Net greenhouse gas balance with cover crops in semi-arid irrigated cropping systems

Climate smart agriculture has been emphasized for mitigating anthropogenic greenhouse gas (GHG) emissions, yet the mitigation potential of individual management practices remain largely unexplored in semi-arid cropping systems. This study evaluated the effects of different winter cover crop mixtures on CO2 and N2O emissions, net GHG balance (GHGnet), greenhouse gas intensity (GHGI), yield-scaled GHG emissions, and soil properties in irrigated forage corn (Zea mays L.) and sorghum (Sorghum bicolor L. Moench) rotations. Four cover crop treatments: (1) grasses, brassicas, and legumes mixture (GBL), (2) grasses and brassicas mixture (GB), (3) grasses and legumes mixture (GL), and (4) a no-cover crop (NCC) control, each replicated four times under corn and sorghum phase of the rotations, were tested in the semi-arid Southern Great Plains of USA. Results showed 5–10 times higher soil respiration with cover crop mixtures than NCC during the cover crop phase and no difference during the cash crop phase. The average N2O-N emission in NCC was 44% lower than GL and 77% lower than GBL in corn and sorghum rotations. Cash crop yield was 13–30% greater in cover crop treatments than NCC, but treatment effects were not observed for GHGnet, yield-scaled emissions, and GHGI. Integrating cover crops could be a climate smart strategy for forage production in irrigated semi-arid agroecosystems.

The Paris Climate Agreement in 2015 aimed to limit global warming by holding global average temperature rise below 2 °C by 2100 compared to pre-industrial levels 1 . The agricultural sector contributes to global warming, emitting 10-12% of the total anthropogenic GHG emissions 2,3 . This number could increase to 20-25% if emissions from land-use change are included, and up to 34% if up and downstream products are included [4][5][6] . In addition, the agriculture sector accounts for approximately 56% of the total anthropogenic non-CO 2 GHG emissions globally 6,7 . Soil emits about ten times more CO 2 than burning fossil fuels, but the soil emission is roughly balanced by a similar amount of C fixation from photosynthesis and it also has the largest reservoir of carbon (C) in terrestrial ecosystems 8,9 . Although GHG emissions via natural processes are inevitable, a substantial reduction in anthropogenic GHG emissions is possible through agricultural innovations 10 . About 60% of the global anthropogenic N 2 O emissions come from agriculture, primarily through synthetic fertilizer and manure application 3 . Although agricultural soils are considered a source of N 2 O on annual or greater time scales, some studies suggest soils can be a sink if climate smart agricultural practices are adopted [11][12][13] . Studies have reported up to 116 g ha −1 day −1 atmospheric N 2 O uptake in nitrogen-limited conditions 12 .
Cover cropping is considered a climate smart strategy to enhance soil health and mitigate global warming because they photosynthetically capture atmospheric CO 2 -C and ultimately store a portion in soils 14 . A metaanalysis highlighted the potential of cover cropping to reduce agricultural GHG by 8% while increasing SOC sequestration by 0.12 Pg C per year 15 . The GHG mitigation potential of cover crops is centered on C sequestration, reduction in fertilizer use with legume cover crops, change in albedo, and enhancing agroecosystem resiliency through a wide range of ecosystem services benefits 14,16 . Besides, cover crops suppress weeds and insect pests and capture residual inorganic nitrogen (N) to prevent it from leaching [17][18][19] . They also reduce soil denitrification potential by scavenging residual N after crop harvest and can reduce GHG emissions by decreasing soil temperature through canopy or residue cover 20,21 . However, the moisture saving due to a mulching effect of cover crop residues could favor CO 2 and N 2 O emissions. This is because more soil water and increased microbial

Results
Average daily and cumulative greenhouse gas emissions. Soil CO 2 -C emissions in cover crop-forage corn rotation were consistently higher when the cover crop was present and during the cash crop growing period (Fig. 1A,F). During the cover crop growing period, the difference in CO 2 -C emissions between cover crop treatments and NCC was visible in later growth stages, i.e., from February to April. Soil CO 2 -C emissions, without accounting for SOC inputs from cover crop above-and below-ground biomass addition, in cover crop phases, were 10.3-10.8 times greater in cover crop treatments than the NCC (Table 1). However, CO 2 -C emissions did not vary among treatments during the corn phase of the rotation (Supplementary Table S2). Among phases in cover crop-corn rotation, the greater average fluxes were observed in the corn phase than in the cover crop phase in both years, with 1.6-2.0 times greater emissions in cover crop treatments and 21 times greater emissions in NCC treatment ( Table 1). The cumulative CO 2 -C emission under cover crop treatments was 1.27-1.38 times higher than NCC ( Fig. 2A). Among years, the average CO 2 -C emission across treatments and sampling dates was 1.98 times higher in 2019/20 than in 2018/19 in the cover crop phase, and no difference between years in the corn phase of the rotation.
