Final report for GNC18-255
Project Information
Up to half of synthetic nitrogen (N) fertilizer applied to grain crops in the Midwest is lost from fields, resulting in environmental problems and economic losses for farmers. Including overwintering legume cover crops in crop rotations can reduce fertilizer inputs by adding biologically fixed N and improving overall soil quality and nitrogen-use efficiency (NUE). Combining legume and grass species in cover crop mixtures can supply new N while also improving soil N retention. Over time, using cover crops that include legumes may increase farm sustainability, maintain or increase crop yields, and reduce fertilizer costs. Despite these potential benefits, adoption rates of cover crops in Michigan are low (<6%) due to social, environmental, and economic barriers. This project linked farmer input with principles of agricultural ecology to conduct an on-farm experiment with real world implications for cover crop management on Southeast Michigan grain farms. Our primary outcome was to increase farmer knowledge about the role of cover crops, particularly cover crop mixtures, in soil fertility management, which can reduce fertilizer inputs over time. Diversifying rotations with cover crops would ultimately reduce nutrient losses from farms in Great Lakes watersheds, with wider implications for water quality.
Our research objective was to determine how multi-species cover crops impact cover crop biomass production, cover crop residue chemistry, soil N cycling processes, and resulting corn outcomes on grain farms in various stages of transition to cover cropping. We hypothesized that mixing legumes with grasses would alter total cover crop biomass and cover crop residue chemistry (e.g., carbon to nitrogen ratio (C:N)); improve synchrony between N release through decomposition and N uptake by the following corn crop via coupled N and C cycling; and result in either positive or no impact on the corn crop compared to a fallow or either cover crop grown alone. Across a nutrient management gradient on five grain farms and seven fields in Southeastern Michigan, we tested the impact of three different cover crop treatments – cereal rye, crimson clover, and a rye-clover mixture - on cover crop residue biochemistry, soil N cycling processes, and corn grain compared to a no-cover fallow. Input costs (seeds) were tracked for each treatment to conduct a cost-benefit analysis for different cover crops compared to leaving the field fallow over winter. Due to Covid-19 restrictions, we were only able to collect cover crop biomass and measure N mineralization rates following cover crop termination on 4 fields, but we were able to collect corn samples on all 7 fields.
Generally, we found that underlying soil conditions varied significantly across farms and that those soil conditions lead to variability in cover crop production and resulting corn outcomes. One of the primary variables we measured, cover crop C:N, varied between treatments with significantly higher C:N in the rye treatment than the clover or mixture treatments across the four farms on which we collected cover crop biomass. Further, we found that soil properties such as particulate organic matter and micronutrient concentrations influenced rye C:N and clover biomass across farms. This suggests that initial soil properties could guide farmer decisions about which cover crop species to plant based on potential benefits. We did not find generalizable results regarding the impact of cover crops on N mineralization or corn yields. Our measurements of N mineralization rates showed different patterns across each farm with no significant differences between cover crop treatments. Corn yield was either positively impacted or not impacted by cover crops on all but one field. The latter field demonstrated the negative impact that cover crops can have on corn production with unfavorable spring weather conditions. This field also had the lowest soil fertility, with cover crops having less of an impact on corn production in higher fertility fields. We developed both farm level and general recommendations from our results.
Farmers were consulted by phone during each phase of the project to plan the next phases. For example, farmer interviews determined the species we planted, the size of the plots, and fertilization methods. Focus groups were not conducted due to the pandemic. Instead, we communicated directly with each farmer over the phone throughout the project, and we are working with each farmer to provide them with a booklet with data and recommendations specific to their farm. I plan to share our results at farmer meetings and conferences over the next year and a half and engage in outreach through Michigan extension and local farmer outreach organizations in the counties where the study participants reside.
The farmers who participated in this study are enthusiastic about continuing to grow cover crops and all of them agreed to participate in a follow up greenhouse study in the Blesh lab, which is assessing the potential for legume-grass cover crop mixtures to sequester carbon in soil. On one farm, we had a difficult experience terminating the legumes in a low fertility soil under unfavorable weather conditions. The legume biomass then stunted corn growth. Instead of being discouraged by this, this farmer proceeded to plant grass and brassica cover crop species for the next growing season, since we had better luck with the rye treatment on his farm given his specific soil conditions. We saw either positive or neutral impacts on corn production at all the other farms, which has helped to encourage the farmers to continue cover cropping. I plan to maintain relationships with these farmers and help them use what we learned from each of their field trials in their outreach. Several of the farmers in this study are influential public speakers in their county and may be able to leverage this experiment to increase cover crop adoption among other farmers in their communities.
