Soil-test biological activity with the flush of CO₂: V. Validation of nitrogen prediction for corn production

Abstract

Production of corn (Zea mays L.) requires significant N availability to reach maximum yield potential. Nitrogen fertilizer recommendations generally have ignored site-specific conditions and have focused more on total N demand for representative soils across a region. Recent evidence suggested that site-specific conditions of the biologically active component (0–10-cm depth) could inform the magnitude of yield response to applied N fertilizer. This approach was tested further on 111 fields in Coastal Plain, Piedmont, and Blue Ridge regions (states of North Carolina, South Carolina, and Virginia). Plant-available N (sum of residual inorganic N and net N mineralization during a 24-d aerobic incubation) was inversely proportional to economically optimum N rate scaled to grain production level (r² = .47, p < .001). Soil-test biological activity as a simple, rapid, and reliable indicator of net N mineralization was also predictive of economically optimum N rate (r² = .46, p < .001) and validated an earlier assessment. Greater soil-test biological activity was obtained from private farms than from research stations (259 vs. 172 mg C kg⁻¹ 3 d⁻¹, respectively; p < .001), as well as from fields with minimum tillage, multi-species cover cropping, and amendment with animal manures. Results imply that some farmers are making choices to improve soil health condition and these choices can lead to lower requirement for N fertilizer inputs. A shift towards site-specific N management should focus on soil biological activity and its association with N mineralization as indicators of soil health to increase profit and reduce environmental impacts.

Abbreviations

CVT

cost/value threshold

EONR

economically optimum nitrogen fertilizer requirement

1 INTRODUCTION

Nitrogen recommendations for corn (Zea mays L.) and other non-leguminous crops vary somewhat from state to state, but in general focus on some defined proportion of total N uptake, for example, calculation of N factor per unit of grain produced [18–22 kg N  Mg⁻¹ grain (1.0–1.2 lb N bu⁻¹ grain)] (Morris et al., 2018). Total N uptake serves as a useful upper limit for fertilizer recommendations to help avoid unnecessary waste of exogenous inputs. Assuming half of the mass of the aboveground corn plant as grain and that whole aboveground corn plant contains 12.5 g N kg⁻¹ (Lentz & Ippolito, 2012), then total N uptake should be 25 kg N Mg⁻¹ grain. Nitrogen factor of 18–22 kg N Mg⁻¹ grain would then be 72–88% of total N need, suggesting that soil N would supply 12–28% of total N uptake. Sources of N available for crop N uptake vary, but can be categorized into three primary components: (a) residual inorganic N in surface soil and/or deeper in the profile, (b) organic N divided into a large passive fraction and a smaller active fraction; and (c) exogenous inputs (inorganic N fertilizers, organic amendments, irrigation water) (Morris et al., 2018).

Biological N₂ fixation from crops in rotation with corn can be a significant source of N as well, and this process could be included with any of the three categories, depending on the boundaries of the accounting system. Since N contained in leguminous plants must first undergo mineralization from organically bound molecules, it would typically be an active organic N fraction. However, it might also contribute to an increase in residual inorganic N, and could be classified as a separate exogenous input. A diversity of exogenous N sources are available to supply crops with optimum N at various stages of the growing season, including pre-plant, at planting, and at sidedress just prior to reproductive development when equipment traffic is still possible, for example, V6–V8 stages. One highly variable component of N sourcing is that supplied by mineralization of soil organic matter. Along a widely variable gradient from active to passive organic fractions, characterization and quantification of soil N mineralization from soil organic matter may be the most important to avoid over-application and to optimize return on exogenous N source investment.