In the cover crop-forage sorghum rotation, the higher soil CO 2 -C fluxes were observed when the cover crop was actively growing (spring season) and during the cash crop phase (Fig. 1F). The highest fluxes were observed in the sorghum growth phase, irrespective of the crop year. The average soil CO 2 -C emission across years in the cover crop phase was 5.38-7.65 times higher in cover crop treatments than in NCC, while emissions from GBL Figure 1. Trend of average CO 2 and N 2 O emissions, water-filled pore space, soil temperature, and air temperature under cover crop treatments during forage corn and sorghum production from 2018 to 2020. (A) Average CO 2 -C emission (kg ha −1 day −1 ), (B) average N 2 O-N emission (g ha −1 day −1 , (C) water-filled pore space (%), (D) soil temperature (°C), and (E) air temperature (°C) in cover crop-corn rotation, and (F) average CO 2 -C emission, (G) average N 2 O-N emission, (H) water-filled pore space (%), (I) soil temperature (°C), and (J) air temperature (°C) in cover crop-sorghum rotation. NCC no cover crops control, GBL grasses, brassicas, and legumes mixture, GB grasses and brassicas mixture, GL grasses and legumes mixture. www.nature.com/scientificreports/ were higher than GB and GL treatments (Table 1). During the sorghum growth phase, no significant differences were observed among treatments, with CO 2 -C emissions ranging between 85.7 to 113 kg CO 2 -C ha −1 day −1 . Also, the cumulative CO 2 -C emission for the entire cover crop-sorghum rotation period did not differ among treatments ( Fig. 2A). Between study years, average soil CO 2 -C emission was 1.76 times higher in 2019/20 than in 2018/19 during the cover crop phase, while no difference was observed during the sorghum phase of the cover crop-sorghum rotation. Averaged across the study years, soil N 2 O-N emissions were inconsistent in the cover crop and corn phases (Fig. 1B). The emissions were negative during the cover crop phase and positive during the main crop phase, with the values between − 10.2 and − 3.42 g N 2 O-N ha −1 day −1 in the cover crop phase and 9.18 to 18.3 g ha −1 day −1 in the corn phase ( Table 1). The average N 2 O-N emission during the main crop phase in the GL mixture was significantly greater (79-99%) than NCC and GBL. The cumulative N 2 O-N emissions in the cover crop-corn rotation were not different among treatments and ranged from − 0.43 to 1.43 kg ha −1 (Fig. 2B). Between study years, soil N 2 O-N emissions were 127% and 78% greater during cover crop and corn growth phases, respectively, in 2019/20 than the respective phases in 2018/19. Soil N 2 O-N emissions in the cover crop-sorghum rotation varied during both phases of the crop rotation and years (Fig. 1G). The average emissions were negative during the cover crop phase and positive during the main crop phase. Also, the emissions were mostly negative in 2018/19 compared to 2019/20 during the cover crop growing period. There was a high variation in N 2 O-N emissions among treatments during the sorghum phase, with average emissions 4.48 times higher in GBL mixture than in NCC but similar to GB and GL in the sorghum phase (p = 0.06) ( Table 1). The emissions during the sorghum phase were 1.64 times higher in 2019/20 than in 2018/19. The cumulative N 2 O-N emissions during the cover crop-sorghum period did not differ among treatments, and the values ranged between − 0.73 and 5.08 kg N 2 O-N ha −1 (Fig. 2B). in the cover crop phase was greater in NCC than in cover crops, except during their early growth phase in 2018/19, and the response was opposite during the corn phase (Fig. 1C). The trend of average WFPS varied among treatments and with cropping phases in both years. Average WFPS in NCC, GBL, GB, and GL treatments in the cover crop phase was 41.0%, 28.5%, 28.3%, and 29.5%, respectively, whereas they were 50.5%, 52.7%, 48.0%, and 52.4%, respectively in the corn phase of the