Rapid nitrate reduction produces pulsed NO and N2O emissions following wetting of dryland soils

Soil drying and wetting cycles can produce pulses of nitric oxide (NO) and nitrous oxide (N2O) emissions with substantial effects on both regional air quality and Earth’s climate. While pulsed production of N emissions is ubiquitous across ecosystems, the processes governing pulse magnitude and timing remain unclear. We studied the processes producing pulsed NO and N2O emissions at two contrasting drylands, desert and chaparral, where despite the hot and dry conditions known to limit biological processes, some of the highest NO and N2O flux rates have been measured. We measured N2O and NO emissions every 30 min for 24 h after wetting soils with isotopically-enriched nitrate and ammonium solutions to determine production pathways and their timing. Nitrate was reduced to N2O within 15 min of wetting, with emissions exceeding 1000 ng N–N2O m−2 s−1 and returning to background levels within four hours, but the pulse magnitude did not increase in proportion to the amount of ammonium or nitrate added. In contrast to N2O, NO was emitted over 24 h and increased in proportion to ammonium addition, exceeding 600 ng N–NO m−2 s−1 in desert and chaparral soils. Isotope tracers suggest that both ammonia oxidation and nitrate reduction produced NO. Taken together, our measurements demonstrate that nitrate can be reduced within minutes of wetting summer-dry desert soils to produce large N2O emission pulses and that multiple processes contribute to long-lasting NO emissions. These mechanisms represent substantial pathways of ecosystem N loss that also contribute to regional air quality and global climate dynamics.


