Nitrogen saturation theory predicts high rates of atmospheric N deposition can increase ecosystem N availability and stimulate ecosystem N losses via soil nitric oxide (NO; an air pollutant at high concentrations) and nitrous oxide (N₂O; a strong greenhouse gas) emissions. However, it remains unclear whether theories developed in mesic ecosystems apply to drylands, where plant and soil N availability are not always coupled in dry soils. NO and N₂O are often produced in soils during the oxidation of ammonia by ammonia oxidizing archaea (AOA) or ammonia oxidizing bacteria (AOB) during nitrification. AOB are thought to emit more NO and N₂O during nitrification than AOA and may be favored in N-rich relative to N-limited environments, suggesting high rates of atmospheric N deposition might produce positive feedback sending more of the N to the atmosphere. To assess how high rates of atmospheric N deposition affect AOB- and AOA-derived N trace gas emissions, we selectively inhibited AOA and AOB nitrifiers and measured NO and N₂O emissions from soils collected from three dryland sites exposed to relatively low (3.8 kg ha^-1 = Low N) or high (11.8 kg ha^-1 = High N1; 15.6 kg ha^-1 = High N2) rates of atmospheric N inputs. We found that while the High N2 deposition site had the lowest AOA:AOB ratio (2.33 ± 0.57), consistent with expectations, this site did not emit the most NO and N₂O. Rather, AOA emitted between 21–78% of the NO from our sites, with higher AOA-derived NO emissions from relatively coarse-textured soils in the Low N deposition site. In addition to nitrification, denitrification also contributed to NO and N₂O emissions, especially in the Moderate N deposition site (where denitrification-derived NO and N₂O emissions were 2.0 – 3.7 time greater than the other sites), which had finer textured soils that may favor denitrification. Interactions between soil texture and N availability, therefore, appears to be the primary mechanism determining whether atmospheric N deposition is retained in the ecosystem or reemitted to the atmosphere as NO or N₂O.

Study Site

We collected soils from three sites exposed to a range of atmospheric N deposition rates in southern California: a Low N deposition site and two high N deposition sites (hereafter, High N1 and High N2; Table 1; EPA, 2021). Vegetation in all sites was dominated by chamise (Adenostoma fasciculatum). Soils from the Low N site are fine sandy loams from the Sheephead series and are classified as shallow Entic Haploxerolls. Soils from the High N1 site are coarse loams from the Shepherdsaddle series and are classified as Ultic Haploxeralfs. Finally, soils from the High N2 site are fine sandy loams from the Trigo series and are classified as shallow Typic Xerorthents (Table S1; Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture, 2019). The climate at all sites is Mediterranean with hot, dry summers and cool, wet winters. Annual precipitation ranges from 500 to 670 mm and average monthly temperatures range from 8 to 40°C.

In the year leading up to the experiment, the High N2 deposition site had the highest ambient atmospheric concentrations of NO_x (NO_x = nitric oxide + nitrogen dioxide; 3.60 ± 1.51 ppm; Figure S1), while the High N1 deposition site had the highest ambient atmospheric concentrations of ammonia (NH₃; 2.98 ± 2.28 ppm), suggesting that they may receive different forms of N deposition. The three sites also have slightly different textures: the High N1 site is more enriched in clay (38.3 ± 2.57 %) compared to the Low N (20.4 ± 2.92 %) or High N2 sites (15.8 ± 4.93 %; Table 1). Given that soil texture and the form of deposited N varied among the sites, we cannot isolate the effects of N deposition and, instead, aim to: i) understand what controls gaseous N losses from dryland soils and ii) whether NO and N₂O emissions derived from AOB consistently associate with high atmospheric N inputs across remote sites varying in soil properties. 

Experimental design

At each of the three sites, we collected dry soils (0 – 10 cm depth; A horizons) from underneath five Adenostoma fasciculatum shrubs across a ~50 m transect (each shrub was roughly 10 m apart) in September 2021. We collected soils in September because this is near the end of the dry season at our sites, allowing us to assess the contribution of AOA and AOB to N emissions after experimentally rewetting soils in the lab. Soils were transported back to the lab where they were sieved to 2 mm and stored at 4 °C until the experiment began (soils were refrigerated for less than one month). Soils were removed from the refrigerator two days prior to beginning the experiment and were incubated at lab temperature (~22 °C). Two soil samples were analyzed at a time. Each soil sample was split into three 50 g subsamples and transferred to 118 mL canning jars. The subsamples in each canning jar were exposed to one of three treatments: AOB inhibition, total nitrifier inhibition, or a control. To inhibit NH₃ oxidation by autotrophic AOB, 1-octyne was added to one subsample to bring the headspace in the jar to 4 µmol L^-1. The 1-octyne was prepared by adding 40 µL of liquid 1-octyne to a 125 mL bottle fitted with a butyl stopper, over pressurizing the bottle with 100 mL of air, and, once the liquid 1-octyne evaporated, removing 2.7 mL of the bottle headspace to inject into the subsample. To inhibit NH₃ oxidation by all autotrophic nitrifiers, acetylene was added to one subsample to bring the jar headspace to 6 µmol L^-1. The acetylene was prepared by first bubbling acetylene through a sulfuric acid trap to remove impurities, diluting the purified acetylene 10-fold with air, and then injecting 0.28 mL into the subsample. The third jar was treated as a control and was incubated under ambient lab conditions. All jars were incubated for 24 h.