rotation. The average soil and air temperature trends were similar among all treatments (Fig. 1D In cover crop-sorghum rotation, WFPS at 0-0.10 m depth in the cover crop phase was higher in NCC than in other treatments, specifically during their later growth period and after termination (April to May), and the result was reversed during the sorghum phase ( Fig. 1H) in both years. Average WFPS in NCC, GBL, GB, and GL treatments were 48.9%, 36.1%, 34.3%, and 34.0%, respectively, in the cover crop phase, while it was 59.0%, 63.8%, 58.6%, and 60.6%, respectively, in the sorghum phase. Soil and air temperatures measured during GHG monitoring were similar among treatments (Fig. 1I,J). In 2018/19, the average soil temperature ranged between 8.7 to 42 °C, while in 2019/20, it was between 6.8 and 31.7 °C. Correspondingly, average air temperatures ranged between 12.7 and 45.2 °C in 2018/19 and 0.6 °C to 42.2 °C in 2019/20.

Soil properties.
The effect of cover crops on SOC min in laboratory incubation was similar to CO 2 -C emission in the field. In cover crop-forage corn rotation, SOC min measured at cover crop termination showed significant differences among treatments, with 134-147% greater SOC min in cover crops than in NCC ( Table 2,  Supplementary Table S3). Conversely, it was comparable among cover crops and NCC at forage corn harvest. Regardless of treatments, SOC min in 2019/20 was 43% greater during cover crop phase than in 2018/19. In cover crop-forage sorghum rotation, SOC min at cover crop termination was 82-125% greater under cover crops than under NCC, while there was no difference among treatments at sorghum harvest.
Soil inorganic N (NH 4 + + NO 3 − ) also varied significantly among treatments at cover crop termination ( Table 2). It was 91-158% and 59-131% higher in NCC than in the cover crop mixtures in cover crop-corn and cover crop-sorghum rotation, respectively. However, no significant treatment differences were observed at cash crop harvest, irrespective of the crop rotations. Comparing study years, inorganic N content was 1.61 times higher in 2018/19 than in 2019/20 at cover crop termination before corn planting, and it was 3.61 times higher in 2019/20 than in 2018/19 at corn harvest. Averaged across treatments, inorganic N was 45% lower in 2019/20 than in 2018/19 at cover crop termination time before sorghum planting and 291% higher in 2019/20 than in 2018/19 at sorghum harvest.
Crop yield, greenhouse gas balance, greenhouse gas intensity, and yield-scaled emissions. Corn biomass yield varied only at p = 0.079, where NCC treatment yielded 18.6% lower than GL but comparable to GBL and GB treatments ( Table 3, Supplementary Table S4). Sorghum biomass yield in NCC was 18.3-23.0% lower than GBL and GB treatments, while it did not differ from GL treatment. Corn yield was 12.6% higher in 2019/20 compared to 2018/19, while there was no difference in sorghum yield between study years. Average aboveground residue C input from cover crop biomass was similar for both rotations and ranged between 2.0 and 2.4 Mg ha −1 . In the corn system, C input through cover crops was 17.1% higher in 2019/20 Table 2. Average soil organic C mineralization (SOC min ) from incubation study and inorganic N (0-0.10 m) under diverse cover crop treatments. NCC no cover crops control, GBL grasses, brassicas, and legumes mixture, GB grasses and brassicas mixture, GL grasses and legumes mixture. ‡ Mean values (± standard error) followed by different lowercase letters in a column indicate significant differences among cover crop treatments or between years (p ≤ 0.05, LSD test   Table S5). In contrast, the regression analysis between CO 2 -C emissions and environmental factors did not show significant relationships ( Supplementary Fig. S2). Soil CO 2 -C emissions were positively correlated with soil temperature (r = 0.27, p < 0.0001) and air temperature (r = 0.18, p = 0.011).