The study was designed to meet the needs of our primary intended audience: Michigan grain farmers who use, or are interested in using, cover crops. Farmers who already plant cover crops may not be harnessing their full potential. For instance, we conducted interviews to solicit input from grain farmers who already grow cover crops and learned that the connection between cover crops and soil nitrogen management is rarely being made. Our proposed learning outcome is to increase knowledge and understanding of the role of cover crops in soil fertility management. This includes understanding the connection between cover crops and soil nitrogen cycling. For example, legumes add new nitrogen to agroecosystems through biological nitrogen fixation. Other cover crops can retain and prevent nitrogen losses. Over time, farmers can apply nitrogen credits to cover crops, slowly reducing their fertilizer inputs. Since the study outcomes varied significantly from one farm to another, we are providing farm level data to each participant with recommendations for using cover crops on their specific soil in addition to a summary of aggregate results across all field sites. Our primary action outcome is increased adoption of cover crops that suit the needs of study participants' farms to meet their soil fertility management goals. The primary audience for this action outcome is farmers who participate in the on-farm study. Over time, we anticipate that a wider group of farmers will adopt these practices through farmer-farmer interactions with participants in our study. The broader outcome is to reduce fertilizer application rates and N losses from Michigan grain farms (particularly in the Lake Erie watershed) by using cover crops to manage soil nutrients
Research
Design: After spending a year attending meetings and recruiting farmers to participate in the cover crop experiment, I identified 12 fields on 7 farms that met our study criteria: (1) fields were in wheat or fallow during summer 2019 and going into corn in summer 2020; (2) the farms were in Monroe and Lenawee counties, which are in the Lake Erie watershed and are within one hour driving distance of the lab where soil and plant samples were processed. Proximity to the lab facilitates accurate measurement of soil nitrate and ammonium concentrations; and (3) fields varied in soil health and duration of past cover crop use. On one end of the spectrum, one field had historically been tilled with limited cover crop usage. On the other end of the spectrum, one field had been in no-till management with cover crops that were grazed in the fall and spring for several decades. This gradient of practices supported our goal of testing how grass and legume cover crops affect the subsequent corn crop depending on differing initial soil conditions. During spring 2019, all farms in the study experienced flooding, delaying our baseline soil sampling, but also showing us where the fields were prone to flooding so that we could plant the experiment where the cover crops were likely to grow well the following spring. We collected baseline soil samples in all 12 fields by treatment using 20 cm x 2.75 cm soil cores (15 cores/plot). The experiment was replicated 4 times on each farm in randomized strips with four treatments: 1) sole crimson clover, 2) sole cereal rye, 3) crimson clover/cereal rye mixture, and 4) no cover crop control (fallow). We decided to include a fallow (no cover crop) treatment instead of an unfertilized cereal rye benchmark at the request of the farmer partners, who are interested in comparing the outcomes for the cover crops to a no cover control.
Soil Sampling: The baseline soil samples were air dried in the lab and analyzed for total organic matter, phosphorus, potassium, magnesium, calcium, cation exchange capacity, and pH by A&L labs in Fort Wayne, Indiana. In our lab at the University of Michigan, we conducted tests for soil texture analysis, particulate organic matter concentrations - both free (i.e., light fraction) and occluded (i.e., physically protected inside aggregates), and total organic C, and total N by dry combustion on a Leco TruMac CN Analyzer (Leco Corporation, St. Joseph, MI). When we planted the cover crops in August 2019, we took a second soil sample, which was immediately analyzed for nitrate (NO3-) and ammonium (NH4+) to get a baseline inorganic N value at the time of planting. Samples were analyzed for NO3- and NH4+ colorimetrically on a discrete analyzer (AQ2; Seal Analytical, Mequon, WI).
Cover Crop Planting: We planted the cover crops on 10 fields in August 2019. We dropped 2 fields due to logistical constraints: (1) we were not able to transport equipment to one very overgrown site; and (2) the field was tilled and put into organic management before we could plant cover crops (Table 1). Planting dates varied during the month of August 2019 due to rain and the timing of wheat harvest. We visited all sites in late October to confirm that the cover crops had established in all fields. One farmer had accidentally plowed under one of the replicates on one of his two fields, but otherwise all experiments remained intact. We took pictures of each treatment in each replicate to create a record of growth.
Challenges: The most challenging part of the study in 2019 was the weather. It rained frequently during spring 2019 and did not warm up until late June. Many farmers were not able to plant crops this year and many acres were planted to cover crops in July through a federal emergency program. Luckily, the farmers we were working with had wheat in the ground during the springtime, which mitigated the impact of the floods compared to farmers with bare ground who struggled to plant corn and soy. The farmers in our study were all able to harvest their wheat in July. After a partly dry July, the rain started up again while we were trying to plant in August. I spent a lot of time cleaning the wheels on the grain drills, getting rained out while planting, and then waiting for the soil to dry out so we could plant. We learned a lot about the practical aspects of farming that can be limitations to cover crop adoption.