Stanford (1973) outlined clearly the components of a sound N management approach: (a) defining the internal N requirement of the crop for expected attainable yield, (b) determining the amount of soil N mineralized during the cropping season, (c) quantifying the amount of residual mineral N present in the root zone early in the cropping season; and (d) estimating the expected efficiency of recovery of plant-available N supply. Internal N requirements of many crops have been generally well established from measurement of total N uptake. Characterizing the soil organic N pool for predicting N requirements of grain crops has been a long-standing and arduous task for generations of soil scientists and agronomists (Hanway & Dumenil, 1955; Schomberg et al., 2009; Stanford, Legg, & Smith, 1973; Waksman & Starkey, 1924). Quantifying the residual inorganic N in the root zone prior to the growing season (Bundy & Malone, 1988) or during the growing season (as pre-sidedress nitrate test; Magdoff, Ross, & Amadon, 1984) has been determined with good success and these residual inorganic N approaches are being used routinely in many regions. Estimating the efficiency of N recovery will depend on a diversity of management (cultivar, tillage system, crop rotation, etc.), environmental (precipitation, temperature, and season), and edaphic factors (soil texture, depth of solumn, parent material, etc.) (Morris et al., 2018).

Core Ideas

 

• Soil N mineralization is a biological process predictable from C mineralization.
• Economically optimum N rate is a function of soil-test biological activity.
• Cost/value thresholds must be part of a fertilizer recommendation system.
• Soil-test biological activity is a key metric of soil health condition.
• Farmers who improve soil health can reduce exogenous N inputs.

Estimating soil N mineralization in the field remains one of the key deficiencies in our ability to accurately adjust N fertilizer requirements for individual fields. Using the successive leaching−incubation method of Stanford and Smith (1972), Cabrera and Kissel (1988) adjusted rate constants for changes in soil temperature and water content with limited success. More rapid chemical extraction tests have been attempted to predict soil N supply and to circumvent the laborious and resource-consuming approach of the successive leaching−incubation method. Some of these tests include the Illinois soil N test to estimate hydolyzable amino sugar N (Khan, Mulvaney, & Hoeft, 2001), hot water-extractable N (Gianello & Bremner, 1986), NaOH-distillable N (Sharifi et al., 2008), and permanganate-oxidizable C (Weil, Islam, Stine, Gruver, & Samson-Liebig, 2003). Some shorter term incubation approaches have also been developed, including 7- or 14-d anaerobic incubation for determination of NH₄ production (Waring & Bremner, 1964), 10-d aerobic incubation for inorganic N accumulation (Franzluebbers, Zuberer, & Hons, 1995), and 3-d aerobic incubation for the flush of CO₂ following rewetting of dried soil (Franzluebbers, Haney, Honeycutt, Schomberg, & Hons, 2000). Soil-test biological activity (as measured by the flush of CO₂) was closely correlated with: (a) net N mineralization under standard laboratory conditions from a variety of soils in North Carolina and Virginia (Franzluebbers et al., 2018a), (b) unamended plant growth under greenhouse conditions (Franzluebbers & Pershing, 2018); and (c) corn yield response to sidedress N applications in an inverse relationship (Franzluebbers, 2018b). Although chemical extraction techniques as N availability indicators can be considered simple and inexpensive, soil-test biological activity deserves attention due to its relatively short incubation period and easy setup for analysis. Importantly, it is a uniquely biological approach to mimic the biological process of N mineralization (Franzluebbers, 2018a). Consistent with results from North Carolina and Virginia, short-term respiration (i.e., 1-d incubation) and corn yield response were highly associated in the Midwest (Yost et al., 2018), and strong association of short-term respiration with N mineralization was found for soils in California (Wade, Horwath, & Burger, 2016).

The lack of a widely acceptable, simple, and reliable soil-specific approach to estimate soil N mineralization has often led to each state characterizing N fertilizer requirements of cereal crops based on major soil types or regions (Rajkovich, Crozier, Smyth, Crouse, & Osmond, 2015). A soil-type/regional approach accounts for an average soil N supplying capacity characterized by some distribution of management conditions via site selection. If site-specific soil condition were an important factor in making N fertilizer recommendations, some historical management details might help to characterize site-specific condition. Soil management can have an immediate and long-lasting effect on soil microbial components and organic C and N fractions, especially in surface soil (e.g., 0–10-cm depth) of fields managed with conservation approaches (Feng et al., 2003; Franzluebbers, 2010a; Mbuthia et al., 2015).