Discussion
Cover crop integration in cropping systems can affect C and N dynamics by improving the diversity and size of the soil microbiome, modifying the soil environment, and reducing the ecological footprint. Evaluation of GHG emissions, C and N inputs from cover crops, the main crop yields, and soil C and N components demonstrated that integrating winter cover crops in irrigated forage corn and sorghum production systems could improve agroecosystem C and N cycling without a significant difference in the net greenhouse gas balance. The average CO 2 emissions were higher with cover crops than NCC during the cover crop growth period (September to April) due to greater total soil (heterotrophic + root) respiration. Plant root respiration can account for 7-90% of the total soil respiration depending on their growth stage, vegetation type, soil, and climatic conditions [48][49][50] . However, microbial heterotrophic respiration and soil organic matter decomposition can also be high at the same time due to the rhizodeposition during the growth of the cover crops or the cash crops. It appears biomass recycling after cover crop termination balanced the GHG emissions during the cover crop phase. The CO 2 -C release during the cash crop phases of both rotations and SOC min did not differ among treatments, while cover cropping increased cash crop yields. Higher biomass production and recycling with comparable CO 2 -C fluxes at the system scale suggested the potential to increase SOC storage in the long-term with cover cropping. Aboveground biomass C Integrating cover crops in semi-arid cropping systems also demonstrated the potential to reduce the net N 2 O balance by acting as a sink of atmospheric N 2 O during cover crop growth. The cover crops utilize mineral N for their growth and prevent its loss to the atmosphere as N 2 O emissions. We observed N 2 O uptake in soil, mainly during the cover crop phase of both rotations, and more uptake in cover crop plots than NCC in cover crop-corn rotation. Mostly, the cover crop period of the first year was the sink, while the second year was the source of N 2 O-N emissions. This could be because the second-year cover crop period had residual mineral N from fertilizers applied during first-year cash crops and cover crop decomposition. In contrast, the first-year cover crop www.nature.com/scientificreports/ did not receive these N credits. The N 2 O uptake occurs when denitrifiers consume N 2 O as an electron acceptor for their respiration or when nitrifiers utilize N 2 O during nitrifier denitrification 12,13 . In N-deficient soils, N 2 O could be the only electron acceptor for complete denitrification leading to N 2 O-N uptake 11 . However, such a phenomenon does not occur when the soil is not N deprived. Nitrogen was not applied in the cover crop phase, but the subsequent cash crops received fertilizer and frequent irrigation. Abundant N in soil and wetting-drying phenomena generally increased N 2 O-N emissions, changing soil from sink to source of nitrous oxide 32 . Legume cover crops can also increase soil N content through N-fixation 29 . Since two of three cover crop mixtures contained legume species, they might have contributed to soil N-accumulation and partly to N 2 O emissions. The soils acted as a source during the cash crop phase when soil moisture was abundant, N was applied, and soil temperature was higher than in the cover cropping phase. A similar N 2 O emissions/uptake response to cover cropping was observed in a silty clay loam under a Mediterranean semi-arid climate that N fertilizer application right after cash crop planting and successive cover crop residue decomposition triggered N loss as nitrous oxide 13 . However, higher N 2 O-N emissions from GL treatment in forage corn and GBL in forage sorghum growing phase than NCC in our study showed a discrepancy in response of various cover crops and highlighted the need for further investigation on factors affecting N 2 O emissions. The N cycling was improved with cover cropping because N 2 O-N emissions from the NCC treatment were similar to GBL and GB in the corn phase and similar to GB and GL in the sorghum phase of the rotation despite a higher inorganic N under NCC. Nitrogen utilized by cover crops might have been recycled back during cash crop growth and contributed to better nutrient cycling than NCC, leading to higher crop yield in cover crop plots 40,41 . High inorganic N in NCC treatment did not support high forage yield because soil inorganic N content remained similar among treatments at harvest. Forage yield was lower under NCC than cover crop treatments. However, higher inorganic N availability in NCC than cover crop treatments improved forage quality, as shown by higher crude protein content 40 . Cover crops efficiently utilized N from fertilizer input and their residue mineralization than NCC, leading to no difference in net GHG emissions at the system scale.