Table 1: Timing and method of planting cover crops, Summer 2019. Seeds purchased from Albert Lea Seed in Minnesota.
|
Farm Code |
Date of Cover Crop Planting |
Planting Method/Notes |
|
F1 |
8/1/19 |
Planted field 1 with farmer’s tractor and 15 ft John Deer no-till drill into a field that never got planted with corn due to flooding allowing us to plant cover crops instead. Only field without wheat stubble - will compare to the wheat stubble fields. |
|
F2 |
8/5/19 |
Planted fields 1 and 2 with farmer’s tractor and 15 ft John Deer no-till drill |
|
F3 |
8/7/19 |
Planted fields 1 and 2: Cereal Rye on 8/6/19, it started raining around noon, returned, and finished planting 8/7/19; Planted with 10 ft drill from the Conservation District (CD). We decided not to plant field 3 because we could not get the equipment there due to a large strip of overgrown weeds between the field and the road. The field also never got baled, so there was a heavy mat of wheat straw that would have made cover crop establishment difficult. |
|
F4 |
8/8/19 |
Planted by farmer with 10 ft CD drill. The farmer pulled field 2 from the study after deciding to start tilling it again to transition it to organic management. Farmer was still excited about the experiment on field 1 though. |
|
F5 |
8/8/19 |
Planted field 1 with farmer’s tractor and 10 ft CD drill, the strips are narrower than other farms due to space limitations (10 ft). |
|
F6 |
8/12/19 |
Planted field 1 with F5’s tractor and 10 ft CD drill. |
|
F7 |
8/26/19 |
Planted fields 1 and 2 with farmers' 10 ft European, no-till drill - clover was accidentally planted at double rate (28 lbs/acre) in Field 2 – we kept this data separate to compare to all the other fields planted at 14-16 lbs/acre (the planned rate). |
Spring Cover Crop Biomass: We had planned to collect aboveground biomass samples from all cover crop treatments in April 2020. However, state travel restrictions in MI prevented us from leaving Washtenaw County during the early part of the cover crop sampling time frame in April. Based on a citizen science project we were developing for another project, over the phone and with the help of written instructions sent via email, I recruited one of the farmer’s wives to take pictures of each cover crop treatment and record plant heights for three of the ten sites that were going to be burned down in April. We then uploaded the pictures into the Canopeo App and calculated % coverage. Height x coverage was used as an estimate for total biomass in these fields. The travel ban was lifted in May, I was able to travel to two more fields to collect pictures and plant heights and one other farmer took pictures of one more field. Later, when we got access to supplies, we were able to collect cover crop biomass, photos, and heights on four of the fields that the farmers waited on to burndown. Cover crop biomass was cut at the soil surface in a 0.25 m2 quadrat, separated by species, air dried and then later fully dried at 60°C for 48 hours, weighed, and coarsely ground (< 2 mm) in a Wiley mill. Shoot biomass was analyzed for total C and N by dry combustion on a Leco TruMac CN Analyzer (Leco Corporation, St. Joseph, MI). We then used a linear regression model between the height x coverage data and biomass cuts (R2 = 0.69, P <0.0001) to predict aboveground biomass for the fields on which we were unable to physically collect cover crop biomass.
Table 2: Timing and method of collecting spring cover crop data.
|
Farm Code |
Date of Cover Crop Sampling |
Method/Notes |
|
F1 |
5/13/20 |
Collected above ground biomass Pictures and plant heights recorded |
|
F2 |
5/3/20 (Field 1) 5/7/20 (Field 2) |
Pictures taken by farmer (Field 1) Collected above ground biomass (Field 2) Pictures and plant heights recorded (Field 2) |
|
F3 |
5/1/20 |
Pictures and plant heights recorded |
|
F4 |
4/27/20 |
Pictures and plant heights recorded |
|
F5 |
4/27/20 |
Pictures and plant heights recorded |
|
F6 |
4/27/20 |
Pictures and plant heights recorded |
|
F7 |
4/8/20 |
Collected above ground biomass (Field 1 & 2) Pictures and plant heights recorded (Field 1 & 2) |
Corn Planting and Fertilizer Application: Five of the seven farmers no-till drilled corn into green cover crop biomass and then burned down the cover crops (Table 3). Farmer 3 decided to plant soy instead of corn, ending his participation in the study. I tried contacting the farmer multiple times and was unable to get back in contact. Farmer 4 told me he could not plant corn anymore and tilled the field and planted soybeans. Interestingly, there was some family discord about this decision, because the daughter wanted to continue the experiment and the father did not. Luckily, all farms for which I collected cover crop biomass cuts completed the full experiment. Generally, I had a hard time managing communications during this period because the pandemic restricted face to face conversations with farmers as well as visits to farms. Phone calls can be effective, but it can also be difficult to get through and details often get lost. Face to face conversations with farmers in the field are a more effective way to ensure that management of on-farm experiments goes well. The greatest impact of failed communication in this experiment was with nitrogen fertilization. In conversations with farmers in Fall 2019, all participants agreed to add nitrogen fertilizer only at planting, and to skip the experimental plots during future side-dressing events. This would allow us to measure nitrogen mineralization from the soil and cover crop biomass following cover crop incorporation, without interruption by addition of inorganic nitrogen fertilizer. I reminded all participants of this individually in the spring. Unfortunately, later in the summer I realized that all but one of the farmers side-dressed their corn with nitrogen. See dates of all fertilizer use in Table 3.
