Research and Demonstration of Precision Planting of Cover Crop Mixtures for Improving Farm Profit and Soil Health

1. Soil Health from Upton Jr. Farm (Measured after six years of cover cropping):

Materials and methods:

Experimental locations and treatments:

A trial was initiated in 2015 and continued until 2022. The field experiment was located on a growers’ farm near Springerton, IL (38º16374'' N, 88º41032'' W). The soil was classified as Bluford silt loam with 0-2 and 2-5 percent slopes (Fine, smectitic, mesic Aeric Fragic Epiaqualfs). The experimental design was a randomized complete block design with three replicates. Treatments were (i) a no-cover crop control (NCC); (ii) no cover on corn row, hairy vetch (V) on middle row, and winter cereal rye (WCR) + annual rye (AR) on the outside row of corn (NOVR); and (iii) oats and radishes on the corn row, V on the middle row, and WCR on the outside row (ORVR).

Cover crop and cash crop managements

The cropping system at the farm was a continuous n-till corn-soybean rotation. In 2015, cover crops were established and in 2016, corn was planted following each cover crop treatment. After corn harvest and prior to planting soybean, WCR was planted in each treatment including the NCC. The planting dates for cover crops ranged between early October to early November. The corn and soybean planted was often occurred in early to late May. Corn and soybean harvest occurred early to mid-October each year up to the year the soil sampling was occurred.

Soil sampling and analysis

In 2020, soil samples were collected by a shovel from three treatments at in and out of the row (corn row vs. cover crop rows) at 0-5 cm (0-2”) and 5-20 cm (2-8”) to assess soil aggregation and aggregate stability, aggregate associated C and N, and soil enzymes along with typical soil fertility and permanganate oxidizable C (POXC). Soil samples were sent to Brookside Laboratory for soil fertility assessment. Methods for measuring soil aggregation and aggregate stability along with aggregate associated soil C and N are described below and are well described in Weidhuner et al. (2021). We collected soil bulk density using a deep core soil probe from 0-5, 5-20, and 20-90 cm (0-2, 2-8, and 8-36”) depths. Soil C stocks was calculated as depth of sampling × bulk density × percent C. We did not measure soil infiltration as our SATURO equipment was not functioning well and we could not trust the data. Soil

Aggregate size distribution and aggregate stability

Stacked sieves of 0.053, 0.25, 0.5, 1, 2, 4.75, and 6.3 mm were used, with a pan to collect soil falling through the smallest, final sieve. Stacked sieves sorted the aggregates into the following categories: <0.053 ,0.053-0.25, 0.25-0.5, 0.5-1, 1-2, and 2-4.75, and 4.75-6.3 mm. Approximately 130g of soil was placed in the center of the top sieve and shaken on a mechanized shaker (Humboldt Mfg. Co., Chicago, IL, U.S.A.) for six seconds modified from the protocol established by Kemper and Rosenaeu (1986). For our specific shaker, six seconds of shaking resulted in the lowest coefficient of variation (CV) while preventing soils from moving to the side of the sieves, unable to fall through (Hanauer et al., unpublished data). An electronic scale was used to record the mass of the contents of each sieve. The percentage of the soil fraction composed of each size aggregate was then calculated as a fraction of initial soil. This test was conducted with three subsamples per experimental unit and was averaged before statistical analysis.

Dry aggregates from the previous analyses were separated into small (0.25-2 mm) and large (2-4.75 mm) aggregate sizes. A wet-sieving apparatus (08.13, Eijkelkamp Agrisearch Equipment, Giesbeek, The Netherlands) was used to determine wet aggregate stability. Approximately 4 g (± 0.02g) of soil were weighed for small aggregates and 2.00 g (± 0.01g) of soil were weighed for large aggregates and placed in the 2 mm sieves. Differences in weight of aggregates were due to lower CV obtained from a preliminary study conducted by Hetrick et al. (2017; unpublished data). The sieves were each submerged in 100 ml of deionized water and soaked for 3 min. The apparatus then ran for a total of 3 min, turned off and allowed to drain for 30 sec, then submerged again and run for an additional 3 min, and finally turned off and allowed to drain for another 3 min. The contents of the can were water and the fraction of the soil that was water-unstable; this fraction was removed, leaving only water-stable aggregates plus sand particles in the sieve. The water was then replaced with an aggregate dispersing solution of 100 mL 2 g/L NaOH and gently stirred for ten revolutions with a glass stirring rod. The apparatus was run for an additional 3 min and was allowed to drain for 3 min. Dispersing solution and aggregate residue that remained  were washed with water into a 150 mL beaker. Only sand remained on the sieve and this was washed with water into another 150 mL beaker. All beakers were then placed in an oven at 50°C until dried. Once dried, each beaker and residue were weighed and recorded. The amount of water stable and unstable aggregates were found by subtracting the initial corresponding beaker weight. These protocols are well described in Weidhuner et al. (2021). 

