Infrequent soil wetting in deserts can induce large nitrogen (N) trace gas pulses; however, how other abiotic mechanisms interactively control the timing and magnitude of these pulses is not clear. In particular, production of nitric (NO) and nitrous (N₂O) oxide may be differentially dependent on temperature, carbon, and nitrogen availability. At a desert field site in southern California, USA, we used an automated sensor system in four years of field campaigns to track NO and N₂O pulse responses to experimental manipulations of carbon (C) and N across a range of ambient temperatures and shrub “islands of fertility”. We observed rapid onset and shorter duration of N₂O pulses immediately after wetting compared to lagged and extended pulses of NO, suggesting preferential incorporation of N initially into N₂O in anoxic microsites and then to NO as soils dry. We identified strong nitrogen limitation and exponential temperature dependence of NO pulses, particularly for soils located under shrubs. N₂O pulses were less responsive to experimental manipulations but showed evidence of C and N co-limitation as well as seasonal temperature differences. As atmospheric N deposition and temperatures continue to increase in desert systems, we can expect larger losses of N from soils as pulse-based emissions.

2.1 Study site and focal species

We conducted five field experiments, spanning four years and encompassing both summer and winter seasons, at Boyd Deep Canyon Desert Research Center (hereafter “BDC”; Palm Desert, California, USA; 33.6480° N, 116.3765° W, 290 meters ASL). BDC is situated in Southern California’s Colorado Desert, a subregion of the Sonoran Desert, and experiences a climate characterized by hot summers and mild winters. Up to 85% of the precipitation at this site occurs in the winter season and predominantly as rain, but sporadic heavy rainfall can occur in summer associated with North American monsoons. Monthly average temperatures in this area range from 12 °C in January and February to 35 °C in July and August, with temperatures occasionally exceeding 45 °C during the summer months. Annual atmospheric nitrogen deposition averages 4.2 kg ha^-1 yr^-1, a relatively low rate for deserts in the southwestern United States (Schwede & Lear, 2014) and globally (Ackerman et al., 2019). BDC soils are alkaline with a pH of 8.36 ± 0.19 and are classified as stony sand Hyperthermic Typic Torriorthent soils in the Carrizo Series. Previous studies conducted at this location have identified high spatiotemporal variability of soil C and N pools (Krichels et al., 2022; Eberwein et al., 2020; Chatterjee and Jenerette, 2011), suggesting hotspots of biogeochemical processes even within experimental replicates.

The primary shrub species at Boyd Deep Canyon and focal species in our wetting experiments was creosote bush (Larrea tridentata). These shrubs play a key role in creating fertile “islands” of herbaceous annual plants and soil organic matter amidst sandy interspaces (Schlesinger et al., 1996). In each field campaign, we used L. tridentata shrubs as experimental replicates. However, for each field campaign, we selected a different group of shrubs within BDC to avoid potential legacy effects from previous wetting and substrate modifications. In experiments where we compared soils under shrubs to soils in shrub interspaces, we chose sandy patches located at least 1.5 meters adjacent to focal shrubs as interspace treatments. All wetting experiments were performed when volumetric soil moisture content was below 10 % and the site had not received significant rain for at least three weeks prior. Therefore, it was assumed that differences in NO and N₂O dynamics between seasons were more likely due to responses to soil temperature rather than antecedent soil moisture levels.

2.2 Soil temperature, moisture, and N trace gas measurement system

             Prior to each field campaign, pairs of polyvinyl chloride (PVC) soil collars measuring 20-cm in diameter were installed to 5-cm depth and adjacent to each other under L. tridentata canopies or in interspaces between shrubs. One collar per pair was designated for trace gas measurements and the other collar was used for soil temperature, moisture, and ancillary soil physicochemistry measurements, where similar climate and edaphic conditions were assumed between collars. Both collars in each pair received identical wetting treatments and collar pairs were situated at least 1.5 meters apart from other pairs.

