Quantification of Nitrogen Lost to Greenhouse Gas Emissions in Cover Crop Systems

Final Report for GNC13-164

Project Type: Graduate Student
Funds awarded in 2013: $9,924.00
Projected End Date: 12/31/2014
Grant Recipient: Michigan State University
Region: North Central
State: Michigan
Graduate Student:
Faculty Advisor:
Dr. Dale Mutch
Michigan State University Extension
Expand All

Project Information


Nitrous oxide (N2O) emissions were measured in a cropping system which included fall-planted oilseed radish, annual ryegrass, and a radish + ryegrass mixture. No fertilizer was applied to the cover crops or the following corn crop. Cover crop fall and spring biomass production varied by site and year, as did N2O emissions. Although cover crop C:N ratios were large enough to cause soil N immobilization at one site, there was no effect on N2O emissions. Cumulative N2O-N emissions were lower than would be expected. Cover crop treatments had minimal effect on corn growth and yield.


Cover crop use provides multiple environmental and economic benefits (Snapp et al. 2005). Among other management strategies, Robertson and Vitousek (2009) advocate their use to improve nitrogen (N) fertilizer uptake into cropping systems. However, in order to increase N use efficiency (NUE), the correct amount of N must be applied at peak cash crop N demand while minimizing N loss to the environment (Ribaudo et al. 2011). In this regard fall-planted cover crops may be problematic because some cover crops which winter-kill (e.g., radish) may decompose and release N before the cash crop needs it while others which do not winter-kill (e.g., annual ryegrass) may cause N to be immobilized during peak cash crop N demand. Furthermore, the early release of N from winter-killed cover crops may provide an opportunity for N to leave the system. Research has focused primarily on the potential for this N to leach. N lost via the emission of greenhouse gases such as nitrous oxide has been far less researched.

A search of the SARE projects database reveals a similar lack of research in the area of greenhouse gas emissions and cover crops. Some SARE regions have received grants for “train the trainer” activities related to climate change (e.g. ENE05-091 “Climate change and agriculture: Preparing educators to promote practical and profitable responses”). This project complemented those efforts by providing data on which to base recommendations.

The SARE projects database includes two projects with direct and immediate relevance to this project. One is student project GNE10-005 “Balancing nitrogen sinks and sources using cover crops on manured fields”. My project differed in that: a) it included a mix of winter-hardy and non winter-hardy cover crops as well as pure stands of each cover crop type, b) gas sampling dates were chosen specifically to elucidate greenhouse gas emissions during the peak cover crop decomposition period, c) no fertilizer or manure was applied, and d) it also investigated the impact of the cover crops on the following cash crop.

The other relevant SARE project is GNC12-158 “Improving resource use efficiency through strip tillage, cover cropping, and deep fertilizer placement”. My project differed in two regards: its primary focus was on cover crops and greenhouse gas emissions, and the work was in an agronomic production system while GNC12-158 was in a vegetable production system. My project aimed to provide data on greenhouse gas (nitrous oxide) emissions in agricultural systems which incorporate fall-planted cover crops, and to characterize the growth of the cover crops and quantify their impact on the following corn crop.

Project Objectives:

This project had three objectives:

  1. Quantify the N lost to greenhouse gas emissions in fall-planted cover crops.
  2. Compare the greenhouse gas emissions in a system using a cover crop which winter-kills with one using a cover crop which over-winters and one with a mixture of the two cover crop types.
  3. Determine cover crop impact on the following cash crop.


Click linked name(s) to expand/collapse or show everyone's info
  • Dr. Dale Mutch


Materials and methods:

Site Description and Experimental Design

A two-year randomized complete block design experiment was initiated in fall 2012 to investigate nitrous oxide emissions in fall-planted cover crops. There were two study sites: one at the MSU Agronomy Farm (East Lansing, MI), and one at W.K. Kellogg Biological Station (KBS) (Hickory Corners, MI). Each site had four replications. Experimental units (treatment plots) were a minimum of 3 x 12 m. Site descriptions are listed in Table 1. Treatments included: bare fallow control, oilseed radish (planted at 11.2 kg ha-1), annual ryegrass (44.8 kg ha-1), and oilseed radish + annual ryegrass (11.2 kg ha-1+ 22.4 kg ha-1, respectively).