Introduction
Soil drying-wetting cycles are widespread and can stimulate large emissions of both nitrous oxide (N 2 O) and nitric oxide (NO) (Scholes et al. 1997;Homyak et al. 2016;Eberwein et al. 2020) with profound implications for Earth's climate, regional air quality, and ecosystem N retention. This is because N 2 O is a potent greenhouse gas (Ciais et al. 2013), NO is a precursor to tropospheric ozone (Crutzen 1979), and both NO and N 2 O represent important pathways for ecosystem N loss (Peterjohn and Schlesinger 1990). While the production of NO and N 2 O is governed by both biological and chemical processes upon wetting dry soil, the magnitude of the emissions vary as a function of aridity (Wang et al. 2014;Liu et al. 2017;von Sperber et al. 2017), with some drylands recording among the highest NO and N 2 O emission pulses globally (Eberwein et al. 2020). However, how these emissions vary across ecosystems experiencing drying-wetting cycles and the biogeochemical processes producing them are still not well characterized. Identifying the underlying processes producing pulsed NO and N 2 O emissions is necessary to predict how ecosystem N cycling may respond to global change factors including high rates of atmospheric N deposition (Fenn et al. 2006), rising temperatures, and changing precipitation regimes (Dai 2013).
Multiple biological and abiotic processes regulate NO and N 2 O emissions after dry soils are wetted. Biological processes include nitrification, the aerobic oxidation of ammonia (NH 3 ) to nitrate (NO 3 − ) with NO and nitrite (NO 2 − ) as intermediates (Caranto and Lancaster 2017;Prosser et al. 2019), and denitrification, the sequential anaerobic reduction of NO 3 − to N 2 gas with NO 2 − , NO, and N 2 O as obligate intermediates (Knowles 1982); both of these processes can release NO and N 2 O as byproducts. In low oxygen environments, some nitrifiers use NO 2 − as the electron acceptor during the oxidation of NH 3 and produce NO and N 2 O via nitrifier denitrification (Prosser et al. 2019). Chemodenitrification-an abiotic nonenzymatic process-can also produce NO and N 2 O through the chemical reduction of NO 2 − and hydroxylamine (NH 2 OH) (Venterea and Rolston 2000;Zhu-Barker et al. 2015;Heil et al. 2016), which can accumulate in dry soils (Homyak et al. 2016). Because both biological and abiotic processes can occur simultaneously, it has been challenging to determine the contribution of individual processes to pulsed N emissions.
To advance understanding of the processes producing NO and N 2 O, the "hole-in-the-pipe" conceptual framework relates the factors that control N emissions to the processes that control N transformation rates (Firestone and Davidson 1989). Under this framework, N transformations are represented by changes in the diameter of the pipe-the diameter varies in proportion to process rates-whereas the factors controlling how much NO or N 2 O leak out of the pipe are represented by the holes (e.g., edaphic or environmental factors such as pH). In this sense, wetting soils could stimulate NO and N 2 O emissions by promoting nitrification, (Placella and Firestone 2013;Homyak and Sickman 2014), denitrification (Parker and Schimel 2011;Soper et al. 2016), or abiotic reactions (McCalley and Sparks 2009;Zhu-Barker et al. 2015;Homyak et al. 2017), thereby increasing the diameter of the pipe. However, the magnitude and timing of pulsed N emissions may vary as a function of environmental and edaphic factors that mediate which gaseous N intermediates are released to the atmosphere (i.e., the holes in the pipe). Understanding how process rates interact with the factors that control how much NO and N 2 O are emitted can help determine how N emissions may vary under future global change scenarios.
Two major challenges have limited progress identifying controls over soil NO and N 2 O emission pulses: (i) multiple biological and abiotic processes occur simultaneously making them difficult to separate, and (ii) traditional static chamber headspace experiments offer low temporal resolution, limiting understanding of the timing and magnitude of N trace gas emissions. To this end, isotope tracers are powerful tools that can help determine which N transformations produce NO and/or N 2 O (Van Groenigen et al. 2015). Moreover, isotope tracers can be coupled with laser-based isotope analyzers and automated soil chambers to detect the incorporation of 15 N tracers into N 2 O insitu and at high resolution (e.g., one measurement per second). While similar instruments do not yet exist for NO, the incorporation of 15 N tracers into NO can be detected using passive samplers (Homyak et al. 2016). By pairing high temporal resolution measurements of N emissions with stable isotopes, we assess the importance of increasing N availability (here used as a proxy for increasing the diameter of the pipe) relative to the factors that control how much NO and N 2 O is released (i.e. the holes in the pipe). Specifically, we ask: (1) what processes are contributing to pulsed NO and N 2 O emissions after wetting dry soils, and (2) how does N availability (both the amount and chemical form) affect the magnitude of pulsed emissions?
To answer these questions, we monitored N emissions at two dryland sites (desert and chaparral) in Southern California characterized by pronounced and frequent transitions from dry-to-wet soils. We chose two sites with contrasting environmental conditions (Table 1) to understand whether meteorological and edaphic factors would overrule the effects of increasing N supply and the form of nitrogen added, nitrate (NO 3 − ) or ammonium (NH 4 + ). We hypothesized that N trace gas emissions are limited by soil N availability, resulting in pulsed NO and N 2 O emissions proportional to the amount of added N. To infer which processes contributed to NO and N 2 O emissions, we added 15 N labeled NO 3 − or NH 4 + and used an automated chamber system connected to a NO and an isotope N 2 O analyzer. We also measured NH 3 emissions using passive samplers as a relative index of the amount of NH 3 in soil pore space that may be available to nitrifiers. We predicted that if pulsed N emissions were from nitrification, then added 15 N-NH 4 + would be captured as NO and/or N 2 O; if they were from denitrification, then added 15 N-NO 3 − would be captured as NO and/or N 2 O; and if they were from the rapid transformation of accumulated nitrification intermediates (e.g., NO 2 − ), then no 15 N label would be incorporated in N emissions.

Sites description
We studied two drylands in Southern California in August 2018 (the end of the summer dry season) with contrasting soils and vegetation. Our chaparral site was located in the Box Springs Reserve (33° 58′ 16.4″ N, 117° 17′ 53.4″ W), a transitional zone between coastal sage scrub and chaparral dominated by chamise (Adenostoma fasciculatum). Our desert site was located in the Boyd Deep Canyon Desert Research Center (33° 38′ 54.7″ N, 116° 22′ 39.4″ W), and was dominated by creosote (Larrea tridentata). Both sites are part of the University of California Natural Reserve System. Since 1980, the chaparral site has received an average of 28 cm of rain per year with an average maximum August daily temperature of 35 °C. During this same time, the desert site received an average of 11.4 cm of rain per year with an average maximum August daily temperature of 39.3 °C. The chaparral soils are sandy loams classified as thermic Typic Haploxeralfs within the Fallbrook series. The desert soils are stony sands classified as hyperthermic Typic Torriorthents within the Carrizo series. Both sites received no rain in the month before our experiments.
The soils at the two sites differed in several ways (Table 1). Soil NO 2 − was over seven times greater in the desert (0.58 ± 0.64 µg N g −1 ) than in the chaparral (0.08 ± 0.03 µg N g −1 , p < 0.05), while extractable NO 3 − and NH 4 + did not differ between sites (Table 1). Total C and N concentrations were Table 1 Soil chemical properties and meteorology prior to beginning experiments in the desert and chaparral sites (n = 8) Statistical significance between the two sites was assessed using student's t-test: *p < 0.10, **p < 0.05, ***p < 0.01. Errors represent standard deviation of the mean both higher in the chaparral (2.03 ± 0.35% C, 0.15 ± 0.02% N) than in the desert (0.92 ± 0.50% C, 0.08 ± 0.03% N). Desert soils were more alkaline (8.4 ± 0.19) than the chaparral (5.8 ± 0.50, p < 0.05).