Cost of N inputs, whether inorganically or organically derived, has risen in recent decades and the prospect for further price increase is likely with heavy reliance on fossil fuels to manufacture and distribute N fertilizers (Etienne, Trujillo-Barrera, & Wiggins, 2016). Commodity prices have also risen in the past decade, but fluctuations are more the norm than movement in a single direction (Santeramo & Lamonaca, 2018; Wright, 2011). Assuming a static cost/value threshold (CVT) (i.e., cost of fertilizer relative to value of grain), or even worse to not even account for a CVT in making N fertilizer recommendations, will not be a suitable approach for serving the best interests of the farming community in the future. Additionally, society in general would not be served well without considering CVTs, given the strong negative consequences of excess N on the environment. With fluctuating conditions, a range of CVTs needs to be offered so that farmers can make informed decisions (Morris et al., 2018).

The objective of this research was to test the validity of using soil-test biological activity as a predictor of N fertilizer requirements for optimizing economic return and minimizing potential for environmental loss. The hypothesis was that site-specific soil condition based on soil N supplying capacity could be used to modify N fertilizer requirements. Although soil type/regional approaches account for some soil differences, they currently do not account for the more dynamic, management-controlled changes in soil organic C and N fractions. More specifically, the hypothesis was that soil-test biological activity (as an indicator of potential soil N mineralization, which is a direct source of soil N supply) could predict changes in N fertilizer requirements, because soil-test biological activity is strongly associated with the biologically controlled process of net N mineralization (Franzluebbers, 2018a).

2 MATERIALS AND METHODS

2.1 Experimental fields

A total of 111 fields planted to corn in 2017 and 2018 were sampled in the Mid-Atlantic and southeastern U.S. region. A number of agricultural agents and farmers were interested in collaborating after having received a summary of previous results that showed potential to optimize profit by modifying N fertilizer requirements from a general recommendation system (Franzluebbers, 2018b). The majority of fields were on private and state-sponsored farms in North Carolina, although eight fields were in South Carolina and 14 fields were in Virginia (Supplemental Table S1). Fields were mostly in Coastal Plain (n = 49) and Piedmont (n = 47) physiographic regions, but also included the Blue Ridge in North Carolina (n = 8) and the Great Valley in Virginia (n = 7). Long-term mean annual temperature ranged from 11.2 to 17.2°C and precipitation ranged from 922 to 1274 mm.

2.2 Soil sampling and analyses

In each field, four replicate blocks of 5 by 30 m each were marked for soil sampling in spring in consultation with collaborators. In 2017, a hydraulically mounted probe (4-cm inside diam.) on a utility vehicle was used to obtain eight cores from each replicate block at depths of 0–10, 10–20, and 20–30 cm. In 2018, the same diameter probe fitted for hand sampling with a t-bar arrangement was used to obtain six cores at depths of 0–10 and 10–20 cm. The 20–30-cm depth was not sampled in 2018 due to lack of influence of this depth in explaining yield response (Franzluebbers, 2018b). Occasionally, soil was too dry at the time of sampling to get all cores or all depths. Cores within a block were composited by depth in a paper bag, transported to the laboratory, and dried by placing in an oven at 55°C for ≥3 d until constant weight (sometimes initially by blowing room-temperature air over the sample on a paper plate due to space limitations, followed by oven drying). Dried soil weight was recorded to determine bulk density (volume of soil sampled was 1005 cm³ in 2017 and 754 cm³ in 2018). Soil was then gently crushed with a pestle over a screen with 4.75-mm openings. Stones and residues not passing the screen were weighed and removed from the sample prior to further processing.