Both CO 2 -C and N 2 O-N emissions were higher in the second year than in the first year at the system scale. This could be attributed to soil moisture availability, inter-annual climatic variability, and increasing residue input in the second year. The second-year cropping had less precipitation and higher summer temperatures than the first year. Hence, more irrigation was provided in the second year than in the first year resulting in total irrigation + precipitation of 810-mm and 1040-mm in the first and second years, respectively. Studies demonstrated positive linear relationships between soil moisture, temperature, and substrate availability with CO 2 and N 2 O emissions 28,32,45,51 . In this study, correlation analysis showed that CO 2 -C emissions positively related to soil and air temperature (Supplementary Table S5). In contrast, N 2 O-N emissions varied more with WFPS than with air and soil temperature suggested by both correlation and regression analysis (Fig. 3, Supplementary Table S5). Soil respiration surges with increasing soil temperatures but often declines after the temperature crosses 30 °C 46 . In this study, soil temperature measured at the time of GHG monitoring ranged between 6.8 and 42 °C in different seasons, suggesting a significant role of temperature in CO 2 emissions in various treatments. Therefore, the average CO 2 fluxes in this study were higher than studies by Sanz-Cobena et al. 13 and Mosier et al. 52 , while it was lower than or comparable with others 45,51 .
The N 2 O fluxes in this study had wide variations between crops and years. Soil releases N 2 O gas during both nitrification and denitrification 30,44 . Denitrification potential of soil increases with higher WFPS, usually above 60%, leading to greater N 2 O-N emissions, whereas aerobic nitrification dominates when WFPS falls between 30 and 60% 53 . In this study, WFPS was between 5 and 90%, indicating both nitrification and denitrification processes controlling N 2 O-N emissions. A significant positive correlation of N 2 O-N emission with WFPS, air temperature, and soil temperature also reflected the complex interaction between soil moisture, temperature, and N 2 O-N emissions. These interactions may have influenced CO 2 -C and N 2 O-N emissions differently. Unlike results reported in some studies (e.g., Guardia et al. 30 ), we did not observe correlations among CO 2 -C and N 2 O-N emissions, suggesting the need for further research on the role of environmental factors in regulating GHG emissions.
Cover cropping could be a climate smart strategy to improve soil health and increase crop production in arid and semi-arid irrigated cropping systems. In this study, cover crop mixtures and NCC had a similar environmental footprint for producing crops in semi-arid irrigated conditions indicated by similar GHG net and yield-scaled emissions. In contrast, cover crops increased forage yield by 13-30% over NCC. A higher yield in cover crops than NCC could be attributed to better nutrient cycling and water conservation 27,40 . Similar yield-scaled emissions among treatments were due to relatively low cash crop yield in NCC compared to cover crops, suggesting a positive relationship between GHG emissions and crop yield. The variability in N 2 O-N fluxes also affected yield-scaled N 2 O-N emissions. Specifically, yield-scaled N 2 O-N emissions were positive with cover crops and negative with NCC under cover crop-sorghum. Studies suggest that using legume cover crops can increase N 2 O-N emissions 28 . However, the response of GBL in cover crop-corn rotation, another legume integrated treatment in our study, does not support such argument. This could be due to the poor performance of legumes in the 2018/19 cropping year; the aboveground biomass of grasses and legumes in GL treatment had an average ratio of 94:6, while the biomass production of grasses, brassicas, and legumes in the GBL was in a 63:36:1 proportion. The GHGI was higher for NCC in both rotations: 13-35% higher than cover crop mixtures in the cover crop-corn rotation and 5-17% higher than GL and GB mixtures in the cover crop-sorghum rotation. Overall, while the response of different cover crop treatments was variable, this study suggested an increased environmental pressure of growing forage crops without cover crops than with cover crop integration. With the addition of organic C inputs through cover cropping, the SOC min was also increased, potentially improving microbial activity and soil biological health.
