrapure Water Purification System (MilliporeSigma, Merck KGaA, Darmstadt, Germany). Mineral media included 10 mM 3-(N-Morpholino)-Propanesulfonic Acid (MOPS) buffer (pH = 7.2, from 1 M stock solution (209.26 g L^-1 MOPS free acid from EMP Millipore Corp., 100 mL L^-1 5 M sodium hydroxide from Spectrum Chemical Mfg Corp.), 2 mM ammonium chloride (Fisher Scientific), 0.2 mM sodium sulfate (J.T. Baker), 0.146 mM dipotassium phosphate (J.T. Baker), and 0.0588 mM monopotassium phosphate (J.T. Baker), and supplemented with trace elements. The final concentration of trace elements included: 20 μM hydrochloric acid (VWR Chemicals), 7.5 μM ferrous ammonium sulfate (Fisher Scientific), 0.48 μM boric acid (Sigma Chemical Co.), 0.5 μM manganese chloride (Fisher Scientific), 6.8 μM cobalt sulfate (Sigma Chemical Co.), 1.0 μM nickel chloride (Acros Organics), 12 nM copper chloride (Acros Organics), 0.5 μM zinc sulfate (Sigma Chemical Co.), 0.15 μM sodium molybdate (Acros Organics), 25 nM metavanadate (Acros Organics), 9 μM sodium tungstate (Acros Organics), 23 nM sodium selenite (Sigma Chemical Co.). Spartina Alterniflora stems, collected from North Inlet salt marshes, were cleaned and chopped to about 3 mm in size, and included as a carbon substrate in vials (1% w v^-1). Complex media were prepared with the same recipe and two additional components, namely 2.5 g L^-1 of yeast extract (Fluka BioChemika) and 2.5 g L^-1 of peptone (Fluka BioChemika). Solid media were prepared by including 2% (w v^-1) agar (Thermo Scientific). After autoclave sterilization and cooling (20 minutes at 121°C), a mixture of penicillin-G sodium (Alfa Aesar) and streptomycin sulfate (Acros Organics) was added to reach a final concentration of 0.2 g L^-1 to minimize bacterial growth.

The composition of anaerobic media used to isolate and maintain fungal cultures was identical to aerobic media with the following exceptions. Resazurin, 1 μg L^-1, (Acros Organics) was added as a redox indicator; 0.1 mM of sodium sulfide (Acros Organics) and 1 g L^-1 of L-cysteine (Acros Organics) were included as reducing reagents. Liquid media were purged with ultra-high purity N₂ (Airgas) for 20 to 30 minutes, and 20-ml were dispensed into N₂-flushed 65-ml serum vials sealed with butyl rubber septa and aluminum crimps. After autoclave sterilization, each vial of anaerobic media was supplemented with a mixture of 0.2 g L^-1 of penicillin-G sodium (Alfa Aesar), 0.2 g L^-1streptomycin sulfate (Acros Organics), and 10 μM sodium nitrite (Fisher Chemicals). For isotopic analysis (described below), the anaerobic media were amended with 100 μM NO 2 − . Anaerobic roll tubes for colony picking were prepared as described previously (Peng et al., 2018).

Triplicate 30-cm sediment cores were taken from two sites, Clambank and Oyster Landing, at the North Inlet-Winyah Bay National Estuarine Research Reserve in Georgetown, South Carolina (33.35°N, 79.20°W) on June 17^th, 2021 ( Figure 1 ). The North Inlet salt marsh is dominated by Spartina alterniflora (Allen et al., 2014). The cores were put on ice during transportation from Georgetown to Columbia, South Carolina, where initial enrichment culture inoculation occurred on the same day.

