Nongrowing season soil nitrous oxide emissions as influenced by cover crops and fall tillage termination

Abstract Cropland soil is a major driver of global nitrous oxide (N2O) emissions. In cold climates, nongrowing season (NGS) emissions can be significant due to high fluxes during freeze–thaw (FT) cycles. Cover crops can alter key soil conditions that govern N2O-producing microbial processes, with multiple potential pathways to either increase or decrease N2O production during FT cycles. Cultivating cover crops in the fall to terminate may further disrupt these processes and the overall impact of cover crops on N2O emissions. Yet, few studies have touched on how termination practices of cover crops impact FT emissions over the NGS. Using the flux gradient method to continuously measure N2O emissions from a conventional corn–soybean rotation, we investigated the effects of summer-established cover crops (perennial ryegrass and crimson clover) (with cover crops, +CC; without cover crops, −CC) when terminated by fall cultivation (with fall cultivation, +FC; without fall cultivation, −FC) over a six-month NGS that was characterized by several freezing and thawing periods. Crimson clover cover crop was completely winterkilled, while the ryegrass survived on the +CC−FC field. Total NGS (Nov–Apr) emissions varied nearly 2.5-fold among treatments from 395.1 (−CC−FC) to 978.1 (+CC+FC) g N2O-N ha−1. Compared with the control treatment (−CC−FC), fall cultivation alone (−CC+FC) and cover crops alone (+CC−FC) increased total NGS N2O emissions, and fall cultivation with cover crops (+CC+FC) increased N2O fluxes even more. Careful CC species selection and management are important to avoid elevated NGS emissions.


Introduction
Cropland soils are a major source of nitrous oxide (N 2 O), a potent greenhouse gas that can persist for over 100 years in the atmosphere and contribute to climate change and destruction of stratospheric ozone (Ravishankara et al. 2009;IPCC 2014).In regions where agricultural soils are seasonally frozen, freeze-thaw (FT) cycles have been shown to contribute 30%-90% of annual N 2 O emission and exclusion of FT effects could lead to a 17%-28% underestimation on global annual N 2 O emission (Wagner-Riddle et al. 2017).Microbial denitrification has been identified through 15 N tracer studies as the dominant contributor to high thaw fluxes (Ludwig et al. 2004;Wagner-Riddle et al. 2008).During thawing events, soil conditions become favourable to denitrification because of increased soil temperature, soil water content (SWC) from snowmelt, and substrate (C and N) availability (Congreves et al. 2018).Cover crops have shown potentials to alter these soil conditions before and during the nongrowing season (NGS) and thus provide opportunities for mitigating thaw N 2 O emission (Wagner-Riddle and Thurtell 1998).Although many have studied the effects of cover crops on soil organic carbon and soil physical properties (Poeplau and Don 2015), much remains unknown in understanding the impact on N 2 O emissions, particularly during the NGS in cold climate regions.
There are several potential pathways in which cover crops can either mitigate or amplify FT emissions, first in the substrate for microbial processes during the FT cycles and second in effects on soil temperature and moisture.As one of its main management functions, cover crops can take up excess nitrogen (N) in soil after the cropping season, and thus can limit nitrate (NO 3 − ) for denitrification (Ussiri and Lal 2013).However, cover crops can only maintain this effect while the plants are alive.Termination of cover crops can occur through cultivation or killing through frost, and the organic N in turn feeds into the N cycle and potential N 2 O emissions, particularly if the cover crop residues have a low C:N ratio (as a conventional rule: less than 25:1) (Robertson and Groffman 2015).However, the fate of substrate inputs from cover crops is likely to differ between mechanical mixing of cover crops into soil during fall cultivation versus from winter-killed cover crops.
Soil temperature affects soil N transformation directly by accelerating organic matter decomposition and microbial metabolic turnover (Ussiri and Lal 2013) and indirectly as both oxygen and N 2 O solubilities in water change with tem-perature.At low temperatures, an increased oxygen solubility may reduce denitrification activity and thus N 2 O production (Coyne 2008), while at temperatures below freezing, ice crystal disruption of soil and cell lysis (i.e., of microbial and plant tissues) may lead to high N 2 O fluxes with subsequent thawing (Wagner-Riddle et al. 2007;Congreves et al. 2018).Cover crops can alter soil temperature by trapping early snowfall that insulates the soil surface in winter and can thus reduce soil freezing depth and severity and/or dampen temperature variation (FT cycles), reducing N 2 O emissions.Cultivation of cover crops in the fall may eliminate the insulation effect of cover crops and decrease overwinter surface soil temperatures (more intense freezing or larger freezing degree days (FDD)) (Wagner-Riddle and Thurtell 1998; Yang et al. 2018).Additionally, during a thawing event, water from melting snow and ice can fill soil pores, transport substrates to microbes, and create an oxygen-deficient environment favouring denitrification and a rapid increase in N 2 O fluxes (Goodroad and Keeney 1984;Prieme and Christensen 2001).Cover crops may impact water content in the growing season/fall leading up to FT by reducing SWC through increased transpiration, or conversely, cover crops may increase soil water by reducing soil surface evaporation, increasing rainfall infiltration and soil water storage capacity (Basche et al. 2016).Because of these multifaceted effects, the impact of cover crops and their termination on N 2 O emissions during FT cycles is not well understood with multiple potential pathways to either increase or decrease N 2 O production.
The majority of N 2 O emission studies have been done with discontinuous measurement methods, mainly through manual chamber techniques performed with nonflow-through nonsteady-state chambers (Bouwman 2002).While manual chamber techniques have advantages in terms of simplicity, affordability, and versatility, the low measuring frequencies (often once per week) may lead to inaccurate estimates of N 2 O fluxes (Rochette and Eriksen-Hamel 2008).In cold environments, winter site accessibility and snow or ice cover may also limit viability of chamber measurements.Indeed, in a meta-analysis of N 2 O emissions using 26 cover crop studies, only 4 had investigated the NGS (Basche et al. 2014).Micrometeorological techniques, such as the flux-gradient method (Wagner-Riddle et al. 2007), allow for field-scale measurements while capturing ephemeral fluxes that characterize FT events.
This study investigated the effects of cover crops and termination practices on soil conditions and N 2 O emissions during the NGS using a micrometeorological method.The objectives were (i) to assess individual and combined effects of cover cropping and fall cultivation on soil conditions and N 2 O emissions during FT cycles during the NGS, and (ii) to evaluate individual and combined effects of cover cropping and fall cultivation on total NGS N 2 O emissions.We hypothesize that fall cultivation (with fall cultivation, +FC; without fall cultivation, −FC) results in more intense freezing and increased NGS N 2 O emissions, with highest values expected when cover crops (with cover crops, +CC; without cover crops, −CC) are cultivated into the soil (+CC+FC), due to more inputs of substrate for denitrification, followed by fall cultivation without cover crops (−CC+FC).Lower emissions were expected when there is no fall cultivation (−CC−FC or +CC−FC), with an expected effect (in direction and magnitude) from cover crops depending on extent that soil temperature, water, and substrate availability are affected in the NGS.