It is important to note that C and N loss from bare soils is solely from antecedent organic matter mineralization or fertilizer input. However, the gaseous loss of C and N from cover crop integrated systems could be contributed by above-and below-ground residue C-inputs, root activity, and root exudates 32  www.nature.com/scientificreports/ environmental cost of production remained similar between cover crops and NCC, a 13-30% increase in forage production with cover cropping encourages farmers to integrate cover crops into their cropping systems. This study did not measure CH 4 , which may change the net GHG balance. This warrants further studies to evaluate the climate change mitigation potential of cover cropping systems and their viability as a climate smart agricultural tool for semi-arid production systems. In addition, high spatial and temporal variability in the data also suggests the need for more research in semi-arid regions. Soil health and other ecosystem services benefits of cover cropping systems should also be accounted for while considering cover crops to mitigate rapid soil health degradation and fertility loss in arid and semi-arid regions.

Conclusions
Cover crop inclusion in forage cropping systems significantly increased CO 2 and N 2 O emissions and cash crop yield while they had no effects on GHG net , GHGI, and yield-scaled CO 2 and N 2 O emissions compared to NCC. Cover cropping did not necessarily reduce GHG emissions in semi-arid irrigated forage production systems. However, the yield benefits from cover crop plots compared to NCC demonstrate its potential as a climate smart strategy for arid and semi-arid agroecosystems. Soil and environmental factors (soil and air temperature and moisture) affected the relative impact of cover crops on CO 2 and N 2 O emissions. Compared to NCC, cover crops may have utilized residual N to prevent it from being lost in the environment and increased SOC min . Adopting such management practices along with no-tillage management could maintain soil health and support forage producers by increasing farm profitability. Considering the environmental footprints and crop yield potential, this study demonstrates the benefits of integrating cover crops in forage crop-fallow systems. However, more research on soil health, GHG emissions, and environmental variables are suggested to warrant the climate change mitigation potential of cover cropping systems in arid and semi-arid regions.

Materials and methods
The experimental site and treatments. The study was established on the Olton clay loam soil (fine, mixed, superactive, thermic Aridic Paleustolls) 54 at New Mexico State University Agricultural Science Center (ASC), Clovis, NM (34° 35′ 59′′ N, 103° 13′ 06′′ W, and elevation 1368 masl). The study area has a hot, dry, semiarid environment with an annual average maximum and minimum temperatures of 22.6 °C and 6.1 °C, respectively, and average yearly precipitation of 462 mm. The field was fallow for a year before establishing the study plots in September 2018. Baseline soil samples (0-0.1 m) had inorganic N 1.31 mg kg −1 , potentially mineralizable N 18.9 mg kg −1 , potentially mineralizable C by aerobic incubation (SOC min ) 141 mg kg −1 , soil pH in 1:1 soil water ratio 7.6, SOC 8.29 g kg −1 , total N 0.93 g kg −1 , and electrical conductivity 0.43 dS m −1 .
