Impact of Cover Crops and Nitrogen Management on Soil Bacterial Alpha Diversity

ABSTRACT

Corn (Zea mays L.) nitrogen (N) strategies that account for fertilizer timing and source may differentially affect soil bacterial communities and require further validation when preceded by cover crops (CC). Field studies were conducted (2014–2016) evaluating the effects of daikon radish [Raphanus sativus (L.)], forage oat [Avena sativa (L.)], and no CC combined with multiple N management strategies on soil bacterial alpha diversity metrics. Nitrogen strategies consisted of 179 kg N ha⁻¹ applied pre-plant incorporated (PPI), poultry litter (PL) (61 kg N ha⁻¹) applied PPI plus sidedress (SD) N at V11 (11 leaf collars), starter N (45 kg N ha⁻¹) applied 5 cm beside and below the furrow followed by SD at V4 (4 leaf collars), V11, or 50:50 (split) V4 and V11, and a zero N control. In 1 of 2 years, radish reduced diversity 47–53% during active growth compared to remaining treatments. Inverse relationships regarding diversity of in-row (IR) soils at R1 (silking) and grain yield indicated that increased bacterial diversity did not correspond to increased grain production. Relative to PL and the zero-N control, starter N strategies reduced R1 rhizosphere bacterial diversity, evenness, and richness. Combining organic with soluble N sources may stabilize changes to soil alpha diversity metrics at corn silking.

KEYWORDS:

• 16S rRNA
• corn
• diversity
• nitrogen
• oat
• radish

Introduction

Increased climate variability and concern for Great Lakes water quality continue to emphasize improved nitrogen (N) management strategies to reduce N loss. Previous investigations in the northern Corn Belt highlighted fertilizer strategies focused on N timing and source combinations to improve synchrony between N availability and plant uptake but required adjustments to account for reduced N availability when used in combination with cover crops (CC) (Rutan and Steinke 2019). However, little is known regarding the effects of N fertilizer timing and source combinations on the soil microbiome when used in conjunction with CCs. The importance of the soil microbiome on soil quality, nutrient cycling, plant fitness, and crop production is well recognized as plants, soil, and microbes simultaneously function to influence plant health and productivity (Chaparro et al. 2012; Hills et al. 2020; Schmidt et al. 2019). Studies suggest increasing bacterial community diversity may influence crop yield, N uptake, and maintain ecosystem services under abiotic disturbance (Awasthi et al. 2014; Eisenhauer et al. 2012; Tautges et al. 2016). While the effects of N fertilizer additions and CCs on belowground microbial communities have been well studied separately, understanding their interaction has important implications for soil functioning and crop production (Dai et al. 2018; Fernandez et al. 2016; Kim et al. 2020). More data are required to discern the microbiome response to N management in medium-textured soils which may lead to best management practices for improving soil fertility.

Results are mixed regarding the impact of N fertilizer to soil bacterial diversity including no impact (Fierer et al. 2012), increased (Ikeda et al. 2014), decreased (Dai et al. 2018; Li et al. 2016), or inconsistent effects (Ramirez et al. 2010) observed. As a direct impact to soil bacteria, N sources containing urea and ammonium are beneficial to ammonium oxidizers and nitrifying organisms which require reduced forms of N (e.g., ammonium) as an electron donor (Geisseler and Scow 2014b). However, banding urea and ammonium fertilizers can result in high concentrations of ammoniacal N increasing soil solution ionic strength detrimental to soil bacteria (Geisseler and Scow 2014b; Müller et al. 2006). Nitrogen fertilizer can also indirectly reduce soil microbial diversity due to changes in soil chemistry such as pH rather than a direct osmotic effect on the soil solution (Li et al. 2016; Lupwayi et al. 2012; Zhalnina et al. 2015). Meta-analyses have indicated N fertilization alone decreased soil pH and bacterial taxonomic diversity but increased soil organic C and available N (Dai et al. 2018). Corn N strategies used in combination with CCs may utilize multiple N sources (Rutan and Steinke 2019). Further research is required for improved understanding of soil biological impact when multiple N sources are utilized to adjust N strategies.