Following the 24 h incubation, NO and N₂O emissions were measured from each jar after experimentally wetting soils. To prevent the growth of new AOB after wetting, the jar that was treated with 1-octyene was wet with a solution containing the antibiotic kanamycin at a concentration of 220 µg g^-1 soil. To prevent the growth of any new nitrifying bacteria or archaea, the jar that was treated with acetylene was wet with a solution containing kanamycin (220 µg g^-1 soil), the archaeal protein synthesis inhibitor fuscidic acid (800 µg g^-1 soil), and the nitrification inhibitor nitrapyrin (200 µg g^-1 soil). The control jar was wet with deionized water only. All jars were wet with enough solution to reach 100% water holding capacity. NO and N₂O emissions were measured from 6 of the jars (representing 2 samples) every 2 hours for 46 hours after wetting. Net nitrification and net N mineralization rates were measured as the difference in NO₃^- (nitrification) or NO₃^- and NH₄⁺ (N mineralization) between the start and end of the 46-hour incubation. This process was repeated twice per week until all 45 jars were analyzed (3 sites   3 treatments = 45 jars total).

The contribution of AOB to NO and N₂O emissions was determined by subtracting how much NO or N₂O was emitted from the AOB inhibition treatment (i.e., treating soils with 1-octyne and kanamycin) from how much NO or N₂O was emitted from control soils (wetted with water only). The contribution of AOA to NO and N₂O emissions was determined by subtracting how much NO or N₂O was emitted from the total nitrifier inhibition treatment (i.e., treating soils with acetylene, kanamycin, nitrapyrin, and fuscidic acid) from how much NO or N₂O was emitted from the AOB inhibition treatment. Finally, heterotrophic NO and N₂O emissions were assumed to be equal to how much NO or N₂O was emitted from soils under the total nitrifier inhibition treatment; acetylene does not inhibit N₂O production from heterotrophic denitrification.

NO and N₂O emissions

            Immediately after wetting, six jars (2 samples x 3 treatments) were connected to a recirculating sample loop joining a multiplexer (LI-8150, LI-COR Biosciences), an infrared CO₂ analyzer (IRGA; LI-8100, LI-COR Biosciences, Lincoln, NE), and an N₂O laser analyzer (Model 914-0027, Los Gatos Research, Inc., Mountain View, CA). To measure N₂O emissions from each jar, air was recirculated through the closed sample loop at a rate of 1.5 L min^-1. Soil N₂O emissions were calculated as the linear change in N₂O concentrations over a 9-minute period. Because the chemiluminescent NO₂ analyzer (LMA 3D; Unisearch Associates, Concord, ON, Canada) consumes NO, the NO₂ analyzer was not connected to the sample loop during this initial 9-minute incubation. Rather, after 9-minutes, an automated 3-way solenoid valve (Parker Hannifin Corp., Series 11/25/26, #991-000539-006) activated so that the NO₂ analyzer pulled from the sample loop at a rate of 1.5 L min^-1. To replace the air that the NO₂ analyzer consumed from each jar (the air from the jars was vented out of the NO₂ analyzer into the lab), a second valve activated at the same time to allow zero air (Ultra Grade Zero Air, Airgas, Radnor, PA) to enter each jar at a rate of 1.5 L min^-1. The NO₂ analyzer used a CrO₃ converter to oxidize NO to NO₂; we did not detect NO₂ when CrO₃ was removed from the sample loop suggesting our measurements were mostly NO. NO emissions were calculated using the following equation: 

