On-farm and isotopic evaluation of deep soil nitrogen capture and cycling by cover crop mixtures

Location

The study was located at the Central Maryland Research and Education Center—Beltsville Facility in Laurel, Maryland USA. The region has a humid climate with annual rainfall relatively uniformly distributed throughout the entire year. The temperature and precipitation for the study period is found in Figure 1. The experiment was performed from September 2015 to October 2016 during the cover crop-cash crop cycle. The study contained six blocks, three at each of two sites located 1.06 km away from each other. The two sites were SANDY (39.01162, -76.83167) and CLAYEY (39.0183, -76.8229). Prior to the study, both fields were planted in winter wheat fall 2014-summer 2015. The fields were in corn in 2014, and in fall 2012-summer 2013 the fields were in winter wheat with a double crop of soybean. Soil characteristics are listed in Table 1.

Table 1. Soil percent sand, silt, clay, pH, percent C and N, NO₃-N (kg N/ha) and NH₄-N (kg N/ha). NO₃-N and NH₄-N values are from soil samples taken in 15 plots (five cores per rep).

Experimental design and treatments

The study included 29 treatments in a randomized complete block design with six replications. Plots were 3 m x 3 m in size. Experimental factors that defined the treatments included cover crop planting date (CC DATE), cover crop species (CC SPECIES), and burial depth of the ¹⁵N tracer (15N DEPTH). The treatments were in an incomplete factorial combination of these factors. The CC SPECIES treatment included: 1) forage radish (RADISH), 2) cereal rye (RYE), 3) forage radish + cereal rye (2MIX), 4) forage radish + cereal rye + Crimson clover (3MIX), and 5) control treatment in which no cover crop was planted (NO CC). These species treatments were chosen in order to investigate how forage radish and cereal rye perform together versus in monoculture, and to asses if the 15N RECOVERY of the rye-radish mix would change with the presence of a legume in the mix.

The CC DATE treatments included: 1) an early planting date of 3-Sep (EARLY PLANTING) and 2) a late planting date of 8-Oct (LATE PLANTING). There was no 2MIX x LATE PLANTING treatment combination, as we did not expect an interaction between CC DATE and the influence of clover on the cover crop mixture. The 15N DEPTH treatments included: 1) 60 cm burial (60 CM), 2) 120 cm burial (120 CM), 3) 180 cm burial (180 CM), or 4) a no burial control treatment (NO BURIAL). There was no 180 CM x LATE PLANTING treatment combination, as we did not anticipate root from the LATE PLANTING reaching even the 120 cm depth (Table 2).

Table 2. Experimental treatment combinations. Experimental factors that defined the treatments included CC DATE, CC SPECIES, and 15N DEPTH. The CC SPECIES indicated in bold text had treatments for EARLY PLANTING and LATE PLANTING. The CC SPECIES indicated in non-bold text had treatments for only EARLY PLANTING. 

Field operations

Winter wheat was harvested from the plots mid-July, and weeds were killed on 1 Sep 2015 using Glyphosate, N-(phosphonomethyl)glycine at the rate of 2.31 kg active ingredient/ha. The EARLY PLANTING were fertilized with 9.07 kg N as urea and ammonium nitrate (UAN) on 3 Sep 2015, and cover crops were planted 3 Sep 2015. The LATE PLANTING were fertilized with 9.07 kg N (as UAN) 28 Sep 2015, and cover crops were planted 8 Oct 2015. Cover crops were chemically terminated using Glyphosate, N-(phosphonomethyl) glycine on 18 May 2016.

At the SANDY site, corn was planted 16 May 2016. Fertilizer was applied 14 June 2016 as 31.8 and 7.3 kg/ha of N and S, respectively in 115 liters of fertilizer solution. At the CLAYEY site, corn was planted 7 June 2016. On 8 June 2016, herbicides were applied (1.85 kg/ha 2-chloro-4-ethylamino-6-isopropylamino-s-triazine (Atrazine), 2.3 L/ha 1,1’-dimethyl-4,4’-bipyridinium dichloride (Paraquat), 4.7 L/ha Acetochlor (Warrant), 0.56 kg/ha Sticker). Fertilizer was applied 29 June 2016 as 31.8 and 7.3 kg/ha of N and S, respectively in 115 liters of fertilizer solution.