In northern Corn Belt states, N fertilizer strategies often include a combination of pre- or at-plant N timings in addition to split applications where reduced early-season N rates are followed by in-season N application between V4 (4 leaf collars) to V10 (10 leaf collars) (Rutan and Steinke 2018). Fertilizer N sources such as urea ammonium nitrate (UAN) are generally surface or sub-surface banded as compared to granular sources such as urea and poultry litter (PL) which may be broadcast and incorporated (Rutan and Steinke 2018). Nitrogen source often dictates placement and decaying CCs may require adjusting N timing and source combinations (e.g., banding N below residue) to meet corn N demands (Bender et al. 2013; Rutan and Steinke 2019). A complex relationship exists between CC residues and fertilizer additions affecting soil biota and mediated through C and N availability (Verzeaux et al. 2016). Cover crop litter quantity and quality (e.g., organic C and C:N) influence soil bacterial community structure as microorganisms acquire C and N from the environment to maintain a C:N ratio of 8:1 (Demoling, Figueroa, and Bååth 2007; Fanin, Hättenschwiler, and Fromin 2014; Marschner, Kandeler, and Marschner 2003). Prior to a corn cash crop, daikon radish CC increased dry matter production 68% more than a forage oat and resulted in 56% greater total N uptake, while both reduced soil N availability 78–84% relative to no CC (Rutan and Steinke 2019). Cover crop residue quality and nutrient availability when combined with multiple N fertilizer sources creates variable mechanisms to influence soil bacterial communities.

Nitrogen fertilizer timing and source combinations when used with CCs may impact soil bacterial communities as an indirect effect on plant growth. Corn roots exude 5–62% of total net belowground C, but C composition may vary by growth stage and differentially affect bacterial communities (Amos and Walters 2006). Copiotrophic bacteria are r-strategists that prefer simple amino acids and predominate prior to R1 corn, whereas oligotrophic bacteria are k-strategists that utilize complex carbohydrates and increase in abundance after R1 (Li et al. 2014). However, fertilizer N sources (e.g., nitric-, ammonium-, and urea-N) have induced corn-mediated structuring of rhizosphere bacterial communities through changes in rhizodepositions (Giagnoni et al. 2016). Zhu, Vivanco, and Manter (2016) observed a positive association between increased N rates, corn root exudate quantity and quality, and abundance of rhizosphere bacteria. Increased root exudation and aboveground biomass due to N fertilization have also increased soil organic C resulting in increased microbial biomass (Geisseler and Scow 2014a, 2014b). Plant N uptake depletes rhizosphere N and has shifted soil bacterial community structure (Bell et al. 2015). Genomic sequencing at multiple corn growth stages using field-derived studies may provide opportunities to assess the impact of CCs when combined with N management strategies on soil bacterial community composition and influences on corn production (Chaparro et al. 2012). A common first approach when evaluating ecological differences between environments is analyzing the alpha diversity of microbial amplicon sequence data (Willis 2019). The objective of this study was to 1) investigate the impact of corn N fertilizer strategies consisting of source and timing combinations on soil bacterial community alpha diversity and 2) determine if effects of N fertilizer strategies are altered when following a CC. Improved understanding regarding changes in soil bacterial communities may assist in choosing corn management practices that improve soil microbiome services.

Materials and methods

Site description and cultural practices

Field experiments were conducted on different sites (183-m apart) in 2014–2015 (designated site year 1 [SY1]) (42°41’19.17“N, 84°28’56.30“W) and 2015–2016 (designated site year 2 [SY2]) (42°41’6.61“N, 84°29’27.35“W) in Lansing, MI, on a Capac loam soil (fine-loamy, mixed, active, mesic Aquic Glossudalf). Each location consisted of a winter wheat (Triticum aestivum L.)-cover crop- corn rotation. Fields were chisel plowed following wheat, disk-harrowed, and leveled (10-cm depth) with a soil finisher. Soils were sampled (20-cm depth) prior to Aug. CC planting, air-dried, and ground to pass through a 2-mm sieve. Soil properties included 5.8–6.0 pH (1:1 soil/water) (Peters, Nathan, and Laboski 2015), 24–55 mg kg⁻¹ P (Bray-P1) (Frank, Beegle, and Denning 2015), 122–136 mg kg⁻¹ potassium (K) (ammonium acetate method) (Warncke and Brown 2015), and 24–34 g kg⁻¹ organic matter (loss-on-ignition) (Combs and Nathan 2015). Triple super phosphate (0–45–0 N–P–K) and muriate of potash (0–0–62) applications occurred prior to corn planting based on soil tests. Cornweed control was acetochlor [2-chloro-N-ethoxymethyl-N-(2-ethyl-6-methylphenyl) acetamide] and glyphosate [N-(phosphonomethyl) glycine] followed by a second application of glyphosate 17–24 d later. Environmental data (moisture and temperature for air and soil) utilized the nearby (approx. 900 m) Michigan Automated Weather Network (http://www.agweather.geo.msu.edu/mawn/, Michigan State University, East Lansing, Michigan; verified 28 Feb. 2020).

Experimental design and treatment application

Eighteen treatments were arranged in a split-plot, randomized complete block design with four replications. The whole plot factor was CC and included ‘The Buster™’ daikon radish, ‘Magnum™’ forage oat (Weaver Seed of Oregon, Crabtree, OR), and no CC. Whole plots measured 55 m by 12 m in length. Cover crops were drill-planted (radish and forage oats at 11.2 and 28.0 kg ha⁻¹, respectively) 14 and 17 Aug. 2014 and 2015, respectively, with a Gandy (Owatonna, MN) Orbit-Air Seeder coupled with John Deere (Moline, IL) double disk openers in 19 cm rows. The no CC received an autumn glyphosate application to ensure no vegetative ground cover. To confirm winterkill, CCs were terminated in Nov. with glyphosate following 79–83 d of growth.