Beginning up to 24 hours prior to wetting, 15 minutes following wetting, and then at 30-minute intervals over the subsequent 24-45 hours, we measured soil temperature, moisture, and NO and N₂O fluxes using a sensor array that has been previously described in Andrews et al. (2023) and is recounted briefly here. The array consisted of eight automated chambers (LI-8100-104; LI-COR Bioscience, Lincoln, NE, USA) installed on soil PVC collars and attached to soil temperature (LI-8150-203 thermistor probe; LI-COR Bioscience) and moisture (LI-GS1 probe; LI-COR Bioscience) probes. Probes were inserted into adjacent collars to 5 cm below the soil surface and provided integrated measurements of 3-5 cm soil depth. Each chamber measurement sequence included a 30-second pre-measurement purge, a 2.5-minute active measurement period of trace gas concentrations and soil temperature and moisture status, and a 30-second post-measurement purge, where concentrations of N₂O and NO were measured simultaneously every 1 and 10 seconds, respectively. Air collected from an actively measuring chamber was passed through a multiplexer (LI-8150; LI-COR Bioscience) and delivered to a sequence of calibrated gas analyzers connected in a semi-closed loop: 1) a N₂O/CO cavity-ringdown infrared spectrometer (Los Gatos Research, San Jose, CA, USA);  2) the LI-8100A; and 3) a coupled nitrogen dioxide (NO₂) converter and NO monitor (Models 401/410, 2B Technologies, Boulder, CO, USA). The multiplexer, N₂O/CO analyzer, and LI-8100 formed a closed loop that returned sample air to the active sampling chamber; however, the NO₂/NO monitor is an open, chemiluminescence-based system, so a portion of air was siphoned from the sample loop through a one-way check valve and was measured as NO (and NO₂, which was assumed negligible at this study site). Given small chamber volume and short measurement time, we assume a small but non-negligible amount of error introduced into flux calculations by incorporating a controlled NO "leak" in the sample loop (Andrews et al., 2022; Davidson et al., 2000; Davidson et al., 2008).

2.3 Soil extractable carbon and nitrogen pools

In four out of five field experiments, we collected 5-cm² x 10-cm deep soil cores from our “ancillary measurements” collars immediately prior to wetting and at 24-48 hours after wetting to track changes in N (ammonium NH₄⁺ and the sum of nitrite+nitrate NO₂^- + NO₃^-). In Summer 2020, we additionally analyzed soil cores for water-extractable C (dissolved organic carbon DOC). Following standard extraction techniques (Carter and Gregorich, 2006), each core of fresh soil was homogenized in a plastic bag, sealed, and transported to the lab on ice. For inorganic N analysis, one subset of each soil sample was extracted in 1:10 soil weight:solution volume ratio of 2 M KCl solution. For DOC analysis, a second subset of each sample was extracted in 1:5 soil weight:solution volume ratio of deionized water. Extracts were shaken for 1 h, centrifuged, and gravity-filtered through 11-micron filter paper at room temperature. NH₄⁺ quantification was performed via phenate method and NO₂^-+NO₃^- quantification was performed via acidification and automated cadmium coil reduction (AQ2 Discrete Analyzer, Seal Analytical Inc.; Mequon, Wisconsin). DOC quantification was performed via acidification and sparging methods (TOC-V, Shimadzu, Kyoto, Japan). All C and N extracted concentrations were normalized to gravimetric water content (GWC), which was measured as the difference between masses of field-moist soil and oven-dried soil, divided by dry soil mass.

2.4 Experimental design

             We conducted five field campaigns at BDC to identify temperature and substrate drivers of NO and N₂O pulses at multiple temporal and spatial scales. These experiments have been previously described in Andrews et al. (2023), for which soil CO₂ pulse dynamics were reported, and will be recounted in brief. Four campaigns explicitly tested interactions among temperature and C and/or N limitation of N pulses at seasonal timescales. An additional summer campaign measured N pulse responses across a diel temperature range to explore fine-scale temperature controls over pulse responses to wetting.