Site description, experimental design, and management
The experiment took place at a long-term N 2 O flux study site established in 2000 at the University of Guelph Elora Research Station, Ontario, Canada (43 • 39 N, 80 • 25 W; 376 m), as described in previous publications (Jayasundara et al. 2007;Wagner-Riddle et al. 2007;Congreves et al. 2018;Ferrari Machado et al. 2020).Measurements for this study were conducted from 31 October 2018 to 30 April 2019.Soil at the site is an imperfectly drained Guelph silt loam (29% sand, 52% silt, 19% clay for the 0-15 cm layer; Jayasundara et al. 2007) and experiences a humid continental climate with cold winters (Köppen-Geiger Dfb).In each of the previous 3 years, there were no significant differences in NGS N 2 O emissions between study fields at the site despite significant differences during the growing season detected between nitrogen treatments applied to corn (Ferrari Machado et al. 2020).Detailed soil description and site history can be found in previous publications (Sulaiman et al. 2017;Ferrari Machado et al. 2020).
N 2 O was measured over four fields (4 ha each), within a 30 ha area of aerodynamically homogeneous land planted with the same cash crop during the growing season, using a fluxgradient method (Wagner-Riddle et al. 1996, 2007).Corn was grown at the site from 2012 to 2018 as described in Baral et al. (2022).On 17 May 2018, the fields were cultivated, and 154 kg N ha −1 of urea was broadcasted and incorporated by disking.Corn was planted in all four fields at 80 000 seeds ha −1 with 76 cm row spacing on the following day.In two of the fields, a two-species cover crop mixture was underseeded to corn at the sixth leaf stage on 20 and 21 June 2018 using a manual seed spreader.The cover crop mixture consisted of crimson clover (Trifolium incarnatum) seeded at 6.7 kg ha −1 and perennial ryegrass (Lolium perenne) seeded at 33.6 kg ha −1 and its choice was based on a consultation with the Ontario Soil and Crop Improvement Association as part of a long-term project of studying diversified crop rotations including cover crops at our flux monitoring site.For soil and plant sampling, each field was divided into nine subplots on a 3 × 3 grid with flags marking the ends of gridlines.On 5 October 2018 and then again on 21 May 2019 (annual rye only as crimson clover was winterkilled and unmeasurable), cover crop samples were taken from the two +CC fields.In each sampling subplot, cover crop samples were taken by clipping all aboveground portion of plants within two randomly placed 0.5 m 2 frames.Samples taken from the same subplot were composited and then the composited sample was separated by cover crop species.This was repeated to obtain nine clover and nine ryegrass samples contained in preweighed paper bags from each of the two +CC fields.After taking fresh biomass weights, the samples were dried at 65 • C for at least 48 h un-til sample weights became stable.Dry biomass weights were then taken.
Corn plants were sampled on 10 October from all four fields to calculate grain and residue yields.Corn plant sampling was done in the same sampling subplots as described above.In each sampling subplot, three separate corn rows were randomly selected and all corn plants along 1 m in each row were clipped at about 5-10 cm above ground.Fresh corn samples from the same subplot were composited and a fresh weight of each composite sample was taken.Each sample was then separated into grain and stover, and the grain portion (on cobs without husks) was weighed separately before all plant samples were oven dried at 65 • C for 72 h.Dry weights of grain and stover were taken to calculate sample moisture and field grain and residue yield.Plant sampling indicated that on average the four fields yielded 12.8 Mg ha −1 of grain (adjusted to 15.5% moisture) and 9.8 Mg ha −1 of residue dry biomass when combine-harvested on 30 October 2018.Fall cultivation with disk plow to a depth of 10 cm took place immediately following harvesting in two fields with and without cover crops, leaving very little visible residue on the surface.Corn residues were left on soil surface of the two −FC fields.Before the fields were covered by snow within the following two weeks, all four fields were weed free.The study consisted of measurements conducted in four 4 ha fields managed according to (i) no cover crops and no fall cultivation (−CC−FC), (ii) no cover crops with fall cultivation (−CC+FC), (iii) cover crops that are not terminated in the fall, i.e., with no fall cultivation (+CC−FC), and (iv) cover crops that are terminated with fall cultivation (+CC+FC).