Total organic C and N were determined with dry combustion using a Leco TruMac CN analyzer. Routine soil nutrient analyses (soil pH, acidity, cation exchange capacity, base saturation, and extractable Ca, Cu, K, Mg, Mn, Na, P, S, Zn) were conducted by Soil Testing Services of the North Carolina Department of Agriculture and Consumer Services in Raleigh, NC. Soil organic C and N fractions were determined according to Franzluebbers et al. (2018). The following analyses were conducted from two 50-g subsamples in the same incubation jar as a starting point in the sequence. Soil-test biological activity was determined from the flush of CO₂ following rewetting of dried soil (3 d) with aerobic incubation of soil at 50% water-filled pore space and 25°C (Franzluebbers, 2016). Soil microbial biomass C was determined with chloroform fumigation incubation without subtraction of a control and using an efficiency factor of 0.41 (Voroney & Paul, 1984) from one of the subsamples removed at 10 d of incubation. Cumulative C and N mineralization during 24 d of incubation were determined from continuation of the incubation with one of the remaining subsamples. Inorganic N was determined by colorimetric techniques using salicylate-nitroprusside for NH₄–N and hydrazine for NO₃–N (Auto-Analyzer 3, Seal Analytical, Inc.) of filtered extracts from 10 g dried soil and 20 ml of 2 M KCl. Plant-available N was the sum of residual inorganic N (NO₃⁻ + NH₄⁺) and mineralizable N during 24 d of incubation. Sand, clay, and particulate organic C and N concentrations were predicted from ball-milled subsamples scanned by near-infrared spectroscopy (Model 5000, Foss NIR Systems Inc.) that was calibrated to a library of laboratory-determined values (Deiss, Franzluebbers, & de Moraes, 2017) specific to similar types of soils in the current study.

Concentrations of soil properties were also calculated across cumulative depths of 0–20 and 0–30 cm where data were available. This required adjustment by soil bulk density in each layer.

2.3 Cover crop and surface residue sampling

In 2017, fields with significant cover crop biomass at the time of soil sampling had biomass >4-cm height removed from two random 0.25 m² areas in Blocks 1 and 4. Biomass was composited for each field, dried, weighed, ground <1 mm, and C and N concentrations determined with dry combustion (TruMac, Leco Corp.). In 2018, surface residue from all fields was estimated from a 0.07 m² area in each of the four blocks at the time of fertilization. Surface residue was composited across blocks for each field, dried, weighed, ground <1 mm, and C and N concentrations determined with dry combustion.

2.4 Experimental design

Generally, selected fields had limited application of inorganic N prior to sampling in spring (details in Supplemental Table S1). Animal manures and other amendments were sometimes applied in fall/winter/early spring prior to sampling. Starter fertilizer was limited to ≤40 kg N ha⁻¹ after sampling and prior to sidedress. Another source of N input for some fields was in the chemical burndown solution to terminate cover crop.

A randomized block design was established in each field with four replications of four N fertilizer rates. Fertilizer was applied at 0, 56, 112, and 168 kg N ha⁻¹ as urea granules stabilized with N-(n-Butyl) thiophosphoric triamide in 2017. Fertilizer was applied at 0, 90, 179, and 269 kg N ha⁻¹ in 2018 as ESN Smart Nitrogen (44% N), which is a urea granule with a polymer coating that releases N over time, depending on temperature. In both years, fertilizer N was spread evenly across 4.5 m width with a manual spin spreader while walking back and forth along the 7.5-m length of each plot. Fertilizer calibrated to field application rates was weighed into sealed plastic bags 1–3 wk prior to application. The investigator applied N fertilizer by hand at approximately V6 growth stage, although the timing varied from as early as shortly before planting when corn was planted late to as late as V8 on a few fields. Private farmers and research station technicians performed all other crop management.