Experimental design
We measured N trace gas emissions from underneath eight chamise shrubs in the chaparral and eight creosote shrubs in the desert. Interspace soils were not sampled as dryland shrubs are considered to be "islands of fertility" where soil nutrients are concentrated ). All shrubs were located within a 10-m radius and were separated from one another by at least one meter. Under each of the eight shrub canopies, we installed two pairs of PVC collars (4 collars, each 20 cm diameter × 10 cm height; inserted 5 cm into the ground) at least 48 h prior to the start of our measurements; the collar pairs were separated from each other by at least 50 cm to avoid cross contamination of isotope tracer and within 50 cm from the base of the shrubs. One pair of collars was wetted with NO 3 − solution, while the other was wetted with NH 4 + solution. Within each pair, one collar was used to measure N emissions, while the other was used to measure soil temperature, moisture, and inorganic N to minimize disturbances to the collars from which we measured emissions.
We wetted soils inside the collars with 500 mL of deionized water; this amount corresponds to about seven mm of rainfall, which is within the range of historically occurring rain events (Boyd Deep Canyon Desert Research Station, https:// doi. org/ 10. 21973/ N3V66D). During wetting, we added eight levels of N spike corresponding to 0, 2, 4, 6, 8, 10, 12, or 15 kg-N ha −1 as either NO 3 − or NH 4 + , covering a range of annual N deposition in Southern California drylands (Eberwein et al. 2020, Fenn et al. 2006). The nitrogen added was isotopically enriched to 2 atom percent 15 N. The labeled NO 3 − was added to two of the collars underneath each shrub starting at approximately 9 am. Soil NO and N 2 O emissions were measured from one collar underneath each shrub every 30 min beginning 15 min prior to wetting. After 24 h, this process was repeated with the NH 4 + label using the remaining collars underneath each shrub.
A separate group of four shrubs was used to measure the emission of NH 3 as well as the isotopic composition of NO. Emissions of NH 3 were used as an index of substrate availability to nitrifiers. These measurements were made using passive samplers (Ogawa pads; Ogawa USA, Pompano Beach, FL) that required soil chambers to be permanently closed, prohibiting integration with our automated chambers. The passive sampling pads are chemically pretreated so that they would collect either NO x , NO 2 , or NH 3 and have been demonstrated to work well under warm and humid conditions expected inside our soil chambers (Coughlin et al. 2017). We did not detect NO 2 on the NO 2 pads, indicating any N accumulation on the NO x pads was mostly NO. Two collars underneath each of the shrubs were wetted with 500 mL of either NO 3 − or NH 4 + solution (2 atom percent 15 N) at a concentration corresponding to 15 kg-N ha −1 . The remaining two collars underneath each of the four shrubs were wetted with deionized water only. Chamber lids were installed immediately after wetting and pads were switched out at the following time intervals: 0 to 15 min, 15 min to 12 h, and 12 to 24 h postwetting to capture NO and NH 3 during periods when we expected N emissions to be high.