¹⁵N solution and burial

Each plot had 0.5 g KNO₃ isotopic tracer 99% enriched in ¹⁵N, which is equal to 0.07345 g ¹⁵N (Equation 1), buried per plot.

The ¹⁵N tracer was divided among five points per plot. Bore holes 2 cm in diameter were made vertically in the soil using a Veihmeyer probe. Five holes were spatially arranged in the shape of an x, with one hole at the intersection and one at the end of each arm with a distance of 43 cm between the center point and the end of each arm (Figure 2).

Immediately following the creation of the five holes in a plot, a PVC pipe was inserted into each hole. Using a funnel, 50 mL of the K¹⁵NO₃ solution was poured in each of the five holes, followed by 10 ml of DI water to rinse the pipe. A total of 250 mL K¹⁵NO₃ solution was buried per plot. Each hole was filled with clean quartz sand to approximately 15 cm above the K¹⁵NO₃, followed by a 2:1 sand/bentonite mix until the hole was filled up to within 30 cm from the surface. The bentonite mix was used to prevent preferential root growth down the backfilled hole. The top 30 cm of the hole was filled with topsoil from the plot.

Cover crop biomass sampling at ¹⁵N burial points and preparation for analysis

Cover crop biomass was sampled between 14 Dec 2015 and 31 Dec 2015. Weeds were not collected because they were estimated to be < 5% of biomass. Spring biomass was sampled between 30 April 2016 and 7 May 2016. A plants was sampled if the point where it emerged from the soil fell within a 20 cm radius of each of the five burial points (Figure 2). The tissue types of radish shoot (RADISH LEAF), radish fleshy taproot (RADISH ROOT), rye shoot (RYE), and clover shoot (CLOVER) were sampled within all treatments.

Forage radish was expected to naturally winter kill within one month of the fall sampling date. Therefore we destructively sampled the radish, but returned >95% of the biomass to the plot to decompose. Forage radish was harvested by hand-pulling the fleshy root from the soil. The RADISH ROOT and RADISH LEAF were separated. The RADISH ROOT was thoroughly washed to remove all soil and then weighed. A 4.40 mm diameter bore was taken horizontally through the RADISH ROOT (perpendicular) at the vertical midpoint of the root. The RADISH ROOT (minus the bore hole) was returned to the exact point in the soil from which it came. The RADISH LEAF was weighed. The RADISH LEAF was split into two equal parts down the midline of the plant. One half of the RADISH LEAF was scattered over the replaced RADISH ROOT. The other half of the RADISH LEAF was blended with 30 ml of DI water until liquefied using a food processor. A sample of 30 ml of the blended RADISH LEAF was saved. The remaining RADISH LEAF puree was immediately stored in the refrigerator and scattered within 48 hours over the burial points from the plot harvested. The RADISH biomass was estimated using percent moisture estimates from a previous study.

Where,

Wet biomass = values taken in field measurements

Percent moisture = values estimated based on a 2012 radish variety trial; average RADISH LEAF percent moisture = 97.07% (N = 20; SD = 1.09), average RADISH ROOT percent moisture = 93.55% (N = 20; SD = 1.89) (Lounsbury and Weil, unpublished)

            RYE and CLOVER are expected to overwinter in the study region. Therefore we took non-destructive samples of the biomass of these species in the fall and in the spring. One leaf was sampled from each RYE clump. A “clump” was defined as all rye leaves coming from a single shoot off a tiller. On areas outside of the plots, we harvested areas using the minimally destructive technique and then completely harvested the area to compare the average % N for the two harvest techniques. For the fall samples (N = 20), the ratio of % N from sample/total harvest was on average 0.991, with each sample differing on average 7.86% from the total harvest. For the spring samples (N = 40), the ratio of % N from sample/total harvest was on average 1.01, with each sample differing on average 10.4% from the total harvest.