As marine fungi are typically hard to isolate (Edwards et al., 2017; Amend et al., 2019), we used the common plate dilution technique (Warcup, 1950) to isolate N₂O-producing fungi that are facultative anaerobes. Sediments from 1 cm and 10 cm depths were diluted in aerobic media and were used to streak agar plates. After three rounds of colony picking, isolated fungal cultures were inoculated into anaerobic media and screened for the ability to grow under sulfidic conditions and produce N₂O. Two fungal strains isolated using complex media and one fungal strain isolated using mineral media demonstrated high potential for N₂O production.

Anaerobic enrichment cultures of salt marsh sediment fungi were established in an Aldrich^® AtmosBag (SKU Z555525) pre-flushed three times with ultra-high purity nitrogen (N₂). Approximately 0.1 g of sediment was placed into the serum vials containing anaerobic media using aseptic techniques. All enrichment cultures were screened for N₂O production in the headspace using a gas chromatograph (described below) equipped with an electron capture detector (ECD). N₂O-producing enrichment cultures were selected for fungal isolation using the anaerobic roll tube technique (Peng et al., 2018). One N₂O-producing fungal strain was isolated from anaerobic enrichment cultures. This was the first known case of anaerobic marine fungi isolated under sulfidic conditions.

Isolated fungal cultures were maintained in anaerobic batch cultures, which were transferred to fresh media every 7-10 days. When inoculating fresh media, 0.2 mL of media containing fungal cells were drawn from the inoculum culture using techniques that avoided oxygen contamination. For yeasts, the batch culture serving as the inoculum was thoroughly homogenized before inoculation. For filamentous fungi, we performed careful visual inspection to include similar amounts of filamentous fungi in each 0.2-ml inoculum to ensure replicate cultures had similar amount of initial biomass.

Fungal cells were harvested by centrifugation and DNA were extracted using the DNeasy Plant Pro kit (QIAGEN). The quantity and quality of the DNA were measured using a Nanodrop 2000C spectrophotometer (ThermoFisher Scientific).

The small (SSU) and large (LSU) subunits of the rRNA genes were amplified using the primers Fun18S1 (5’-CCATGCATGTCTAAGTWTAA-3’) (Lord et al., 2002) and FR1 (5’-ANCCATTCAATCGGTANT-3’) (Vainio and Hantula, 2000) targeting the V1 to V8 regions of the SSU rRNA gene and LR0R (5’-ACCCGCTGAACTTAAGC-3’) and LR5 targeting the D1 to D3 regions of the LSU rRNA gene (Tedersoo et al., 2015). The PCR reactions were performed using Phusion^® high-fidelity DNA polymerase (New England BioLabs, M0530, Ipswich, MA). The thermal cycle started with 30 seconds at 98°C, followed by 30 cycles of 10 seconds at 98°C, 30 seconds at 60°C, and 30 seconds at 72°C. The final elongation at 72°C was 5 minutes long. The PCR products were gel purified using Zymo DNA Clean & Concentrator -5 following the manufacturer’s protocol and sent to Etonbio (Research Triangle Park, North Carolina, USA) for Sanger sequencing.

SSU and LSU sequences were searched against the NCBI nt database using the web portal for blastn (Johnson et al., 2008). Sequences for phylogenetic analysis were retrieved from the NCBI GenBank. Sequences were aligned using MUSCLE and manually trimmed using MEGA version 11 (Edgar, 2004; Tamura et al., 2021). Maximum-likelihood trees were constructed using FastTree 2.1 with default settings (1,000 bootstrap replicates, Jukes-Cantor model) (Price et al., 2010). The tree was then imported to Interactive Tree of Life (iTOL v6.6) for visualization (Letunic and Bork, 2019).