The subplot factor was N management and consisted of five N fertilizer strategies with a zero N control. Subplots measured 4.6 m by 12.2 m in length. Total N rate was equalized to the maximum return to N rate (MRTN) using a 0.10 N:corn price ratio (179 kg N ha⁻¹) (Steinke 2015). Fertilizer strategies consisted of 179 kg N ha⁻¹ applied as pre-plant incorporated (PPI) urea, poultry litter (PL) applied at 2.2 Mg ha⁻¹ (61 kg N ha⁻¹ first-year available N) PPI plus sidedress (SD) N at V11 (i.e., 11 leaf collars), and starter N (45 kg N ha⁻¹) (urea ammonium nitrate [UAN; 28-0-0]) applied 5 cm beside and 5 cm below the furrow (5 × 5) followed by SD at V4, V11, or 50:50 (split) V4 and V11. Corn V4 N was UAN coulter injected 5 cm deep and 38 cm to the side of each row. At V11 SD, an urease inhibitor (CO(NH₂)₂ + n-(n-butyl) thiophosphoric triamide) (Koch Agronomic Services, LLC, Wichita, KS) was mixed with UAN to prevent N volatilization and banded 10–15 cm to the side of each row. Corn was seeded on 1 and 17 May 2015 and 2016, respectively, in 0.76 m rows at 84,510 seeds ha⁻¹ with Dekalb DKC48–12 (98 d relative maturity) (Monsanto Co., St. Louis, MO).

Data collection and processing

Corn measurements

To indicate ear-leaf N status, chlorophyll content was assessed at R1 using a Minolta SPAD 502 chlorophyll meter (SPAD) (Konica Minolta, Tokyo, Japan) (Scharf, Brouder, and Hoeft 2006). The center two rows of each plot were harvested for grain yield, moisture, and test weight using a research plot combine 14 and 18 Oct. 2015 and 2016, respectively. Final grain yields were corrected to 155 g kg⁻¹ moisture.

Soil bacterial DNA genomic sequencing

Soils were sampled (0–10 cm) for bacterial DNA analysis at CC planting and termination, corn planting, corn R1, and corn harvest (i.e., R6). Two soil sampling zones were sampled at growth stage R1 and R6: 38 cm mid-way between two corn rows to represent bulk soil (BR) and between two corn plants within a row to represent root-impacted soil (e.g., corn rhizosphere) (IR). At R1 and R6, six cores per plot were collected and combined for a single sample for both BR and IR soil. Soils were air-dried and ground to pass through a 2-mm sieve (Dolfing et al. 2004).

Amplicons and libraries were constructed according to Kozich et al. (2013). Briefly, soil genomic DNA was extracted from 0.25 g soil using the MO-BIO Power Soil® kit (MO BIO Laboratories, Carlsbad, CA) and stored at −20°C. Polymerase chain reaction (PCR) was used to amplify the V4 hypervariable region (length 250 bp) of the 16S gene using genomic DNA as a template and high fidelity polymerase (Accuprime™ Pfx Supermix, ThermoFisher Scientific, Waltham, MA). The MiSeq Illumina platform (San Diego, CA) was used to sequence amplicon libraries submitted to the Michigan State University Research and Technology Support Facility for next-generation sequencing (East Lansing, MI). Sequence data were processed through the Michigan State University High Performance Computing Center (East Lansing, MI) and analyzed using a previously described analysis pipeline (Kozich et al. 2013) and protocol (http://www.mothur.org/wiki/MiSeq_SOP, verified 19 Apr. 2022) with the Mothur software package (ver. 1.33.2b; Schloss et al. 2009). Operational taxonomic units (OTUs) were based on a 97% sequence identity and classified into the SILVA database.

Community analysis

Sequences were subsampled (9500 and 6000 sequences randomly selected in SY1 and SY2, respectively, 1000 times) to standardize data. Soil bacterial community composition was analyzed using alpha diversity metrics (i.e., membership of taxa within a single sample) and OTU relative abundance. Alpha diversity metrics included estimates of diversity, richness, and evenness. Community richness was assessed as the number of detected OTUs (SOBS) where greater SOBS indicates greater OTU richness. Evenness was assessed using the Shannon evenness index (SEI; 0–1 index). As SEI approaches 1, evenness increases indicating the number of each organism is relatively the same. The inverse Simpson’s diversity index (ISD) was used to assess taxonomic diversity (Schloss and Handelsman 2006). As ISD increases, diversity increases.