The study was conducted in a no-tillage corn and sorghum production system with winter fallow in rotation, fallow starting late September to early May of the subsequent year. Both corn and sorghum were present each year, and cover crops were planted to replace the winter fallow. Four cover crop treatments and four replications were arranged in a randomized complete block design. Treatments were cover crop mixtures of grasses, brassicas, and legumes (GBL), grasses and brassicas (GB), grasses and legumes (GL), and a fallow (no cover crop, NCC). Grasses included annual ryegrass (Lolium multiflorum Lam.) and winter triticale (Triticale hexaploid Lart.), brassicas included turnip (Brassica rapa subsp. rapa), and daikon radish (Raphanus sativus var. Longipinnatus), and legumes included pea (Pisum sativum subsp. arvense L.) and berseem clover (Trifolium alexandrinum L.). Individual plot size was 9.1 m × 12.2 m.
Cover crops were planted each year in mid-September using a double-disc drill opener (Model 3P600, Great Plains Manufacturing, Inc., Salina, KS, USA), maintaining 0.15-m row spacing and 0.02-m seeding depth. Seeding rates for cover crops varied among treatments (Supplementary Table S1). They were determined based on individual seed size and germination potential to maintain a comparable plant population for each species combination. All the cover crops were terminated using a mixture of herbicides, as described in Paye et al. 40 , and the residues were left on the ground.
Forage corn (Pioneer P1828AM, 61,776 plants ha −1 ) and sorghum (Mojo Seed OPAL, 123,553 plants ha −1 ) were planted in mid-May, about three weeks after cover crop termination using a John Deere planter (Deere and Company, Moline, IL, USA) adjusted to a row spacing of 0.76-m. Each year the field was fertilized with a single dose of N fertilizers (urea and ammonium nitrate) at 168.1 kg ha −1 , P (ammonium phosphate) at 42.0 kg ha −1 , S (ammonium sulfate) at 28.3 kg ha −1 , and Zn (chelated zinc) at 7.02 L ha −1 within 2 days of cash crop planting. Fertilizer rates were based on the recommended dose for irrigated corn and sorghum silage production adjusted with soil test recommendation in the first year, and the same rate was applied for the rest of the study years. Therefore, the N from cover crop residue mineralization was not accounted for in this study. Irrigation was uniformly provided using a center pivot system for all treatments based on the soil moisture content and crop need. Irrigation was provided only to facilitate cover crop planting, germination, and establishment during the cover crop phase. It was adjusted based on precipitation to meet the crop water needs during the corn and sorghum phases of the crop rotation. First-year (2018/19) cover crops and cash crops received 231 and 272 mm of precipitation, respectively, whereas the second year (2019/20) cover crops and cash crops received 294 and 144 mm of precipitation ( Supplementary Fig. S1). Therefore, the first-year cover crops and cash crops received 25-and 301-mm of irrigation, respectively, whereas the second-year cover crops and cash crops received 40-and 551-mm irrigation, respectively. In mid-September, cash crops were harvested using a pull-type forage harvester (Model 3960, John Deere, Moline, IL, USA) with an attached wagon, and biomass samples were collected from 4.57 m in length on two rows. All the forage samples were oven-dried at 65 °C until a constant weight to estimate the dry matter yield. where R is the gas emission rate (CO 2 or N 2 O flux in g m −2 h −1 ), G 0 is the gas concentration (CO 2 /N 2 O) at the time of gas chamber installation (T = 0), G n is the gas concentration at time T n (200 s), A is the area of soil exposed in m 2 , and V is the system volume in m 3 . The cumulative emission of CO 2 -C and N 2 O-N was estimated by linear interpolation of weekly/bi-weekly emission rates and numerical integration of individual data points. Hydraprobe SDI-12 (Stevens Water Monitoring Systems, Inc., Portland, OR, USA) was used to estimate air and soil temperatures and moisture content at the surface (0-0.10 m).