Table 3: Timing of corn planting, cover crop burndown, and fertilizer addition.
|
Farm Code |
Date of corn planting |
Date of Burndown |
Fertilizer |
|
F1 |
5/13/20 |
5/22/20 |
73 lb N/acre at planting |
|
F2 |
5/4/20 (Field 1) 5/7/20 (Field 2) |
5/5/20 (Field 1) 5/27/20 (Field 2) |
26 lb N/acre at planting 128 lb N/acre at side-dress (June 8) |
|
F3 |
Farmer planted soybeans instead of corn |
N/A |
N/A |
|
F4 |
Farmer planted soybeans instead of corn |
N/A |
N/A |
|
F5 |
5/5/20 |
5/29/20 |
150 lb N/acre at side-dress (June 8) |
|
F6 |
5/2/20 |
5/27/20 |
30 lb N/acre at planting 140 lb N/acre at side-dress |
|
F7 |
5/9/20 |
5/13/20 |
30 lb N/acre at planting 46 lb N/acre at side-dress (June 1) 46 lb N/acre at side-dress (July 2) |
N mineralization rates: We measured in situ soil N mineralization potential starting 3-7 days after cover crop termination (to follow pesticide safety protocols) for 2-3 months on 4 fields using a buried bag incubation method (Babbar and Zak 1994; Nadelhoffer et al. 1984; Robertson et al. 1999). This method compares initial soil NO3- and NH4+ concentrations, measured roughly every 28 days, to concentrations in soil cores that are incubated in the field in buried, polyethylene bags. In each replicate plot on each field, we collected 10-12 soil cores to 10 cm depth, homogenized and sieved the soil, and extracted NO3- and NH4+ in triplicate with 2 M KCl for initial concentrations. Samples were analyzed for NO3- and NH4+ colorimetrically on a discrete analyzer (AQ2; Seal Analytical, Mequon, WI). At the same time as the baseline sampling, we took an additional 8 soil cores to 10 cm depth, placed them in sealed polyethylene bags, and buried the bags immediately at 10 cm depth in the spot where the core was taken. Each soil core was incubated in the field for ~28 days before being removed and analyzed for NO3- and NH4+. This process was repeated for three months. Net N mineralization was estimated for each time period by comparing the sum of NO3- and NH4+ in each set of incubations to the respective baseline measurements.
Table 4: Dates of N mineralization measurements. We ended measurements on three of the farms in July because these fields had been fertilized after the experiment started, compromising the data quality. We continued until the third week of August on Farm 1, which did not receive fertilizer after planting.
|
Farm Code |
N min start 1 |
N min start 2 |
N min finish (or start 3) |
N min finish |
|
F1 |
5/26/20 |
6/25/20 |
7/24/20 |
8/21/20 |
|
F2 |
6/1/20 |
6/29/20 |
7/27/20 |
|
|
F7A |
5/20/20 |
6/15/20 |
7/13/20 |
|
|
F7B |
5/21/20 |
6/16/20 |
7/13/20 |
|
Sources for N mineralization Method:
Babbar, L.I., and D.R. Zak. 1994. Nitrogen cycling in coffee agroecosystems: net N mineralization and nitrification in the presence and absence of shade trees. Agricult. Ecosys. Environ. 48:107-113.
Nadelhoffer, N.J., J.D. Aber, and J.M. Melillo. 1984. Seasonal patterns of ammonium and nitrate uptake in nine temperate forest ecosystems. Plant Soil. 80:321-335.
Robertson, G.P., D.C. Coleman, C.S. Bledsoe, and P. Sollins. 1999. Standard Soil Methods for Long-Term Ecological Research. Long-Term Ecological Research Network Series. Oxford University Press, New York, NY.
Corn Sampling: We sampled corn by hand in October 2020. We were hoping to work with the farmers to use their combines to collect yield data, but that only worked out on F1. Other farmers had much larger combines that were not able to accurately measure yield at the small scale of the experimental plots. Instead, we collected corn grain from 2-meter row-lengths of corn plants and counted the number of plants. The cobs were shucked in the field to remove the husks and silks.
Grain biomass was dried at 60°C for 48 hours. Grains were then separated from the cobs, dried again at 60°C for > 24 hours, weighed, and coarsely ground (< 2 mm) in a Wiley mill, followed by analysis for total C and N concentration by dry combustion on a Leco TruMac Analyzer (Leco Corporation, St. Joseph, MI).
Table 5: Dates and methods of corn harvest in Fall 2020.
|
Farm Code |
Corn Harvest |
Method |
|
F1 |
10/30/20 |
Collected yield data in bushels/acre from combine |
|
F2 |
10/3/20 |
Hand-picked corn in lbs grain/linear meter and counted corn plants |
|
F5 |
10/6/20 |
Hand-picked corn in lbs grain/linear meter and counted corn plants |
|
F6 |
10/6/20 |
Hand-picked corn in lbs grain/linear meter and counted corn plants |
|
F7 |
10/26/20 |
Hand-picked corn from 5 plants per plot after failed yield collection with combine and we had to do all the corn sampling in twenty minutes so the farmer could harvest as a storm was rolling in. |
Despite the relatively large plot size (15 x 130 ft), after consulting with the farmers in the study, we do not believe that the plots were large enough to accurately estimate yield. However, since we used the same harvesting method between treatments on each farm, we can still detect differences in corn production between cover crop treatments.
Statistical analysis was conducted in RStudio and in JMP Pro 15 software.