All data were subjected to analysis of variance using the GLIMMIX procedure in SAS (ver. 9.4; SAS Institute 2011). Statistical models considered CC, N strategy, and soil sampling zone as fixed effects and assumed random block × CC effects. Data were processed separately by soil sampling time and year due to computational load of microbial DNA sequencing data (Smith et al. 2016). Normality and homoscedasticity of residuals were tested with the UNIVARIATE procedure and Levene’s test, respectively (P ≤ 0.05). When ANOVA generated a sig. F-value treatment least square means were separated using the LINES option of the LS_means statement. Metagenomic sequence and agronomic data were assessed at α = 0.05 and 0.10, respectively. The SLICE statement was used to investigate interacting effects. The CORR procedure was used to generate Pearson correlation coefficients for investigating relationships between alpha diversity metrics and grain yield each year (P ≤ 0.05).

Environmental conditions

At corn planting cumulative May and June rainfall was 128 mm above normal in 2015 but 103 mm below normal in 2016 resulting in contrasting wet and dry soils, respectively (Table 1). In both years, May air temperatures were near normal, while June to August air temperatures were 0.8–1.2°C below and 0.3–1.6°C above 30-yr means in 2015 and 2016, respectively.

Table 1. Monthly cover crop, corn-growing season, and 30-yr mean precipitation and temperature departures from normal data for Lansing, MI, 2015–2016.

	Precipitation^†			Air temperature^‡
Month	2015	2016	30-yr mean	2015	2016	30-yr mean
	Difference (mm) from avg.		mm	Difference (°C) from avg.		℃
Apr.	−50	+2	73	−0.5	−1.3	8.7
May	+25	−32	85	+1.6	0.0	14.7
June	+103	−71	89	−0.8	+0.3	20.0
July	−21	+13	83	−1.2	+1.0	22.1
Aug.	+39	+79	84	−1.0	+1.6	21.3
Sept.	+3	+14	92	+2.1	+0.5	16.9
Oct.	−22	+12	70	+0.7	−1.2	10.4
Nov.	−40	−22	71	+2.3	+0.1	4.6

†Precipitation and air temperature data were collected from Michigan Automated Weather Network (http://www.agweather.geo.msu.edu/mawn/).

‡30-yr mean sources for precipitation and air temperatures: NOAA (https://www.ncdc.noaa.gov/cdo-web/datatools/normals; verified 17 Aug. 2017).

In the 14-d period prior to each soil sampling time, soil min. and max. moisture and temperature were recorded for the study site (Table 2). Prior to corn planting time, soil moisture and temperature were often greater in SY2 relative to SY1. In SY2 soil moisture at R1 was reduced (relative to SY1) likely due to deficit May and June rainfall followed by decreased precipitation in the 14-d period prior to R1 observations. Soil moisture and soil temperature can shift soil bacterial community composition (i.e., β-diversity) with taxa responding differently to moisture stress and altering CO₂ efflux in soils under corn management (Bainard, Hamel, and Gan 2016; Ding et al. 2010). Environmental conditions encountered in the current study likely influenced soil bacterial communities differently each year.

Table 2. Mean soil moisture (30-cm) and soil temperature (5-cm) and cumulative precipitation (mm) data for the 14-d period prior to each soil sampling event (cover crop planting [CC plant], autumn cover crop harvest [CC harvest], corn planting [corn plant], corn silking [corn R1], and corn harvest [corn R6]), in site year 1 (SY1) and 2 (SY2), Lansing, MI.

	CC plant^†	CC harvest	Corn plant	Corn R1	Corn R6
SY1	14-Aug	3-Nov	1-May	20-Jul	9-Oct
Max. soil moist. (cm^3 cm^−3)	0.110	0.211	0.197	0.273	0.174
Min. soil moist. (cm^3 cm^−3)	0.110	0.202	0.190	0.257	0.170
Max. soil temp. (℃)	24.0	4.0	12.0	27.0	16.5
Min. soil temp. (℃)	18.4	2.5	6.3	16.5	13.2
Precip. (mm)	17.0	17.8	7.1	46.2	14.0
SY2	17-Aug	30-Oct	17-May	20-Jul	11-Oct
Max. soil moist. (cm^3 cm^−3)	0.200	0.232	0.250	0.085	0.237
Min. soil moist. (cm^3 cm^−3)	0.179	0.224	0.235	0.079	0.222
Max. soil temp. (℃)	22.9	11.4	16.5	30.0	19.3
Min. soil temp. (℃)	18.9	8.1	10.5	21.4	14.6
Precip. (mm)	65.3	32.5	37.3	42.2	80.8

†Cover crop planting and harvest relative to 2014 and 2015, respectively. Corn planting, R1, and R6 relative to 2015 and 2016, respectively.