Nitrous oxide flux
N 2 O was measured with the flux-gradient method, and the fluxes were determined using (Arya 2001): where u * is the friction velocity, k is the von Karman constant (=0.41),C is the difference in measured N 2 O concentration between two sampling heights, z 1 and z 2 , d is the displacement height, and 1 and 2 are the integrated Monin-Obukhov similarity functions for heat for the two sampling heights.Two 3D sonic anemometers (CSAT3, Campbell Scientific) were installed at the site, one that captured −CC fields and one for +CC fields, which provided near-continuous measurements of u * .A five-cup anemometer tower (cups at 67, 102, 155, 236, and 354 cm) was also installed to provide backup estimates of u * when the sonic anemometers did not yield data (e.g., under heavy rain or snow conditions, or equipment maintenance).A detailed description of the system can be found in Wagner-Riddle et al. (2007).
To sample gas and measure N 2 O concentration difference ( C), one tower with two intakes drawing air at two heights (spaced 50 cm apart) was installed in each of the four fields.The heights of air intakes were adjusted over the study period, with the lower intake heights at approximately 1.9 times cover crop canopy height or winter snow cover depths, and the upper intake 50 cm above the lower intake.When soil surface was bare, intake heights were at 0.4 and 0.9 m for all four fields.Air intakes were programmed to switch between the two heights every 15 s through the activation of solenoid valves.Air samples from the four fields were drawn with a central vacuum pump into a manifold controlled by a datalogger (CR1000, Campbell Scientific).A second vacuum pump subsampled air from one of the sampled fields over 30 min from the manifold and directed it through a tunable diode laser trace gas analyzer (TGA100A, Campbell Scientific), which continuously measured N 2 O concentrations at 10 Hz in the passing gas stream.Air sampled from the other three fields was discarded from the manifold through bypass valves.With this system, N 2 O concentration differences for each field were measured for 30 min every 2 h, 12 times per day.Raw N 2 O concentration data were filtered according to a list of criteria, including poor analyzer performance or site maintenance.From each of the 15 s intervals over which N 2 O concentrations were measured, values measured during the transition between the two intake heights were eliminated, leaving approximately 5400 concentration values per height per 30 min period.Fluxes were filtered based on friction velocity (<0.1 m s −1 ), poor analyzer performance, fetch conditions, and stability parameter outside a range of −5 to 2 (Wagner-Riddle et al. 2007).

Concentrations of available mineral N (NO 3
− and ammonium (NH 4 + )) in the surface soil at 0-15 cm depth were determined.Soils were sampled once before harvesting of corn, on 19 October 2018, and once during the FT when soils were defrosted and uncovered, on 4 April 2019.Each of the four fields were divided into nine sampling subplots by a 3 × 3 grid as per the plant sampling scheme.A total of 36 soil samples were taken at each sampling event.Within each subplot, soil cores were taken at five locations between two rows of corn stubble.One subsample was then taken from the composite sample of the five cores.Collected soil samples were placed in a chest cooler for transportation and stored in a −20 • C freezer until extraction.Frozen samples were taken to a 4 • C refrigerator a day before extraction to thaw (Esala 1995).Soil NO 3 − and NH 4 + were then extracted with 2.0 mol/L KCl solutions as described in Maynard et al. (2007).Extracts were analyzed for NO 3 − and NH 4 + concentrations using a continuous flow analyzer (Seal AutoAnalyzer 3-AA3, Seal Analytical).

