Cropland soil nitrogen oxide emissions vary with dairy manure incorporation methods

Abstract

Soils contribute 15%–75% of total atmospheric nitrogen oxide (NO_x) emissions in agricultural regions during the growing season. However, the impacts of cropland fertilizer management on spatially heterogeneous, temporally episodic NO_x emission patterns are highly uncertain. We examine the effects of liquid slurry dairy manure application practices on soil NO_x emissions in rainfed, corn-soybean rotations during spring 2016 and 2017. Daily soil NO_x emissions and weekly soil inorganic N measurements were performed in a randomized split–split plot design for 1–4 weeks following manure applications. NO_x emissions and soil N with shallow-disk injection and chisel-disk manure incorporation methods were compared with unincorporated broadcast practices. Injected manure and chisel-disk incorporation exhibited two–four times larger mean NO_x emissions than those with unincorporated broadcast manure. Larger soil NO_x emissions with manure incorporation practices were driven by the predominance of nitrification in these treatments with evidence of soil nitrate production. Soil NO_x emission differences between treatments were detectable across order of magnitude changes in daily NO_x emissions during two growing seasons. Larger soil NO_x emissions associated with manure incorporation practices compared with unincorporated broadcast practices occur alongside larger N₂O and smaller NH₃ emissions, highlighting important air quality and climate impact tradeoffs for cropland manure fertilizer management choices.

Abbreviations
• ANOVA
• analysis of variance
• NRCS
• National Resource Conservation Service
• PM-2.5
• fine particulate matter
• USDA
• United States Department of Agriculture
• WFPS
• water-filled pore space

INTRODUCTION

Atmospheric nitrogen oxides (NO_x ≡ NO + NO₂) are major precursors to tropospheric ozone and nitrate (NO₃⁻) aerosol formation, with implications for air quality, climate change, and ecosystem nitrogen (N) deposition (IPCC, 2013; Paulot et al., 2013; Pinder et al., 2012). Soil emissions comprise ∼15% of total annual NO_x emissions (Weng et al., 2020), yet regional soil emission hotspots can contribute up to 75% of total growing season NO_x emissions (Vinken et al., 2014). As other anthropogenic NO_x emission sources have declined, soil NO_x emissions are contributing an increasingly larger portion of the NO_x budget with important implications for tropospheric ozone production (Almaraz et al., 2018; Romer et al., 2018; Sha et al., 2021).

Along with NO_x, agricultural soils emit a suite of other reactive N species that are relevant for air quality and climate change. First, nitrous oxide (N₂O) is both a potent greenhouse gas and a stratospheric ozone depleting substance with increasing global emissions dominated by agricultural soils (Tian et al., 2020). Second, ammonia (NH₃) emissions represent a primary component of cropping system N losses to the atmosphere, where it is a precursor to fine particulate matter (PM-2.5) and inorganic N deposition (Bauer et al., 2016). Ammonia represents a key target species for mitigating agricultural soil N emissions because it is associated with the largest soil reactive N emission-related health burden and economic losses due to volatilization of crop-available N. The societal damages associated with PM-2.5 formation from precursor soil NH₃ emissions exceed those from soil NO_x and N₂O emissions in the United States (Luo et al., 2022). Finally, nitrous acid (HONO) is a short-lived photo-chemically reactive species that contributes to hydroxyl radical formation, though its contribution to soil reactive N emissions by mass is very small (Oswald et al., 2013).

Gaseous reactive N emissions from agricultural soils are spatially heterogeneous and temporally episodic because they originate from pathways regulated by fertilizer types, application methods, and soil and meteorological conditions (Cooter et al., 2012; Pilegaard, 2013). Nitric oxide (NO) is primarily produced in fertilized soils by nitrifying bacteria oxidizing ammonium (NH₄⁺) under aerobic conditions; denitrifying bacteria reducing NO₃⁻ or nitrite (NO₂⁻) under anaerobic conditions; and abiotic NO₂⁻ reduction (Homyak et al., 2017; Pilegaard, 2013). Soil moisture, aeration, temperature, and N substrate availability are major factors that regulate soil NO emission magnitudes (Davidson et al., 2000). N₂O is also produced during both nitrification and denitrification, with larger denitrification N₂O emissions and NO consumption favored at greater soil moisture conditions (Davidson et al., 2000). Nitrogen from NH₄⁺ or urea-based fertilizers is volatilized from soils as gas phase NH₃ (Dell et al., 2012; Paulot et al., 2013; Xu et al., 2019). Quantifying gaseous NO_x, NH₃, and N₂O emissions remains a major challenge due to large spatial variability across fertilizer management practices and soil conditions (Duncan et al., 2019; Liu et al., 2016; Xu et al., 2019).