We will present the results for and discuss the soil fertility gradient that we found across farms, cover crop quality and differences we found between treatments across farms, the impact of soil quality on clover and rye, N mineralization rates over time, and resulting corn production. We have only had a few weeks to conduct an initial analysis of the complete dataset. We will continue to analyze the data in new ways as we produce presentations and publications from this study. Figures are attached in this PDF: Results with Figures
Soil Fertility Gradient: At the beginning of the project, we collected soil samples and analyzed them for soil fertility properties including soil texture, micro and macro nutrients, and different fractions of soil organic matter. We found a gradient of soil fertility levels, driven by different soil metrics, indicating that we achieved our goal of conducting this cover crop experiment across a range of baseline soil fertility. Tables 1-5 show all baseline soil data across the seven fields that completed the study (F1, F2, F5, F6, F7). Here we have included the mean, standard error, and range for each soil property.
Table 1: Baseline means, standard error (SE), and range for bulk density and soil texture for soils collected across all farms in the study.
|
Bulk Density |
% Sand |
% Clay |
% Silt |
|
|
Mean |
1.56 |
57.47 |
21.00 |
21.52 |
|
SE |
0.014 |
2.751 |
1.830 |
1.208 |
|
Range |
0.37 |
56.03 |
38.92 |
27.15 |
Table 2: Baseline means, standard error (SE), and range for pH, organic matter (%), phosphorus (P), potassium (K), magnesium (Mg), calcium (Ca), cation exchange capacity (CEC), N and C for soils collected across all farms in the study.
|
pH |
Organic Matter (%) |
P (ppm) |
K (ppm) |
Mg (ppm) |
Ca (ppm) |
CEC |
%N |
%C |
|
|
Mean |
6.66 |
2.61 |
70.79 |
111.54 |
175.00 |
1344.64 |
9.73 |
0.14 |
1.52 |
|
SE |
0.09 |
0.143 |
7.06 |
7.85 |
7.95 |
59.26 |
0.51 |
0.008 |
0.079 |
|
Range |
2.00 |
3.30 |
165.00 |
197.00 |
190.00 |
1350.00 |
11.8 |
0.18 |
1.73 |
Table 3: Baseline means, standard error (SE), and range for the amount, N, and C of free particulate organic matter (FPOM) for soils collected across all farms in the study.
|
g FPOM/kg soil |
kg FPOM/ha |
% N |
% C |
mg N FPOM/ kg soil |
mg C FPOM/ kg soil |
|
|
Mean |
2.48 |
7653.72 |
1.50 |
27.28 |
37.36 |
669.29 |
|
SE |
0.139 |
419.512 |
0.028 |
0.391 |
2.246 |
36.250 |
|
Range |
3.75 |
11135.78 |
0.86 |
12.14 |
53.97 |
817.43 |
Table 4: Baseline means, standard error (SE), and range for the amount, N, and C of occluded particulate organic matter (OPOM) for soils collected across all farms in the study.
|
g OPOM/kg soil |
kg OPOM/ha |
%N |
%C |
mg N OPOM/ kg soil |
mg C OPOM/ kg soil |
|
|
Mean |
3.87 |
11674.39 |
1.67 |
25.32 |
64.11 |
968.51 |
|
SE |
0.19 |
514.05 |
0.03 |
0.40 |
3.23 |
44.29 |
|
Range |
5.70 |
16280.00 |
1.09 |
12.33 |
104.74 |
1310.01 |
Table 5: Means, standard error (SE), and range for soil inorganic nitrogen in Kg N/ha and in concentrations from extractions (EXT) and ammonium from incubations (INC) for soils collected across all farms in the study at the time of cover crop planting.
|
Kg NO3-N/ha (EXT) |
ug NO3-N/g soil (EXT) |
Kg NH4-N/ha (EXT) |
ug NH4-N/g soil (EXT) |
Kg NH4-N/ha (INC) |
ug NH4-N/g soil (INC) |
|
|
Mean |
165.29 |
55.03 |
27.05 |
8.74 |
240.35 |
79.57 |
|
SE |
17.08 |
5.82 |
1.60 |
0.52 |
15.10 |
5.37 |
|
Range |
389.17 |
134.40 |
40.66 |
13.10 |
413.61 |
162.32 |
Cover Crop Litter Quality: Across farms (F1, F2, F7A, and F7B), we found that cover crop biomass quality varied between treatments. In a mixed effects model with treatment as a fixed effect and block nested in field as a random effect, we found that rye had the highest mean C:N of 28.7, which is a typical value for grasses, and both treatments with clover had significantly lower C:N ratios of 16.8 for clover grown alone and 17.5 for both species combined in mixture (p < 0.0001) (Figure 1). Rye on average made up 43% of total biomass in the mixture treatments, with a range from 14% to 68%. Adding rye to the mixture treatment did not significantly increase C:N compared to clover grown alone. This indicates that the mixture provided the same quality cover crop litter inputs as the clover biomass.