Soil water content and soil temperature
Soil volumetric water content (SWC) and soil temperature were continuously measured at 5 and 25 cm using time domain reflectometry probes (Model CS616, Campbell Scientific) and lab-built copper-constantan thermocouples, respectively, and averaged over 10 min intervals in all fields.The sensors were installed in pairs, each consisting of one SWC and one temperature sensor.Both SWC and temperature sensors in a pair were connected to a data logger (CR23X, Campbell Scientific).In each field, two pairs of sensors were installed at each measurement depth for a total of eight measurements (four temperature and four SWC) every 10 min.
Two data loggers were installed on the shared boundary between the two +CC fields and between the two −CC fields, each connected with all eight pairs of sensors from the two adjacent fields.Collected data were manually downloaded from the data loggers monthly throughout the period.
SWC was calculated using equations derived from Kulasekera et al. (2011).SWC and soil temperature measurements were filtered for any sensor or data logger malfunctions or damages by identifying and deleting erroneous values (0-1 m 3 m −3 for SWC, or −50 to 50 • C for temperature).For sensors measuring the same field, filtered data points measured at the same depths were inspected again and then averaged.Both filtered SWC and temperature data were averaged into daily means, and any gaps in the daily averaged data were filled by linear interpolation with the function "na.approx" from package "zoo" (v1.8.7) in R (v3.6.3).

Other supporting data
Other data required for N 2 O flux calculations and interpretation, such as air temperature, precipitation, and on-theground snow depth, were obtained from the Environmental and Climate Change Canada weather station (WMO ID: 71352) located within 50 m from the research area.Snow depth measurements within the research area were also taken manually through the winter on a weekly basis.Ten measurements were taken across the four fields each time by walking a rectangular loop with the intake towers as the corners and taking measurements using a measuring stick along the path, including one taken near each intake tower and six taken at random locations within a few meters along the rest of the path.An average depth was recorded and used for intake height adjustments and to derive timing of snow melts.

Data and statistical analyses
Half-hourly N 2 O fluxes were calculated using filtered data.Due to the sequential sampling setup, a maximum of 12 halfhourly flux values were calculated for each plot per day.The average of the half-hourly flux values was used as daily N 2 O flux and then linearly interpolated where data gaps existed.Cumulative emissions over the NGS were calculated using the gap-filled daily N 2 O fluxes for each field.
Harsh weather, equipment maintenance or testing, and equipment removal due to field activities such as cultivating, planting, harvesting, and spraying were the main causes of continuous data gaps that varied from 1 to 9 days during the NGS.The latter was due to an intake valve leak found and fixed in the +CC−FC field 9 days following harvesting and reinstallation of equipment.The percentage of daily flux data capture during this NGS was 90.6% for −CC−FC, 89.6% for −CC+FC, 84.6% for +CC−FC, and 89.6% for +CC+FC.
An uncertainty test was performed to evaluate potential errors resulting from gap-filling when deriving cumulative N 2 O emissions following procedure developed for carbon studies (Falge et al. 2001;Taki et al. 2018).Based on the existing gaps in the daily N 2 O flux means for each field, five artificial gap scenarios were developed, each with a unique frequency of gaps of lengths varying from 1 to 5 days each.The total length of created gaps was set to not surpass 25% of the existing gap length in the measured daily flux time series.Artificial gaps, based on the scenarios developed, were created by removing measured daily N 2 O flux data at random positions in the time series.The artificial gaps inserted were kept at least one location away from other gaps to ensure the individual gaps would not combine with each other.These time series with artificial gaps were then linearly interpolated to obtain a new filled daily flux series.Using a Monte Carlo method, this process was repeated 50 times for each of the five scenarios, resulting in a total of 251 sets of filled daily flux data, including the measured and filled dataset.The cumulative N 2 O emission for each of the 251 sets of data was calculated and standard deviations between each of 250 generated sums and the sums from the original filled data were calculated as errors resulting from linear interpolation.
Distribution of N 2 O fluxes were nonnormal (Yates et al. 2006).Therefore, paired-sample Wilcoxon's signed-ranks test was used to examine statistical differences in medians of the half-hourly N 2 O fluxes among the treatments, with the Bonferroni correction applied to reduce Type 1 errors of false positivity due to multiple comparisons (Ferrari Machado et al. 2020).Four treatment pairs were compared to determine effects of cover crops and fall cultivation.Among the treatments, −CC−FC was considered the control treatment and compared with the other three treatments.We evaluated effect of fall cultivation (vs.−CC+FC), effect of cover crops (vs.+CC−FC), and combined effect of cover crops that are fall cultivated (vs.+CC+FC).Comparisons were done for the six-month NGS and for specific freezing and thawing events.Freezing events were defined as periods when the mean daily soil temperature measured at 5 cm depth for the four fields was below 0 • C for at least 48 h.Thawing events were defined as periods with air temperature above 0 • C for longer than 48 h since we have observed increased N 2 O fluxes within this time frame of increased air temperature, when the top few millimeters of the soil are likely thawing (Wagner-Riddle et al. 2010).Note that shorter (<48 h) thawing events during prefreezing or freezing periods were not included in the defined FT periods.
Comparisons of medians of daily measured soil temperature, volumetric water content, and half-hourly N 2 O fluxes were carried out using paired-sample Wilcoxon's signedranks test in R with the "wilcox.test"(R, v3.6.3)function.An independent sample t test was carried out to compare NO 3 − levels measured in the four fields in pairs.Gap-filled daily flux values were not considered for the tests to avoid biasing treatment comparisons but are presented to discuss NGS emission totals.The uncertainty in gap-filling detailed above was then used to determine differences in emission totals, verifying if totals overlapped between treatments taking the uncertainty into account.