Core Ideas

• Soil NO_x emissions were characterized following rainfed liquid dairy manure applications in the northeastern United States.
• Soil NO_x emissions with manure incorporation were two–four times larger than with unincorporated broadcast manure.
• Larger soil NO_x emissions with manure incorporation were associated with the predominance of nitrification.

Because diverse N fertilizer sources and management practices are increasing in use, their reactive N impact profiles are important to characterize. Manures have always been an important N source for crop production, and the use of no-till and other conservation tillage practices has increased in recent decades in order to reduce soil erosion (Amin et al., 2018; Cooter et al., 2012; Han et al., 2017). The degree to which manure is incorporated into the soil or left on the surface influences the balance of oxidized and reduced N emitted and nutrient losses in runoff. Compared with unincorporated manure, low-disturbance manure incorporation methods (e.g., shallow-disk injection) can significantly reduce NH₃ volatilization while retaining most of the soil conservation benefits of no-till (Dell et al., 2012; Duncan et al., 2017). However, injection increases N₂O emissions because it creates high soil moisture and rich soil carbon (C) zones that promote rapid N mineralization and incomplete denitrification (Dell et al., 2011; Duncan et al., 2017; Ponce de León et al., 2021).

Although soil NO_x emission patterns have been quantified using remote sensing, in situ observations, and model simulations, these estimates are limited in scope across manure application practices. Satellite remote sensing studies have captured regional-scale, daily to seasonal emission pulsing events, yet they are too limited in spatial resolution to discern field-scale emission heterogeneities (Huber et al., 2020; Lu et al., 2021). Model simulations have estimated field-scale reactive N emission patterns across tillage and fertilizer types (Luo et al., 2022; Rasool et al., 2019). However, models are based on emission factors from limited field measurements that do not include all manure incorporation methods. Field-based soil NO_x emission reductions have been observed in tilled systems with partial substitution of synthetic fertilizer with manure (Zheng et al., 2003) and for synthetic N-fertilizer with no-till compared with tillage (Liu et al., 2005, 2006). However, the effects of chisel-disk incorporation or shallow-disk manure injection compared with unincorporated broadcast manure on soil NO_x emissions remain unknown. New in situ emission observations are needed to grow the limited body of soil NO_x emission observations and understand the influences of manure application practices on soil NO_x emissions.

In this study, we characterized the effects of liquid slurry dairy manure incorporation practices on in situ cropland soil NO_x emissions during a significant portion of the assumed emission pulse period (typically 1–4 weeks after manure application). The corn-phase of a rainfed corn–soybean rotation was sampled following spring applications of dairy slurry manure. During 2016, unincorporated broadcast manure and shallow-disk manure injection methods were investigated. During 2017, unincorporated broadcast manure, chisel-disk incorporation, and shallow-disk manure injection methods were investigated. We performed daily soil NO_x emission and weekly soil NH₄⁺ and NO₃⁻ concentration measurements collocated with emission observations. We assumed our relatively short-duration, high temporal frequency sampling scheme captures the key emission period for manure N processing and peak NO_x emissions. We investigated the potential drivers of temporal and treatment emission variations, including daily water-filled pore space (WFPS), temperature, and N substrate evolution. Finally, we discuss the potential benefits and tradeoffs of evolving manure incorporation practice choices for air quality and climate impacts.