C and N by depth

Air-dry soil samples by depth were pulverized using a bowl mill (Spex Sample Prep LCC, Metuchen, NJ) for C, and N analysis. Prior to analysis, samples were air-dried at 50 C for at least 48 hours to ensure soil moisture was not a confounding factor in the analysis. Contents of C and N by depth were determined by dry combustion (950°C) using a Thermo Scientific, Flash 2000 C and N analyzer (CE Elantech, Inc., Lakewood, NJ).

Soil hydrolytic enzyme activities

Soil enzymes that catalyze depolymerization of C-containing substrates were assayed to assess potential differences in maximum rates of hydrolysis: cellobiohydrolase (Enzyme Commission 3.2.1.91) and β-glucosidase (EC 3.2.1.21). Assays were performed using field-moist soils based on Tabatabai (1994) as modified by Margenot et al. (2018): 1.00 g of oven-dried soil equivalent was incubated for 1 h (β-glucosidase) or 2 h (cellobiohydrolase) at 37°C in 5 mL of 18.2ΩM·cm water. No buffer was used, for several reasons: modified universal buffer does not maintain assay pH than water (Li et al., 2021) and because buffer requires an assumption of a single pH optimum that is incorrect for many soil enzymes, including cellobiohydrolase and β-glucosidase (Niemi and Vepsäläinen, 2005; Turner, 2010; Wade et al., 2021). A final substrate concentration of 10 mmol L^-1 for β-glucosidase and 5 mmol L^-1 for cellobiohydrolase was used to ensure substrate saturation (Malcolm, 1983; Margenot et al., 2018). Reactions were immediately alkalized after 1 h by the addition of 4 mL of 0.1 mol L⁻¹ Tris (pH 12.0) and 1 mL of 2 mol L^-1 CaCl₂. Assays were centrifuged to remove sediment from the supernatant and an aliquot was used to quantify pNP colorimetrically using absorbance at 410 nm.{Turner, 2010 #60} Mean absorbance of negative controls (i.e., substrate but no soil) were subtracted from absorbance of soil assays to account for non-enzymatic hydrolysis of substrate during the incubation (Neal et al., 1981; Turner et al., 2002).

Permanganate-oxidizable carbon

Permanganate oxidizable carbon (POXC) was determined on air-dried soils using the method described by Weil et al. (2003a) as adapted by Culman et al. (2012). In 50 mL centrifuge tubes, 2.50 g of oven dried equivalent of soil was combined with of 0.2 M KMnO₄ and 9 ml MΩ cm^-1 water were added, yielding 20 mL of 0.02 M KMnO₄. The mixture was immediately shaken at 120 rpm for 2 min, then allowed to settle for 10 min. The supernatant was diluted (1:50) and absorbance at 550 nm was quantified by spectrophotometry. POXC was calculated assuming 9,000 mg C oxidized mol⁻¹ permanganate (Weil et al., 2003b).

Statistical analysis

All data were analyzed with PROC Mixed in SAS (SAS Institute, 2017). All variables were tested for normality prior to analysis using a continuous distribution function. Based on the Shapiro-Wilk test and Anderson-darling test used to determine the normality of data. Data that were not normally distributed were log-transformed. Data were back-transformed to normal to report in figures. To analyze soil parameters (aggregate size fractions, stability, C and N, POXC, enzymes, and bulk density) by depth, block and treatment nested within block (indicating plots) were random effects. In addition, an autoregressive covariance structure was specified for the plots being repeatedly measured over the sampling depths. The fixed effects in the model were sampling depth, cover crop, sampling row, and their interactions. LSMEANS in SAS was used for mean separation among treatments when P > 0.1. At each depth, we also ran a contrast analysis between CC (NOVR + ORVR) vs. NCC treatments to assess if CC improves soil properties.

2. On-farm trials at two sites (five site-yrs in two years) plus two site-yrs in an extra site: 

Materials and methods:

Experimental locations and treatments

Two on-farm trials were conducted in Marion and Springerton, IL in 2020-2021 and replicated in 2021-2022. Soil type was silt loam at both locations. Before trial initiation, composite soil samples were taken by a soil probe from 0-15 cm soil depth at each site and some important physical and chemical properties were determined. Data from soil analysis are not included but will be added.

The experimental design was a randomized complete block design with five treatments and four replicates. Five treatments were: (i) a no-CC control (ii) skipping the corn row, vetch beside the corn row, winter-rye-annual ryegrass-vetch mixture on the middle rows (iii) oat/radishes mix on the corn row, vetch on the row besides corn, winter-rye-annual ryegrass-vetch mixture on the middle rows (iv) skipping the corn row, crimson clover beside the corn row, winter-rye-annual ryegrass-crimson clover mixture on the middle rows (v) oat/radishes mix on the corn row, crimson clover on the row besides corn, winter-rye-annual ryegrass-crimson clover mixture on the middle rows. 