Agronomy and soil nitrate levels
Total cover crop biomass was low (<20 g m −2 or 0.2 t ha −1 ) and was dominated by crimson clover on both CC fields (Table 1).From field observations in December 2018 and   March 2019, all crimson clover was winterkilled, which suggests that part of the crimson clover biomass on the field may have entered the soil N pool during the NGS.Perennial ryegrass in +CC−FC survived well and continued to grow until termination in May 2019 when the average ryegrass biomass yield (dry, aboveground biomass) on the field was 17.4 g m −2 .
Surface soil (0-15 cm depth) available NO 3 − was measured on 19 October 2018 before corn harvest, and on 4 April 2019 when the fields became dry enough (Table 2).No significant differences in NO 3 − levels were found between fields in fall sampling.In the spring, the +FC fields measured significantly lower spring NO 3 − availability in comparison to −FC with cover crops (+CC+FC vs. +CC−FC, p value < 0.01) or without (−CC+FC vs. −CC−FC, p value < 0.01).+CC−FC measured significantly higher soil NO 3 − level in the −FC comparison (+CC−FC vs. −CC−FC, p value < 0.01).

Environmental conditions and daily N 2 O emissions
The NGS was characterized by cold weather with a mean temperature of −2.6 • C (Fig. 1a) and early snow cover from 31 October (Fig. 1b), the first day following corn harvesting and cultivation.From 31 October 2018 to 30 April 2019, a prefreezing period was followed by three main freeze periods (F-1, F-2, F-3), which were interrupted by two short thaw periods (T-1, T-2) and followed by the final thaw period (FT) (Fig. 1a).There were very short warming periods during the first freeze period (F-1) that did not meet our criteria (air temperature above zero for longer than 48 h) (Fig. 1a).The prefreezing period between 31 Oct and 26 Nov 2018 was mild, with an average air temperature of −0.3 • C and soil temperatures slightly above 0 • C in all fields (Fig. 1a).During this period, N 2 O daily fluxes measured on all fields were small (Fig. 2).The first freeze (F-1) started at day of year (DOY) 331 of 2018 and lasted until the first thaw (T-1) occurred on DOY 33 of 2019 (Fig. 1a).During this 68-day freezing period, mean air temperature was −6.0 • C and mean soil temperature and liquid SWC progressively dropped in all fields (Fig. 1).A few very short thaws with small flux events were recorded on all fields during F-1 with the two +FC fields having higher flux peaks (Fig. 2).The first freezing period was followed by a 3day thaw (T-1) from DOY 34 to 36 of 2019, when average air temperature rose to 2.1 • C (Fig. 1a).
The second freezing period (F-2), from DOY 37 to 71 of 2019, was the coldest period of the NGS, with an average air temperature of −7.2 • C and average soil temperature below −2 • C in all fields (Fig. 1a).In the following 3-day thawing event (T-2) (DOY 72 to 74 of 2019), air temperature rose to an average of 2.5 • C and surface soil temperatures were close to 0 • C.
The third freezing period (F-3) from DOY 75 to 85 of 2019 was mild with air temperatures peaking above zero briefly on some days.This was followed by the final thawing (FT) period (DOY 86 to 120 of 2019), in which air temperature continued to increase with highest daily temperature of 8 • C and mean temperature of 4.1 • C and increasing liquid SWC (Fig. 1).Increased N 2 O emissions above baseline values were observed throughout the FT (Fig. 2).