SRI Greenhouse Gas Monitoring Gas Chromatograph, model 8610-0040, equipped with an ECD and a flame ionized detector (FID) was used to measure N₂O and CO₂ headspace gas production. Prior to gas extraction, 5 mL of ultra-high purity N₂ gas was added to vials with a gas-tight syringe and a sterilized needle to avoid air contamination. A gas-tight syringe and a sterilized needle was used to extract 5 mL of headspace gas, which was manually injected into the sampling port on the instrument. Ultra-high purity N₂ gas was used as the carrier gas, maintained at 30 psi. The column oven temperature was set at 90 °C. The ECD and FID, coupled to a methanizer for CO₂, measurements, were set at 300 °C. N₂O standard curves were constructed using 5 ppm, 10 ppm, and 100 ppm N₂O (GASCO, Cal Gas Direct Incorporated, Huntington Beach, California, USA). CO₂ standard curves were constructed using 1%, 5%, and 10% CO₂ calibration standards from GASCO. Multiple reference samples served as controls, including uninoculated media and N₂O producing fungal cultures terminated with concentrated sodium hydroxide and purged with N₂.

Analysis of N₂O stable isotopes and isotopomers were made using an Elementar Americas Inc. isoprime precisION continuous flow, multicollector, isotope-ratio mass spectrometer (CF-MC-IRMS) equipped with a custom purge-and-trap and gas extraction systems, as described in Charoenpong et al. (2014). The CF-MC-IRMS has the necessary collector configuration for simultaneous determination of masses 30, 31 for the NO⁺ fragment of N₂O (determination of δ¹⁵N^α) and 44, 45, and 46 (determination of bulk δ¹⁵N and δ¹⁸O). Cultures for isotopic analysis had a final NO 2 − concentration of 100 μM. A gas tight syringe was used to extract 3-mL to 5-mL headspace gas, which was then injected into the CF-MC-IRMS injection port. N₂O was purified in a purge and trap system under helium continuous flow (40 mL/min), CO₂ was chemically removed, and H₂O vapor was eliminated with both chemical and cryogenic traps. N₂O was cryofocused with two liquid N₂ traps and passed through a capillary GC column prior to IRMS analysis. These latter steps, including GC column backflushing to eliminate interferences in the SP determination, were nearly identical to what was described by McIlvin and Casciotti (2010). Helium flow was optimized to achieve quantitative extraction and reproducible results, even at low N₂O concentrations. N₂O concentrations in our samples were calculated from relative peak heights between the samples and a dilution series of pure N₂O gas mixtures (in N₂) of known N₂O concentrations (500 and 7500 ppm). The reproducibility of bulk δ¹⁵N and δ¹⁸O and SP as well as any instrumental drift were determined from measurements of an internal reference gas distributed through the analytical run. The measurements were calibrated from a four-point calibration correction using N₂O standards covering a large range of SP (-92.7‰ - 18.9‰), as well as bulk δ¹⁵N and δ¹⁸O composition calibrated by S. Toyoda (Tokyo Institute of Technology), and obtained from Joaquim Mohn (EMPA, Swiss Federal Laboratories for Materials Science & Technology). These standards were analyzed in duplicate for each run to quantify the scrambling effect, potential offset and iteratively solve for the different calibration parameters (Frame and Casciotti, 2010; Mohn et al., 2014). Correction for isobaric interference from ¹⁷O was included in these procedures. Standard deviations for triplicate measurements of our N₂O standards were typically below 0.1‰ for δ¹⁵N-N₂Obulk, 0.2‰ for δ¹⁸O-N₂O and 1‰ for SP, which were comparable to values reported by Mohn et al. (2014).

The four fungal strains we isolated and investigated in this study were Purpureocillium lilacinum BL2022, Trichoderma virens XP2022, Trichoderma harzianum MB2022, and Rhodotourla glutinis MT2022, as identified by blastn and phylogenetic analysis ( Supplementary Figures S1, S2, and S3 ; Supplementary Table 1 ). T. harzianum and T. virens were closely related to terrestrial and marine derived species, such as T. polysporum and T. citrinoviride. R. glutinis MT2022 was closely related to other species which have been isolated from both marine (R. diobovata) and terrestrial environments (R. graminis and R. babjeavae). P. lilacinum was closely related to terrestrially derived species but was distinct from P.lilacinum CBS284.36. T. harzianum was the only strain isolated using mineral media ( Table 1 ), which did not contain yeast extract or peptone.