We took measurements of rye patchiness, height, and percent cover that could be used to estimate biomass. Patchiness was determined by counting the number of clumps. Height was estimated to be the average height of rye leaves within each clump. The leaves were pulled vertical beside a measuring stick to determine the average height. To reduce human error, a single investigator determined the patchiness and height of all plots within a block. To estimate percent cover, a bird’s eye photo was taken of each sampling point; any RADISH in the plot was removed before the photo was taken. Two investigators made independent visual estimates of percent cover from the photo images; the two estimates were averaged for each point.

            On areas outside of the plots, we measured rye patchiness, height, and percent cover and then harvested the area and dried and weighed the biomass in order to correlate measurements to actual biomass. We ran a regression analysis to correlate the measured rye parameters to the dry biomass. The analysis selected for the combination of independent variables with the highest adjusted R². We forced the equation to pass through the origin (i.e., have no y intercept) (Eisenhauer, 2003). Correlation curves were made for each sampling group.

One stem was collected at its base from each CLOVER clump. A “clump” was defined as all clover stems coming from what appeared to be a single root. To reduce human error, a single investigator identified “clumps” for all plots within a replication. Crimson clover biomass was not estimated as we did not expect Crimson clover to pick up ¹⁵N tracer.

For all treatments, TISSUE TYPE dry matter per m² and CC SPECIES dry matter per m² was calculated. The samples of cover crop biomass was dried at 40° C and ground to < 0.1 mm size.

Corn sampling at ¹⁵N burial points and preparation for analysis

Corn plants were sampled at SANDY site on 8 Jun 2016 and at CLAYEY site on 29 Jun 2016, when plants had reached the five leaf collar growth stage (CORN V5). Corn grain (CORN GRAIN) was sampled at SANDY site on 2 Sep 2016 and at CLAYEY site on 7 Oct 2016. Corn was sampled from a 1.8012 m² circular area (0.7572 cm radius) encompassing the five sampling points. The sampling area was split in half; the corn from one half was harvested at V5 stage and the corn from the other half was harvested for grain (Figure 5). The corn biomass was dried at 40° C and weighed. The dried biomass was ground to < 0.1 mm size.

Soil sampling at ¹⁵N burial points and preparation for analysis

In order to investigate NO₃ leaching patterns, we took soil cores at either 10 cm or 20 cm distance to the side of the buried ¹⁵N (DISTN15) at several times during the year (Table 4). The first set of samples was taken on the sandy site on 20 February 2016 and on the clayey site on 10 April 2016. We intended to take these soil cores in late fall, but were delayed until February and April by weather and field soil conditions. Soil cores were taken to 210 cm deep and divided into 15 cm increments. Soil samples were dried at 40°C for at least 48 hours. The weight of the soil at the field moisture level, the weight of the soil after drying and the weight of the gravel that did not pass through the 2 mm sieve was determined. The soil was sieved through a 2 mm sieve and then ground to < 0.1 mm size. We analyzed soil samples from selected depths for AT% 15N/14N (Table 4).

Table 4 Soil samples taken (and depths analyzed in parentheses) per block. The number of asterisks indicate the number of composite cores per sample. Two cores from the same DISTN15 were from two different burial points within a plot; two cores from different DISTN15 were from one burial point within the plot. The letter m indicates samples taken in 3MIX treatment; the letter r indicates samples taken in RYE treatment

¹⁵N analysis and calculations

Biomass and soil samples were analyzed for ¹⁵N at Cornell University Stable Isotope Laboratory using an isotope ratio mass spectrometer (ThermoFinnigan Delta Plus) integrated with an elemental analyzer (Carlo Erba NC2500) through an open split interface (Conflo II). The ¹⁵N is reported as atom AT% 15N/14N. The 15N UPTAKE was calculated using Equation 3 and 4. The 15N RECOVERY was calculated using Equation 5 and 6.

Equation 3. TISSUE TYPE 15N UPTAKE

Equation 4. CC SPECIES 15N UPTAKE

Equation 5. TISSUE TYPE 15N RECOVERY

Equation 6. CC SPECIES 15N RECOVERY

Data analysis

Evaluating if cover crops from experimental treatments contain higher AT% 15N/14N than background level— The background levels (unenriched) of AT% 15N/14N of cover crop tissue were determined by measuring the AT% 15N/14N for each TISSUE TYPE in NO BURIAL plots. We performed an ANOVA for EARLY PLANTING and for LATE PLANTING to test if the background AT% 15N/14N differed according to SITE, TISSUE TYPE, or TISSUE TYPE within CC SPECIES (e.g., RADISH LEAF in RADISH compared to RADISH LEAF in 3MIX).  