NO and N 2 O emissions
We used an automated chamber system to simultaneously measure NO and N 2 O emissions from one of the collars under each of the eight shrubs sequentially over a 24-h period post-wetting. Collars were equipped with automated chambers (8100-104/C, LI-COR Biosciences, Lincoln, NE) connected to a multiplexer (LI-8150, LI-COR Biosciences) to sequentially measure emissions from each of the eight collars. We measured gas concentrations for two minutes, during which time gas from the chamber was recirculated through a sample loop connecting the multiplexer, an infrared gas analyzer (IRGA; LI-8100, LI-COR Biosciences), an isotope N 2 O analyzer (Model 914-0027, Los Gatos Research, Inc., Mountain View, CA), and a NO analyzer (Model 410 and Model 401, 2B Technologies, Boulder CO). The IRGA, N 2 O analyzer, and NO analyzer all sampled air from the recirculating sample loop, and each instrument, except for the NO analyzer, returned air back into the sample loop. Since the NO analyzer consumed NO, this air was vented to the atmosphere at a rate of 0.75 L min −1 . While this open system dilutes the concentration of trace gases emitted from the soil with atmospheric air, flux rates are not appreciably affected after accounting for our chamber volume (~ 6 L) and the short incubation period (Davidson et al. 1991). All instruments were housed inside an air-conditioned box made of five cm thick housing insulation (5 × 2 × 2 m). To prevent condensation in the lines, the sample loop included a water trap to remove moisture by cooling the hoses with ice water. Soil temperature (Model 8150-203, LI-COR Biosciences) and moisture sensors (Model 8150-205, LI-COR Biosciences) were installed under each shrub and were connected to the IRGA, which also measured relative humidity.
Fluxes of NO and N 2 O were calculated as the linear change in trace gas concentrations inside the chamber headspace over the last 90 s of the twominute incubation (script available on https:// github. com/ handr 003/ Trace GasAr ray). This timeframe was chosen to allow for even mixing of chamber air throughout the sample loop. The N 2 O analyzer recorded concentrations once every second and the NO analyzer recorded every ten seconds. If the linear correlation between time and trace gas concentration was not statistically significant (p > 0.1), the net flux was reported as zero. The change in NO concentration over time was highly linear over the 90 s window for all measurements (R 2 = 0.96). The change in N 2 O concentration over time was close to linear for all measurements (R 2 = 0.56) and was highly linear when N 2 O fluxes were greater than 10 ng N-N 2 O m −2 s −1 (R 2 = 0.97).
Flux values were corrected for the volume in the sample loop, soil temperature, and chamber volume. The isotopic N 2 O analyzer also recorded [δ 15 N]N 2 O, which requires five minutes of averaging time to report δ 15 N values within 1-sigma precision. Given the short incubation period of our measurements (2 min) and the fact that our measurements were diluted with ambient air, we do not attempt to calculate absolute [δ 15 (Coplen et al., 2012). Briefly, the Ogawa pads were extracted in 8 mL of deionized water and shaken overnight to extract NO as NO 2 − ; no NO 3 − was detected in the filtered extracts. The NO 2 − was then converted to N 2 O using Pseudomonas aureofaciens (Sigman et al. 2001). δ 15 N and δ 18 O values were measured using a Thermo Delta V isotope ratio mass spectrometer (Thermo Fisher Scientific, Woltham, MA) at the Facility for Isotope Ratio Mass Spectrometry (FIRMS; https:// ccb. ucr. edu/ facil ities/ firms) at the University of California, Riverside. Due to isotopic fractionation associated with NO collection with passive samplers, isotopic fractionation associated with the denitrifier method, exchange of oxygen atoms between NO 2 − and water (Casciotti et al. 2007;Dahal and Hastings, 2016;Yu and Elliott 2017), and potential interactions between volatile organic compounds (VOC) with NO (Walters and Michalski 2016) it is unlikely we measured the actual [δ 15 N]NO and [δ 18 O]NO emitted from soil. However, these fractionation and oxygen exchange effects are generally uniform across samples (Dahal and Hastings, 2016) and even if NO and VOCs interacted within our chambers, the passive samplers can still inform when 15 N tracers added to soils are detected as NO (Homyak et al. 2016 (Casciotti et al. 2007).
We used Ogawa pads to measure NH 3 emissions as an index of NH 3 availability in soil pore space that may be available to nitrifiers. The NH 3 pads were extracted in 8 mL of deionized water and shaken overnight to extract NH 3 as NH 4 + . Extracted NH 4 + was quantified using a colorimetric assay (SEAL methods Environmental Protection Agency (EPA)-126-A) using a SEAL AQ-2 discrete analyzer (SEAL analytical, Mequon, WI).