Cover Crop Biomass Variability Between Treatments:
Using the pictures and plant heights we took across all farms, combined with the cover crop biomass data we collected on four farms, we were able to estimate cover crop biomass for the farms on which we couldn’t sample due to the pandemic. Using these data for all farms in a mixed effects model with treatment as a fixed effect and block nested in field as a random effect, we found that the mixture produced higher biomass (mean = 2731 kg/ha) than the rye (mean = 1818 kg/ha) or clover (mean = 2151 kg/ha) treatments (p < 0.0001). All cover crop treatments produced more aboveground biomass than the weedy fallow control (mean = 956 kg/ha). These results support our hypothesis that the mixture would produce higher overall biomass inputs than either species grown alone. While this supports our hypothesis, we did not find a different C:N between mixture and clover. As visualized in the boxplots in Figure 2, we found high variability in cover crop biomass across farms indicating that factors other than cover crop treatment impact cover crop outcomes. The next section explores soil properties that were most highly correlated with cover crop outcomes.
Soil properties affect cover crops:
The occluded particulate organic matter (OPOM) fraction was a predictor of quality (i.e., C:N) and quantity for crimson clover when grown alone and in mixture across farms (F1, F2, F7A, and F7B). OPOM is a fraction of soil organic matter that is physically protected inside soil aggregates and changes slowly with a turnover time of decades. We found that the OPOM pool (Mg OPOM/ha) was significantly correlated with clover biomass (kg/ha) and quality (C:N). As the OPOM pool increased, we found a decrease in C:N (the proportional amount of N increased). Since legumes fix nitrogen, legumes growing in higher quality soils with larger intermediate soil carbon pools, may be fixing more nitrogen than legumes in poorer quality soils. This corresponds with higher total N in clover biomass and higher overall biomass production. We found a similar trend with potassium (K), a critical nutrient for legumes (Figure 3).
While OPOM pools and K significantly improved clover outcomes across farms, we did not see the same strong positive trends with mg N in OPOM/kg soil, which was not a predictor of clover biomass or quality. This observation is reasonable given that the legume, clover, fixes its own N and is thus not dependent on soil N pools for biomass production. We also expect legume biomass to be lower in the mixture in soils with more POM N because it is less competitive with grasses that do need soil N. Soil carbon storage, an indicator of soil quality, and nutrients such as K are more important metrics for predicting legume success. Farmers with higher OPOM and K pools may benefit more from planting clover as a cover crop than farmers with lower quality soils in the early years of cover cropping.
We found that a range of soil properties impacted cereal rye quality across farms (F1, F2, F7A, and F7B). First, we found that the concentrations of OPOM and FPOM N were significantly and negatively correlated with the C:N ratio of cereal rye when grown alone and in mixture, with a stronger relationship in the mixture. We found a similar negative relationship between the FPOM pool and C:N (Figure 4).
POM was not the only soil property that helped explain cereal rye quality. We also found significant and positive relationships between calcium (Ca) and cation exchange capacity (CEC) and cereal rye C:N and a negative relationship between P and C:N (Figure 5), indicating that as P increases, N concentration also increases in rye biomass. Since P is a limiting nutrient, rye may be able to absorb more N from the soil when P is not limiting. This relationship was stronger in the mixture than in the monoculture, which also suggests that higher P levels may be allowing the rye to compete more against the clover. We did not find any effect of these nutrients on total rye biomass, though, indicating that these soil properties influenced the quality, rather than the quantity, of the cereal rye biomass. Cover crop quality matters to future cash crop growth in addition to just cover crop biomass, so it is an important metric for farmers to consider when managing their cover crops.
Effect of Crimson Clover Seeding Rate on Biomass:
On farm 7, clover was accidentally planted at 28 lbs/acre in F7B, while F7A, on the same farm, was planted to 16 lbs/acre (the intended rate for the study). This mistake allowed us to compare cover crop biomass outcomes based on seeding rates (Figure 6). In the clover monoculture treatment (T2), we found higher (but not significantly, p = 0.0799) biomass (kg/ha) and significantly higher total N (kg/ha) (p = 0.0463) in the treatment that was planted double (F7B). There was no difference between total cover crop biomass, C, or N (clover + rye) in the mixture treatment, or in clover biomass in the mixture treatment. Competition with rye likely reduced the effect of clover seeding rate on clover biomass in the mixture. Since we found no difference when clover was planted in mixture, farmers may not benefit from increasing seeding rates of legumes in cover crop mixtures. However, increasing seeding rates when legumes are planted alone has the potential to moderately increase total biomass although may not be worth the higher price.
Nitrogen Mineralization Following Cover Crop Treatments:
Our approach to estimating nitrogen mineralization in the field worked well on one of the farms that carefully followed the study design and only fertilized once at corn planting (73 lbs N/acre). This fertilization occurred 13 days before we collected and buried cores to measure N mineralization, and 4 days after cover crops had been burned down. The differences between treatments are not statistically different, however, the fallow treatment had higher N immobilization (mean = -0.77 µg N/g soil/day) than the treatments with cover crop inputs in the first month, and the clover treatment had higher N mineralization rates later in the growing season (mean = 0.69 µg N/g soil/day). Notably, the fallow and rye treatments did not shift from net immobilization (below the 0 line) to net mineralization (above the 0 line) until the end of the measurement period, while the clover and mixture treatments shifted to net mineralization about halfway through the growing season. Based on these results, farmers would benefit most from planting cover crops that include legumes specifically in terms of N availability. Added N from legumes can help drive higher soil mineralization rates through decomposition of organic matter that provides enough N to microbes such that they release excess N back into the soil (mineralization) (Figure 7).