The AT% 15N/14N of the TISSUE TYPE in an enriched plot was compared to the AT% 15N/14N of the same TISSUE TYPE, SITE, and SAMPLE DATE in an unenriched plot (background level). A one-sample t-test was performed to determine if an experimental treatment value of AT% 15N/14N was significantly higher than the background level. Because we were only interested in knowing if cover crops from experimental treatments had a higher AT% 15N/14N than the background level (not lower), we used a “lower one-sided t-test”, with the null value (H0) being the AT% 15N/14N of the experimental plot in question. The null hypothesis is that the mean of control values is equal to the H0; the alternative hypothesis is that the mean of control values is less than the H0 (SAS help, T-Test procedure, syntax). Levels of AT% 15N/14N significantly above the NO BURIAL AT% 15N/14N values were interpreted to mean that the cover crop captured some of the buried ¹⁵N tracer.

            Evaluating how much ¹⁵N buried tracer the cover crops captured—Due to the incomplete factorial design (no LATE PLANTING  x 2MIX treatment and no 180 CM 15N DEPTH x LATE PLANTING treatment), EARLY PLANTING and LATE PLANTING results were analyzed separately. For the EARLY PLANTING 15N RECOVERY and the LATE PLANTING 15N RECOVERY, an ANOVA was performed including the independent variables of CC SPECIES, 15N DEPTH, the interaction of CC SPECIES x 15N DEPTH, and the interaction of SITE x CC SPECIES, SITE x 15N DEPTH, and SITE x CC SPECIES x 15N DEPTH. The 15N RECOVERY data were not normally distributed so it was log10 transformed; the log10 transformation normalized the distribution. Thus the dependent variable in the ANOVA was LOG10 15N RECOVERY.

            To test for differences of 15N RECOVERY between CC DATE treatments, an ANOVA was performed including the independent variables of CC DATE, CC SPECIES, 15N DEPTH, all possible interactions, and the interaction of SITE with all of the above factors. The 3MIX CC SPECIES treatment and the 180 CM 15N DEPTH treatment were not included since they were not represented in the LATE PLANTING treatment. The dependent variable was LOG10 15N RECOVERY.

            To test if RYE from spring sampling (SPRING) had higher 15N RECOVERY than RYE from fall sampling (FALL), an ANOVA was performed including the independent variables of cover crop sampling date (SAMPLE DATE) and the interaction of SAMPLE DATE x SITE. A separate analysis was performed for EARLY PLANTING 60 CM, EARLY PLANTING 120 CM, EARLY PLANTING 180 CM, LATE PLANTING 60 CM, and LATE PLANTING 120 CM. The dependent variable was LOG10 15N RECOVERY. For all ANOVA tests, the level of probability considered significant is p < 0.05, unless otherwise stated.

            Evaluating if corn from experimental treatments contain higher AT% ¹⁵N/¹⁴N than background level—A one-sample t-test was performed to determine if 1) CORN V5 and 2) CORN GRAIN samples had significantly higher AT% 15N/14N than the background level. We analyzed CORN V5 and CORN GRAIN for the presence of AT% 15N/14N in plots from all EARLY PLANTING treatments and all NO CC treatments. In addition, for the EARLY PLANTING, 3MIX, 60 CM treatment, a one-sample t-test was performed to determine if corn stalk (at corn grain harvest time) had significantly higher AT% 15N/14N than the background level.

            For CORN V5 and CORN GRAIN 15N RECOVERY, an ANOVA was performed using the 60 CM treatment N15 DEPTH data including the independent variables of CC SPECIES and the interaction of SITE x CC SPECIES. A separate ANOVA was performed using the NO CC treatment CC SPECIES data including the independent variables of 15N DEPTH and the interaction of SITE with 15N DEPTH. The level of probability considered significant is p < 0.05, unless otherwise stated.