Field operation and data collection dates are listed in Table 2. Wheat was harvested in July of each year, then fields were treated with glyphosate to control weeds prior to cover crop planting. Cover crop treatments were planted each August. Cover crops were no-till drilled in East Lansing. At KBS, fields were tilled with a chisel plow prior to planting. In early spring glyphosate + 2,4-D ester was applied as per recommendations to control surviving cover crops prior to corn planting. The 102-d corn cultivar ‘DeKalb 52-59’ (Monsanto, St. Louis, MO) was planted on 76-cm rows at a population of 74,100 seeds per ha-1 using a four-row planter in spring 2013 and spring 2014. To elucidate the impact of the cover crops on the N status of the corn, no fertilizer was applied to the cover crops or the field corn. The last fertilizer input received by the plots came several months before cover crop planting, when the wheat was fertilized. No irrigation was applied.

Soil Sampling

To determine baseline site soil characteristics, a 2-cm diameter soil probe was used to collect 20 subsamples to a depth of 15-20 cm in each replicate of each field in the fall. The samples were stored at 4 °C until they were processed. They were ground to pass through a 2-mm mesh screen and then sent to the Michigan State University Soil and Plant Nutrient Laboratory to determine pH and soil organic matter content. Values were averaged to create a composite for each field-year. Soil samples were collected periodically throughout the study. Fall and spring soil nitrate (NO3) data during the cover crop phases were collected in Oct.-Nov. and Apr.-May, respectively. Post-corn harvest soil NO3 samples were collected in Nov.-Dec. The soil samples was sampled to a depth of 15-20 cm with a 2-cm diameter soil probe. Five samples were collected in an “H” pattern from each treatment plot. Samples were dried at 60° C and ground to pass through a 2-mm mesh screen. In 2012-2013, extractions were performed and extracts were analyzed for NO3-N and NH4-N as per Robertson et al. (1999). In 2013-2014, extracts were sent to the Michigan State University Soil and Plant Nutrient Laboratory for analysis. Using the same soil sample collection method as for the NO3 protocol above, subsamples were collected at corn V6-V8 for pre-side dress NO3-N testing (PSNT). Samples were aggregated by treatment, ground to pass through a 2-mm mesh screen, and sent to the Michigan State University Soil and Plant Nutrient Laboratory for NO3 analysis.

Gas Sampling

Gas samples were taken 10-12 times from fall through spring (per the advice of Dr. Neville Millar,Senior Research Associate at KBS researching greenhouse gas emissions in cropping systems). Starting in late Oct., a set of three greenhouse gas samples taken 2-3 weeks apart served as a baseline. The next set of samples was taken after oilseed radish winterkill, when the radishes were expected to be rapidly decomposing (during a warm spell when temperatures reach 5? C for several days). During this 10-14 day period 3-4 greenhouse gas samples were taken. Three more sampling dates 2-3 weeks apart after radish decomposition served as a spring baseline. Finally, samples were taken once per month during corn growth.

Round steel gas chambers with a diameter of 28.5 cm were installed to a depth of 5-cm in the soil after cover crop planting. Each chamber was centered over one cover crop row. Gas chambers were removed prior to tillage and corn planting, and immediately reinstalled between corn rows thereafter. The static chamber protocol was used to determine greenhouse gas fluxes (Holland et al., 1999). On each sampling date, chamber lids were installed and then headspace gas was immediately extracted using a 10-mL nylon syringe and a 23-gauge needle. At 20-min intervals over a 60-min period, three more samples were collected from each chamber. Gas samples were placed in 6 mL vials which had been flushed with 10-mL of extracted air. Each vial was filled with a 10-mL gas sample to over-pressurize it so as to avoid contamination and facilitate analysis. Soil temperature near each experimental unit was also collected, along with the depth of the chambers as measured at four points along the circumference of each chamber. Calculations to determine N2O (nitrous oxide) flux rates (µg N2O-N m-2 hour-1) were made using the following equation:

N2O = (α x V*WA*60)/(A*MVcorr)

where α represents the change in headspace N2O concentrations during the period when the chamber is closed, V is the chamber headspace volume, WA is the atomic mass of the N present in a molecule of N2O, 60 is a conversion factor from minutes to hours, A is the soil surface area covered by the chamber, and MVcorr corrects for temperature and pressure mole volume. N2O-N fluxes were then converted from µg N2O-N m-2 hour-1 to g N2O-N ha-1 day-1 for each sampling date. Samples were analyzed at the W.K. Kellogg Biological Station using an Agilent 7890A gas chromatograph (Agilent Industries, Inc.; Wilmington, DE) fitted with a 63Ni electron capture detector and a Gerstel MPS2XL autosampler (Gerstel; Linthicum, MD) (Kahmark and Millar, 2008). Interpolation was used to calculate cumulative N2O-N emissions for the fall, late winter, spring, and summer sampling periods, which were standardized to respective lengths of 29, 8, 26, and 36 d.

Cover Crop Biomass and C:N Ratios

Cover crop, weed, and volunteer wheat aboveground biomass was harvested from two randomly placed 0.25 m2 quadrats per treatment plot in Nov. and again the following Apr.-May. Biomass was separated into its component cover crop, weed, and wheat fractions before processing. Fall biomass data were collected at the point of peak biomass production before the onset of hard frosts. Spring biomass data were collected immediately prior to cover crop termination. Harvested plants were dried at 70° C for 5-10 d and then weighed. The dried cover crop biomass was ground using a Wiley mill (Thomas Scientific, Swedesboro, NJ). The C:N ratio was determined via combustion using a Costech Elemental Combustion System ECS 1040 (Costech Analytical Technologies, Inc., Valencia, CA) in 2012-2013. In 2013-2014, samples were sent to Midwest Laboratories, Inc. (Omaha, NE) for total C and N analysis.

Corn Growth and Yield

Corn heights were collected when corn was at the V6-V8 stage. A minimum of ten plants per experimental units were measured by holding the newest fully mature leaf up against a meter stick. Corn N status was determined after the onset of tasseling by collecting 25 corn ear leaves total (the leaf directly below the lowest ear of corn on each plant) from the two data rows in each treatment plot. Corn ear leaves were dried at 70 °C for 5-7 d and then ground with a Wiley mill (Thomas Scientific, Swedesboro, NJ). Two grams of this ground material were then sent to A&L Great Lakes Laboratories, Inc. (Fort Wayne, IN) for corn ear leaf N analysis. A Minolta SPAD-502 chlorophyll meter (Spectrum Technologies, Inc., Aurora, IL) was used to collect chlorophyll content data at corn stage VT. The meter was placed on the newest fully mature leaf of 15 plants in each experimental unit. SPAD values were also collected from nearby reference plots which had been fertilized as per PSNT testing recommendations. To determine corn grain yields, 9 m of each data row (outside of the area in which corn ear leaves were collected) were harvested using a two-row research combine. Yield was adjusted to 15.5% moisture.

Data Analysis

Data were analyzed using analysis of variance with PROC MIXED in SAS v. 9.3 (SAS Institute, Cary, NC) with a significance level of α ≤ 0.05. Interactions between site, year, and cover crop treatments were tested for significance. For most of the data sets, the three-way interaction (site*year*treatment) was significant while the two-way interactions (site*treatment and year*treatment) were not. For the sake of uniformity, we chose to combine data across years at each site. Once the data were tested for site, year, and treatment interactions, year and replicate were treated as random effects. Means were separated using Tukey’s test and the PDMIX 800 macro (Saxton, 1998). Soil nitrate values were converted from ppm to kg ha-1 using an average soil bulk density of 1.6 g cm-3 (Crum and Collins, 1995; USDA-NRCS, 2015).

In the fall radish + ryegrass treatments, the C:N ratio was calculated by taking a weighted average using the C:N ratios of the component cover crops and the percentage of dry aboveground biomass of each cover crop in each replicate. To normalize corn height data, corn heights were converted into a percentage of the control by dividing the heights of the plants in each treatment in each replicate by the average height of the corn in the corresponding replicate control treatment. Corn SPAD values were similarly normalized and were also compared to the reference plot SPAD values using one-sample t-tests to determine if values were significantly different.