P. lilacinum produced the greatest amount of N₂O and CO₂ (22.8 ± 7.8 nmol of N₂O and 179 ± 24 μmol of CO₂, Figures 2A, E ). T. virens, T. harzianum, and R. glutinis. produced an average of 17.4 ± 5.4 nmol, 4.45 ± 0.48 nmol, and 4.24 ± 1.57 nmol of N₂O ( Figures 2B–D ) and 39.0 ± 20.0, 5.00 ± 0.44, and 88.1 ± 37.5 μmol of CO₂ ( Figures 2F–H ), respectively. The production of N₂O by P. lilacinum did not plateau until the 22nd day. The production of N₂O by T. virens plateaued after the 16th day. The production of N₂O by T. harzianum and R. glutinis reached a maximum in only two to three days. N₂O yield for each fungal strain was calculated using the average maximum N₂O produced, where N₂O yield is the fraction of NO 2 − N converted to N₂O-N (see Supplementary Information for additional details). P.lilacinum, T. virens, T. harzianum, and R. glutinis had respective yields of 22.8 ± 7.8%, 17.4 ± 5.4%, 4.45 ± 0.48%, and 4.24 ± 1.57%. There was a significant correlation between N₂O and CO₂ production (p< 0.01) for each replicate culture of the filamentous fungi P. lilacinum, T. virens, and T. harzianum ( Figure 3 ). N₂O and CO₂ production by R. glutinis were not significantly correlated.

To further explore the effect of concentration on fungal N₂O production, the species with the highest cumulative N₂O production, P. lilacinum, was cultivated with 0 μM, 10 μM, 100 μM NO 2 − in triplicate. When grown without NO 2 − , P. lilacinum produced <1 nmol of N₂O ( Figure 4 ). The addition of 100 μM to growth media resulted in a >10-fold increase in N₂O production for two of the three replicates when compared to the three replicates grown with 10 μM NO 2 − (285.0 ± 20.1 nmol and 22.8 ± 7.8 nmol of N₂O, respectively).

Site preference (SP) values of N₂O from salt marsh sediment fungi ranged from 7.49 ± 1.57‰ for T. harzianum, to 33.41 ± 1.20‰ for R. glutinis ( Table 1 ), which were all lower than N₂O SP values (37.1 ± 2.5‰) measured from the model fungal denitrifier F. oxysporum (Sutka et al., 2008).

We used both aerobic and anaerobic cultivation techniques to isolate N₂O-producing fungi from salt marsh sediments in North Inlet, South Carolina, USA. The aerobic technique is similar to previous efforts to isolate N₂O-producing fungi from soil (Jirout et al., 2013; Mothapo et al., 2013) except that our media were prepared with a seawater base, which selected for fungi adapted to salinity in the range of 30 to 35 ppt. All filamentous species (P. lilacinum, T. virens, and T. harzianum) we isolated from salt marsh sediments have been isolated from soil and shown to produce N₂O (Lavrent’ev et al., 2008; Jirout, 2015; Maeda et al., 2015). A comparative genomics studies showed that all these three species possess the diagnostic gene for fungal denitrification, the cytochrome P450 nitric oxide reductase (P450nor) (Higgins et al., 2018). The positive correlations between N₂O and CO₂ production by filamentous fungi indicate respiratory denitrification is responsible for N₂O production. Variability in N₂O production, within individual strains’ cultures, is likely due to small differences in the number of cells in the inoculum between triplicates. While this study was not intended as an exhaustive search for N₂O-producing fungi from salt marsh sediments, our results indicate that at least a subset of N₂O-producing fungi from terrestrial environments (Mothapo et al., 2015) are present in salt marsh sediments and have the potential to produce N₂O at high yields.