Soil chemical properties
We measured soil extractable NH 4 + and NO 3 − prior to wetting, two hours after wetting, and 24 h after wetting. NO 3 − and NH 4 + were measured by extracting soils (5 g) in 2 M KCl (30 mL). Soil solutions were shaken for one hour, filtered (Whatman 42 filter paper; 2.5 µm pore size), and frozen until analysis. We also measured NO 2 − prior to wetting; NO 2 − was extracted in deionized water to minimize its loss via gaseous N products (Homyak et al. 2015). We used colorimetric assays to measure soil extractable NH 4 + (SEAL method EPA-126-A), NO 3 − (SEAL method EPA-129-A), and NO 2 − (SEAL method EPA-137-A). Additionally, we measured total C, total N, and pH in dry soils (0-10 cm depth) collected from underneath each shrub prior to adding water or N. Soil total C and total N was measured in an elemental analyzer (Flash EA1112; Thermo Scientific, Woltham, MA) at the Environmental Sciences Research Laboratory at the University of California, Riverside (https:// envis ci. ucr. edu/ resea rch/ envir onmen tal-scien ces-resea rchlabor atory-esrl). Soil pH was measured in a 1:1 soil to water ratio with a pH meter (Orion VersaStar Pro; Thermo Scientific, Woltham, MA).

Statistical analyses
All statistics were conducted in R version 3.6.1 (R core development team, 2019). We used linear regression to evaluate the relationship between the amount of added N and soil NO emissions, N 2 O emissions, and peak *[δ 15 N]N 2 O. This was accomplished by first calculating the cumulative NO or N 2 O emissions measured at each shrub using the "trapz" function. Peak *[δ 15 N]N 2 O was calculated as the highest *[δ 15 N]N 2 O value recorded from underneath each shrub. We then used the "lm" function to determine the linear relationship between the amount of added N and cumulative NO or N 2 O emissions and peak *[δ 15 N]N 2 O. We report the R 2 of each linear regression where p < 0.10 to avoid type II error associated with high spatial variation in field experiments. However, we consider linear regressions with p > 0.05 as "weak" and include alternative explanations for these relationships. A block in the sample loop prevented us from measuring fluxes from two of the collars in the chaparral (2 and 10 kg-N NO 3 − ha −1 ) and these data were omitted from our analyses.
We used mixed effects models to determine when 15 N tracers were detected in NO collected using passive samplers. The models included *[δ 15 N]NO as the response variable, collection time as the predictor variable, and a random effect to account for measuring the same collar repeatedly. Models were run using the "nlme" package in R. We used the anova.lme function to determine if time was a significant model term and Tukey corrected multiple comparisons to determine which times differed compared to ambient *[δ 15 N]NO. We used the same approach to determine if *[δ 15 N]NO and *[δ 18 O]NO changed in collars that were amended with water only.

Soil N 2 O emissions
In the desert, peak N 2 O emissions averaged 529 ± 469 ng N-N 2 O m −2 s −1 after wetting soils with NO 3 − and NH 4 + amended solutions, and returned to prewetting levels within four hours (Fig. 1a,c). However, desert N 2 O emissions did not increase in proportion to adding more NO 3 − (p = 0.12) or NH 4 + (p = 0.89, Table 2). In contrast to the desert, peak chaparral N 2 O emissions averaged only 38.0 ± 72.0 ng N-N 2 O m −2 s −1 after wetting with NO 3 − and NH 4 + amended solutions (Fig. 1b,d). As observed in the desert, N 2 O emissions did not increase in proportion to adding NO 3 − (p = 0.25) or NH 4 + (p = 0.10,  (Fig. 2a,b), whereas the *[δ 15 N]N 2 O of from soils amended with water only did not surpass 32.4 ‰ at either site. Peak *[δ 15 N]N 2 O increased in proportion to NO 3 − addition in the desert (R 2 = 0.62, slope = 107 ‰ (kg N ha −1 ) −1 , p = 0.013) and to a smaller degree in the chaparral (R 2 = 0.31, slope = 6.2 ‰  -amended soils at both sites (Fig. 2c,d). Peak *[δ 15 N]N 2 O was positively correlated to NH 4 + addition in the desert only (R 2 = 0.50, slope = 2.85 ‰ (kg N ha −1 ) −1 , p = 0.03), but the slope of this relationship was over 37 times smaller compared to NO 3 − addition (Table S1).