Research results and discussion:

Cover Crop Fall Biomass

Dry aboveground cover crop biomass was collected each fall. The treatment*year interaction was not significant (P = 0.35), so data were combined across years at each site (Table 3). There were no significant differences in fall biomass production at KBS. At the low end of the range, annual ryegrass produced 1318 kg ha-1 while at the high end of the range, radish produced 1564 kg ha-1. At MSU, radish produced significantly more biomass in the fall than annual ryegrass (but not the mixture), with respective values of 2631, 1862, and 2460 kg ha-1. The cover crops produced more biomass at MSU than KBS.

Cover Crop Spring Biomass

There was a treatment*year interaction for spring cover crop dry aboveground biomass (P = 0.03). Year was treated as a random effect, and data were combined across years at each site (Table 3). The radish failed to overwinter as expected, so there was no radish biomass at either site in the spring. At KBS, annual ryegrass produced more biomass than radish + ryegrass. The same was true at MSU. Given the difference in seeding rates between the two treatments, this was not unexpected. In contrast to the fall trend, in the spring the cover crops produced more biomass at KBS than at MSU. This may be because the higher winter precipitation at KBS (Figures 1 – 4) provided insulation that protected the ryegrass from the cold.

Fall and Spring C:N Ratios and N Accumulation

Cover crop aboveground biomass was analyzed to determine treatment C:N ratios. The fall treatment*year interaction was not significant (P = 0.67), so data at each site were combined (Table 3). There were no differences at KBS, where the C:N ratios were 19:1, 23:1, and 22:1 respectively for the radish, annual ryegrass, and radish + ryegrass treatments. Nor were there differences at MSU, where the respective C:N ratios for radish, annual ryegrass, and radish + ryegrass were 13:1, 19:1, and 16:1. In spite of the lack of statistical significance, the C:N ratios seen at both sites followed the expected pattern. Radish had the lowest ratio, annual ryegrass the highest ratio, and the radish + ryegrass treatment was intermediate between the two. In the spring, the treatment*year interaction was significant (P = 0.03) for spring C:N ratio data. Year was treated as a random effect and data were combined at each site (Table 3). At KBS, the annual ryegrass C:N ratio was higher than the radish + ryegrass C:N ratio, with respective values of 28:1 and 25:1. At MSU, there was no difference. The C:N ratios of the cover crops was lower at MSU than KBS. At MSU, the annual ryegrass C:N ratio was 18:1 while the radish + ryegrass ratio was 16:1. The typical threshold above which N immobilization is expected to occur is 20:1 to 25:1 (Cochran et al., 1980; Kuo and Jellum, 2002). At KBS, only the fall radish C:N ratio was low enough (19:1) to avoid likely net N immobilization in the soil. Since its C:N ratio was 25:1, in the spring it is probable that even the radish + ryegrass treatment had net N immobilization.

Cover crop N accumulation was calculated based on the biomass production and tissue N content data (Table 3). The cover crops at KBS accumulated less N in the fall than at MSU. At KBS, N accumulation ranged from 21-33 kg ha-1. At MSU, radish accumulated significantly more N in the fall than annual ryegrass and radish + ryegrass. There were no differences in N accumulation in the spring at either site. Cover crops again accumulated more N at MSU than KBS. This is in spite of the fact that biomass production was larger at KBS in the spring than MSU, and is due to the fact that tissue N concentrations were higher at MSU in the spring than at KBS.

Soil NO3 Levels

Soil samples were collected in the fall and spring during cover crop growth to test for NO3-N. In the fall of 2013, NO3 levels were greater in the control and radish treatments than in the annual ryegrass and radish + ryegrass treatments at KBS (Table 4). At MSU, NO3 levels were greater in the control than in the cover crop treatments. The interaction between treatment and year was not significant for spring NO3 data (P = 0.56), so data were combined (Table 4). At KBS in the spring, NO3 levels were higher in the radish treatment than in the annual ryegrass treatment, while the radish + ryegrass and control values were intermediate. There were no differences in the spring at MSU. Soil NO3 levels ranged from 12–17 kg N ha-1. Soil samples were collected twice during the corn phase of the experiment to test for soil N, once at corn V6-V8 for PSNT and again after corn harvest. PSNT data were pooled, so no statistical analysis was performed (Table 1). In 2013, soil NO3 levels were higher at KBS than MSU. At KBS, control and radish plots tested at 20–30 kg N ha-1, while annual ryegrass and radish + ryegrass plots tested at 11–14 kg N ha-1. The range was narrower at MSU in 2013 in the spring, with 11–16 kg N ha-1. In 2014, soil NO3 values were lower at KBS than MSU, ranging from 9-16 and 16-24 kg N ha-1, respectively. After corn harvest, there were no differences in soil NO3 values at either site. At KBS there were 11-12 kg N ha-1, while at MSU there were 18-26 kg N ha-1.