Soil sampling and laboratory analyses. Soil samples were collected from 0 to 0.10 m depth of each plot using a core sampler at the time of cover crop termination (mid-April) and cash crop harvest (mid-September) each year. Four cores were collected from each plot, mixed well, and composite subsamples of ~ 300-g were brought to the laboratory for estimating 72-h SOC mineralization (SOC min ) by aerobic incubation 55 and soil inorganic N by ammonia analysis method in a Timberline Instruments, Boulder, CO, USA. Three cores of diameter 0.023-m and 0-0.10 m in depth were collected and oven-dried for 24 h at 105 °C to determine the dry bulk density and gravimetric water content. Inorganic N and SOC min concentrations were converted to a volume-area-based unit using the bulk density values. The water-filled pore space (WFPS) was calculated using Eq. (2): where, θ g = gravimetric soil moisture (%), ρ b = bulk density (Mg m −3 ), and ρ s = particle density of 2.65 Mg m −3 .
Net greenhouse gas balance, greenhouse gas intensity, and yield-scaled emission. The GHG net from CO 2 and N 2 O was calculated by using Eq. (3) 56 below: where CO 2 eq. of farm operations included installation and use of the center pivot; farm inputs included production, transportation, storage, transfer, and application of fertilizers, pesticides, and herbicides; cash crop and cover crop planting and cash crop harvesting. The CO 2 eq. of farm operations and farm inputs were calculated using literature values from Lal 57 . Similarly, the heterotrophic respiration was calculated by multiplying measured soil respiration by 0.307 value 48 . The methane emissions were not estimated in this study. Agricultural soils in arid and semi-arid regions are often low emitters or serve as a small sink for CH 4 52,58 . The CO 2 equivalent of N 2 O emissions (310) was estimated based on Smith 7 on a 100-year timescale.
Cover crop biomass yield was estimated by hand-clipping biomass samples from four 0.25-m 2 areas (total 1-m 2 ) per experimental plot and oven drying at 65 °C for 72 h. Approximately 100-g subsamples were ground in a Thomas Wiley laboratory mill (Arthur H. Thomas Company, Swedesboro, NJ, USA) to pass through a 1-mm screen. The ground samples were analyzed for C content in a CN Analyzer (LECO Corporation, St. Joseph, MI, USA) using the dry combustion procedure. CO 2 eq. of residues returned from cover crops was calculated using biomass C estimate for each treatment. Since cash crops were harvested without leaving residues, we did not account for the C returned from the cash crops. GHG net was divided by the total annual forage yield (Mg ha −1 ) to calculate GHGI. Also, yield-scaled CO 2 -C and N 2 O-N were calculated by dividing the cumulative yearly CO 2 -C and N 2 O-N emissions by the annual forage yield of cash crops. Cover crop yield was not included in the calculation because they were chemically terminated and not harvested as forage.
GHG net Mg CO 2 eq. ha −1 year −1 = CO 2 eq. of farm operations + farm inputs + soil heterotrophic respiration www.nature.com/scientificreports/ Statistical analysis. All GHG, soil, and water data were analyzed separately for cash crops (forage corn and sorghum) and crop phase (cover crop and cash crop) using the Mixed procedure in SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). An unstructured covariance was used with Kenward-Rogers adjustment for the degrees of freedom (ddfm = kr). Crop yield, C-input from cover crops, yield-scaled CO 2 -C, and N 2 O-N, GHG net , and GHGI data were analyzed for cover crop-forage corn and cover crop-forage sorghum rotations considering treatment and years as fixed factors and rotation × year as a repeated term. The relationship between soil variables and GHG emissions was analyzed using simple regression analysis in SigmaPlot (V.15 Systat Software Inc., UK). Pearson's correlation coefficient was determined using the PROC CORR procedure in SAS. All the data were tested for normality of residuals and equality of variance. The non-normal data were log-transformed, and the back-transformed means were reported. The means were separated using Fisher's protected LSD at p ≤ 0.05 unless otherwise stated.
Statement for standard protocol. All methods were carried out in accordance with relevant guidelines and regulations of the United States Department of Agriculture and New Mexico State University. The study was conducted following the standard GHG emissions and soil and plant sampling protocol.

Data availability
All data generated or analyzed during this study are included in this published article and its Supplementary Information files.