The ability to produce N₂O was widespread among the hundreds of strains tested by Maeda and colleagues (Maeda et al., 2015), but only one (Mucorales) of the 70 N₂O-producing strains in their study was not from the phylum Ascomycota. A survey of P450nor in over 700 fungal genomes also showed that this diagnostic gene for fungal denitrification was primarily found in Ascomycota (Higgins et al., 2018). While Ascomycota seem to be the main lineage capable of N₂O production, fungi from other phyla, particularly early diverging lineages, are poorly represented in these studies, partially due to their low abundance in terrestrial environments (Berbee et al., 2017). There is evidence that early diverging fungi play a more important role in marine environments than in terrestrial environments, and their potential to produce N₂O remains to be examined (Peng and Valentine, 2021).

To the best of our knowledge, this is the first study that used strictly anaerobic techniques to isolate fungi from salt marsh sediments, where oxygen is typically depleted a few millimeters below the surface (Peng et al., 2021). The only N₂O-producing fungus isolated anaerobically was the basidiomycetous yeast R. glutinis, which is not known to possess P450nor in its genome (Higgins et al., 2018), but it does possess a fungal nitrite reductase (fungal nirK). The decoupling between N₂O and CO₂ production in R. glutinis cultures ( Figure 3 ) suggests that N₂O production by R. glutinis may not be from respiratory denitrification. Instead, energy production by R. glutinis, an oleaginous yeast, was likely from fermentation (Yeeh, 1999; Xue et al., 2008). This could also explain the low N₂O yield by R. glutinis (3.25 ± 1.20%). While this is the first report of N₂O production by R. glutinis grown under sulfidic conditions, an unidentified Rhodotorula species grown under aerobic conditions produced N₂O after it reached stationary phase (Bleakley and Tiedje, 1982). Therefore, regardless of the mechanism of N₂O production by R. glutinis, it can contribute significantly to N₂O production from salt marsh sediments where the redox conditions fluctuate due to daily tides. A strong positive correlation was observed between N₂O production rate and the relative abundance of Rhodotorula in estuarine sediments in Xiamen, China (Su et al., 2021).

Nearly all previous studies on N₂O production by fungi used media containing 10 mM of NO 2 − (Shoun et al., 1992; Mouton et al., 2012; Jirout et al., 2013; Mothapo et al., 2013; Maeda et al., 2015; Zou et al., 2021), which is orders of magnitude higher than the in situ NO 2 − concentrations in soil or marine sediments. It is known that different nutrient levels can cause different physiological responses in fungi, including changes which affect the ability of the cell to transport nutrients and degrade carbon (Ozcan and Johnston, 1999; Zaman et al., 2008). In this study we cultivated N₂O-producing fungi using sulfidic media containing NO 2 − at a level (10 μM) much closer to in situ conditions. Consequently, the total amounts of N₂O production in our study were lower than previously reported value, but the yield of N₂O from NO 2 − by salt marsh sediment fungi (up to 22.8 ± 7.8%) was comparable to some of the highest levels in previous studies (Mothapo et al., 2013; Jirout, 2015; Maeda et al., 2015). Furthermore, the salt marsh sediment fungi were cultivated under sulfidic conditions, demonstrating their relevance in N₂O production even at sulfidic depths of the sediments.

The ~10-fold increase in both N₂O and CO₂ production by P. lilacinum grown with 100 μM NO 2 − (compared to 10 μM ) further supports the notion that respiratory reduction of NO 2 − was the primary mechanism for N₂O production, although we cannot rule out the possibility that secondary metabolisms also contribute to N₂O production (Higgins et al., 2018). Such a response by P. lilacinum implies that fungal N₂O production from salt marsh sediments will scale linearly with nutrient inputs from anthropogenic sources. A recent report demonstrated the drastic increase in N₂O production from salt marsh sediments under long-term fertilization (Peng et al., 2021), and fungi may have played a major role in the observed N₂O production by the bulk sediment community.