Effect of Cover Crops on Corn Yield:
We found variable corn biomass production across farms, each being impacted by soil fertility, weather, and success of growing and terminating cover crops. When corn grain biomass was compared across all farms, there were no differences between cover crop treatments. We did find a difference in corn quality between cover crop treatments across farms. Corn C:N was lower in treatments with clover (the monoculture and the mixture) compared to the rye and fallow treatments (p = 0.0012) with the highest % corn N in the mixture treatment (mean = 1.35) followed by the clover treatment (mean 1.34), both significantly higher than the rye treatment (mean = 1.28) (p = 0.0033).
We also analyzed the corn grain production by farm. Except for on farm 1, we generally found positive or neutral relationships between cover crop biomass and chemistry and corn production. On farm 7, across both fields and all treatments, we found that total cover crop biomass (kg/ha) predicted corn grain quality (%N) (Figure 8). On farm 2, we found no impact of cover crops on corn production (Figure 9). While on farm 7 and farm 2, we did not find a negative effect of cover crops on corn production and quality with some positive outcomes, on farm 1, we found the opposite trend. As cover crop biomass increased, corn biomass decreased (Figure 10). In fact, we found that the mixture treatment resulted in significantly lower corn production than the fallow treatment, while the sole clover and rye treatments produced the same corn biomass as the fallow treatment (p = 0.03). The unpredictable impact of cover crops on yield is often cited as a reason not to grow cover crops. While we did have one unfavorable outcome on one farm, all other farms saw either positive or neutral impacts of cover crops on corn. Based on this, although there is some risk of damaging corn yield, the overall benefits outweigh the risks.
Educational & Outreach Activities
Participation Summary:
Due to Covid dominating the last year and a half of this study, we were not able to hold the in-person outreach activities we were planning. We considered Zoom but given how difficult it has been to communicate with many of the older farmers (many in their 70s and 80s) over the phone, we determined that Zoom would not be an effective way to share information. In the spring of 2020, we collected all the data we had for each farm and mailed a copy to each farmer, along with maps of their fields and a reminder of the experimental design and next steps for the summer. I also kept in touch with farmers over the phone.
We are planning outreach this summer with farmers who participated in the study by sending each farmer a personalized booklet in the mail with data and lessons learned from their farm, along with overall takeaways from the study. I will send email copies to a few of the farmers who use email, so they have the option to use images at outreach events. This booklet includes a model for estimating N inputs from legume cover crops on their farm with estimates for % N in cover crop biomass and % N from biological nitrogen fixation. The booklet concludes with a formula for estimating cost/benefits of cover cropping that each farmer will be able to apply to their farming operations. I plan to invite all the farmers who participated to presentations that I plan to give in the future (see conference plans below).
Since Covid delayed sample analysis by over 6 months, I received the complete dataset for the project in June 2021 and thus have not been able to present this work at conferences yet. I plan to present this project at the Underground Innovations meeting this winter, the Midwest Cover Crop Council meeting in Ada, OH in March 2022, and at the Ecological Society of America conference in Montreal, Canada in August 2022. I also plan to present at the next Society and Natural Resources conference to discuss our experiences with farmer outreach and engagement in field trials.
Beyond these formal speaking engagements, I plan to work with leaders in farmer engagement in Michigan to share appropriate portions of the project with a general farmer audience. I am developing a collaboration with the Cooperative Institute for Great Lakes Research (CIGLR) who have expressed interest in increasing farmer outreach to mitigate the impacts of conventional agriculture on Great Lake’s water quality. After identifying how to best use outreach materials to share with a local audience in Michigan, I will look for broader scale opportunities to share what we learned from this study through avenues such as farmer magazines and podcasts. I hope to take my career in this direction of engaging in farmer outreach to increase cover crop adoption in the Great Lake’s watershed.
Project Outcomes
This project was unique in that a year-long experiment was conducted across a range of working farms in Michigan. We aimed to manage the projects in a way that was consistent with typical farm management in the region and worked closely with farmers to use existing equipment and infrastructure to conduct the study, 7 of which made it through to the end of the experiment. Ideally, this method would provide farmers with more realistic outcomes and recommendations for managing cover crops on their own unique soil type. As farmers often mention in outreach talks – ‘this worked on my farm but might not work on yours’ – it is important to ask questions about cover crops across a range of soils to provide more targeted information to individual farmers.