2.5 Plant sampling and analyses

At physiological maturity, corn grain was harvested from 3.83-m length of one of the center rows of each plot by placing each dehusked ear in a labeled cloth bag. Bags were dried for several days in open air or forced air oven, followed by shelling grain from the cob, and drying grain in an oven at 50°C for two more days before recording constant weight. After oven drying, grain moisture was assumed 7% based on observations from a previous dataset. Yield calculations were made with adjustment of moisture to 15.5% standard.

Regression of yield on the four N rates was determined across the four replications for each of the field trials. All data within each field were first tested for fit to a non-linear function of the form:

$$Y=Y_0+a(1-\mathrm{ex}{\mathrm{p}}^{-\mathrm{b}\times \mathrm{N}})$$

where Y = yield (Mg ha⁻¹), Y₀ = baseline yield without sidedress N (Mg ha⁻¹), a = additional yield potential with limitless N input (Mg ha⁻¹), b = non-linear rate constant, and N = sidedress N fertilizer rate (kg N ha⁻¹). When the non-linear equation produced a negative yield response or an unrealistically rapid rise at the first instance of N input followed by no change thereafter, then a linear regression was fitted to the data. If the linear regression had negative slope, then mean yield across N rates was calculated to assume no response to N input. These were the only three choices used to calculate the following parameters of interest for further statistical evaluation of each field: (a) maximum yield based on regression at the highest N rate tested (Mg ha⁻¹), (b) relative yield without N fertilizer derived from the best-fit regression equation at 0 kg N ha⁻¹ divided by maximum yield (Mg Mg⁻¹), (c) yield response to initial dose of N (empirically derived from the instantaneous yield produced at the first instance of N, that is, the product of regression parameters a and b based on best-fit regression [kg grain kg⁻¹ N]), (d) economically optimum nitrogen fertilizer requirement (EONR) at low cost/value ratio (kg N ha⁻¹), (e) EONR at medium cost/value ratio (kg N ha⁻¹); and (f) EONR at high cost/value ratio (kg N ha⁻¹). Threshold cost/value ratios were calculated from the cost of N fertilizer (US$ kg⁻¹) and value of corn grain ($ kg⁻¹). Urea N prices from 2010–2019 varied from US$0.73–$1.83 kg⁻¹ N (www.dtn.com). Corn grain prices during the same period varied from $0.12  to $0.32 kg⁻¹ grain (www.macrotrends.com). Low threshold cost/value ratio was calculated at equivalent of $1.00 kg⁻¹ N and $0.20 kg⁻¹ grain (=5 kg grain kg⁻¹ N). High threshold cost/value ratio was calculated at equivalent of $2.00 kg⁻¹ N and $0.10 kg⁻¹ grain (=20 kg grain kg⁻¹ N). Medium threshold cost/value ratio was calculated similarly for target of 10 kg grain kg⁻¹ N. These three targets of EONR (kg N ha⁻¹) were also calculated as N factor for economically optimum production (i.e., scaled to yield; kg N Mg⁻¹ grain). Field-specific data can be found in Supplemental Table S2.

2.6 Statistical analyses

Plant response variables were single observations for each of the 101 fields harvested (10 fields were abandoned prior to harvest due to complications). Soil variables were also averaged across replications to obtain single observations for each field and to associate with plant variables. Standard deviation and coefficient of variation of soil variables were calculated for each of the 111 trials. Linear and non-linear regressions were performed between plant and soil variables (i.e., field-specific means) using SAS v. 9.4 and SigmaPlot v. 14. As a way of smoothing responses across numerous, diverse fields, means of six consecutive fields in ranked order of independent soil properties (e.g., plant-available N and soil-test biological activity) were calculated and used in regressions (n = 6 fields in each group). For analysis of regional effects, orthogonal contrasts were arranged according to: (a) eastern (Coastal Plain and Piedmont) vs. western (Blue Ridge and Great Valley), (b) within eastern, and (c) within western. Data were separated into different sets of groups to test for differences in soil properties and yield responses (e.g., between private farms and research stations, between minimal and severe soil disturbance via tillage management, between previous crops of legumes and non-legumes, among previous cover crop types of multi-species, single-species, and none, and between management with and without animal manure application). Unpaired t tests were performed between different groupings of data. Effects were considered significant at p ≤ .05. Significance of correlations among plant and soil variables was set at stricter threshold of p ≤ .01 to avoid spurious associations. Data were plotted with SigmaPlot v. 14.