MATERIALS AND METHODS Field sampling approach

Field measurements were conducted as part of the Pennsylvania State University Sustainable Dairy Cropping Systems Experiment (Northeast Sustainable Agriculture Research and Education [NESARE], 2009) at the Russell Larson Agricultural Research Center site in Rock Springs, PA (40.7234°N, 77.9244°W) (Amin et al., 2018). This field site is located in the Chesapeake Bay Watershed, and soils are classified as Murrill channery silt loam (fine-loamy, mixed, semiactive, mesic Typic Hapludults) with a pH of ∼6.2, ∼13.0 g total C kg⁻¹, and 1.3 g total N kg⁻¹ in the upper 15 cm (Ponce de León et al., 2021). Corn (Zea mays L.) following soybeans (Glycine max L. Merr.) received liquid slurry dairy manure (∼8%–11% solid) containing NH₄⁺ (∼40% of N mass) and organic N (∼60% of N mass) (Table 1). Liquid dairy slurry manure was obtained from a neighboring dairy farm herd of milking cows and heifer cattle. Dairy slurry was agitated and removed from a bottom-loaded, aboveground storage lagoon, which contributed to a relatively large dry matter fraction (Table 1). The dairy manure slurry was applied via three methods: (1) broadcast with chisel-disk incorporation (tillage to 20- to 25-cm depth), (2) unincorporated broadcast manure, and (3) shallow-disk injection at ∼10-cm deep bands with injector units spaced at 76-cm intervals (e.g., Atiyeh et al., 2015; Dell et al., 2011, 2012; Maguire et al., 2011). The chisel-disk process took place ∼4–5 h after manure application and involved a chisel plow, a pass with a disk followed by harrow, and finally a cultimulcher. In chisel-disk plots, soil was not completely inverted and 30% or more of residue was left on the soil surface. Unincorporated broadcast manure plots had all manure left on the soil surface. Injected manure plots had a fraction of the 15- to 18-cm wide manure band exposed to the surface that varied with soil moisture conditions and manure application rate. Manure N application rates varied by ∼14% between treatments in the same year and were smaller during 2017 compared with those during 2016 (Table 1). Corn seed was planted 9 and 12 days after spring 2016 and 2017 manure applications, respectively. During corn planting, blended starter fertilizer was also applied at 21.5 kg N ha⁻¹ in all treatments in bands 5-cm to the side of corn seeds at ∼ 5-cm depth. Due to pre-side-dress nitrate test results below 21 ppm for soil samples at 0- to 15-cm depth (Penn State, 2015), inorganic N fertilizer was side-dressed after our sampling periods in late June in unincorporated broadcast and unmanured plots (both years) and injected manure plots (2017 only). We also sampled unmanured plots prior to mid-growing season side-dress application of urea ammonium nitrate inorganic fertilizer.

TABLE 1 Liquid slurry dairy manure composition and N application rates.

Manure application date	Dry mattera (%)	Total N contentb (g N kg−1)	Ammonium (g N kg−1)	Broadcast manure N applied (kg N ha−1)	Injected manure N applied (kg N ha−1)
May 9, 2016	10.6	3.98	1.54	175	183
May 11, 2017	10.8	4.42	1.76	129	112

a

Dry matter fraction was measured using methods detailed in Hoskins et al. (2003).

b

Total N content was measured using methods detailed in Watson et al. (2003).

We sampled soil NO_x emissions using static flux chambers in the corn–soybean rotation of a randomized, nested split plot design (Ponce de León et al., 2021). Each nested split plot represented a fertility source treatment. A 29 cm × 50 cm × 8 cm aluminum chamber with pressure vents was deployed in each plot by placing the flux chamber onto a chamber base inserted ∼8 cm below the soil surface and positioned perpendicularly between two corn rows within each 27 m x 9 m split plot (Figure S1). Flux chambers covered at least one of the starter fertilizer side bands. Flux chambers in injected manure plots were positioned perpendicular to and including one full injection band (∼5-cm width with manure N) to include a representative fraction (10%) of the plot surface area with injected manure N. At least three emission measurements were conducted in each treatment (Table 2; Table S1). During spring 2016, adjacent split plots in one corn–soybean rotation block were sampled for each treatment with limited replicates due to isotopic sample collection constraints (Figure S1). During spring 2017, split plots were sampled in three–four field blocks (Figure S1) and some treatments were sampled from four blocks due to isotopic sample collections. The emphasis on isotopic measurements yielded a limited number of replicate emission measurements over limited sampling periods. Nevertheless, we focus on the statistically significant emission rate differences observed across treatments.

We sampled during 2–3-week periods in May 2016 and May–June 2017 (Table ). Our sampling periods targeted 1–4 weeks following manure applications to capture a key portion of the largest manure-induced NO_x emission pulses as observed in previous studies (Section S2) (e.g., Vallejo et al., 2005; Venterea et al., 2005). Measurements commenced at 6 and 9 days after manure application during spring 2016 and 2017, respectively. Our sampling periods also coincided with and preceded the period from 14 to 45 days following manure application that produced the largest portion of total growing season N₂O emissions at our field site (Ponce de León et al., 2021). Although limited in total duration, our daily sampling frequency for NO_x emissions was advantageous to capture short term, day-to-day emission variations during a key portion of the growing season that may be missed by sparser weekly sampling (Ponce de León et al., 2021).

TABLE 2 Soil NO_x emission sampling size and coverage per manure fertilizer treatment.