Soil NO and NH 3 emissions
Prior to wetting, NO emissions were greater in the chaparral (14.7 ± 4.31 ng N-NO m −2 s −1 ) than in the desert (0.39 ± 9.27 ng N-NO m −2 s −1 ) ( Table 1). In the desert, NO emissions steadily increased for 10 h post-wetting, and remained elevated for the remainder of the experiment (Fig. 3a); peak NO emissions averaged 221 ± 269 ng N-NO m −2 s −1 in NO 3 − -amended soils (Fig. 3a) and 254 ± 200 ng N-NO m −2 s −1 in NH 4 + -amended soils (Fig. 3c). In contrast to the desert, chaparral NO emissions reached their peak within only 5 h of wetting and decreased at faster rates; NO emissions averaged 114 ± 204 ng N-NO m −2 s −1 in NO 3 − amended soils (Fig. 3b) and 202 ± 154 ng N-NO m −2 s −1 in NH 4 + amended soils (Fig. 3d).
Soil NO emissions increased in proportion to incremental NH 4 + additions in the chaparral (R 2 = 0.58, p = 0.03, Fig. 6d), whereas in the desert, the relationship was positive but weak (R 2 = 0.45, p = 0.07, Fig. 6c). Adding NO 3 − may have increased NO emissions in the chaparral, but the relationship was weak (p = 0.09, Table 2); adding NO 3 − did not increase NO emissions in the desert (p = 0.28, Table 2).
Soil NH 3 emissions increased immediately after wetting both sites but remained higher in the desert Table 2 Linear relationship between the added amount of NO 3 − or NH 4 + and cumulative N 2 O or NO emissions over the course of 24-h from either the desert or chaparral Linear regression was performed to assess the relationship between the amount of N-added (kg N ha −1 ) and cumulative N 2 O or NO emissions (µg N m −2 ). Coefficient of determination (R 2 ) and p-value of the linear regression are reported. Significance of the relationship is noted as: *p < 0.10, **p < 0.05, ***p < 0.01 relative to the chaparral (Fig S1). In the desert, NH 3 emissions averaged 27.3 ± 24.6 µg N-NH 3 m −2 h −1 between 0.25 and 12 h in NO 3 − amended soils; the NO 3 − treatment did not increase NH 3 emissions compared to soils amended with only water (Fig S1a). In NH 4 + amended desert soils, NH 3 emissions averaged 52.5 ± 45.0 µg N-NH 3 m −2 h −1 between 0.25 and 12 h, compared to 16.7 ± 10.6 µg N m −2 h −1 in soils amended with only water (Fig S1c).

Discussion
We investigated the dynamics of and mechanisms driving pulsed NO and N 2 O emissions during drying-wetting cycles in two contrasting drylands. We found that soil NO emissions increased in proportion to the amount of NH 4 + added in both sites, although this relationship was weaker in the desert, partially supporting the hypothesis that increasing biological process rates would increase N emissions and solutions. Each black dot represents flux measurements over a 2-min interval for each of the 8 chambers suggesting that nitrification may control NO emission magnitude in these coarse-textured soils. In contrast, increasing N supply did not increase N 2 O emissions at either site, which does not support the hypothesis that N 2 O emissions are limited by NO 3 − or NH 4 + . While N addition did not stimulate N 2 O emissions, N 2 O was produced in part by the near-instantaneous reduction of NO 3 − , raising questions as to the from each treatment within each site. Dots represent individual measurements using passive Ogawa samplers. Asterisks indicate if the mean for a given time differed from the control (*p < 0.1, **p < 0.05, ***p < 0.01) mechanisms driving NO 3 − reduction in these dryland soils and how factors controlling these emissions could help explain variation in N emissions across ecosystems.