In 2021, those treatments were replicated in Marion but three of those were used in the Springerton site. The three treatments used in the Springerton site were (i) a no-cover crop control (NCC); (ii) no cover on corn row, crimson clover (C) on middle row, and winter cereal rye (WCR) + annual rye (AR) on the outside row of corn (RCSKIP); and (iii) ) no cover on corn row, hairy vetch (V) on middle row, and winter cereal rye (WCR) + annual rye (AR) on the outside row of corn (RVSKIP). In 2021 growing season, corn was planted following the cover crops and in 2022, soybean was planted after the cover crops in Marion but corn was planted following cover crops in Springerton to allow us to evaluate a three-site-yr assessment of cover crop effects on corn. In 2021, we also evaluated next season volunteer cover crops in the fallow season after cover-crop-corn rotation and prior to planting soybean in the Springerton site which had a WCR cover crop over all treatments.  

We added an experimental site (Agronomy Research Center, ARC, Carbondale, IL) to our assessments in both years that evaluated three treatments including (i) a no-cover crop control; (ii) clover-WCR mixture (RCM); and (iii) clover on the corn row and WCR on the middle rows (RCS).

Cover crop and cash crop management and measurements

Planting dates for cover crops in 2020 were September 25, in Marion's site, October 16 in the Springerton site, and September 23 at the Carbondale site. Planting dates for cover crops in 2021 were October 10 in Marion's site, October 8 in the Springerton site, and October 15 at the Carbondale site. 

Cover crops’ aboveground biomass was sampled with grass shears (GS model 700; Black and Decker Inc., Towson, MD) in all sites and years. Cover crop sampling dates in spring 2021 were May 12, in Marion's site, May 13 in the Springerton site, and April 13 at the Carbondale site. Cover crop sampling dates in spring 2022 were May 4 in Marion's site, May 10 in the Springerton site, and May 4 at the Carbondale site. The harvesting area for all trials and in both years was 0.675 m² (three frames of 0.225 m² from the center of plots to avoid edge effects). Biomass samples were first separated into species and weeds and then placed into an air-forced oven for 72 h at 48°C to determine DM yield. Biomass sub-samples were then ground to pass through a 1 mm sieve for nutrient analysis (Weidhuner et al., 2019). Nitrogen, P, K, along with Ca, Mg, S, B, Fe, Mn, Cu, Zn, Al, and Na were determined using near-infrared reflectance spectroscopy (NIRS) at the Brookside Laboratories (New Bremen, OH, USA). Carbon was measured using the combustion method as explained in Weidhuner et al. (2019). Nutrient uptake, and C accumulation were calculated by multiplying DM biomass of cover crop species with the percent concentration of each element as explained in Vaughn et al. (2022).   

Planting dates for corn in 2021 were in early May in Marion's site, June 1 in the Springerton site, and May 5 at the Carbondale site. Planting dates for cash crops in 2022 were May 11 in Marion's site, in the Springerton site, and May 18 at the Carbondale site. Harvesting dates for grain corn in 2021 were September 22 for the Marion site, October 5 for the Springerton site, and October 5 in Carbondale site.

In 2022, harvesting dates were September 29 (soybean) for the Marion site, October 6 for the Springerton site, and September 22 at the Carbondale site. At each site, corn stand count was recorded and 10 ears were hand harvested to assess corn yield components including total kernel weight, kernel weight per ear, number of kernels per ear, and 1000-grain weight. A subsample of corn grain was sent to Brookside Lab for grain nutrient analysis. Corn nutrient removal and balances were calculated based on Sadeghpour et al. (2017):

Nutrient applied (kg ha^-1) – nutrient removed (kg ha^-1) = nutrient balance (kg ha^-1).

Statistical analysis

Prior to analysis, data were evaluated for normality of the residuals using the Shapiro-Wilk test reported from Proc Univariate (SAS Institute, 2017). Data were then analyzed using Proc Mixed in SAS.

Data for cover crops that followed by corn (three site-yrs) were evaluated separately. Fixed effects were site-yrs and cover crop treatments and block was considered a random effect. In 2022, two trials were also conducted in which cover crops followed by soybean. Due to differences in the two trial, we analyzed those separately. Therefore, fixed effect was cover crop treatments and block was considered a random effect. Linear and non-linear models including quadratic, two, and three parameters exponential regression in JMP software (JMP Pro 14; SAS Institute, 2015) were used to evaluate relation among dependent variables. The best trend (model) was used based on lower P values, lower root mean square error (RMSE) values and higher R-squared (R²) values. If linear was a good fit (R² > 0.95 and low RMSE), we used linear and did not overfit the data. When significant at p ≤ 0.05, Fisher’s least significant difference (LSD) test was used for mean separation. Principal component analysis was used to assess the relation among corn yield components.  

Outreach activities:

The trials will be conducted on-farm which is the best form of outreach due to increased peer-to-peer interactions. We will share the results with growers in multiple ways including two field days in 2021. We have initiated writing a factsheet on use of precision mixed cover cropping which will become a part of the Soil Fertility & Fertilizes Course (CSEM/PSAS 447) at SIU-C. We have initiated a “school to farm” approach to bring novel ideas through our students to farms. We expect to publish at least one peer-reviewed journal article. We are partnering with Illinois Farm Bureau and especially Lauren Lurkins (Director of Research for Natural Resources in Illinois) and Austin Omer to reach as many growers as possible.  We have presented our results in numerous opportunities including field days, invited talks, North Central Soil Fertility Conference, and ASA-CSSA-SSSA annual meeting conferences.