Cultivation increased N 2 O fluxes at thaw
Cultivation increased N 2 O fluxes during F-1, T-1, and F-2 (p < 0.05; Fig. 3), with −CC+FC showing higher emissions than −CC−FC.During F-1 and F-2, a few short thaws were observed, which contributed to the elevated N 2 O fluxes observed for −CC+FC compared with −CC−FC (Fig. 2).The impact of cultivation was clearly seen with the much higher N 2 O flux for −CC+FC compared with −CC−FC during T-1 (Fig. 3).b) indicate prefreeze, freezing (indicated by "F"), and thawing (indicated by "T") or "Final Thaw" periods.Note that SWC values when soil temperature is below 0 • C are not accurate but are shown here to indicate the freezing and then thawing periods (i.e., when SWC apparently decreases and then increases).
No significant differences were found for the remainder of the NGS, which can also be observed in the cumulative values of −CC+FC and −CC−FC that followed a very similar trend after F-2 (Fig. 4).Cultivation did lead to significantly lower soil temperature during prefreezing, F-1, and F-2 (Fig. 3).More intense freezing can result in increased N 2 O emissions at thaw (Wagner-Riddle et al. 2007;Congreves et al. 2018), which supports the observed higher N 2 O fluxes during T-1 and the short thaws during F-1 and F-2 (Fig. 3).In contrast, the higher soil temperature for −CC+FC compared with −CC−FC during F-3 and the FT (Fig. 3) was not associated with significant increases in flux.Since the SWC sensors only measure liquidstate water, the low SWC measured on −CC+FC during prefreeze and F-1 also reflects that more soil water was frozen early (Fig. 3) and relates to the intensity of soil freezing.During F-3 and FT, when snow melted and soil surface was exposed to direct sunlight, a lower surface albedo caused by cultivation could have led to earlier/faster warming and drying observed likely explaining the decrease/no effect on −CC+FC fluxes compared with the control (Fig. 3).

Cover crops increased N 2 O fluxes at thaw
Cover cropping reduced soil moisture content in the prefreeze period, which could have contributed to the lower N 2 O fluxes observed on +CC−FC than −CC−FC (Fig. 3).Cover crop-driven N 2 O emissions started accumulating after T-1, as +CC−FC had observed higher N 2 O fluxes than −CC-FC during F-2, F-3, and FT (p < 0.05; Fig. 4).Short but intense flux events were observed in +CC−FC in the F-2 and F-3 periods, potentially from the short warming periods that did not meet our cut-off to define the start of a thaw period but still may have led to partial thawing of the soil and FT emissions, which ultimately contributed to higher mean fluxes during these periods than −CC−FC (Fig. 4).A few of the highest single-day flux events were recorded on +CC−FC during the FT (Fig. 2; 32.2, 52, and 35.2 ng N m −2 s −1 on DOY 97, 101, and 103, respectively), resulting in significantly higher fluxes than for the control (p < 0.01; Fig. 3).
During the FT, once the soil surface was exposed from snow cover, we observed that crimson clover cover crops had been winterkilled, which suggests that part of the crimson clover   biomass on the field may have entered the soil N pool.Considering the timing of the extremely cold periods toward the end of F-1, and when N 2 O fluxes became elevated on +CC−FC (i.e., F-2), the crimson clover biomass could have been winterkilled during F-1, which might have temporarily boosted N-mineralization during short warming periods during F-2.This is also supported by the higher nitrate concentration measured in the surface soil (0-15 cm) for +CC−FC than −CC−FC on DOY 94 (6.9 vs. 4.6 mg NO 3 -N kg −1 soil, respectively; p < 0.05).Byers et al. (2021) showed a winter-sensitive clover cover crop accelerated nitrification by providing extra N through biomass decomposition, which depleted oxygen in soil, increased denitrification, and thereby elevated N 2 O fluxes.In a laboratory chamber experiment, Parkin et al. (2006) concluded that rye (Secale cereale) reduced N 2 O emission under 15-18 • C temperature.From a winter field experiment, Parkin and Kaspar (2006) reported nonsignificant impact by an overwinter rye cover crop compared with the no-till, no cover crop management.On the contrary, other studies have indicated that a rye cover crop can increase winter N 2 O emissions (Mitchell et al. 2013;Thomas et al. 2017).Our results and field observations suggest that the significantly higher N 2 O emission measured on +CC−FC was likely attributed to the clover rather than the ryegrass in the cover crop mix used in this experiment.From a year-long field chamber experiment, Brozyna et al. (2013) observed that periods with high N 2 O fluxes overlapped with the timing of ryegrass-clover residue turnover.
Cover crop establishment on +CC−FC did not keep surface soil warmer than −CC-FC during all freezing periods as anticipated.On the contrary to observations from studies included in Blanco-Canqui and Ruis (2020), +CC−FC observed lower (more negative) soil temperatures at 5 cm depth than −CC−FC during prefreeze, F-1, and F-2 (Fig. 3).The low cover crop biomass (<20 g m −2 ; Table 1) may not have been sufficient to create significant insulation effect on the surface soil.Higher (more positive) temperatures were measured on +CC−FC than −CC−FC in F-3 after a sharp increase in air temperature in T-2 that brought soil temperature from all fields to around zero (Figs. 1 and 3).Surface SWC measured from +CC−FC were lower than −CC−FC going into the winter prefreeze and consequently F-1 periods (Figs.1b and 3), and could be related to increased evapotranspiration in cover crop fields during fall (Sharma et al. 2017).However, during the F-3 and the FT, SWC was not different between these fields indicating that this effect was lost over the winter.