N 2 O emissions: controls and dynamics
While we expected N 2 O to increase within minutes after wetting to produce large emission pulses (Eberwein et al. 2020), the incorporation of 15 N-NO 3 − tracer into N 2 O within 15 min of adding water was unexpected (Fig. 2a,b)-denitrification is an anaerobic process not thought to dominate in well-aerated coarse-textured soils during dry summer months (Werner et al. 2014). Possibly, rapid onset of microbial respiration (Birch 1958;Jenerette and Chatterjee 2012) consumed sufficient O 2 to stimulate N 2 O production via denitrification immediately after adding water, or soil aggregates may have sustained a viable denitrifier population within anoxic microsites throughout the hot and dry summer (Sexstone et al. 1985). Indeed, laboratory studies show denitrification enzyme activity can be maintained in dry soils (Peterjohn 1991;Parker and Schimel 2011), perhaps suggesting this process is viable in deserts. However, because NO is produced as an obligate intermediate during denitrification, and our 15 NO 3 − tracer was not incorporated into NO within 15 min post-wetting in the desert (Fig. 4a), denitrification may not have contributed to rapid N 2 O emissions. Besides biological processes, chemodenitrification can produce N 2 O (Zhu-Barker et al. 2015;Heil et al. 2016;Harris et al. 2021), but the abiotic reduction of NO 3 − has only been reported under manipulated laboratory settings (Davidson et al. 2003;Matus et al. 2019) and is yet to be demonstrated to occur in-situ (Colman et al. 2007(Colman et al. , 2008. The detection of the 15 N-NO 3 − label in N 2 O within 15 min of wetting dry soils at our site shows that dryland soils have the capacity to reduce NO 3 − immediately after wetting and argues for additional work identifying which processes contribute to rapid N 2 O emissions. Even though 15 N-NO 3 − was rapidly reduced to N 2 O, adding more NO 3 − did not increase the magnitude of pulsed N 2 O emissions. This suggests that the processes reducing NO 3 − to N 2 O are not limited Isotopic NO composition is presented from shrubs receiving water-only additions by soil N availability (i.e., the size of the pipe), and that other factors regulate the magnitude of N 2 O emissions. For example, more 15 N-NO 3 − tracer was reduced to N 2 O in the desert ( Fig. 2; Table S1), where soils had higher pH and warmer temperature compared to the chaparral (Table 1). These soil properties and environmental conditions can determine which N intermediates are released to the atmosphere, and cumulative NO emissions. Shaded gray areas represent the 95% confidence interval for statistically significant linear regressions (p < 0.1) potentially explaining variation in the magnitude of N 2 O emissions between sites. For example, higher pH desert soils may have increased denitrification rates (Knowles 1982), or warmer temperatures in the desert may have favored abiotic reactions that can produce N 2 O (McCalley and Sparks 2009;Zhu-Barker et al. 2015). Average peak N 2 O emissions from the desert were slightly higher compared to emissions measured in tropical forests (66.4 ng N-N 2 O m −2 s −1 ; Hall and Matson 2003) and temperate agricultural systems (355 ng N-N 2 O m −2 s −1 ; Smith et al. 1994), which are thought of as denitrification hotspots. In addition to differences in pH and temperature between sites, variation in soil properties underneath each shrub could override any effect of experimental N addition on N 2 O emissions. Indeed, N 2 O emissions are notoriously difficult to predict since they are often driven by high rates of microbial activity within microsites where soil C and N are concentrated (Sey et al. 2008;Harris et al. 2021). As such, greater replication may be needed to detect effects of N addition over the inherent variability in N 2 O emissions. Despite this variation, documenting the rapid reduction of NO 3 − to form N 2 O is an important step in identifying controls over dryland N 2 O emissions.