Stable isotope mass balance has become a useful tool to determine the contribution of fungi to total N₂O production (Wankel et al., 2017; Rohe et al., 2021; Su et al., 2021). This approach constrains the contribution of different sources and sinks of N₂O by measuring N and O isotopes and isotopomers of N₂O from a complex system, which requires the knowledge of endmembers for each pathway (e.g. bacterial denitrification, fungal denitrification). However, studies employing this method so far have relied primarily on the N₂O isotopomer signature determined for the model organism F. oxysporum (Sutka et al., 2008), which is not necessarily representative of N₂O produced by marine fungi from salt marsh sediments. A recent study using an isotope mass balance approach found that the site preference value and δ¹⁸O of N₂O from estuarine sediments sometimes exceeded the values for the fungal endmember based on F. oxysporum (Su et al., 2021). We interpret this as additional evidence for the need to evaluate the N₂O isotopologue composition produced by fungi from salt marsh sediments.

A study on soil fungi showed that N₂O SP ranged from 15.8‰ - 37.1‰, depending on the species (Maeda et al., 2015). Yet, work by Maeda and colleagues (Maeda et al., 2015) has since not been used in any published isotope mass balances calculations. Our findings add to evidence presented by Maeda et al. (2015) that N₂O SP depends on the fungal isolate at the species and even strain level. We further present evidence that N₂O SP may differ based on growth conditions, even for fungi of the same species. N₂O SP values measured by Maeda et al. (2015) for several strains of T. harzianum ranged from 30 - 33.4‰. In contrast, the SP values of N₂O produced by the salt marsh sediment T. harzianum from this study (7.46 ± 1.57‰) was much lower, and was the lowest fungal N₂O SP reported to date. This may be attributed to two factors. Firstly, the T. harzianum isolated from salt marsh sediments in this study is the first and only N₂O-producing fungal culture grown on mineral media to the best of our knowledge. It has been shown that different organic nitrogen sources can impact N₂O emissions, though little is known about how N₂O SP values would be impacted (Pelster et al., 2012). Secondly, fungal culture media we prepared did not include any soluble sugars (e.g. dextrose), which was a staple ingredient in all previously published studies (Sutka et al., 2008; Rohe et al., 2014; Maeda et al., 2015). Instead, the stems of Spartina alterniflora (lignocellulose) was provided as the sole carbon source to T. harzianum grown on mineral media. It has recently been shown that the type of carbon substrates used to cultivate freshwater bacteria has a significant impact on the SP values of bacterial N₂O (Li et al., 2022). It is possible that this is true for N₂O-producing fungi as well, though future work is needed to verify.

Along with T. harzianum, other salt marsh sediment fungi in this study produced N₂O SP values lower than that from the model fungal denitrifier F. oxysporum (Sutka et al., 2008), indicating that it is inaccurate and oversimplifying to use the N₂O SP from one single strain as an endmember for stable isotope mass balance calculations. Many of the N₂O SP values measured from core incubations by Wankel and colleagues (Wankel et al., 2017) are within the range of the N₂O SP values reported in this study and by Maeda et al. (2015). Future work is needed to determine how carbon and nitrogen substrate types influence fungal N₂O SP values. While the range of fungal N₂O SP values overlaps with the N₂O SP range from abiotic N₂O production, abiotic denitrification is favored in high pH conditions with solid iron (III) or copper (II) catalysts (Zhu-Barker et al., 2015) and hence irrelevant or negligible in our cultures (pH buffered at 7.2) and in salt marsh sediments (typically < 7.5). Given the wide range of fungal N₂O SP values, the selection of endmember is therefore critical in studies using stable isotope mass balance calculations. The endmember selection can be informed by metabarcoding analysis of the fungal community in general (e.g. targeting the rRNA gene) and the functional genes for fungal denitrification (e.g. nirK gene) (Long et al., 2015; Chen et al., 2016; Maeda et al., 2017).