Results and discussion

DNA sequence analysis

In SY1 and SY2, 7.1–9.2 × 10⁶ valid sequences were obtained and covered 91,369 and 49,176 OTUs using 97% identity as the cutoff. Collectively, singletons and doubletons (i.e., sequences observed 1× and 2× per sample, respectively) accounted for 73% and 56% of all sequences identified in SY1 and SY2, respectively, indicating rarity was common in the agricultural soils sampled. While soils were dried prior to DNA extraction, many studies have often used fresh and frozen soil samples (Vestergaard et al. 2017). Few differences have been found between fresh and dried soils. Differences have included reduced richness and diversity based on the Shannon index with dried soils but did not affect detectable bacterial responses in regard to agricultural management (Tzeneva et al. 2009). The use of air-dried soils in the current study provides meaningful data, but storage conditions may have impacted amplitude of alpha diversity indices and resulted in slight variation from those reported in literature using fresh or frozen soils.

Alpha diversity metrics

Soil alpha diversity metrics were observed from soils collected in bulk soils of whole plots prior to corn planting and from BR and IR soils collected from subplots at R1 and R6. Additional study data on soil bacterial community beta-diversity as affected by daikon radish and forage oat cover crops in combination with N management strategies have previously been reported (Rutan, Rosenzweig, and Steinke 2022). Data are discussed pertaining first to whole plots (immediately following cover crops but prior to corn planting) for SY1 and SY2 followed by subplots (during corn growth and harvest) at R1 and R6 each site year.

Impact from cover crops prior to corn planting

During autumn SY1, the radish CC resulted in an ISD of 206 which was a significant reduction (47–53%) relative to 392 and 435 for no CC or oats, respectively. Unlike SY1, CCs did not affect diversity in the autumn of SY2 resulting in similar ISD values of 339, 343, and 367, respectively. Other alpha diversity metrics (SOBS and SEI) were not affected by CC and data are not presented. Results contrast the common assertion that increased plant diversity increases soil biological diversity. A reduction in ISD does not indicate a positive or negative affect on the ensuing corn crop. Gains and losses of plant growth promoting bacteria or pathogens can equally impact indices. Diversity differences due to CCs indicate that CC species selection may be more important than the decision to plant a CC. Inconsistent impact to diversity each autumn indicates daikon radish or forage oat CCs may not have the same impact to soil bacterial community composition each year or on different sites and warrants further investigation.

COrn impacts at silking growth stage

No interactions between CC and N strategy (P = 0.11–0.99) were observed in either year. No interaction contrasted previous observations (Verzeaux et al. 2016) and suggests soil bacterial communities responded similarly to N timing and source combinations despite previous CCs. Results of ANOVA indicated soil sampling zone had a large impact on alpha diversity metrics due to the number of significant observations (8 of 12 observations; P ≤ 0.05) and often affected response to CC (3 of 12 observations) and N strategy (6 of 12 observations) (data not shown). Spatial effects due to soil sampling zone indicate that corn roots influenced bacterial communities and can impact community response to N management.

At corn R1 in SY1, N management affected diversity, evenness, and richness when averaged across cover crops and soil sampling zones (Table 3). In general, results support the assertion that soluble N inputs reduce soil bacterial alpha diversity (Li et al. 2016; Lupwayi et al. 2011; Ramirez et al. 2010; Zhou et al. 2017). However, no differences were observed between the zero-N control and PL + V11 SD treatments. The lack of differences suggests that a soluble + organic N source may reduce the impact of N fertilization on soil bacterial community composition. Zhao et al. (2014) suggested including livestock manure with NPK fertilizers as a best management practice to sustain the biodiversity of agricultural soils. Soil bacteria metabolize C, and C additions from manure (e.g., PL) may shift C utilization patterns and increase soil microbial functional diversity (Zhong et al. 2010). Strategies containing urea- or UAN-N sources contributed negligible total C (78 and 38 kg/ha, respectively) relative to PL plus V11 SD (1443 kg/ha). Other studies have also observed increased diversity from organic plus soluble fertilization (e.g., PL plus V11 SD) but attributed results to stabilized soil pH due to manure additions (Zhao et al. 2014). Relative to the PL plus V11 SD strategy and zero-N control, 5 × 5 starter N strategies reduced diversity (6.6–8.6%) and evenness (0.6–0.9%) while PPI N reduced diversity (8.3–13.5%). Strategies utilizing V4 SD reduced OTU richness 4.1–6.8% relative to PL plus V11 SD strategy and the zero-N control, respectively. Results demonstrate that altering N management by adjusting timing and source combinations can influence soil bacterial OTU response to N inputs (Zeng et al. 2016). Supplementing N management with an organic N source such as PL may stabilize negative changes to bacterial communities.