	Spring 2016	Spring 2017
Measurement periods	May 18–26	May 17–June 4
Number of sampling days	6	14
Number of blocks	1	3–4
Number of chamber bases per split plot	3	2
Number of sample size per treatment	3	6–8
Number of manure treatments	2	3

Measurement methods

Soil NO_x emissions were determined using a static flux chamber and luminol-based chemiluminescence NO₂/NO_x analyzer (Figure S2; Drummond et al., 1990), with emission uncertainties of 0.1 ng N-NO_x m⁻² s⁻¹ (lower detection limit) ±8% (calibration slope drift). We measured NO_x emissions to quantify all direct soil NO and NO₂ emissions, including a fraction of NO that is quickly oxidized to NO₂ inside the flux chamber. It would be challenging to quantify soil NO and NO₂ emissions separately because direct NO₂ emissions are not easily distinguished from NO₂ produced via NO oxidation. The flux chamber design and measurement technique are reported in detail by Miller et al. (2018). Briefly, NO_x mixing ratios were measured every 3 s inside the flux chamber, immediately after placing the chamber on its base, for 3 min sampling periods. Emission rates were derived per unit area of soil surface with linear regression. The flux chamber was not left on long enough to determine nonlinear chamber effects. Flux chamber bases were installed during the full sampling periods, except for temporary removal for corn planting. We sampled static flux chambers daily between 10:00 and 13:00 local time (Eastern Daylight Time). Based on repeated measurements within minutes at select flux chamber locations, emissions did not vary significantly within the ∼1 h needed to measure all flux chambers. During each NO_x emission measurement, plot-level soil temperature was measured at 0- to-10 cm soil depth near the chamber with a Digi-Sense thermometer probe (±1°C accuracy).

Soil NH₄⁺ and NO₃⁻ mass concentrations were derived weekly from three 2.5-cm-diameter soil cores (0- to 5-cm soil depth for tilled and no-till treatments) collected outside of and adjacent to each flux chamber to avoid holes in soil within the chamber footprint. For soil sampling of injected manure plots, five cores were obtained ∼15 cm apart across the 76 cm spaced injection bands and adjacent soil, consistent with the methods of Meinen et al. (2023). Fresh soil cores were stored at −20°C, homogenized to represent each soil flux chamber footprint, and extracted in 2 M KCl for analysis as detailed in Miller et al. (2018) during spring 2017 and as detailed in Ponce de León et al. (2021) for spring 2016. Soil N mass concentrations were referenced to dry soil using gravimetric water content and their analytical uncertainties are reported in Miller et al. (2018). The soil core sampling approach is expected to exhibit relatively large variances due to spatial heterogeneities, especially across injection bands.

Hourly, site-level volumetric water content and soil temperature from 0- to 5-cm soil depth and rainfall rates were also measured continuously at the United States Department of Agriculture (USDA) National Resource Conservation Service (NRCS) Soil Climate Network (SCAN) site at Rock Springs, PA, ∼0.5 km from our experimental site. WFPS was calculated using the volumetric water content at 0- to 5-cm depth, assuming an average soil bulk density of 1.36 g cm⁻³ corrected for >2-mm-diameter stones at a volume fraction of 11.8%. These site-level WFPS measurements represent a proxy for general soil water content trends at the nearby site. However, we do not use them to represent variations in WFPS across manure treatments.

Data analysis methods

We evaluated soil NO_x emission rates and temperature differences among treatments and their temporal dependence (within subject factor) using repeated measures analysis of variance (ANOVA) with Tukey pairwise comparison tests in MATLAB. Tukey pairwise multiple comparison tests were used to determine which treatments exhibited statistically different emissions. We apply lower-bound corrections to account for sphericity. We also log-transformed emission data because of suspected high emission outliers and a limited sample size for normality tests. We found consistent results using log-transformed compared with untransformed data, suggesting that high outliers do not impact our ANOVA results. Nonparametric ANOVA (Kruskal–Wallis) with Tukey pairwise comparison tests was also performed to evaluate median NO_x emission and soil temperature differences, on each day separately, assuming independent observations during each daily sampling period. The Kruskal–Wallis test is the nonparametric equivalent of the one-way ANOVA. The larger value of the emission measurement uncertainty or the standard error of daily mean emissions from ANOVA was used to represent mean daily emission uncertainty in each treatment.

RESULTS Temporal trends in soil NO_x emissions

Soil NO_x emissions with manure incorporation treatments exhibited an order of magnitude day-to-day variations (Figures 1 and 2). Statistically significant temporal dependence of soil NO_x emissions (p < 0.05) during both spring sampling periods illustrates the episodic nature of soil emissions. Daily mean soil NO_x emissions with chisel-disk incorporation and injected manure in both spring sampling periods exhibited maximum values during drier soil conditions with WFPS < 55%. Soil NO_x emission minima followed heavy rainfall events that increased WFPS (e.g., 14 days after manure in 2016), whereas light rainfall events on drier soils coincided with increased soil NO_x emissions (e.g., 10 days after manure in 2016; 8 days after manure in 2017) (Figures 1 and 2). These emission differences between WFPS regimes are consistent with higher soil NO production via nitrification at ∼15%–65% WFPS (depending on soil texture), and light rainfall promoting gaseous displacement to the soil surface (Davidson et al., 2000; Pilegaard, 2013). Finally, soil NO_x emission temporal trends across all treatments were not significantly correlated with soil or air temperature measurements within the range observed during the study period at the nearby USDA NRCS site (Figure S3).