Cultivating cover crops amplified N 2 O fluxes at thaw
Fall cultivation of cover crops resulted in significantly higher N 2 O fluxes in all periods except during the prefreeze and the FT (p = 0.406 and 0.173, respectively; Fig. 3).The difference in N 2 O fluxes were remarkably large during T-1 and T-2, when mean N 2 O fluxes from +CC+FC were over 15 and 19 times higher than from the −CC−FC field (mean flux of 22.20 vs. 1.37 ng-N m −2 s −1 and 11.02 vs. 0.54 ng-N  m −2 s −1 ), respectively.Comparing to the +CC−FC field, fall cultivation of cover crops increased total N 2 O emissions by 33.1% over the NGS (Fig. 4).The difference in cumulative N 2 O emission started occurring early in the season during F-1 and T-1 (Fig. 4).A larger difference was found between the two fields during T-1 (p < 0.01), when +CC+FC had its highest single-day flux in the NGS (46.2 ng N m −2 s −1 on DOY 35) and a high mean flux of 25.3 ng N m −2 s −1 over the 3-day thaw, while +CC−FC had negligibly increased fluxes during the same thaw (mean of 4.98 ng N m −2 s −1 ).
In terms of soil temperature, the combined effects of cultivation and cover crops (+CC+FC) resulted in lower average temperatures than −CC−FC during the prefreeze, F-1, and F-2, and higher average temperatures during F-3 and FT (Fig. 3).Lower SWC were recorded on +CC+FC than on −CC-FC during the prefreeze, F-1, and F-2 (Fig. 3), which meets the expectations because both cultivation and cover crops can reduce soil SWC due to reasons discussed above.A higher SWC was recorded on +CC+FC than on −CC−FC (p < 0.001) during F-3, indicating an earlier melt of surface accumulated ice, snow, and frozen soil water in this field.A lower SWC was recorded on +CC+FC than on −CC−FC (p < 0.001) during FT, which could have been a result of accelerated evaporation and drainage associated with soil profile melting caused by the higher temperature on +CC+FC.

Cumulative nongrowing season N 2 O emissions
Overall NGS N 2 O emissions among the fields varied almost 2.5-fold from 395 g N ha −1 (−CC−FC) to 978 g N ha −1 (+CC+FC) (Table 3).On the −CC−FC field (i.e., the control), corn residue left on the soil surface contributed to the low FDD over the NGS, which likely resulted in the low total NGS emissions associated with less freezing.This is in line with the findings from Wagner-Riddle et al. ( 2007) that no-tillage with crop residues remaining on the soil surface significantly reduced NGS N 2 O emissions, which were found well correlated to FDD.In both +CC and −CC pairs, +FC significantly increased total NGS emissions, and in both −FC and +FC pairs, +CC significantly increased total NGS emissions, as indicated by the nonoverlapping ranges derived from the uncertainty test (Table 3).Noncultivated fields with crimson clover and ryegrass cover crops (+CC−FC) increased total N 2 O emissions by 85% compared with the −CC−FC field over the NGS (Fig. 4; Table 3).Over the NGS, total N 2 O emission measured on the +CC+FC field was 147% higher than the −CC−FC field (Fig. 4; Table 3).
Between the +CC fields, the addition of fall cultivation (+FC) significantly increased total NGS N 2 O emissions.While fall cultivation cover crops amplified the emissions in the NGS, total NGS N 2 O emission from the +CC+FC field was only 43% higher than that from the −CC+FC field (Fig. 4), suggesting fall cultivation contributed more heavily than the incorporated cover crop biomass to the large difference in N 2 O emissions between +CC+FC and −CC−FC.According to Robertson and Groffman (2015), C:N ratio of corn stover is roughly 2-3 times that of the ryegrass and clover cover crop residue.However, in our study, the quantity of corn residue biomass was nearly 50 times greater than cover crop biomass (Table 1), which suggests that the effect of fall cultivation on soil conditions and corn residue turnover could have been the dominant source of the elevated N 2 O emission measured on +CC+FC.The increased freezing during cold months and increased supply of N-rich cover crop biomass combined likely resulted in the high fluxes during this NGS.
Based on our results that show increases in N 2 O emissions in +CC−FC tended to occur later in the NGS and from our field observations of crimson clover being winterkilled, we found evidence that the winter-killed legume cover crop counteracted the possible effects that cover cropping might have had on reducing NGS N 2 O emissions.This effect occurred despite the low legume cover crop biomass established on both +CC fields (Table 1).Such result indicates that nonlegume cover crops or potentially a multispecies cover crop mixture would be more beneficial for N 2 O emission reduction during spring thaw.