Table 3. Alpha diversity metrics including inverse Simpson’s diversity index [ISD], richness (i.e., number of observed operational taxonomic units [SOBS]), and evenness (Shannon evenness index [SEI]) as affected by N management^† (pre-plant incorporated [PPI] N, poultry litter [PL], or 5 × 5 starter N in combination with sidedress [SD] timings) averaged across cover crops and soil sampling zones at corn growth stage R1 in site year 1 (SY1), and the interaction of N management and soil sampling zone (in-row [IR, between two corn plants within a row] or between row soils [BR, between two corn rows]) averaged across cover crops with ISD (P < 0.01), SEI (P < 0.01), and SOBS (P < 0.01) in site year 2 (SY2), Lansing, MI.

N management	ISD					SEI					SOBS
	SY1
Zero N	363^‡ a^†					0.8852 a					2228 a
Urea PPI	339 b					0.8809 b					2165 ab
PL + V11 SD	371 a					0.8854 a					2211 a
5×5 + V4 SD	327 b					0.8776 c					2121 bc
5×5 + V11 SD	333 b					0.8799 bc					2175 ab
5×5 + split SD	321 b					0.8776 c					2076 c
P > F	<0.01					<0.01					<0.01
	SY2
	Soil sampling zone
	IR		BR		P > F	IR		BR		P > F	IR		BR		P > F
Zero N	305	aB^‡	336	aA	0.03	0.8903	aA	0.8930	abA	0.34	1603	aB	1682	aA	0.01
Urea PPI	281	aB	343	aA	<0.01	0.8841	bB	0.8930	abA	<0.01	1518	bB	1643	aA	<0.01
PL + V11 SD	288	aB	321	aA	0.03	0.8853	abA	0.8909	abcA	0.06	1584	aA	1629	abA	0.15
5×5 + V4 SD	199	bB	310	aA	<0.01	0.8613	dB	0.8861	cA	<0.01	1357	cB	1560	cA	<0.01
5×5 + V11 SD	225	bB	344	aA	<0.01	0.8704	cB	0.8943	aA	<0.01	1367	cB	1660	aA	<0.01
5×5 + split SD	221	bB	318	aA	<0.01	0.8685	cB	0.8884	bcA	<0.01	1381	cB	1574	bcA	<0.01
P > F	<0.01		0.10		–	<0.01		0.04		–	<0.01		<0.01		–

†Means within each column followed by same lower case letters are not significantly different at P ≤ .05.

‡Means within each row followed by same upper case letters are not significantly different at P ≤ .05.

At corn R1 in SY2, soil sampling zone affected OTU diversity, richness, and evenness (P < .01) responses to N management in addition to affecting richness response to CC. However, CCs did not affect richness IR or BR soils and data are not shown. Data are presented by sampling zone for N management (Table 3). Regardless of sampling zone, results were similar to SY1 where PL + V11 SD produced similar diversity, evenness, and richness as the zero-N control. Regarding IR soils, application of UAN-N at planting via 5 × 5 starter N reduced diversity 19.9–34.8%, evenness 1.5–3.3%, and richness 9.0–15.3% from PPI N, PL, and the zero-N control. Additionally, full SD at V4 reduced evenness 0.8–1.0% relative to 5 × 5 + V11 or split SD. In BR soils, V4 SD reduced evenness 0.7–0.9% and richness 5.2–6.0% relative to 5 × 5 + V11 SD. However, ISD was not affected in BR soils. Additionally, trends were also observed among 5 × 5 strategies indicating IR OTU evenness and richness were reduced from the zero-N control. While 5 × 5 N strategies reduced OTU evenness and richness of IR soils from the zero-N control, BR soils were unaffected by the 5 × 5 + V11 SD strategy. No effect on ISD in BR soils from N management and trends among 5 × 5 strategies suggest that differences due to N management of IR soils could have been a result of N management affecting corn growth and the subsequent influence to specific communities in the root zone. Plants impart selection pressure and shape rhizosphere microbial communities through root morphology, exudate release, and nutrient uptake (Bell et al. 2015; Berg and Smalla 2009; Kowalchuk et al. 2002). A rhizosphere effect was likely in SY2 as evidenced by 12 of 18 contrasts where diversity, richness, or evenness of IR soils were reduced relative to BR soils and consistent with Peiffer et al. (2013). Corn rhizosphere composition is influenced by the interaction of corn growth and N fertility (Zhu, Vivanco, and Manter 2016). Nitrogen strategies which affect corn growth may have influenced a corn-mediated rhizosphere response detected in IR soils. Relative to SY1, dry soils at R1 in SY2 likely reduced PL mineralization and resulted in variable N supply to corn. Chlorophyll meter data provide evidence of a 3.1–8.1% reduced SPAD index when PL plus V11 SD was used relative to strategies receiving V4 SD while no differences were observed among strategies receiving N in SY1 where soil moisture was increased (Table S1; S2). In addition, chlorophyll meter data were inversely proportional to IR soil diversity, richness, and evenness (r= −0.59 to −0.66; P < .01) and proportional to grain yield (r = 0.80; P < .01). Reduced N uptake likely affected the influence of corn to IR soil OTU evenness and richness. In most instances, IR soil diversity, richness, and evenness were reduced relative to BR soils suggesting a rhizosphere effect. Differences in OTU metrics due to N strategy in the current study suggests N timing and source are important considerations when predicting effects on belowground ecosystems. Inclusion of PL may stabilize bacterial community response to fertilizer N inputs while including SD prior to V11 may enhance N uptake in dry soils. Future studies are warranted to investigate additional corn SD timings with PL.