N2O Fluxes

N2O fluxes are presented in Figures 1-4. Colored columns on each graph denote the different sampling periods. N2O emissions were smallest during the fall sampling period of late October - December, peaked in March – June, and then returned to near-fall levels in July – August. The fluxes followed similar patterns at both sites over the course of each site-year. N2O emissions were higher at MSU than KBS, and higher in 2013-2014 than in 2012-2013. Analysis of cumulative N2O-N emissions is ongoing, but based on Figures 1-4 it appears that there were not many differences in emissions among treatments. Even though N2O emissions appear to be different, it is worth noting that overall N2O emissions were so low that there may not be a practical difference.

The N2O flux data in our study were highly variable. Soil fertility and physical properties, weather and season, crop residue quality, and field operations can all influence N2O fluxes (Novoa and Tejeda, 2006). Cover crop biomass production and C:N ratios (Table 3) may explain the difference in the size of N2O emissions between KBS and MSU. Spring biomass production and C:N ratios were both larger at KBS than MSU, while N accumulation was smaller. Net immobilization of soil N likely decreased the amount of N available to the microbes that facilitate denitrification and thus also decreased N2O emissions. At MSU, spring cover crop biomass production was smaller and C:N ratios were under the 20:1 to 25:1 thresholds at which N immobilization becomes probable. Given the availability of both C and N from the cover crops, and the anaerobic soil conditions typical of a wet Michigan spring, it is likely that most of the N2O emissions resulted from the denitrification process. Other research has also found N2O fluxes to be correlated with excess soil N (McSwiney and Robertson, 2005). Based on modeling by Novoa and Tejeda (2006), roughly 1% of the N contributed by crop residue in a system would be expected to be lost to N2O emissions. A study by Gomes et al. (2009) supported this estimate, with 1% or less of the N in legume cover crop residue lost to N2O emissions. The cover crops in our study accumulated 21-67 and 16-28 kg N ha-1 in the fall and spring, respectively (Table 3). It would thus be expected that N2O-N emissions over the course of our study would be in the range of 1.6-6.7 kg N ha-1 (160-670 g N ha-1), since no fertilizer was applied. Note that the y-axes in Figures 1-4 are in g and not kg. It appears that N2O-N emissions were lower even than would be estimated. It could be concluded that in our study, N2O emissions did not represent a major pathway for N loss from the cropping system. It is also possible that, due to our choice of sampling dates, we may not have captured periods of peak N2O flux. However, other studies such as that of Parkin and Kaspar (2006) have also found cover crops to have no impact on N2O emissions in a rotation that included corn.

Corn Growth and Yield

Corn height, SPAD chlorophyll meter readings at stage VT, ear leaf N, and grain yield were collected to evaluate the impact of the cover crops on corn. At KBS, corn in the radish and radish + ryegrass treatments was taller, as a percentage of the control, than in other treatments (Table 5). Corn in the annual ryegrass treatments was the shortest. At MSU, corn in the radish treatment was taller than in the annual ryegrass and control treatments. Corn in the annual ryegrass treatment was, as at KBS, shortest. At KBS, corn in the radish treatment had higher SPAD values than in all other treatments (Table 5). The only difference at MSU was that corn in the annual ryegrass treatment had smaller SPAD values than in all other treatments. At both sites, all treatment SPAD values were smaller than SPAD values in nearby reference plots which had received N fertilizer (P = 0.001) (data not shown). In spite of the differences in corn height and SPAD values, there were no differences in corn ear leaf N or corn grain yield at either site (Table 5). Corn at KBS yielded 4368-4671 kg ha-1, while corn at MSU yielded 8331-9786 kg ha-1.