Conclusion
During a six-month NGS following corn, a crimson clover and perennial ryegrass cover crop mix increased seasonal total N 2 O emission, although we note the limited growth of cover crops prior to the NGS.High N 2 O flux peaks were observed during F-3 and FT on +CC−FC, which may be explained by the nitrogen substrate released from winter-killed clover biomass.Cover crops (+CC) effectively reduced SWC throughout most of the NGS, which could have reduced den-itrification and N 2 O emissions if the fields were not cultivated (−FC) and the cover crop species could survive through the winter.Fall cultivation alone increased N 2 O emissions, with or without the presence of cover crops.Cultivation (+FC) lowered soil temperature and water content during most of the beginning two-thirds of the NGS (Nov-Feb), increased soil temperature during the last one-third of the NGS (Mar-Apr), and increased SWC because of the earlier and accelerated melting in that period.Cover crop biomass incorporated (+CC+FC) to soil contributed to the highest seasonal total N 2 O emissions, with elevated N 2 O fluxes through all FT cycles except the prefreeze when cover crops were still living, and the FT period, during which the added nitrogen substrate from incorporated cover crop biomass was likely depleted.Our results emphasize the importance of cover crop selection and termination practices in mitigating N 2 O emissions.We focused on NGS emissions in a cold climate region, which to date has received little attention, but integration into full year or full crop rotation emissions will identify the overall effect of cover crops and inform management interventions from planting to the subsequent cropping season that mitigates N 2 O emissions.

Note:
An independent sample t test was carried out to compare NO 3 − levels measured in the four fields in pairs.Differences are indicated by different letters following the mean paired values: +CC+FC vs. +CC−FC, p value < 0.01 or −CC+FC vs. −CC−FC, p value < 0.01 or +CC−FC vs. −CC−FC, p value < 0.01.No significant difference was observed in the fall results.

Fig. 1 .
Fig. 1.(a) Mean daily soil temperature measured at 5 cm (colored solid lines) and mean daily air temperature (dashed line) and (b) mean daily soil water content (SWC) at 5 cm (solid lines), average snow depth (bars), measured from 31 October 2018 to 30 April 2019 with labeled day of year representing the first of each month between November 2018 and April 2019.The legends for line colors in panel (b) are common for both panels (a) and (b), and indicate values for −CC−FC (red), −CC+FC (blue), +CC−FC (orange), and +CC+FC (green).Grey, solid vertical lines in both panels (a) and (b) indicate prefreeze, freezing (indicated by "F"), and thawing (indicated by "T") or "Final Thaw" periods.Note that SWC values when soil temperature is below 0 • C are not accurate but are shown here to indicate the freezing and then thawing periods (i.e., when SWC apparently decreases and then increases).

Fig. 2 .
Fig. 2. Daily average N 2 O flux measured from 31 October 2018 to 30 April 2019 with labeled day of year representing the first of each month between November 2018 and May 2019, measured for each field, with segments of period divided by grey, vertical dashed lines representing prefreeze, freezing ("F"), or thawing ("T" or "Final Thaw") periods.

Fig. 3 .
Fig. 3. Median and standard error of the difference in N 2 O emissions (ng N 2 O-N m −2 s −1 , top panel), soil temperature ( • C, middle panel), and SWC (m 3 m −3 soil, bottom panel) between treatments and the control field (no cover crops and no fall cultivation, −CC−FC).Asterisk symbols indicate significance level at p ≤ 0.05 following Wilcox test of treatment versus control.

Fig. 4 .
Fig. 4. Cumulative N 2 O emissions (g N 2 O-N ha −1 ) measured from 31 October 2018 to 30 April 2019 with labeled day of year representing the first of each month between November 2018 and May 2019.Colored solid lines represent daily main N 2 O flux measured from each field, and segments of period divided by grey, vertical dashed lines represent prefreeze, freezing ("F"), or thawing ("T" or "Final Thaw") periods.

Table 1 .
Corn grain (15.5% moisture) and residue yield (dry matter) measured on 10 October 10 2018; cover crop biomass yield (aboveground dry matter) in g m −2 measured on 5 October 2018.Ryegrass biomass was 17.4 ± 3.8 g m −2 on 21 May 2019 and the clover plants were winterkilled and unmeasurable on this date. *

Table 3 .
Freezing degree days (FDD) of surface soil temperatures measured at 5 cm depth between 31 October 2018 and 30 April 2019.
Total nongrowing season (NGS) N 2 O emissions derived from linearly filled daily flux measurements, and a statistical summary of total NGS N 2 O emissions derived from the uncertainty test, showing the mean, standard deviation, relative standard deviation (RSD), and first and third quartile of the sums generated for each field.Measured NGS emission is derived from the linearly gap-filled measured daily mean flux.Mean is the average of total emissions calculated from the 250 generated datasets with artificial gaps that were linearly filled. Note: