Cover Crop Effects on Greenhouse Gas Emissions and Global Warming Potential in Furrow-Irrigated Corn in the Lower Mississippi River Valley

Abstract

Corn (Zea mays) production systems are described as high risk for emissions of greenhouse gases (GHG) due to large fertilizer inputs. Conservation practices, such as cover crop (CC), can limit the effects of agricultural activities on GHGs while increasing carbon and nitrogen storage. The objective of the study was to assess the effects of cover crops, i.e., with CC and no-CC, on GHG (i.e., CO₂, CH₄, and N₂O) emissions and global warming potential (GWP) in furrow-irrigated corn in the Lower Mississippi River Valley. Gas sampling was conducted with an automated system that measured GHGs four times daily during the 2024 growing season in furrow-irrigated corn on a loam soil in Mississippi. Only CO₂ emissions differed (p < 0.05) by CC treatment, with soil respiration ~30% greater from CC than no-CC. Season-long emissions ranged from −0.22 to 0.30 kg CH₄ ha⁻¹ season⁻¹, 5.53 to 7.28 kg N₂O ha⁻¹ season⁻¹, with a GWP between 12,888 and 15,053 kg CO₂ eq. ha⁻¹ season⁻¹ from no-CC and CC, respectively. The role of CC as a conservation practice needs to be evaluated with soil and plant parameters. The beneficial effects of CC on soil physical and chemical properties likely outweigh a predictable increase in GHG emissions.

1. Introduction

The Delta region of the Lower Mississippi River Valley (LMRV) represents vast crop-producing areas that have recently experienced a series of challenges leading to the reduction of cropland from 66,000 to 62,000 km² since 2018 [1]. Cropland decreases of 14 and 7% were recently reported in Mississippi and Arkansas, respectively, partially due to unsustainable agricultural practices, such as intensive row cropping and excessive fertilizer use, that have substantially reduced topsoil quality [1]. Reduced soil health has resulted in a limited ability of the soil to absorb water and retain nutrients, therefore enhancing flood, drought, and runoff risks, along with negative environmental effects such as eutrophication [1]. For example, Daniels et al. [2] reported that runoff volume generated from rainfall events from cotton fields in Southeastern Arkansas can be as high as 63% of the total annual rainfall. Climate-smart practices, such as cover crops (CC) and reduced tillage practices, have been developed and evaluated to address soil fertility, water retention, yield potential, and greenhouse gas (GHG) reduction [3,4,5,6,7].

Agricultural activity accounts for ~10% of the total direct GHG emissions in the United States (US), and direct emissions tend to be relatively large in intensively managed cropland, particularly where corn (Zea mays) is the main crop, to the point that areas like the Corn Belt of the upper-midwest and the LMRV have been classified as large emissions regions [8]. Corn fields are a predominant source of carbon dioxide (CO₂) due to the substantially large production of plant biomass that results in large levels of root respiration [8]. Most reports in corn systems indicated neutral or negative methane (CH₄) emissions, suggesting that corn production systems, like other upland crop systems, behave as CH₄ sinks [9,10,11,12]. Corn production systems are further described as high risk for both direct and indirect emissions of nitrogen (N)-gases due to the large fertilizer-N inputs required to maximize corn yields [8]. The adoption of climate-smart practices in corn production systems represents an ideal option to limit the effects of agricultural activities on GHGs while increasing carbon (C) and N storage, restoring soil fertility, and improving soil health.

The effects of cover crops on GHGs in corn production systems are often complex and dynamic, making evaluation of cover crops as a climate-change-mitigation practice challenging [13]. Several studies highlighted the ability of cover crops to enhance soil health and soil quality through increased soil C and N inputs, decreased nutrient leaching, enhanced soil fertility, and improved soil aggregation in various agricultural settings, but the impact of cover crops on GHG emissions has not yet been thoroughly evaluated [14,15,16]. One of the challenges is that the effects of cover crops on GHGs are not limited to the period when cover crops are growing but also to the period following cover crop termination and subsequent decomposition [14]. Legume cover crops can fix N from the atmosphere into the soil through symbiotic relations between roots and bacteria, therefore somewhat reducing the need for cash-crop fertilization. Non-legume cover crops can increase C sequestration and reduce nutrient leaching by increasing organic material decomposition in the soil [17,18,19,20]. Field studies across the US have shown that cereal cover crops can reduce nitrate (NO₃⁻) leaching and losses between 13 and 94%, while non-leguminous cover crops, in general, average 70% NO₃⁻-N reduction during cover crop growth [21]. Reduced NO₃⁻ leaching, however, varies across soils, cropping systems, agricultural management practices, and rainfall amount and intensity [21].

Cover crop root biomass produced between cash-crop growing seasons can enhance root respiration compared to agricultural practices without CC [22,23,24]. After termination, C inputs from CC biomass can also enhance microbial processes and, therefore, CO₂ emissions, resulting in potentially greater CO₂ emissions on an annual basis than without CC [22]. Legume CC can enhance nitrous oxide (N₂O) emissions compared to non-legume or no CC due to soil-N inputs through N fixation that can act as substrate for nitrification-denitrification, while a non-legume CC can reduce N₂O emissions due to plant nutrient uptake that limits substrate available for microbial activity, although the magnitude of increased or decreased N₂O emissions can vary according to CC species, biomass, C:N ratio, lignin content, and management practice [25,26].

Cover-crop decomposition rates are correlated to the C:N ratio of CC’s residue after termination [25,26]. Residues with low C:N ratios generally decompose faster than residues with larger C:N ratios, potentially resulting in greater and earlier peaks of CO₂ emissions [27]. The C:N ratios of CC residue often vary by growth stages, making CC termination timing an important factor potentially impacting GHG emissions [28]. Residue incorporation tends to accelerate decomposition, often resulting in greater CO₂ and N₂O emissions compared to management practices where residues are left on the soil surface [29].

Cover crops, especially deep-rooted species such as cereal rye, also alter soil water status, often decreasing evaporation from the soil surface and enhancing infiltration and water movement within the soil profile, specifically during CC root growth [29]. However, the presence of CC can favor the potential for short-term anaerobic conditions, especially on somewhat poorly drained soils, by intercepting and slowing water infiltration and percolation, potentially increasing N₂O emissions, and enhancing the episodic nature of N₂O emissions with pulses of N₂O, specifically observed during CC growth [30]. In contrast to CO₂ and N₂O, CC effects on CH₄ emissions are much less understood [22].

To date, limited studies in the LMRV have addressed the impact of CC as a climate-smart practice to reduce GHGs in corn production systems. Therefore, the objective of this field study was (i) to assess the effects of CC [i.e., with cover crop (CC) and without CC (no-CC)], measurement date [day of the year (DOY)], measurement time of the day (i.e., 0300, 0900, 1500, and 2100 h), and their interactions on CO₂, CH₄, and N₂O fluxes and soil volumetric water content (VWC) and soil temperature, and (ii) to assess the effects of CC on season-long GHG emissions, emissions intensity, global warming potential (GWP), and GWP emissions intensity in furrow-irrigated corn on a loam soil in the LMRV. It was hypothesized that GHG fluxes would be greatest at the 1500 and lowest at the 0300 h sampling time, respectively, due to the role of temperature on microbial activity. It was also hypothesized that GHG fluxes from CC would peak earlier than from no-CC due to greater organic biomass in the CC treatment. It was hypothesized that soil VWC and soil temperature would experience less variability with CC due to the presence of the CC residues that slow the percolation of water into the soil profile and reduce the amount of solar radiation reaching the soil surface. It was also hypothesized that GHG fluxes, season-long emissions, GWPs, and yield would be greater from CC than from no-CC due to the organic substrates C mineralization released by CC residue. It was hypothesized that season-long GHG emissions and GWP intensities would not differ between CC treatments due to potentially greater yields under CC than no-CC, which would offset differences in GHG emissions between CC treatments.

2. Material and Methods

2.1. Site Description

This field study was conducted during the 2024 growing season at the Vayda Regenerative Farm near Clarksdale, Mississippi (34.216950° N, 90.658806° W, Figure 1). The study area was on Quaternary-aged alluvial terrace of the Mississippi River, mapped as Dubbs silt loam (fine-silty, mixed, active, thermic Typic Hapludalfs), which are well-drained, moderately permeable soils [31,32]. The 30 yr (i.e., 1991 to 2020) average monthly air temperature in the area is 16.5 °C, while the 30 yr mean annual precipitation is 135.8 cm [33].

The specific study area was within a ~64 ha field, 1000 m long and 175 m wide, with 15 cm tall raised beds that were 38 cm wide on top of the beds, with north–south orientation and an average slope of 0.2%. In 2024, the eastern half of the field was planted with soybean (Glycine max L.), while the western half was planted with corn. The entire 64 ha field was divided into various strips that encompassed several raised beds and differed by CC mix planted during Fall 2023. The specific study area in which GHG data were collected included four raised beds without CC (i.e., no-CC) and adjacent four raised beds planted to a CC mix, described below.

2.2. Field Management

Prior to experiment initiation, from 2018 to 2022, the entire 64 ha field was under soybean production, while corn was planted during the 2023 growing season. On 10 October 2023, after corn harvest, one pass of a shallow disc across the entire study area was used to manage crop residues. On 23 October, the CC strip was treated with one pass of a groover implement and planted with a mix of cover crops, seeded at a rate of 49.3 kg ha⁻¹, composed of 69% cereal rye (Secale cereale), 16% crimson clover (Trifolium incarnatum), 7% balansna clover (Trifolium michelianum), and 8% radish (Raphanus sativus) by mass, while the adjacent strip (i.e., the no-CC treatment) was left with no CC. The two clover varieties are members of the Fabaceae family and, thus, are considered legumes, while rye and radish are non-leguminous [34]. Both treatment strips were under reduced-tillage management for at least the previous two years.

On 18 April 2024, the CC was chemically terminated with an application of glyphosate [Roundup Power Max3, N-(phosphonomethyl)glycine] (Bayer, St. Louis, MO, USA) applied at a rate of 2.4 L ha⁻¹ and flumioxazin (Valor SX, Valent, San Ramon, CA, USA) applied at a rate of 140 mL ha⁻¹. On 28 April 2024, the corn hybrid 1627-TC (Revere, Memphis, TN, USA) was seeded with a planter at a rate of 65,558 seeds ha⁻¹ to a depth of 2.5 cm as single rows on the center of the raised beds with ~19 cm seed spacing. Following soil test results and best practices for plant nutrient management developed by Mississippi State University [35], at planting, the study area was broadcast with 22 kg N ha⁻¹ as granular urea (46-0-0), 3.4 kg S ha⁻¹ of sulfur (S) and zinc (Zn) as zinc sulfate (0-0-0-36Zn-20S), 39 kg ha⁻¹ of phosphorus (P) as triple superphosphate (0-46-0), and 67 kg ha⁻¹ of potassium (K) as muriate of potash (0-0-60). On 11 May 2024, the study area was broadcast with an additional 154 kg N ha⁻¹ as urea. On 28 May 2024, the field was sprayed with a tank mix of post-emergence herbicides composed of Bayer Roundup Power Max3 at a rate of 369 mL ha⁻¹, Helena (Collierville, TN, USA), Atrazine 4L [atrazine, (2-chloro-4-ethylamino)-6-(isopropylamino)-s-triazine] at a rate of 5 L ha⁻¹, Estes (Waukegan, IL, USA) NIS 80-20 (alkyl polyethoxy ethers) at a rate of 947 mL ha⁻¹, and Halex GT [mesotrione, (S-metolachlor, N-(phosphonomethyl) glycine; Syngenta, Basel, Switzerland] at a rate of 4.7 L ha⁻¹. On 30 May and 26 June 2024, a second and third application of urea was broadcast-amended at a rate of 69.5 and 31 kg N ha⁻¹, respectively, for a total fertilizer-N application of 276.5 kg N ha⁻¹ applied over the course of the growing season.

Irrigation was applied through a polyethylene polypipe. On 2–24 and 26 June, irrigation applications amounted to a total of 11.2 cm of water ha⁻¹, while from 26 June to 2 July 2024, 12.3 cm of water ha⁻¹ was used to irrigate the study area. On 23 August 2024, the study area was combined-harvested.

2.3. Soil Sampling and Analyses

On 3 May 2024, two sets of soil samples were collected from the top 15 cm of the top of the four raised beds. Five subsamples were collected with a 2 cm diameter push probe for soil physical and chemical analyses from the top of the raised beds ~30 cm downslope of each chamber and combined for one composite sample per chamber, while a second set of four samples was collected with a 4.8 cm diameter, stainless steel core chamber and slide hammer for bulk density (BD) determinations. All soil samples were oven-dried at 70 °C for at least 48 h, weighed, crushed, and sieved through a 2 mm mesh. A modified 12 h hydrometer test was used to determine particle-size distribution [36] and a 1:2 soil mass/water volume suspension was used to potentiometrically measure soil electrical conductivity (EC) and pH. Weight-loss-on-ignition following combustion at 360 °C for 2 h was used to determine soil organic matter (SOM) concentration, and a VarioMax CN analyzer (Elementar Americas Inc., Mt. Laurel, NJ, USA) was used to determine total C (TC) and total N (TN). All measured TC was assumed to be organic C due to the lack of effervescence upon treatment with dilute hydrochloric acid. The C:N ratio was calculated from measured TC and TN concentrations. Following extraction in a 1:10 soil mass/extractant volume ratio and analysis by inductively coupled argon–plasma spectrophotometry, Mehlich-3-extractable soil nutrients concentrations were determined (i.e., P, K, Ca, Mg, Fe, Mn, Na, S, and Zn) [37]. Measured BDs were used to convert measured TC, TN, SOM, and extractable nutrient concentrations to contents (kg or Mg ha⁻¹) for data reporting and analyses.

2.4. Gas Sampling and Analyses

On 3 May 2024, two 21 cm inner-diameter, 12 cm tall PVC base collars were installed in each of the two CC treatments, for a total of four base collars from which GHG measurements were collected. Collars were installed on top of adjacent raised beds within the two CC treatments between corn plants. Two raised beds separated the two adjacent pairs of collars. Collars had a beveled bottom to facilitate installation and were installed to a depth of 7 to 8 cm. Collars were located ~15 m in the downslope direction from the up-slope edge of the field. Collars did not contain plants or residues at any point of the growing season.

Gas sampling was conducted through an automated system powered by solar panels (LI-COR Environmental, Inc., Lincon, NE, USA). Four 335-W solar panels, 200 cm long and 120 cm wide, were mounted on an aluminum frame with mounting feet and telescoping legs. To maximize solar energy absorption, the length of the telescoping legs was adjusted to have a tilt (i.e., angle between ground surface and panel surface) of ~34°, which represented the latitude of the study area location. Solar panels were wired to a battery control enclosure equipped with a 15 A charge controller (Morningstar Prostar PS-15, MorningStar Inc., Chicago, IL, USA), a 60 A charge controller (Morningstar Prostar TS-60, MorningStar Inc., Chicago, IL, USA), an electromagnetic compatibility (EMC) filter (TDK Lambda RSHN-2016, TDK-LAMBDA Americas, National City, CA, USA), a series of eight, 6 V batteries (Concorde PN-002783, Concorde Battery Corporation, West Covina, CA, USA), and one load of terminal blocks. Solar panels and the battery control enclosure were placed and anchored on six wooden pallets to maintain a level position throughout the growing season. The solar panels and battery enclosure were placed at the field edge, up-slope of the irrigation polypipe, and were oriented toward true south.

The battery control enclosure was then wired to an Eddy-flux enclosure (LI-COR 7900-050, LI-COR) and a 24 V power distribution kit (LI-COR 7900-235, LI-COR). The power distribution kit provided power to an ethernet switch connected to a cellular modem (AirLink RV50X, SEMTECH, Colorado Spring, CO, USA) equipped with an external high-gain cellular antenna to communicate with cellular signals. The power distribution kit was wired to provide current to a CO₂/CH₄ analyzer (Li-7810, LI-COR), a N₂O analyzer (Li-7820, LI-COR), and a multiplex system (Li-8250, LI-COR). The multiplex system was also connected to four opaque, long-term chambers (Li-8200-104, LI-COR) equipped with a Stevens hydraprobe (LI-COR 90017114) installed at a depth of 10 cm that recorded soil volumetric water content (VWC) and soil temperature. The multiplex system and the four chambers were connected through 15 m cables that provided gas-flow inlets and outlets, electrical power, and digital transfer of data. The four chambers were installed over the PVC collars in a leveled position via adjustment legs. The entire system was powered on in the evening of 4 May, and data collection started on the morning of 5 May 2024.

The system was configured to collect and analyze gas samples every day at four specified times: 0300 (3 a.m.), 0900 (9 a.m.), 1500 (3 p.m.), and 2100 (9 p.m.) hours, local time. Sampling times were determined to potentially capture the daily average (i.e., 0900), minimum (i.e., 0300), and maximum (i.e., 1500) flux according to internationally adopted gas sampling protocols [8,38]. At each sampling time, chambers were programmed to rotate over the collars from an initially open position and lowered to a sealed position on top of the base collars. Chambers were programmed to operate in sequential order from chambers 1 to 4 and collect gas samples for 320 s (i.e., observation length). During the observation length, gas flow from a single chamber was pumped and directed toward the multiplex systems, and the two GHG analyzers recorded CO₂, CH₄, and N₂O concentrations every second through optical-feedback, cavity-enhanced, absorption spectroscopy (OF-CEAS). Right before and after each chamber’s sampling, a purge of air was performed throughout the tube lines for 45 s to eliminate any possible contamination in the inlet and outlet tubes. Outside the sampling times, chambers were programmed to remain in an open position to avoid interference with the sampled area inside the collars.

The multiplex system was integrated with SoilFluxPro software (version 5.3, LI-COR) with settings customized for gas sample analysis in the climatic conditions of the LMRV. Using the stop-analysis and deadband-analysis options in SoilFluxPro on field test trials conducted in the LMRV, the observation length was set at 320 s for N₂O but only 120 s for CO₂ and CH_4, while all three GHGs had a deadband period of 45 s, indicating that the data collected during the first 45 s of the observation length were not used to calculate gas fluxes. The deadband allowed potential initial perturbations created within the sealed chamber-collar-soil environment to be excluded from the sampling procedures. The SoilFluxPro software was set to calculate the slope of the best-fit linear regression among the 275 (N₂O) and 75 (CO₂ and CH₄) concentration points per sampling on a chamber-by-chamber basis and to convert the slope into a flux (µmol m⁻² s⁻¹) using the volume of the entire multiplex system and the area encompassed by the collars. Fluxes were then converted to g or mg m⁻² h⁻¹ using the molecular mass of the GHGs. Negative slopes were retained and considered in this study. The multiplex system was checked routinely, connecting remotely to assess diagnostic parameters, such as cavity, thermal enclosure, and laser phase pressure, temperature, and input voltage for the two analyzers, the four chambers, and the overall system. In case of malfunction of one of the system components, SoilFluxPro allowed for manual modification of the observation length and deadband to improve the fit of linear models.

Season-long GHG emissions (kg ha⁻¹ season⁻¹) were calculated on a chamber-by-chamber basis through linear interpolation between hourly fluxes for consecutive measurement times and dates. A three-gas GWP was calculated on a chamber-by-chamber basis, using conversion factors of 1, 28, and 265 for CO₂, CH₄, and N₂O, respectively, to convert to CO₂ equivalents (CO₂ eq.) according to the Intergovernmental Panel on Climate Change (IPCC) 6th assessment [38,39]. A reduced, two-gas GWP (GWP*) was also calculated from only N₂O and CH₄. Season-long emissions, on a chamber-by-chamber basis, were divided by the measured yield (described below) for a specific treatment (Mg ha⁻¹ season⁻¹) to calculate emissions intensity [EI; kg GHG (Mg yield)⁻¹]. The same procedure was used to calculate GWP emissions intensities [kg CO₂ eq. (Mg yield)⁻¹].

The entire system was powered off on 18 August 2024 after continuous recording for 105 consecutive days. No system malfunctions occurred during the entire 105-day measurement period; thus, no adjustments to flux calculations in SoilFluxPro were necessary.

2.5. Yield Estimation

On 19 August 2024, a day after the multiplex system was shut down, corn ears were manually collected from a 3 m row length in the downslope direction from the position of each GHG chamber. Corn ears were manually husked, and grain was manually removed from the cobs and weighed. Grain moisture content was measured on a consistent sub-sample mass with a grain moisture analyzer (GAC-2500-UGMA, Dickey-John Corporation, Auburn, IL, USA). The recorded corn grain mass was then corrected to 15.5% moisture for reporting purposes. Yield was estimated on a chamber-by-chamber basis by dividing the corrected grain mass by the area of the experimental unit, which was 3 m long by 0.76 m wide.

2.6. Statistical Analyses

Based on a repeated measures design, a three-factor analysis of variance (ANOVA) was conducted to evaluate the effects of the measurement date, measurement time (i.e., 0300, 0900, 1500, and 2100 h), CC treatment (i.e., CC and no-CC), and their interactions on GHG fluxes, soil VWC, and soil temperature. Measurement date and time and CC treatment were considered fixed effects. All analyses were conducted with the R-package ASREML (R version 4.3.2, R Foundation for Statistical Computing, Vienna, Austria). Analysis of GHG fluxes, soil VWC, and soil temperature was conducted by modeling the variance–covariance structure of the residuals in relation to measurement date and time. The repeated samplings of the within-subject (i.e., chamber) factor (i.e., measurement date and time) were equally spaced in time. The between-subject factor (i.e., CC treatment) was randomized in a completely randomized design (CRD) at the beginning of the growing season. A residual maximum likelihood (REML) was used as the convergence method for the GHG flux, soil VWC, and soil temperature models.

Different distributions (i.e., normal and gamma) and variance–covariance structures were evaluated, including identity, diagonal, autoregressive first order, autoregressive second order, and unstructured. The best fit was determined by the evaluation of the Akaike Information Criteria (AIC) and the likelihood ratio test (LRT). Heteroskedasticity was included for each variance–covariance structure and statistically compared to the equivalent model defined by homoskedasticity. As a result, the three GHG models, along with soil property models, were defined by a normal distribution, homogenous variance, and autoregressive first-order variance–covariance structure for measurement date and time.

Based on a CRD design, a one-factor ANOVA was conducted with the ASREML package to evaluate the effect of CC treatment on initial soil properties, season-long emissions, GWPs, emissions intensity, GWP intensity, and yield. For all one-factor ANOVA analyses, normality of the residuals was checked with qqplots, and no substantial deviance from normality was observed for all response variables. Each response variable dataset was complete and balanced, and no outliers were identified for any response variable.

A Wald test was performed on the best-fit model in the repeated measures and one-factor ANOVA for all response variables to extract the ANOVA-like table for the fixed effects. Pairwise multiple comparisons were performed according to Tukey’s honest significant difference (HSD). Significance for the LRT test, ANOVA, and multiple comparisons were judged at α = 0.05.

A multiple linear regression analysis was performed in JMP (version 14.3.0, SAS Institute, Inc., Cary, NC, USA) using a stepwise, forward regression function with soil VWC and soil temperature as predictors to explain the variability of the three GHG fluxes. The categorical variable “cover crop” was not included in the multiple regression to simplify models by using only continuous variables. The maximum likelihood of the models was fit using the Bayesian Information Criterion as the stopping rule. A continuous, normal distribution was used for the three GHGs in the multiple regression analyses. Data for the multiple regression were reviewed in JMP for outliers and influential values using the jackknife distance method and Cooks distance, respectively, where no data points were removed from any data set. Significance was determined using α = 0.05.

3. Results

3.1. Initial Soil Properties

Initial soil properties were assessed to evaluate potential inherent differences among CC treatments. All initial soil properties in the top 15 cm did not differ (p > 0.05) between CC treatments (

Table 1. Summary soil physical and chemical property means and standard errors (SE) (n = 4) in the top 15 cm of a silt loam soil at the Vayda Regenerative Farm near Clarksdale, MS, from the beginning (15 May) of the 2024 growing season at the up-slope position of a furrow-irrigated corn field.

Soil Property	Mean (SE)
Sand (%)	46.8 (0.20)
Silt (%)	44.2 (0.56)
Clay (%)	9.0 (0.71)
Bulk density (g cm−3)	1.26 (0.10)
pH	6.2 (0.04)
Electrical conductivity (dS m−1)	0.192 (0.01)
Extractable nutrients (kg ha−1)
P	118.4 (11.7)
K	367.1 (52.6)
Ca	2650 (193)
Mg	429.1 (40.1)
S	23.78 (3.6)
Na	16.34 (1.2)
Fe	448.7 (49.6)
Mn	101.1 (12.3)
Zn	2.77 (0.28)
Total N (Mg ha−1)	5.75 (0.13)
Total C (Mg ha−1)	20.55 (0.48)
Soil organic matter (Mg ha−1)	38.25 (0.23)
C:N ratio	10.0 (0.1)

). Consequently, any subsequent differences between CC treatments were attributable to the field treatments themselves instead of being due to inherent soil property differences.

Although mapped as silt loam surface texture, laboratory analyses confirmed a loam surface texture with 47% sand and 9% clay (Table 1). Soil test results indicated P (65 to 128 kg ha⁻¹) and K (259 to 455 kg ha⁻¹) in the high category and Mg in the very high category (>85.7 kg ha⁻¹, Table 1) [35]. The ideal soil pH for corn production in the LMRV ranges between 6 and 7 when the availability of micronutrients, such as Zn, Fe, and Mn, is moderate (

Table 2. Thirty-year (1991–2020) regional mean and local monthly rainfall and monthly minimum (Min), maximum (Max), and mean air temperatures (NOAA-NCEI, 2024) from weather stations located near the Vayda Regenerative Farm, Clarksdale, MS, for the 2024 growing season.

Month	30-Year Mean				2024 Local
	Rainfall (cm)	Air Temperature (°C)			Rainfall (cm)	Air Temperature (°C)
		Min	Max	Mean		Min	Max	Mean
May	14.0	16.2	27.3	21.7	4.7	13.9	33.3	24.5
June	10.1	20.3	31.3	25.8	6.2	15.0	36.7	27.4
July	9.5	21.9	32.6	27.2	13.1	18.9	37.2	27.9
August	6.9	21.3	29.6	26.9	6.1	17.2	38.3	26.6
Season total	40.5	-	-	-	30.1	-	-	-

) [35]. Soil test analyses within the study area suggested that plant response to fertilizer additions other than N was likely not expected (Table 2). The calculated mean C:N ratio indicated that, upon N additions, rapid SOM mineralization and release of inorganic N was expected (Table 2) [40].

3.2. Growing-Season Weather Characteristics

Rainfall and air temperature variations during the 2024 growing season differed somewhat from the 30-year averages for the region (Table 2). Rainfall amount and distribution, along with air temperature ranges, during the growing season influence microbial activity and, thus, can impact C- and N-gas losses [41]. At the beginning of the growing season, rainfall during May and June was more than 50% lower than the 30-year monthly averages (Table 2). In contrast, July had 38% more rainfall than the 30-year monthly average (Table 2). Similarly, the air temperature during the growing season had a lower minimum and greater maximum during every month compared to the 30-year average (Table 2). However, the mean air temperature was only 13, 6, and 3% greater during May, June, and July, respectively, and only 1% lower during August than the 30-year monthly averages, suggesting that air temperatures roughly resembled average conditions (Table 2).

3.3. Greenhouse Gas Fluxes

3.3.1. CO₂

Based on 1680 observations across all treatment combinations, CO₂ fluxes differed among measurement date–time (p < 0.01) and among CC treatment–measurement time (p = 0.04) combinations (

Table 3. Analysis of variance summary of the effects of measurement date (Date; n = 105), measurement time of day (Time; n = 4), cover crop treatment (TRT; n = 2), and their interactions on carbon dioxide (CO₂), methane (CH₄), and nitrous oxide (N₂O) fluxes and soil volumetric water content (VWC) and soil temperature (Soil T) measured at 0300, 0900, 1500, and 2100 h each day during the 105 days of the 2024 growing season at the up-slope position of a furrow-irrigated corn field at the Vayda Regenerative Farm near Clarksdale, MS.

Source of Variation	CO2	CH4	N2O	VWC	Soil T
	p
TRT	<0.01	0.04	0.01	<0.01	0.07
Date	<0.01	0.96	<0.01	<0.01	<0.01
Time	<0.01	0.99	<0.01	<0.01	<0.01
TRT × Date	0.34	<0.01	0.99	<0.01	0.09
TRT × Time	0.04	0.99	<0.01	0.75	0.02
Date × Time	<0.01	0.11	<0.01	0.99	<0.01
TRT × Date × Time	0.12	0.59	<0.01	0.99	0.99

). Across all measurement date–time combinations, CO₂ fluxes were all greater than zero (p < 0.05).

Averaged across CC treatments, the greatest CO₂ flux (2.53 g m⁻² h⁻¹) was measured at the 1500 h on DOY 229 (16 August) and did not differ (p > 0.05) from the flux reported on DOY 168 (16 June) at the 1500 h (Figure 2a). A significant difference among measurement times of day occurred on 26 measurement dates over the growing season (i.e., DOY 130, 135, 146, 147, 148, 153, 154, 157, 158, 159, 161, 167, 168, 169, 171, 172, 174, 176, 183, 192, 193, 199, 203, 204, 228, and 229), where almost 80% of the differences occurred during the first half of the growing season (Figure 2a,b). On 20 of the 26 measurement dates when a significant difference among measurement time occurred (i.e., DOY 130, 135,146, 147, 148, 154, 157, 159, 167, 168, 169, 171, 172, 176, 183, 192, 193, 203, 228, 229), the CO₂ flux at 1500 h was greater (p < 0.05) than at 0900, 2100, and 0300 h, which did not differ (p > 0.05; Figure 2a,b). The numerical CO₂ flux peaks at 0900 (0.87 g m⁻² h⁻¹), 2100 (0.75 g m⁻² h⁻¹), and 0300 (0.62 g m⁻² h⁻¹) hours occurred on DOY 178, 130, and 126 (i.e., 26 June, 9 May, and 5 May, respectively; Figure 2a,b).

Averaged across measurement dates, the greatest CO₂ flux occurred from CC at 1500 h, while the lowest CO₂ flux was from no-CC at 0300 h, which did not differ from CO₂ fluxes from CC at 0300, CC and no-CC at 0900, and no-CC at 2100 h (Figure 3). The lowest CO₂ flux was also lower (p < 0.05) than from CC at 2100 and no-CC at 1500 h, which also differed (p < 0.05) from each other (Figure 3). Carbon dioxide fluxes only differed between CC treatment at 1500 h but were unaffected (p > 0.05) by CC treatment at 0300, 0900, and 2100 h. However, at all measurement times, CO₂ fluxes were at least numerically greater from CC than no-CC (Figure 3).

3.3.2. CH₄

In contrast to CO₂, CH₄ fluxes differed between CC treatments over time (p < 0.01) but were unaffected (p > 0.05) by measurement time (Table 3). Across the 105 total measurement dates, 14 and 44% of CH₄ fluxes were numerically negative from CC and no-CC, respectively, highlighting a potential substantial difference among CC treatments for methanogenic and methanotrophic processes (Figure 3). During the first half of the season, CH₄ fluxes from CC were all positive, except at DOY 175 (i.e., 23 June) when the first negative fluxes occurred. From no-CC, more than 27% of the CH₄ fluxes were negative during the first half of the season (Figure 4). However, CH₄ fluxes differed (p < 0.05) between CC treatments on three dates (i.e., DOY 177, 178, and 195; 25 and 26 June and 13 July, respectively) when CC had a steep increase in methanogenic activity and CH₄ release compared to a likely steep increase in methanotrophic activity and CH₄ consumption under no-CC (Figure 4).

Methane fluxes measured on DOY 177, 178, and 195 were the only fluxes that differed (p < 0.05) from a flux of zero from CC, and all three were positive. From no-CC, CH₄ fluxes measured on DOY 177, 178, 184, 185, 194, 195, and 220 (i.e., 25 and 26 June; 2, 3, 12, and 13 July; and 7 August) were the only dates when CH₄ fluxes differed (p < 0.05) from a flux of zero and were all were negative, except for the CH₄ flux on 12 July, which was the only significantly positive CH₄ flux from no-CC over the entire season (Figure 4).

3.3.3. N₂O

In contrast to CO₂ and CH₄, N₂O fluxes differed (p < 0.01) among CC treatment–measurement date–measurement time-of-day combinations (Table 3). Despite a complex interactive effect, N₂O fluxes differed (p < 0.05) among CC treatment–measurement date–measurement time combinations on only eight measurement dates (i.e., DOY 127, 144, 145, 146, and 153, 168, 178, and 191; 6, 23, 24, and 25 May, 1, 16, and 26 June, and 9 July respectively), which all occurred within or right at the end of the first half of the growing season (Figure 5a,b and Figure 6a,b; Table 2). On six of the eight dates (i.e., DOY 127, 144, 153, 168, 178, and 191; 6 and 23 May, 1, 16, and 26 June, and 9 July respectively), mainly at 1500 (i.e., DOY 127, 153, and 168) and 2100 (i.e., DOY 144, and 178) hours, N₂O fluxes were greater from CC than from no-CC (Figure 5a,b and Figure 6a,b).

Among all treatment combinations, >20% of measured N₂O fluxes from CC (i.e., 14 dates for 0300 and 0900, 41 dates for 1500, and 16 dates for 2100 h) and almost 13% (i.e., 10 dates at 0300, 11 dates at 0900, 22 dates at 1500, and 11 dates at 2100 h) from no-CC were greater (p < 0.05) than a flux of zero (Figure 5a,b and Figure 6a,b). The largest mean N₂O flux was recorded on DOY 168 (i.e., 16 June) at 1500 h under CC (4.9 mg m⁻¹ h⁻¹), which did not differ (p > 0.05) from the flux on DOY 145 (i.e., 24 May) at 1500 h under no-CC (3.8 mg m⁻¹ h⁻¹, Figure 5b).

3.3.4. Soil Volumetric Water Content and Soil Temperature

Somewhat similar to CH₄ fluxes, VWC differed among measurement times of day (p < 0.01) and among CC treatment–measurement date combinations (p < 0.01; Table 3). Averaged across measurement times of day throughout the growing season, 12 discernible wet and dry cycles were observed in both CC treatments, with lower VWC generally occurring under no-CC during the drying phase, especially throughout and following June (Figure 6). Soil VWC was greater (p < 0.05) under CC than no-CC for 27 consecutive days starting on DOY 202 (i.e., 20 July) to DOY 230 (i.e., 17 August), with the exception of DOY 220 (7 August) when VWC was similar between CC and no-CC (Figure 7a), confirming the ability of CC residues to maintain a greater and more stable near-surface VWC.

Although the study area remained mostly aerobic throughout the entire growing season, the greater VWC under CC between 26 June and 4 July, when VWC remained between 0.38 and 0.40 cm³ cm⁻³ (Figure 7a), approached saturation and likely contributed to the largest peak CH₄ flux from the CC treatment (Figure 3). The wider VWC fluctuations during the first half of the growing season (Figure 7a) created soil conditions that alternated between fully aerobic and potentially O₂-limited, enhancing nitrification followed by denitrification, thus resulting in N₂O flux spikes that temporally followed the VWC oscillation (Figure 5a,b).

Similar to effects on CO₂ fluxes, soil temperature differed among CC treatment–measurement date (p < 0.01) and among CC treatment–measurement time combinations (p < 0.01; Table 3), suggesting a potentially strong correlation between soil respiration and soil temperature. Over the course of the growing season, soil temperature differed (p < 0.05) among measurement times on >50% of measurement dates, with the largest soil temperature generally occurring at 1500 and the lowest at 0300 h (Figure 8a,b). The peak soil temperature (49.6 °C) was measured on DOY 167 (15 June) at 1500 h, which did not differ from the soil temperature measured on DOY 168 (16 June) at 1500 (Figure 8a).

Averaged across measurement dates, soil temperature was greatest at 1500 h under no-CC, which was greater than under CC, and lowest at 0300 h from both CC treatments, which did not differ (Figure 8b). Soil temperature differences among measurement times followed the pattern of 1500 > 0900 > 2100 > 0300 h, where soil temperature only differed between CC treatments at 1500 h (Figure 8b).

3.4. Multiple Regression Analyses

Multiple regression analyses were conducted to further assess the role of environmental parameters on GHG fluxes. Soil temperature and VWC explained 27, 1, and 10% of the overall variation in CO₂, CH₄, and N₂O fluxes, respectively (

Table 4. Summary of multiple regression analyses to predict carbon dioxide (CO₂; n = 1680, g m⁻² h⁻¹), methane (CH₄; n = 1680, mg m⁻² h⁻¹), and nitrous oxide (N₂O; n = 1680, mg m⁻² h⁻¹) fluxes from soil volumetric water content (VWC, cm³ cm⁻³) and soil temperature (Soil T, °C) measured in the top 6 cm combined across all cover crop treatment–measurement times of day combinations for the 2024 corn growing season at the Vayda Regenerative Farm near Clarksdale, MS.

Response Variable	Model Parameter	Model Parameter Coefficient (Standard Error)	Model Parameter Coefficient p-Value	% Explanation of the Total Sum of Squares	Overall Model p-Value	Overall Model R2	Adj. R2 †	RMSE †
CO2	VWC ***	0.88 (0.07)	<0.001	6.4	<0.01	0.27	0.27	0.28
	Soil T ***	0.03 (<0.01)	<0.001	23.6
	Intercept	−0.48 (0.04)	-	-
CH4	VWC	0.07 (0.02)	0.058	0.2	0.09	0.01	0.01	0.14
	Soil T	<0.01 (<0.01)	0.054	0.1
	Intercept	−0.04 (0.02)	-	-
N2O	VWC ***	1.54 (0.11)	<0.001	9.1	<0.01	0.10	0.10	0.45
	Soil T ***	0.01 (<0.01)	<0.001	2.0
	Intercept	−0.49 (0.07)	-	-

^† Adj. R², adjusted R²; RMSE, root mean square error. *** p < 0.01.

). However, only the multiple regression models for CO₂ and N₂O fluxes were significant (p < 0.01), and in both models, both soil temperature and VWC significantly (p < 0.01) explain the gas flux variability over the course of the season (Table 4).

Figure 8. Soil temperature measured in the top 6 cm, averaged across cover crop treatments, among measrement times of day (i.e., 0300, 0900, 1500, and 2100 h) over time (panel (a)) and among CC treatment–measurement time of day combinations (panel (b)) averaged across measurement dates during the 2024 corn growing season at the Vayda Regenerative Farm near Clarksdale, MS. For the top panel, the standard error extracted from Tukey’s mean separation analysis was 0.23 °C. For the bottom panel, different letters on top of bars denote a significant difference at the 0.05 level.

Figure 8. Soil temperature measured in the top 6 cm, averaged across cover crop treatments, among measrement times of day (i.e., 0300, 0900, 1500, and 2100 h) over time (panel (a)) and among CC treatment–measurement time of day combinations (panel (b)) averaged across measurement dates during the 2024 corn growing season at the Vayda Regenerative Farm near Clarksdale, MS. For the top panel, the standard error extracted from Tukey’s mean separation analysis was 0.23 °C. For the bottom panel, different letters on top of bars denote a significant difference at the 0.05 level.

For the CO₂ flux model, soil temperature represented 23.6% of the total sum of squares, partially explaining the strong visual correlation and similarities between CO₂ fluxes and soil temperature variations across treatment combinations (Table 4; Figure 2, Figure 3 and Figure 8). The root mean squared error (RMSE) for the CO₂ flux multiple regression model was 0.28 g m⁻² h⁻¹, which was numerically lower than the standard deviation (SD) calculated across all hourly fluxes measured over the season (0.32 g m⁻² h⁻¹), substantiating the validity of the resulting model (Table 4). The CO₂ flux model indicated an intensification in soil respiration as soil temperature and VWC increased (Table 4). The model parameter coefficients indicated an increase of 1 g CO₂ m⁻² h⁻¹ for each 0.03 °C increase in soil temperature and 0.88 cm³ cm⁻³ increase in VWC, which reflects how soil respiration was more sensitive to changes in soil temperature than VWC in the current study (Table 4).

For the N₂O flux model, soil VWC and soil temperature represented only 9.1 and 2.0% of the total sums of squares, respectively, indicating that more N₂O flux variance was attributed to changes in soil VWC than temperature (Table 4). The RMSE for the N₂O flux multiple regression model was 0.45 mg m⁻² h⁻¹, which was slightly lower than the SD calculated across all hourly fluxes measured over the season (0.47 mg m⁻² h⁻¹), suggesting the presence of other explanatory variables not measured in this study. Consequently, consideration of including additional soil and environmental parameters should be made for the multiple regression analysis for N₂O fluxes (Table 4). The parameter coefficients indicated an increase of 1 mg N₂O m⁻² h⁻¹ for each 0.01 °C increase in soil temperature and a 1.54 cm³ cm⁻³ increase in VWC (Table 4).

3.5. Season-Long Emissions, Emissions Intensity, and GWPs

Season-long emissions were calculated by linear interpolation between measured hourly fluxes among treatment combinations and, thus, represent an estimation of the actual net GHG release. Season-long CO₂ emissions do not constitute a measure of the net ecosystem balance but the overall estimation of soil respiration. Among GHG-emissions-related response variables tested, only season-long CO₂ emissions and EI-GWP* differed (p < 0.05) between CC treatments (

Table 5. Summary of the effect of cover crop treatment on gas emissions and emissions intensity (EI-) for carbon dioxide (CO₂), methane (CH₄), nitrous oxide (N₂O), global warming potentials calculated according to the conversion factors reported in the Intergovernmental Panel on Climate Change 6th (CO₂ = 1, CH₄ = 28, and N₂O = 265) assessment (2021) where CO_2, CH₄, and N₂O were included (GWP) and where only CH₄ and N₂O were included (GWP*), and corn yield from the 2024 corn growing season at the Vayda Regenerative Farm near Clarksdale, MS.

Parameter	Cover Crop	No Cover Crop
CO2 (Mg ha−1 season−1)	15.50 a †	11.94 b
CH4 (kg ha−1 season−1)	0.30 a	−0.22 a
N2O (kg ha−1 season−1)	7.28 a	5.53 a
GWP (kg CO2 eq. ha−1 season−1)	15,053 a	12,888 a
GWP* (kg CO2 eq. ha−1 season−1)	1936 a	1460 a
Yield (Mg ha−1 season−1)	11.7 a	10.5 a
EI-CO2 [kg (Mg yield−1)]	1.29 a	0.99 a
EI-CH4 [kg (Mg yield−1)]	0.03 a	−0.02 a
EI-N2O kg (Mg yield−1)]	0.68 a	0.47 a
EI-GWP [kg CO2 eq. (Mg yield−1)]	1473 a	1113 a
EI-GWP* [kg CO2 eq. (Mg yield−1)]	181.8 a	124.7 b

^† Different letters next to means in a row are different at p < 0.05.

). Soil respiration was ~30% greater from CC than no-CC (Table 5). In contrast, season-long CH₄ and N₂O emissions were unaffected by CC treatment and ranged from −0.22 under no-CC to 0.30 kg CH₄ ha⁻¹ season⁻¹ under CC and from 5.53 under no-CC to 7.28 kg N₂O ha⁻¹ season⁻¹ under CC, respectively (Table 5). Despite not differing significantly, GWP, GWP* yield, EI-CO₂, EI-CH₄, EI-N₂O, and EI-GWP were all numerically greater under CC than no-CC (Table 5).

Corn yield in Mississippi averaged 11.9 Mg ha⁻¹ in 2024 [42], which was only slightly greater than the yields measured in the current study (Table 5). From both CC treatments, CO₂ emissions represented >90% of the season-long GWP, while N₂O emissions represented >99% of GWP* (Table 5). The numeric differences in season-long N₂O and CH₄ emissions between CC treatments resulted in a 45% greater EI-GWP* for CC compared to no-CC, likely due to the numeric contribution of N₂O emissions that represented >90% of the EI-GWP* in both CC treatments (Table 5).

4. Discussion

Environmental conditions, particularly moisture inputs to control VWC, played a large role in resulting GHG emissions. Over the entire course of the 2024 growing season, rainfall near the study area was 35% below average (Table 2). In contrast to the conditions during the current study, drought periods in northwest Mississippi commonly occur late in the summer and early in the fall. However, when irrigation applications over the course of the growing season (i.e., 23.5 cm ha⁻¹) were added to the 2024 local rainfall, the overall water amount potentially available for plant growth was slightly above the 30-year regional mean (Table 2). Climatic variation at the regional and local scale can have a profound effect on GHG production and release, and characterization of the local weather conditions experienced provides essential information to contextualize GHG results.

4.1. CO₂ Fluxes

The role of air and near-surface soil temperature on soil respiration has been widely evaluated in laboratory and field studies, but CO₂ fluxes have rarely been measured at different times of the day, regardless of measurement technique [43]. In a study conducted in the Loess Plateau in China, Liu et al. [44] measured CO₂ fluxes every two hours over a period of 24 h from a corn field on a silty clay loam and reported similar diurnal trends as in the current study. In the current study, narrower and less frequent CO₂ flux fluctuations were measured at 0900, 0300, and 2100 h, likely due to lower solar radiation reaching the soil surface, highlighting the predominant role of soil temperature in regulating soil respiration (Figure 2).

Similar to that hypothesized, wider CO₂ flux fluctuations occurred at 1500 h compared to the other measurement times, likely due to greater variations in soil temperature that sparked microbial and root respiration (Figure 1) [43]. During the first half of the growing season (i.e., DOY 126 to 176), relative CO₂ flux peaks and lows were recorded at 7- to 10-day intervals, suggesting that irrigation applied on a weekly schedule likely played a relevant role in CO₂ production and release (Figure 2). Additionally, the greater soil temperatures in the mid-afternoon (i.e., 1500 h) each day likely contributed to substantial variations in VWC that could have resulted in an enhanced CO₂ release. Large VWC fluctuations, which commonly occur during wet and dry cycles, can cause a burst of CO₂ production in what is known as the Birch effect [45]. Furthermore, results clearly indicated large CO₂ flux variability across the four measurement times of day, suggesting that measurement protocols with only one measurement per day, on a weekly or monthly basis, should avoid early afternoon hours when large air temperatures and solar radiation can enhance soil respiration to a rate that is not representative of the daily average CO₂ flux [46]. Contrary to that hypothesized, CO₂ fluxes did not peak earlier under CC, as no significant difference between CC treatments across measurement dates occurred in the current study, which was likely related to the large C:N ratio of the cereal rye CC residue that did not provide immediate substrate availability (Figure 2) [47].

However, somewhat supportive of that hypothesized, when CO₂ fluxes were averaged across measurement dates, greater CO₂ flux from CC at each measurement time was observed (Figure 2). Although CC use is considered a climate-smart practice, the role of the CC plants themselves during and after the cash-crop season can be substantially different. During CC vegetative growth, plant competition for nutrients can limit microbial activity, reducing GHG production [14]. Once terminated, the CC could be considered an organic soil amendment, closely resembling the impact that organic additions can have on the microbial community. Greater rates of nutrient mineralization and immobilization with the addition of organic amendments have been reported in laboratory incubator studies [48] and from farming systems in South Africa [49], Kansas [50,51], and Pennsylvania [52], although rates of mineralization and immobilization were related to residue quality, particularly C:N [48,49,50,52].

4.2. CH₄ Fluxes

Methane fluxes measured in the current study closely matched ranges and trends reported by Hernandez-Ramirez et al. [53] in a study conducted in corn on silt loam and silty clay soils under various management practices where GHGs were assessed. Hernandez-Ramirez et al. [53] reported that the corn systems assessed oscillated from a minor CH₄ source to a small sink, as observed in the current study (Figure 4). As hypothesized, CH₄ fluxes peaked earlier and with greater magnitude under CC than no-CC, but, contrary to that hypothesized, no CH₄ flux difference occurred across measurement times (i.e., dates, Figure 4; Table 3). The two large CH₄ flux peaks from CC were likely related to the two largest rain events during the growing season that occurred in a timespan of less than 24 h [54]. While the rainfall events provided enough water to at least temporarily approach saturated soil conditions, the presence of CC residues favored the formation of soil-microsite anaerobiosis through limited evaporation, enhancing CH₄ production (Figure 4) [55,56]. The lack of CC residues in the no-CC treatment resulted in no surface residue cover to limit evaporation, thus likely not achieving proper reducing conditions for CH₄ production (Figure 4) [55,56]. Field research has shown that, similar to the current study, corn production systems often behave as small CH₄ sinks, even when CC is present [57].

4.3. N₂O Fluxes

Contrary to that hypothesized, N₂O fluxes from CC did not peak earlier than from no-CC, but, as hypothesized, the CC generally had greater N₂O fluxes than no-CC, particularly at 1500 h (Figure 5 and Figure 6). Wet and dry cycles at the beginning of the season created soil disturbances that likely forced rapid and substantial reactions from the microbial community. Laboratory studies have reported N₂O flux peaks when the soil surface experiences rapid water infiltration followed by rapid drainage where air enters the pore space, allowing N₂O produced during the soil wetting stage to escape the pedosphere [58]. While denitrification rates commonly increase during the soil rewetting, gaseous N can remain trapped in the topsoil due to the water-filled pore space, and when pore water evaporates, a N₂O flux spike can occur [58,59]. The soil hysteresis phenomena likely contributed to the lower frequency of N₂O flux spikes from wet–dry cycles during the second half of the growing season when the addition of irrigation water and rainfall did not result in as large of a VWC change as occurred during the first half of the growing season (Figure 5 and Figure 6) [58]. The concentration of significantly positive N₂O fluxes at 1500 h reinforces the importance of appropriate measurement time selection for the commonly used weekly GHG measurements, as measurement times in the middle of the day can lead to misrepresentation of the average environmental conditions and result in distorted interpretations and conclusions [60].

4.4. Soil Volumetric Water Content and Soil Temperature

The presence of CC residues in the current study not only allowed for greater soil VWC but, as hypothesized, contributed to greater VWC for a longer period of time, affecting the GHG-producing environment (Figure 7). The role of VWC in regulating GHGs has been highlighted in various studies and under various cropping systems [61,62,63]. However, monitoring near-surface soil temperature at regular, daily intervals showed a large range of soil temperatures experienced during corn production in Mississippi, highlighting that dynamic environmental conditions of furrow-irrigated systems are not strictly attributed to VWC but also to soil heating and cooling cycles (Figure 8) [60].

4.5. Multiple Regression

Few studies have assessed the relative contributions of VWC and soil temperature in explaining GHG flux variations in corn systems in Mississippi, making comparison with the current study difficult. In a study conducted in furrow-irrigated rice on a silt loam in east-central Arkansas, Della Lunga et al. [60] reported positive significant correlations between CO₂ fluxes and VWC and soil temperature, although the correlation coefficient highlighted a greater dependency of soil respiration on VWC than on soil temperature, likely due to the soil conditions that approximated saturation at numerous instances during the growing season. Similar to CO₂, the multiple regression model for N₂O fluxes suggested an increase in the coupled nitrification–denitrification processes as soil VWC and soil temperature increased (Table 4). Considering the fluctuations of soil VWC, nitrification likely provided the nitrate (NO₃⁻) substrate during the dry phase, which was subsequently reduced to N₂O by denitrification during the wet phase (Figure 7; Table 4).

4.6. Season-Long Emissions, Emissions Intensity, and GWPs

The lack of difference in CH₄ and N₂O emissions, EI, GWP, yield, and EI-GWP between CC treatments stood in contrast to that hypothesized and, thus, was likely related to having only two replicate GHG observations per CC treatment that reduced the ability and statistical power to properly characterize the true variance between CC treatments. Although two replicates represent the minimum to calculate the variance [64], the power to detect significant differences was greatly reduced in the current study, and influential values likely affected results and conclusions. Future studies should include a greater number of replicates to better characterize CC effects on GHG flux and emissions in any natural or agroecosystem. However, greater replication has to be balanced with the increased costs associated with more long-term GHG chambers, which is not a trivial investment.

Season-long CO₂ emissions measured in the current study were similar to the 3-year average soil respiration rates reported by Drury et al. [13] in corn field trials in Ontario, Canada, but were substantially greater than soil respiration rates reported from US corn production systems with CC residues [65] and without residue [66]. The greater season-long CO₂ emissions of the current study may be related to the greater gas measurement frequency, highlighting how weekly measurements commonly used in GHG assessments represent a potential underestimation of actual CO₂ emissions. Season-long CH₄ emissions were negligible in the current study, reinforcing that upland crop production systems are often characterized by aerobic conditions, where methanogenesis is restricted to only sporadic, temporally limited episodes.

Studies on CC effects on N₂O emissions during the cash-crop season reported inconsistent results depending on CC type, termination time, management practices, and environmental conditions [29]. In a study conducted over two years in Piacenza, Italy, Fiorini et al. [67] reported N₂O emissions from no-tillage corn with winter rye as a cover crop (6.94 kg N₂O-N ha⁻¹ season⁻¹) that were almost three times greater than in the current study. In contrast, a study conducted in the United Kingdom in corn on silt loam, where rye residues were incorporated as green manure, reported N₂O emissions lower (0.9 kg N₂O ha⁻¹ season⁻¹) than in the current study [68]. A study conducted over one calendar year in New York reported cumulative N₂O emissions between 18.1 and 21.0 kg N₂O ha⁻¹ season⁻¹ from corn fields with a winter clover cover crop that were greater than season-long N₂O emissions in the current study (Table 5 [69]. In contrast, Abdalla et al. [70] reported that a CC mix with a high C:N ratio does not impact gaseous-N losses during the cash-crop season from various crop varieties and under different management. The CC effect on gaseous-N losses has often been evaluated during CC vegetative growth, coupled with the cash-crop season after CC termination, making a direct comparison with the current study difficult [71].

The GWP and GWP* reported in the current study were almost five and three times greater than the net GWP calculated by Adviento-Borbe et al. [72] and the GWP* calculated by Doberman et al. [73] in studies conducted in Nebraska in a continuous corn system on a silty clay loam over 6 years. The variety of methods used in scientific research to estimate, calculate, and model GWP from agricultural systems increases the difficulty of properly comparing results [74]. The results of the current study are to be considered preliminary, and further research should investigate the long-term effect of CC on GHGs. Additionally, the results reported in the current study need to be contextualized within site-specific parameters, such as environmental and weather conditions. Finally, the limited number of replicates (i.e., 2) per treatment might have skewed the trend of the data toward specific directions, resulting in the potential inflation of significant differences among fixed factors of the study. However, although results from the current study have the limitation of being restricted to just one cash-crop growing season, these results provide valuable information that can be used for further comparisons, particularly given the large, simultaneously measured, three-gas dataset generated and evaluated in the current study.

5. Conclusions

Corn is a major summer cash crop in the LMRV, as well as in other major row-crop-producing regions in the US. Thus, corn systems in the LMRV play a significant role in the GHG footprint and potential climate-change mitigation strategies. Cover crops have been identified as a climate-smart practice that can reduce GHG emissions and potentially GWP. However, studies evaluating CC effects on GHGs in corn production in the LMRV have been limited, particularly those using an automated, long-term, in-field measurement system and under a mixed CC scenario involving both legumes and non-legumes. This study aimed to assess the effects of CC treatment (i.e., CC and no-CC), measurement date (DOY), measurement time of the day (i.e., 0300, 0900, 1500, and 2100 h), and their interaction on CO₂, CH₄, and N₂O fluxes, emissions, GWP, EI, VWC, and soil temperature in a furrow-irrigated corn production system on a loam in the LMRV.

Results did not support the hypothesis that CC would have earlier peak GHG fluxes but partially supported the hypothesis that GHG fluxes at 1500 h would be greater in magnitude compared to the other three measurement times, particularly for CO₂ and N₂O. Results did not support the hypothesis that GHG fluxes would be lowest at 0300 h, as no difference in GHG fluxes occurred among 0300, 0900, and 2100 h. Results supported the hypothesis that soil VWC and/or soil temperature would experience greater fluctuations under no-CC due to the lack of surface residue to limit solar radiation and slow infiltration and drainage. Results did not support the hypothesis that GHG fluxes, season-long emissions, GWPs, and yield would be greater with CC than without CC. Results supported the hypothesis that EIs would not differ between CC treatments.

The high temporal resolution of GHG measurements in the current study highlighted the variable nature related to plant–soil–microbial interactions. Beneficial CC effects on soil physical and chemical properties likely outweigh an expected increase in GHG emissions and/or GWP. Future studies should evaluate longer-term CC effects as a climate-smart practice in corn systems to develop mitigation recommendations for the US agricultural sector and beyond.