Corn impacts at physiological maturity

In SY1, OTU richness was affected by N management 161-d after planting when averaged across cover crops and soil sampling zones (Table 4). When N management included V4 SD, richness decreased 3.3–4.6% relative to PL + V11 SD and the zero-N control, a similar trend that continued from R1. Reduced richness indicated banding UAN subsurface at V4 SD was decreasing the number of bacterial OTUs detected. Soil sampling zone also affected diversity (P = .03) and evenness (P = .01) response to CC (data not shown). Slicing the CC × soil sampling zone interaction indicated no statistical difference between BR and IR soil diversity following a radish CC suggesting radish residues reduced spatial variability between IR and BR soil samples. Lack of differences suggests that a grower’s choice of CC may be more important than the decision to simply plant a CC. Unlike N management, decaying CC residues did not affect communities until R6. If the goal of radish and oat CCs is to impact soil diversity, richness, or evenness and in turn influence plant health or soil resilience, changes observed at R6 and not R1 are likely too late in a single growing season. Future studies which include additional soil sampling beyond R6 may improve understanding of CC impacts to soil resiliency and rotational crops.

Table 4. Main effect of N management (pre-plant incorporated [PPI] N, poultry litter [PL], or 5 × 5 starter N in combination with sidedress [SD] timings) averaged across cover crops and soil sampling zones at corn growth stage R6 on richness (number of observed operational taxonomic units [SOBS]) in site year 1, Lansing, MI.

N management	SOBS
Zero N	2253 a^†
Urea PPI	2207 ab
PL + V11 SD	2243 a
5×5 + V4 SD	2168 b
5×5 + V11 SD	2219 ab
5×5 + split SD	2150 b
P>F	0.03

†Means within each column followed by same lower case letters are not significantly different at P ≤ 0.05.

Similar to SY1, N management affected OTU diversity, richness, and evenness 148-d after planting during the second year of study, but the response was dependent on soil sampling zone. The interaction with soil sampling zone was likely mediated by contrasting moisture conditions between sampling years at R6 (Table 2). Unlike diversity and evenness indices, N inputs did not increase OTU richness at R6, a trend which continued from R1 and also noted in SY1. The 5 × 5 + SD N strategies reduced richness 4.4–7.2% relative to PL of IR soils, while splitting SD timings reduced richness 5.0% as compared to the zero-N control (Table 5).

Table 5. Interaction of N management (pre-plant incorporated [PPI] N, poultry litter [PL], or 5 × 5 starter N in combination with sidedress [SD] timings) and soil sampling zone (in-row [IR, between two corn plants within a row] or between-row soil [BR, between two corn rows]) averaged across cover crops on soil bacterial diversity (inverse Simpson’s diversity index [ISD]) (P < 0.01), evenness (Shannon evenness index [SEI]) (P < 0.01), and richness (number of observed operational taxonomic units [SOBS]) (P = 0.02) at corn growth stage R6 in site year 2, Lansing, MI.

	ISD					SEI					SOBS
Zone	IR		BR		P > F	IR		BR		P > F	IR		BR		P > F
N management
Zero N	310	b^†B^‡	366	aA	<0.01	0.8943	bB	0.9013	aA	<0.01	1744	abA	1758	aA	0.56
Urea PPI	371	aA	353	aA	0.24	0.9003	aA	0.8985	aA	0.34	1740	abA	1680	bB	0.01
PL + V11 SD	366	aA	360	aA	0.67	0.9004	aA	0.8989	aA	0.43	1786	aA	1707	bB	<0.01
5×5 + V4 SD	385	aA	344	aB	0.01	0.9001	aA	0.8960	aB	0.04	1696	bcA	1629	cB	<0.01
5×5 + V11 SD	391	aA	358	aB	0.03	0.9013	aA	0.8994	aA	0.33	1707	bA	1688	bA	0.43
5×5 + split SD	381	aA	357	aA	0.11	0.8999	aA	0.8977	aA	0.25	1657	cA	1669	bcA	0.63
P > F	<0.01		0.59		–	0.01		0.14		–	<0.01		<0.01		–

†Means within each column followed by same lower case letters are not significantly different at P ≤ 0.05.