EQ1: NO flux = ([NO]outlet - [NO]inlet) x flow x mass N / soil wt. / R / temp

Where [NO]_outlet is the concentration of NO leaving the jar headspace, [NO]_inlet is the concentration of NO entering the jar (assumed to be 0), flow is the flow rate of the sample loop (1.5 L min^-1), mass N is the molar mass of N in NO, soil wt. is the mass of soil in the jar, R is the molar gas constant (0.0821 L atm K^-1 mol^-1), and temp is the room air temperature. We measured NO concentrations for 10 minutes while zero air was flowing through the jar, allowing NO concentrations to reach equilibrium within the sample loop. NO fluxes were calculated using the average NO concentrations during the final 30 seconds of the 10-minute incubation. After 10 minutes, the sample loop was exposed to ambient lab air for one minute to purge zero air from the sample loop. After this 20-minute incubation, the multiplexer connected the next jar to the sample loop and this process was repeated, allowing us to measure each of the 6 jars once every ~2 hours. The continuous stream of dry air decreased soil moisture throughout the 46-hour incubation (Figure S2).

            We modified a publicly available script to calculate N₂O and NO emissions (https://github.com/handr003/TraceGasArray). N₂O emissions were calculated as the change in N₂O concentrations over the last 7 minutes of the incubation when the NO analyzer was not connected to the sample loop. Emissions were considered 0 if the linear relationship between time and N₂O concentrations was not statistically significant (p > 0.05). NO emissions were calculated using EQ1. Both the N₂O and the NO analyzer recorded trace gas concentrations every second. We used the trapezoidal integration to calculate cumulative N₂O and NO emissions over the 46-hour period post-wetting (final units were ng N g dry soil^-1 hr^-1).

Net nitrification and net N mineralization rates

            We measured soil extractable NO₃^- and NH₄⁺ before wetting soils and immediately after the 46-hour incubation to calculate net nitrification and net N mineralization rates. Briefly, 3 g of soil (dry weight equivalent) were extracted in 30 mL 2M KCl for 1 hour, filtered (Whatman 42 filter paper; 2.5 µm pore size), and the extracts were then frozen until analysis. Extracts were analyzed for NO₃^- (SEAL method EPA-136-A) and NH₄⁺ (SEAL method EPA-129-A) using colorimetric assays. Net nitrification and net N mineralization rates were calculated as the difference in inorganic N (NO₃^- for nitrification and NO₃^- plus NH₄⁺ for mineralization) before and after the incubation divided by the length of the incubation (~46 hours). We also measured soil gravimetric water content by drying soil samples (~10 g) at 104 °C for 24 hours. We estimated 100% soil water holding capacity (WHC) as the amount of water held by soils after saturating them with water and allowing them to drain in an air-tight container (to limit evaporation) for 8 hours.  

amoA gene quantification

            A subsample (~5g) of each soil sample (n = 15) was frozen (-20 °C) as soon as soils arrived in the lab. Within one month of freezing, DNA was extracted from 0.25g of each subsample using a DNA extraction kit (Qiagen DNeasy PowerSoil Pro, Hilden, Germany) following the manufacturer’s guidelines after an overnight incubation to enhance DNA extraction (700µl CD1 + 100µl ATL at 4°C; Qiagen). Quantitative polymerase chain reaction (qPCR) was used to estimate the abundance of bacterial and archaeal amoA genes; the AmoA1F/amoA2R primer set was used for bacteria and the Arch-amoAF/ArchamoAR primer set was used for archaea. Each qPCR was run in 10 µL reactions containing 5 µL master mix (Forget-Me-Not EvaGreen qPCR Master Mix, Biotium, Inc., Fremont, CA), 0.8 µL of 2 mM MgCl₂, 0.25 µL of 0.5 mg ml^-1 bovine serum albumin, 0.125 µL of 0.25 µM forward and reverse primer, 2.5 µL H₂O, and 1.2 µL sample DNA. Bacterial amoA was amplified using the following protocol: 5 minutes at 95 °C, followed by 40 cycles of 45 seconds at 95 °C, 30 seconds at 56 °C and 60 seconds at 72 °C (CFX384 Touch Real-Time PCR Detection System, Bio-Rad, Hercules, CA). Archaeal amoA was amplified using the following protocol: 4 minutes at 95 °C, followed by 40 cycles of 30 seconds at 95 °C, 45 seconds at 53 °C and 60 seconds at 72°C. The standard sequences were chosen from well-known archaeal (crenarchaeota genomic fragment 54d9) and bacterial (amoA gene of Nitrosomonas europaea ATCC 19718) ammonia oxidizing microorganisms. Standard curves were prepared using serial dilutions for bacterial amoA (10⁶ to 10² copies) and archaeal amoA (10⁷ to 10³ copies). The bacterial amoA standards had efficiencies of 83.5% (R² = 0.998) and archaeal amoA standards had efficiencies of 66.9% (R² = 0.997).