FIGURE 1. Time series of hourly water-filled pore space (WFPS) and precipitation (top panel); daily mean soil NOx emissions ± standard error (middle panel); and daily soil inorganic N concentrations (bottom two panels) during spring 2016. Daily soil inorganic N concentrations are offset on the horizontal axis to distinguish between treatments and each data point represents soil cores from one flux chamber footprint.

FIGURE 2. Time series of hourly water-filled pore space (WFPS) and precipitation (top panel); daily mean soil NOx emissions ± standard error (middle panel); and daily soil inorganic N concentrations (bottom two panels) during spring 2017. Daily soil inorganic N concentrations are offset on the horizontal axis to distinguish between treatments and each data point represents soil cores from one flux chamber footprint.

Large interannual soil NO_x emission differences were associated with a combination of soil moisture and manure N application rate differences between growing seasons. Mean soil NO_x emissions during spring 2017 were 21 times smaller in injected manure and 13 times smaller in unincorporated broadcast manure compared with those in spring 2016 (Figure 3). Mean soil moisture observations indicated wetter soils during spring 2017 compared with those during spring 2016. WFPS values were >70% during approximately half of the spring 2017 sampling period. During spring 2017, N application rates were 24% smaller for unincorporated broadcast and 39% smaller for injected manure compared with N applications during spring 2016. During 2016, corn grain yields were comparable for unincorporated broadcast and injected manure plots (11.1 and 10.4 ± 0.7 Mt ha⁻¹). During 2017, corn grain yields were smaller than those in 2016, yet comparable for unincorporated broadcast manure, chisel-disk, and injected manure plots (7.9, 7.7, and 6.8 ± 0.8 Mt ha⁻¹).

FIGURE 3. Mean soil NOx emissions across field management practices and sampling periods on a logarithmic scale. Error bars denote standard errors from repeated measures analysis of variance (ANOVA). Chisel-disk incorporation plots were only sampled during spring 2017. Sample sizes for each bar are detailed in Table S1. Asterisk (*) denote statistically significant differences in mean emissions compared with those with unincorporated broadcast manure.

Treatment dependence of soil NO_x emissions

Manure incorporation methods increased soil NO_x emissions relative to those in unincorporated broadcast manure. Chisel-disk manure incorporation exhibited the largest mean soil NO_x emissions as follows: 4 ± 2, 3 ± 2, and 5 ± 4 times those in unincorporated broadcast manure, injected manure, and unmanured plots, respectively (p < 0.05) (Figure 3). Injected manure exhibited larger mean soil NO_x emissions than those in unincorporated broadcast manure by 2.6 ± 1.6 times (p > 0.05) during spring 2016 and 1.5 ± 1 times (p > 0.05) during spring 2017 (Figure 3; Tables S2 and S3). Larger emissions with injected manure compared with unincorporated broadcast manure were consistent across sampling day and season. The small number of replicates and large interquartile ranges and skewness for daily injected manure emissions limited the ability to establish statistically significant differences for injected manure. Larger emissions in chisel-disk incorporation occurred on 4 of the 13 sampling days during spring 2017 (p < 0.01) (Figure 2). Mean soil temperatures were 1.0°C higher in chisel-disk incorporation than in unmanured plots (p < 0.05) (Table S3). These warmer conditions have been associated with enhanced NO_x emissions (Venterea et al., 2005; Yamulki & Jarvis, 2002). Mean soil NO_x emissions with unincorporated broadcast manure were not statistically different from those in unmanured plots (Figure 3; Table S3).

Our observed treatment effects on NO_x emissions are consistent with several studies reporting two–four times larger NO_x emissions for tillage incorporation of synthetic N fertilizers (Liu et al., 2005, 2006; Yamulki & Jarvis, 2002; Yao et al., 2009), though literature meta-analysis reported only ∼30% larger emissions (Liu et al., 2016). In contrast, Vallejo et al. (2005) reported no significant differences for flux chamber-based NO_x emission measurements in shallow injection (5 cm) compared with surface broadcasting of pig slurry during the emissions peak 10–15 days after slurry application. Although Vallejo et al. (2005) provide the most direct comparison with our study, their irrigated grassland soils were generally wetter (70%–80% WFPS) and exhibited continuous N uptake compared with conditions prior to corn growth during our study period. Finally, other studies found opposite emission effects from fertilizer incorporation depending on synthetic fertilizer type (Venterea et al., 2005) or soil texture and crop residue incorporation (Yao et al., 2010).