‡Means within each row followed by same upper case letters are not significantly different at P ≤ 0.05.

Suppression of taxonomic richness across years and sampling times could result from loss of particular taxa or increase in dominant taxa followed by loss of rare taxa (Coolon et al. 2013). Additionally, full SD N applied at V4 reduced richness 4.6% relative to PL, similar to R6 in SY1. Full V4 SD was the only strategy that increased IR soil community metrics relative to BR soils and suggested coulter-injected SD N was still affecting community metrics. Adjusting corn N timing and source combinations affects corn N availability and N uptake but suggests a corn-mediated soil bacterial response is more likely during active plant growth at R1 and an unlikely mechanism at R6 corn (Giagnoni et al. 2016). Two years of data indicate that soil moisture during corn growth may affect consistency of CC and N management impacts to diversity and evenness, while OTU richness may be more sensitive to N management and resilient to weather variability. The duration of bacterial community impacts from N fertilizer application may be extended under periods of reduced May–June rainfall. Utility of CCs to manipulate bacterial soil populations may depend on seasonal rainfall for decomposition or other factors such as CC termination date which may influence residue decomposition rates. Soil bacterial communities appear more sensitive to changes in N management as compared to planting a CC.

Correlations between grain yield and alpha diversity

Within the zero-N control, multiple relationships were observed between grain yield and alpha diversity indices. At corn R1 in SY1, correlation analysis indicated grain yield was inversely proportional to IR soil diversity (r= −0.63; P = .03), evenness (r= −0.73; P ≤ 0.01), followed by richness in BR samples at R6 (r= −0.75; P < .01). At corn R6 in SY2, similar correlations occurred with diversity (r= −0.72; P ≤ 0.01) and evenness (r= −0.80; P ≤ 0.01) in BR samples. Results contrast those of Wu et al. (2018) where Pearson correlations indicated rice yields were sometimes positively associated with Shannon–Weiner and Inverse Simpson indices (r = 0.58 to 0.72; P < .05). However, rice paddies were fertilized which may have confounded the relationship with yield. Inverse relationships between alpha diversity indices and grain yield suggest that functional diversity (i.e., range of organisms that influence ecosystem functions) as opposed to taxonomic diversity may be a better indicator of crop yield and warrants further investigation.

Conclusions

The current study provides an analysis of soil microbial alpha diversity, a common first approach to evaluate ecological differences created between environments (Willis 2019). Results provide evidence that various forms and application timings of N fertilizer will differentially affect soil biology. Both site-years of data indicate increased plant diversity with radish or oat CCs preceding a corn crop did not mitigate effects of N inputs on soil bacterial OTU diversity, richness, or evenness in medium-textured soils. Annual soil alpha diversity indices appear more sensitive to changes in N management strategy rather than CC planting. Increased plant diversity with the use of CCs did not translate into increased soil bacterial diversity. Results suggest a grower’s choice of CC was more important than the decision to plant a CC. Variable rainfall creates variable soil moisture which was a key factor affecting soil bacterial response to CCs and N management. Growers who choose to plant a CC may not expect the same influence to soil bacterial OTUs each year due to inconsistent weather variably affecting CC growth.

Manipulating N fertilizer timing and source influenced the distribution of soil bacterial communities among N strategies. A key finding from 1 year indicated N strategies that increased plant N status likely increased corn’s capacity to mediate a microbial community response. Application of UAN at planting through the 5 × 5 placement is a recommended corn N management strategy in northern climates to satisfy early-season N requirements. This subsurface applied N strategy influenced the corn rhizosphere by reducing diversity, richness, and evenness under limited May–June rainfall conditions. The inverse nature of diversity indices and grain yield relationships suggested that functional diversity rather than taxonomic may be a better indicator of crop yield. While the importance of microbial diversity is still debated, recent evidence suggests that the preservation of community diversity is coupled with ecosystem functioning (Maron et al. 2018). Despite rainfall variability in the current study, including an organic N source with a soluble N strategy prevented declines in soil alpha diversity metrics during corn growth. As genomic sequencing technology improves, future investigations that discern the relationship between ecosystem functioning and stability with diversity may become more common. Manipulating soil biological health to manage a crop will likely require a holistic, long-term approach to the soil microbiome, including impacts to soil fungi as well.

Acknowledgements

The authors would like to thank Andrew Chomas, undergraduate research assistants, and research farm staff for their technical assistance in the field.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/00103624.2023.2211605.

Additional information

Funding

The authors would like to thank the USDA National Institute of Food and Agriculture, Michigan State University AgBioResearch, Michigan State University College of Agriculture and Natural Resources, and the Corn Marketing Program of Michigan for partial funding and support of this project.