3 RESULTS AND DISCUSSION

3.1 Soil characteristics

Soil characteristics varied considerably among the 111 fields sampled in this study, and this variation was desirable to create a gradient of conditions that could potentially alter yield response intensity to exogenous N inputs. Most directly relevant to potential yield response variation was thought to be that of net N mineralization and plant-available N (residual inorganic N + net N mineralization). Net N mineralization averaged 73, 24, and 10 mg kg⁻¹ soil 24 d⁻¹ at 0–10-, 10–20- and 20–30-cm depths, respectively. Plant-available N averaged 90, 32, and 18 mg kg⁻¹ soil 24 d⁻¹ at 0–10-, 10–20- and 20–30-cm depths, respectively. Net N mineralization adjusted to bulk density in the field was 56–117 kg N ha⁻¹ 24 d⁻¹ (middle 50% of values) at 0–10-cm depth, 81–161 kg N ha⁻¹ 24 d⁻¹ at 0–20-cm depth, and 85–170 kg N ha⁻¹ 24 d⁻¹ at 0–30-cm depth.

Other soil properties also varied considerably among fields (Table 1). Stratification of many soil properties occurred with soil depth. Depth stratification is common in soils of the southeastern United States, and even more dramatically expressed in soils managed with conservation agricultural approaches (Franzluebbers, 2013).

Table 1. Soil properties (middle 50% of values) among field sites in North Carolina, South Carolina, and Virginia as affected by soil depth

	Soil depth, cm
Soil property	0–10	10–20	20–30
Number of fields sampled	111	104	40
Bulk density, Mg m^−3	1.17–1.33	1.35–1.53	1.32–1.54
Sand, g kg^−1	425–628	360–613	307–560
Clay, g kg^−1	112–268	129–304	218–344
pH (–log [H+])	5.8–6.4	5.6–6.2	5.5–6.0
Cation exchange capacity, cmolc kg^−1	6.3–11.2	4.7–8.0	5.0–7.7
Residual inorganic N, mg N kg^−1	8–20	4–9	6–8
Mehlich-III-extractable P, mg P kg^−1	86–258	35–137	12–120
Mehlich-III-extractable K, mg K kg^−1	121–238	60–132	63–134
Total organic C, g C kg^−1	13.7–26.0	6.3–11.9	4.6–7.8
Total soil N, g N kg^−1	1.02–2.16	0.49–0.97	0.42–0.66
Particulate organic C, g C kg^−1	3.1–5.9	0.9–1.4	0.6–1.0
Soil microbial biomass C, mg C kg^−1	490–1020	184–378	117–225
Cumulative C mineralization, mg C kg^−1 24 d^−1	439–832	146–260	79–151
Net N mineralization, mg N kg^−1 24 d^−1	43–97	14–30	6–12
Plant-available N, mg N kg^−1 24 d^−1	57–118	19–40	11–21
Soil-test biological activity, mg C kg^−1 3 d^−1	154–293	55–102	30–59