2.4.1 Interactive temperature-substrate manipulation experiments

Four of five field experiments were substrate addition experiments conducted at BDC during 2018-2021 to evaluate individual and interactive seasonal temperature and substrate controls on soil NO and N₂O flux responses to wetting. In 2018, two campaigns evaluated non-factorial combinations of wetting with water alone, with water plus N (Winter 2018 only), and with water plus C and N (Summer 2018 only). In Summer 2020 and Winter 2021, we extended these studies by conducting two fully-factorial experiments which quantified pulse responses to wetting alone, wet+C, wet+N, and wet+C+N amendments in summer and winter seasons. All wetting experiments began during mid-morning hours (8:30-10:00) and tracked soil temperature, moisture, and NO and N₂O fluxes over subsequent 45-hour periods under 3-4 L. tridentata shrubs and in adjacent interspaces. All soils were wetted by hand to simulate a 2-cm rain event (1.5L), representing a medium-sized event for the site (5-year event size mean = 0.6 cm, min = 0.03 cm, max = 6.3 cm; WRCC, 2021). C and N amendments were added as aqueous dextrose (30 g/L; 2406.4 kg C ha^-1) and ammonium-nitrate (13 g/L; 912 kg N ha^-1) solutions, respectively. These C and N concentrations are much larger than naturally-occurring deposition rates and were chosen to help overcome C and N limitation, allowing us to compare the potential sensitivity of nitrification and denitrification between endpoints of resource availability.

2.4.2 Fine-scale temperature manipulation experiment

Expanding on observations of strong pulse responses to summer wetting in 2018, we conducted a fifth field experiment in Summer 2019 to explore temperature-pulse relationships at fine temporal scales. In this experiment, we wetted soils at different hours of the day and quantified how temperature at time of wetting influenced NO and N₂O pulses. Soil collars were installed in cardinal directions around 8 focal L. tridentata shrubs, where 2 shrubs constituted a treatment replicate containing all hour-of-day treatments. Within each 2-shrub replicate, collars were wetted with water at 3-hour intervals (0:00, 3:00, 6:00, 9:00, 12:00, 15:00, 18:00, 21:00), one collar per hour treatment (collars were not wetted more than once), and were measured for 24 hours post-wetting.

2.5 Data processing and statistical methods

Raw NO and N₂O concentration and soil probe measurements were batch processed into fluxes and associated soil temperature and moisture data using algorithms adapted from previous work using this chamber array (Andrews et al., 2023; Andrews et al., 2022; Krichels et al., 2022). Instantaneous fluxes of NO and N₂O were calculated as the regression coefficient of linear increase in gas concentration data during the 2.5-minute active chamber measurement period, accounting for soil collar dimensions and atmospheric parameters following the Ideal Gas Law (Davidson et al., 2000). Fluxes were compiled and integrated with instantaneous soil temperature and moisture measurements using a publicly-available R script (Andrews and Krichels, 2021). Additional post-processing filtering steps were conducted when data failed to cross a threshold of data quality and control due to chamber or analyzer malfunctions. Our final five-campaign dataset consisted of 7972 NO and N₂O fluxes with corresponding soil temperature and/or moisture measurements. From continuous measurements of each chamber, we constructed 24-hour time series following wetting and extracted the magnitude and timing of maximum instantaneous flux. We also calculated 24-hour total NO and N₂O fluxes using linear trapezoidal integration of chamber observations which occurred at 30-minute intervals. 24-hour total flux measurements were removed from further analysis if, due to previous quality control steps, integration of a total flux was calculated using less than 5 instantaneous flux measurements.

Statistical analyses were conducted in JMP 16 and data management and visualization were performed in R 4.1.3 and RStudio. All p-values are reported at ɑ = 0.05. NO and N₂O fluxes were transformed using the Box-Cox family of transformation as needed to ensure normality of measurements. For boxplots describing pulse behavior following wetting, we included all field measurements. We compared temperature sensitivity of NO and N₂O pulses from nutrient addition treatments using models (linear, quadratic, or exponential) that generated the highest AIC values. To evaluate interactive season and substrate effects on N pulses, we included field measurements from fully-factorial C and N amendment experiments (Summer 2020, Winter 2021). We constructed general linear models to describe peak and 24-hour total N fluxes as described by season, nutrient addition treatment (wet, wet+C, wet+N, wet+C+N), and their two-way interaction. Post-hoc Tukey’s HSD tests were used to evaluate pulse differences across individual treatments.