All farmers who participated in the experiment were excited to watch as the cover crops grew in the fall and spring and impacted their corn. Generally, we found that cover crops did not negatively impact corn yield compared to the fallow, while providing ecosystem services – adding biomass and newly fixed nitrogen. This study was a good start to engage farmers in Southeast Michigan in a research experiment that uses ecological practices. Only one farmer had engaged in an on-farm experiment before participating in this experiment (a seed test for a biotechnology company). Since this group of farmers helped us develop the research questions surrounding cover crop litter composition, decomposition, and resulting corn quality, they were all engaged in the project. Even the two farmers who had to plant soy in 2020 and drop out of the project are working to transition their farms to be more sustainable by diversifying rotations, adding cover crops, and switching to more organic practices. Two of the farmers who completed the study had not used cover crops extensively on their farms yet. One farmer is now trying to grow grasses on all his acres to regenerate the soil before trying more complex mixtures, and the other is focusing on diversifying his farm with both cover crops and adding vegetable production. The farmers who had already been growing cover crops had mostly just been growing winter kill varieties (broadleaves and brassicas) and grasses and were excited to see what would happen with crimson clover and the mixture. One farmer had frost seeded red clover before, but no farmers in the study had tried overwintering crimson clover before. This exposure to this new cover crop may lead to future adoption, although we do not have data to support this yet. One participant is so excited about his cover crops that he sent me pictures of them in late Fall 2020 to show me how they were affecting his soil.
Future plans to solicit feedback from participants: I am still putting together individual booklets that will be shared with farmers later this summer. The booklet includes farm level data for each farmer and a formula for calculating clover N inputs and a formula for calculating economic benefit of different cover crops over time. I plan to call each farmer after they have time to look through the booklet to discuss the results from their farm and talk about how it will affect their cover crop management in the future.
Based on the social science research I have done concurrently with this study evaluating the National Wildlife Federation’s Cover Crop Champion’s program, I predict that the early adopter farmers who participated in these experiments will share what they have learned with other farmers. I will provide them with slides to share data visualizations. Through formal presentations and informal farmer-farmer interactions, it is possible that adoption will increase in the counties where these experiments occurred creating a larger “innovation niche” of a conservation practice that is not currently widely used. This, combined with the more extreme weather we have had the last few years, and resulting payments from the government to plant more cover crops, will likely increase acres that are cover cropped. The community really seems interested in learning about cover cropping, and hopefully the data that comes out of this study will give the participating farmers more evidence that cover crops can work to convince other farmers that it is worth trying. If adoption of cover crops increases in these communities, farmers will benefit economically in the long-term by mitigating the impacts of climate change on their soils while reducing the negative environmental impacts of commodity agriculture in the Lake Erie watershed. Over time, increasing use of cover crops will benefit the whole community by reducing the shocks of extreme weather events and improving community drinking water, which provides social benefits for the whole community.
I have learned that in order to successfully implement research experiments on farms, close communication and building trusting relationships with farmers over many years is important. I started getting to know these farmers in 2017, when I first interviewed them about their management practices and have further developed relationships with all of them to the point that they trusted me coming on to their farms to take samples and use their machinery to plant cover crops, take many soil samples, and harvest corn.
Through the process of physically implementing the experiment on working farms, I have learned a lot about farm management and associated logistics of planting and terminating cover crops. Several of the farmers generously spent a lot of time teaching me about their practices and why they plant certain cover crops on certain plots of land. I have learned a lot about planting green and how much easier it is to no-till drill into living plants rather than dead ones. I was really concerned the first time we planted the cover crops into a field of weeds – following wheat – that had been hit with roundup a few days before, but 2 weeks later, it looked amazing, and the cover crops established well. I think I have really learned that no matter how bad it looks, to just trust that it will be alright and that cover crops are strong and will grow. A similar lesson was learned when planting corn into green cover crops before terminating the cover crops days-weeks later. Of all farms that planted green, we only had negative outcomes on one. I am sold on planting green and will promote it in future outreach. I have also learned how to calibrate and maintain a no-till grain drill, drive a tractor, and collect corn samples from a combine to add to my on-farm research skill set.
Generally, this project drove home how difficult it is to farm under increasingly unstable weather conditions and helped me empathize with farmers who are trying to adopt new practices like cover cropping on their farms. I believe that cover cropping is critical to successfully adapting to climate change by improving soil quality over time so that farmers can keep farming their land in the future and am now more aware of how difficult it is to add another element, such as cover cropping, to a farm management plan. I will continue to advocate for farmers so that they are provided with the financial resources and social support to transition to more ecologically based management.
On one farm, we had a difficult experience terminating the legumes in very poor-quality soils under unfavorable weather conditions. The legume biomass then stunted corn growth. Instead of being discouraged by this, this farmer proceeded to plant grass and brassica cover crop species for the next growing season, since we had better luck with the rye treatment on his farm given his specific soil conditions. The grasses he planted worked out better this year than the legumes we planted last year, and he plans to keep planting cover crops to improve his soils. My experiences with this farmer demonstrated how determined and resilient farmers are, even when things do not work out as planned.
Finding a no-till drill appropriate for planting the cover crops in research plots was tricky. It might be helpful to create a sort of tool bank or library/list to facilitate researchers accessing equipment to implement farm experiments.
This study had limitations in terms of what we learned due to the one-year time frame in which it was conducted. The year during which we conducted this experiment was a flood year which impacted the experiment’s outcomes. Flooding is becoming more common though, so this may be representative of average years moving forward. Cover crops are unlikely to have an effect on yield in just one year and the variability between farms was so high that it is difficult to draw conclusions from this study. Future studies on this topic should be on the order of 5 or more years and include cover crops between each cash crop.