Representativeness of NO_x emission rates

We report soil NO_x emission observations from 1 to 4 weeks following manure application. This sampling period represents a key period for manure N transformation and maximum soil emissions, as reported in previous, relevant field-based observations (Huddell et al., 2021; Laville et al., 2011; Liu et al., 2005, 2006; Vallejo et al., 2005; Venterea et al., 2005). Although our daily, short-duration NO_x emission observations are not representative of the entire growing season, our observed soil NO_x emission rates (3.5–14 ng-N m⁻² s⁻¹) are within the hourly or daily timescale emission rate ranges reported in the studies referenced above.

Soil inorganic N treatment profiles

Soil NO_x emission differences across treatments were associated with larger soil NH₄⁺ concentration extremes. During 2017, soil NO_x emissions distributions exhibited two regimes. First, relatively small soil NO_x emissions, some of which are indistinguishable from the lower detection, occurred when soil NH₄⁺ was below 2 mg N kg⁻¹ dry soil, including soil NH₄⁺ values indistinguishable from the lower detection limit (Figure S5). Second, a larger soil NO_x emission distribution occurred when soil NH₄⁺ ranged from 3 to 50 mg N kg⁻¹ dry soil (Figure S5). Soil NH₄⁺ concentrations exhibited larger extreme concentrations in injected manure compared with those in unincorporated broadcast manure by 2–14 times (Figures 1 and 2). Relative to soil samples in unincorporated broadcast manure, soil NH₄⁺ samples in injected manure exhibited larger positive skewness and variability of over an order of magnitude. Soil NH₄⁺ in chisel-disk incorporation was larger early in the spring 2017 sampling period before declining to soil NH₄⁺ levels in unmanured plots (Figure 2). Larger soil NO_x emissions with chisel-disk incorporation occurred with sufficient NH₄⁺ available in aerated soils. Finally, unincorporated broadcast manure plots exhibited low soil NH₄⁺ levels similar to those in unmanured plots. This is expected with large NH₃ volatilization rates during the first 48–72 h after unincorporated broadcast manure application (Dell et al., 2012; Duncan et al., 2016).

The presence of elevated soil NO₃⁻ concentrations with manure incorporation during the latter periods of sampling suggests nitrification of mineralized N was a dominant process over denitrification. During 2016, injected manure exhibited larger median and two times larger maximum soil NO₃⁻ concentrations than those with unincorporated broadcast manure at 18 days following manure application (Figure 1). During 2017, injected manure and chisel-disk incorporation exhibited larger median soil NO₃⁻ than those in unmanured plots at 12 and 21 days following manure application (Figure 2). Miller et al. (2018) reported approximately equivalent median soil NO₃⁻ and NH₄⁺ concentrations in the same injected manure plots sampled during spring 2017 compared with those reported in the present study. Residual soil NO₃⁻ was also detected in unmanured plots from starter fertilizer at corn planting, inorganic N fertilizer applications during the previous growing season, and/or soybean crop residue N mineralization. Because dairy slurry manure contains negligible NO₃⁻, the presence of soil NO₃⁻ above levels in unmanured plots on a subset of latter sampling days suggests soil NO₃⁻ was produced with manure incorporation. The lack of significant correlation between soil NO_x emissions and soil NO₃⁻ concentrations indicates that soil NO₃⁻ levels were also dependent on soil N processes independent of those producing soil NO_x, including denitrification of soil NO₃⁻.

DISCUSSION

Larger soil NO_x emissions with manure incorporation practices were associated with the nitrification of soil ammonium. Our soil N substrate observations suggest that NO₃⁻ production via nitrification exceeded NO₃⁻ consumption via denitrification in injected manure. These results are consistent with those based on the stable isotopic composition (δ¹⁵N) of NO_x measurements (Miller et al., 2018). During our sampling periods, sufficient time elapsed after liquid slurry manure incorporation for the mineralization of organic N and manure NH₄⁺ retained in the soil to produce available N (Ponce de León et al., 2021). Injected manure bands contain locally large C and N substrate concentrations, favoring enhanced microbial activity (Dell et al., 2011; Duncan et al., 2017, 2019). Chisel-disk incorporated treatments also exhibit sufficient soil NH₄⁺, higher soil temperatures, and drier, tilled soils favoring nitrification over denitrification. Tilled, aerated soils also exhibit enhanced surface heating that promotes higher microbial activity in surface soil layers (Venterea et al., 2005).