The corn height and SPAD meter data suggest that radish was beneficial to corn growth, while annual ryegrass was detrimental. These differences were not enough to translate to grain yield differences. Grain yield was higher at MSU than KBS (Table 5), perhaps as a result of the different soils at each site (Table 1) and the varying cover crop biomass production and C:N ratios (Table 3). Given the C:N ratio of the annual ryegrass (Table 3), the negative impact of annual ryegrass on corn height and SPAD values likely reflected the presence of net N immobilization in those plots.

Participation Summary

Educational & Outreach Activities

Participation Summary:

Education/outreach description:

This work was presented in poster format in December of 2013 and December of 2014 at the Great Lakes Fruit, Vegetable, and Farm Market Expo in Grand Rapids, MI as well as in February of 2015 at the Midwest Cover Crops Council annual meeting in Des Moines, IA. A presentation was given in November of 2014 at the Agronomy Society of America, Crop Science Society of America, and Soil Science Society of America (Tri-Societies) annual meeting in Long Beach, CA. Results from the first year of this study were presented at the “Building Healthy Soils” workshop organized by the Shiawassee (MI) Conservation District in February of 2014. The results will be published in written form in a chapter of my dissertation and in a peer-reviewed journal; both manuscripts are currently in progress.

Project Outcomes

Project outcomes:

Results from our study indicate that radish and annual ryegrass biomass production varies by planting location and is dependent upon factors including soil fertility levels and the weather. The C:N ratios of radish, annual ryegrass, and radish + ryegrass cover crops were large enough to cause net N immobilization at the KBS site, but not at the MSU site. The C:N ratios at both sites were annual ryegrass > radish + ryegrass > radish. The detrimental effects that annual ryegrass had on corn growth were likely due to N immobilization. N2O emissions were lower at the KBS site, where spring cover crop biomass and C:N ratios were larger than at the MSU site. N2O emissions in our study were smaller than would be expected from calculations based on the work of other authors and did not represent a major pathway for N loss from this cropping system. It is probable that N released from the radish biomass prior to corn planting was either integrated into soil organic matter or leached. Based on this study, farmers should manage annual ryegrass carefully to avoid the risk of N immobilization after cover crop termination. If farmers choose to grow radish with the intent of scavenging residual soil N, they should plant it in a mixture with a cover crop with a higher C:N ratio to decrease the likelihood of N loss over the winter. Another alternative would be to plant it into residues such as corn or wheat, which also have higher C:N ratios.

Farmer Adoption

There is no way to know how exactly many people stopped to view my posters. The Great Lakes Fruit, Vegetable, and Farm Market Expo attracts over 4,000 attendees each year (GLEXPO, 2015). Roughly 75% attendees at the annual Midwest Cover Crops Council meeting are farmers, and attendance this year was about 400 people. Approximately 20 people viewed my presentation at the Tri-Societies meeting.

When I gave a presentation at the Shiawassee (MI) Conservation District workshop on February 25, 2014 I was also able to administer an anonymous survey to the participants (Appendix 3). Results are reported in the table in Appendix 4. Twelve attendees completed the survey. The majority farmed 100 or more acres. Most grew row and/or vegetable crops. All but one had used cover crops in the last three years, and the majority planted about 50 acres to cover crops. There was no single overwhelmingly popular cover crop type, with grasses, legumes, and brassicas all represented. Most respondents planted cover crops to improve the soil, manage water relations in their cropping system, and/or manage N. Time and monetary costs and the risk to a following cash crop were equally cited as concerns/problems related to cover crops. Most of the participants were aware of the benefit of cover crops to soil health, but all were interested in learning more about multiple cover crop topics. Before my presentation, four participants were not aware that N could be lost as part of greenhouse gas emissions while seven participants were aware of this concern. All but one participant was interested in learning more about the relationship between greenhouse gas emissions and cover crops. This survey suggests that farmers would welcome further opportunities to learn about a range of topics related to cover crops.


Areas needing additional study

This work could be expanded upon by repeating the experiment as a split-plot design with N-fertilizer application as a factor to replicate “real-world” conditions. Gas and soil sampling should be conducted more frequently to better trace the movement of N through the cropping system.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.