NO emissions: controls and dynamics
Nitrification produced NO at our sites as supported by the detection of 15 N-NH 4 + in NO (Fig. 4c,d) and the positive response of NO emissions to adding NH 4 + (Fig. 6c,d). In addition to nitrification, denitrification also produced NO at both sites; 15 N-NO 3 − was reduced to NO 12 h after wetting dry soils in the desert, and within 15 min in the chaparral (Fig. 4a,b). Denitrifiers can initiate NO 3 − reduction within hours of decreasing soil O 2 concentrations (Liu et al. 2019) and maintain this activity once aerobic conditions return (Roco et al. 2016). We also observed a simultaneous decrease in *[δ 18 O]NO and *[δ 15 N]NO over the course of the incubation at both sites, perhaps suggesting other processes produced NO (Fig. 5). While changes to *[δ 18 O]NO and *[δ 15 N]NO could have been caused by interactions between NO and VOCs (Walters and Michalski 2016), these observations may also indicate nitrifier denitrification activity as observed in a Mediterranean grassland (Homyak et al. 2016). Nitrifier denitrification produces NO from NO 2 − , which contains O from both water and air, whereas nitrification produces NO from NH 2 OH, which contains only O from air (Andersson and Hooper 1983;Buchwald et al. 2012;Medinets et al. 2015;Boshers et al. 2019). As such, the change in *[δ 18 O]NO may reflect incorporation of 18 O from the NO 2 − produced prior to and after wetting these dry soils (Homyak et al. 2016). Furthermore, biological NO production pathways-including nitrifier denitrification and nitrification-fractionate against 15 N by 28-60 ‰ (Robinson 2001), consistent with the simultaneous decrease in *[δ 15 N]NO observed throughout the incubation. Abiotic reactions may have also contributed to soil NO efflux by converting nitrification intermediates-such as NO 2 − or NH 2 OH-to NO (McCalley and Sparks 2009;Heil et al. 2016;Homyak et al. 2017). Regardless of the mechanism, our work suggests that multiple pathways, including those requiring anaerobic conditions, produce NO after wetting these dry coarse-textured soils.
Soil NO-producing pathways were likely limited by soil N availability, since adding more N was associated with higher NO emissions. The positive response of cumulative NO emissions to adding NH 4 + is consistent with N limitation of N trace gas production via nitrification (Davidson et al. 2000;Vourlitis et al. 2015;Prosser et al. 2019), as has been observed in other drylands (Hartley and Schlesinger 2000;Eberwein et al. 2020). However, other factors besides N limitation likely contributed to the magnitude of the NO pulse since NO emissions diverged between sites; the tracers were reduced to NO more quickly in the chaparral (Fig. 4), while cumulative NO emissions had a larger positive relationship with NH 4 + addition in the desert (Fig. 6). These differences between sites may be explained by background microbial activity. For example, chaparral soils were exposed to fog (Table 1) and were already producing NO before we added water, whereas desert soils were not (Fig. 3b,d). In this sense, non-rainfall water inputs via fog (McHugh et al. 2015) may have influenced the magnitude of pulsed N emissions by resuscitating microbes and priming them for the N we added, helping to explain the rapid NO emission pulse (Fig. 3b,d) and the rapid incorporation of 15 N-NO 3 − into NO (Fig. 4b). In contrast to chaparral, microorganisms in the relatively drier desert took hours to activate before producing the more delayed, but relatively long-lasting, NO emission pulse (Fig. 3a,c). In the desert, we measured higher NH 3 emissions relative to the chaparral (Fig S1), consistent with higher soil pH favoring NH 3 production from the equilibrium between NH 3 and NH 4 + (pKa = 9.25; Avnimelech and Laher 1977). This suggests the longer NO emission pulse in the desert could have been sustained by greater NH 3 diffusion through soil pore space and supply to nitrifiers even as drying soils may have limited nitrifier access to NH 4 + in soil pore water (Stark and Firestone 1995). The role of NH 3 diffusion to nitrifiers may also help explain why the relationship between NH 4 + addition and NO emissions was weaker in the desert; variable background NH 4 + concentrations may have supplied NH 3 to nitrifiers even when little N was added to soils. Taken together, our observations support the hypothesis that wettinginduced NO emissions are limited by soil N availability but suggest that environmental and edaphic factors contribute to variation in NO production among ecosystems.

Conclusion
We demonstrate that rapid NO 3 − reduction (within 15 min) can occur even in coarse summer-dry desert soils under temperature extremes to produce N 2 O. However, the N 2 O emissions produced were insensitive to experimentally adding N. Identifying the processes that govern the rapid NO 3 − reduction pathway will help constrain variation in N emissions across dryland soils as these ecosystems expand with expected changes in climate (Huang et al. 2016). In contrast to N 2 O, NO emissions were governed by N limitation of multiple N cycling processes, suggesting that N-limited NO production pathways may increase in response to higher rates of atmospheric N deposition (Fenn et al. 2006). These wetting induced N trace gas production pathways appear widespread across ecosystems that experience repeated drying-wetting cycles and will likely become increasingly important sources of atmospheric NO and N 2 O as global precipitation regimes become more variable.
Funding This research was funded by the National Science Foundation (DEB 1916622 and DEB 1656062).

Data availability
The datasets generated during this study are available in the Dryad repository, https:// doi. org/ 10. 6086/ D1C39X.
Code availability Custom code used for calculating trace gas fluxes is available at https:// github. com/ handr 003/ Trace GasAr ray

Declarations
Conflict of interest The authors have no conflicts of interest to declare that are relevant to the content of this article.

Ethical approval Not applicable.
Informed consent Not applicable.

Consent for publication Not applicable.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.