The magnitude of NO_x emission increases (>2 times) with manure incorporation suggests that adopting these practices has important implications for the contributions of soils to the regional NO_x emissions budget. Because total annual soil NO_x emission inventories (EPA, 2014) are known to underestimate growing season emissions by factors of 4–6 times (Hudman et al., 2012; Vinken et al., 2014), agricultural soils represent at least the third largest growing season NO_x emission source (10%–15%) in our study region (Centre County, PA). With the potential for future adoption of manure injection in this region, cropland soils could become the second largest regional NO_x source sector below on-road vehicles and contribute >14% of the NO_x emissions budget during the post manure application period. Thus, future studies are important to derive representative growing season NO_x emission factors per N applied across all manure application methods to inform regional NO_x emission modeling.

Relative treatment differences in soil NO_x, NH₃, and N₂O emissions indicate there are important air quality and climate impact tradeoffs for manure injection compared with unincorporated broadcast manure application. We compiled the relative differences in mean soil NO_x, NH₃, and N₂O emissions between each management practice during the relatively short period (1–28 days) following N application (Table 3; Section S1). This metric represents a short time window during the post manure application emission pulse and does not represent daily emissions rates later in the growing season that may be driven by rainfall events, side-dress N application, and/or freezing. Because the temporal sampling year, frequency and duration of emission estimates are somewhat different for each reactive N species, we do not compare absolute emission estimates among different reactive N species. From an air quality perspective, the adoption of injected manure practices could result in relative mean daily soil NO_x emissions increases of 60%–160%, while NH₃ emissions are estimated to decrease by 88% during the post-manure application period (Table 3). Similar air quality tradeoffs exist for chisel-disk incorporation with mean daily NO_x emission increases of 290% in our study, and 52%–82% decreases in NH₃ emission reported by Dell et al. (2012). Finally, from a greenhouse gas perspective, mean daily N₂O emissions with manure injection could also increase by 120%–380% (Table 3).

TABLE 3 Mean soil reactive N emission estimates in injected manure and chisel-disk incorporation expressed as a ratio relative to emissions in unincorporated broadcast manure.

Species	N emissions in injected manure relative to unincorporated broadcast	N emissions in chisel-disk incorporation relative to unincorporated broadcast
NOx	2.0 (1.6, 2.6)a	3.9 (2.7, 6.3)b
N2Oc	3.5 (2.2, 4.8)d	–
NH3	0.12 (0.11, 0.12)e	–

Note: Ratios are computed using emission rates reported for the 1–4 weeks following dairy manure applications. Uncertainty ranges in parentheses represent temporal variability and are detailed in the footnotes below.

a

Uncertainty range represents variability between spring 2016 and 2017 sampling periods (this study).

b

Spring 2017 sampling period only (this study) with uncertainty range based on standard error.

c

Spring 2016 measurements 1–4 weeks following manure application (Ponce de León et al., 2021).

d

Uncertainty range represents variability across 1-to-4-week time windows following manure application (Ponce de León et al., 2021).

e

Uncertainty range is standard errors of the least squares mean propagated through the relative NH₃ emissions.

Alternative management approaches could be investigated in future studies to evaluate their potential to mitigate both air quality and climate impacts of reactive N emissions. For example, because manure injection leads to larger crop-available soil N and crop N uptake as well as enhanced microbial activity with larger soil C (Duncan et al., 2016, 2019), manure injection at N application rates less than those developed for unincorporated applications could preserve crop yield levels. Furthermore, strategies that better synchronize manure N availability with crop N uptake could leave less soil N for potential reactive N emissions (Dell et al., 2012; Han et al., 2017). In our study region, these strategies could include delaying manure injection until corn is growing or applying manure during early spring or the previous fall to growing cover crops, winter annual forage crops, or perennials (Binder et al., 2020; Milliron et al., 2019). However, manure injection also has several limitations for mitigating soil reactive N emissions. Injected manure would not be advantageous in regions where soil NO_x and/or N₂O emissions and soil NO₃⁻ leaching contribute a major proportion of societal damages (Luo et al., 2022). Energy requirements associated with the manure injection process relative to those for unincorporated broadcasting are also important to assess in future studies. It is uncertain whether smaller N application rates with manure injection are sufficient to reduce N₂O and NO_x emissions. Future studies are warranted to quantify this dependence. Minimum N application rates for manure incorporation should be determined to achieve emission reduction benefits and meet crop yield goals (Roy et al., 2021). Finally, side-dressing manure after corn becomes established has not been widely adopted, and requires both new technology and educational outreach (Ponce de León et al., 2021).

Our study illuminates important gaps in knowledge on soil NO_x emissions across manure application methods that warrant future investigations. First, field-based soil NO_x emission measurements could be expanded to other climatic regions with different manure application rates and soil types to quantify the regional variability in treatment practice effects. Second, automated flux chambers could be used to sample soil NO_x and N₂O emissions simultaneously with larger temporal coverage. Higher frequency, automated chamber sampling could better capture episodic emissions across the entire year (e.g., Anthony et al., 2023) that are missed with periodic sampling and reduce uncertainty ranges for estimating the dependence of emissions on management practices. Alternative open-path measurement approaches that do not require flux chambers, including integrated horizontal flux (e.g., Goedhart et al., 2020) and open-path eddy covariance flux (e.g., Pan et al., 2022) methods, are also promising. Third, further investigations could focus on treatment effects of additional manure incorporation methods (e.g., strip till) and nitrification inhibitors (Halvorson et al., 2016; Venterea et al., 2005). Fourth, nitrification and denitrification rate measurements and controlled laboratory experiments would be useful to quantify contributions of different soil NO_x emission mechanisms among field management scenarios. Finally, our results illuminate the need for soil emission model validations to bridge field and regional emission scales and explicitly represent temporally-varying manure incorporation management scenarios.

CONCLUSIONS

This study reports the first characterization of chisel-disk tillage and shallow-disk injection treatment effects on soil NO_x emissions from dairy manure-fertilized cropland during the key emissions pulse period (i.e., 1–4 weeks following manure application). We conclude that manure incorporation methods in a corn production system exhibit factors of 2–4 larger mean daily soil NO_x emissions, during the critical post-application period, than those with unincorporated broadcast manure. These emission differences between treatments are consistently detectable across large daily soil NO_x emission variations during two sampling years. Our observations suggest that larger soil NO_x emissions for manure incorporation via chisel-disk or shallow-disk injection are driven by nitrification of soil NH₄⁺ during the sampling periods.

This analysis highlights important implications for air quality and climate impacts of manure application practices in manure N-rich, mid-latitude agricultural regions. We find tradeoffs in the use of manure incorporation methods between the societal health and crop-N benefits of reduced NH₃ emissions, and the environmental damages associated with larger soil NO_x and N₂O emissions. Future analyses could investigate the potential for alternative management approaches, including N application rate and timing adjustments, to reduce soil NO_x, N₂O, and NH₃ emissions.

AUTHOR CONTRIBUTIONS

David J. Miller: Conceptualization; data curation; formal analysis; funding acquisition; investigation; methodology; project administration; software; supervision; validation; visualization; writing—original draft; writing—review and editing. Jiajue Chai: Conceptualization; data curation; formal analysis; investigation; methodology; writing—review and editing. Felix Guo: Data curation; formal analysis; investigation; methodology; software; visualization; writing—review and editing. María A. Ponce de León: Data curation; investigation; methodology; writing—review and editing. Rebecca Ryals: Conceptualization; investigation; methodology; writing—review and editing. Curtis J. Dell: Conceptualization; methodology; project administration; resources; supervision; validation; writing—review and editing. Heather Karsten: Conceptualization; project administration; resources; supervision; writing—review and editing. Meredith G. Hastings: Conceptualization; funding acquisition; methodology; project administration; resources; supervision; validation; writing—review and editing.

ACKNOWLEDGMENTS

The authors are grateful for the helpful guidance and suggestions of David Murray, Ruby Ho, and the entire Hastings lab group at Brown University. The authors would like to thank two anonymous reviewers for helpful feedback. The authors are grateful for the support of Emad Jahanzad and the entire Pennsylvania State University sustainable dairy cropping study team for access to the field site and invaluable guidance; Sarah Fishel for providing laboratory space at USDA-ARS; and Brooke Osborne for guidance on soil extract methods. This work was funded by the United States Department of Agriculture National Institute of Food and Agriculture Postdoctoral Fellowship Award 2016-67012-24707 to David J. Miller and the National Science Foundation CAREER Award AGS-1351932 to Meredith G. Hastings. Heather Karsten acknowledges funding support from National Institute of Food and Agriculture Awards LNE09-291, LNE13-329 and LNE16-354, and Hatch Appropriations Awards PEN04600 and PEN04425.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT

Data presented in this study are available on the Brown University Repository at https://doi.org/10.26300/ttq1-ja78 and data from the USDA NRCS Rock Springs, PA site are available at https://wcc.sc.egov.usda.gov/nwcc/site?sitenum = 2036.