3.2 Yield and soil nitrogen supply

Maximum corn grain yield from each of the 101 fields in this study was 9.6 ± 3.7 Mg ha⁻¹ (mean ± standard deviation). No statistical difference occurred in maximum grain yield among physiographic regions (Supplemental Table S2). Without sidedress N, corn grain yield was lower in Blue Ridge fields than in Great Valley fields (5.0 vs 9.7 Mg ha⁻¹, respectively; p = .008). Expressed as relative yield without sidedress N application to that of maximum yield, Coastal Plain fields had lower values than Piedmont fields (0.69 vs 0.80, respectively; p = .01). Mirroring differences in relative yield, soil-test biological activity (174 vs 282 mg kg⁻¹ 3 d⁻¹, respectively; p < .001) and net N mineralization (51 and 90 mg kg⁻¹ 24 d⁻¹, respectively; p < 0.001) were also lower in Coastal Plain locations than in Piedmont locations. Relative yield in Great Valley locations was greater than in Blue Ridge locations (0.94 vs.  0.51, respectively; p = .001). In synchrony, soil-test biological activity and net N mineralization were greater in Great Valley than in Blue Ridge locations (402 vs.  184 mg CO₂-C kg⁻¹ 3 d⁻¹, respectively; p = .003 and 128 vs.  52 mg N kg⁻¹ 24 d⁻¹, respectively; p = .007). The lower the relative yield without sidedress N, the greater the demand for exogenous N supplied by N fertilizer (Morris et al., 2018).

Demand for N can be satisfied with residual inorganic N, mineralization of organic N from soil and plant residues at the soil surface, or from application of fertilizer N (Cabrera, Kissel, & Vigil, 2005). In this dataset, relative yield of 1 indicates full satisfaction of N demand by inherent soil N supply (inorganic and organic N) and relative yield near 0 indicates little satisfaction of N demand by inherent soil N supply. These conditions were the fundamental basis for conducting yield response trials to sidedress N application in relation to soil N supply indicators. These varying soil N supply conditions were also an expression of soil health from a nutrient supply perspective. Across all fields, relative yield was significantly correlated with residual inorganic N, net N mineralization during 24 d of aerobic incubation, and plant-available N (sum of residual inorganic N and mineralizable N), confirming the assumption that N sufficiency could be indicated by some measure of available N supply. For plant-available N and net N mineralization, strongest associations with relative yield were when restricting soil sampling to the surface 0–10 cm (r² = .25 and .19, respectively; n = 101, p < .001). Many of the fields tested in this study deployed conservation agricultural practices that concentrated soil organic matter at the surface, and therefore, the immediate surface of soil was most revealing of biologically active N. However for residual inorganic N, strongest association with relative yield was when sampling the surface 0–30 cm (r² = .20, n = 39, p = .005). Residual soil nitrate is mobile throughout the profile, and so it was logical that inorganic N of the entire profile sampled was most revealing. Across locations, 90% relative yield was achieved when plant-available N (0–10-cm depth) was 173 kg N ha⁻¹. At 100% relative yield, plant-available N was 214 kg N ha⁻¹, and therefore, 41 kg N ha⁻¹ of plant-available N provided yield support for an additional 0.96 Mg ha⁻¹ of grain yield (based on 10% of average maximum yield). The additional yield of 0.96 Mg ha⁻¹ required an additional 27 kg N ha⁻¹ of residual inorganic N at a depth of 0–30 cm. Assuming 15 g N kg⁻¹ contained in corn grain, the additional grain should have required only 14 kg N ha⁻¹, suggesting that N-use efficiency was 35 and 53% for plant-available N and residual inorganic N, respectively. These efficiency values are consistent with many observations of fertilizer-uptake efficiency (Cassman, Doberman, & Walters, 2002). On irrigated corn in Colorado, N fertilizer requirement to achieve near maximum yield was 19.5 kg N Mg⁻¹ grain (Halvorson, Mosier, Reule, & Bausch, 2006).

Sites with low relative yield caused by insufficient N supply were expected to have large yield response to initial dose of N fertilizer, whereas sites with high relative yield were expected to have low yield response intensity. In this dataset, there was an inverse relationship, but it was not as strong as expected (r² = .29, p < .001). Therefore, data were separated into fields with different yield response model types (i.e., non-linear, linear, and mean) to see if this might have affected responses. Surprisingly, fields with non-linear yield response to N fertilizer application had even weaker association between relative yield and yield response to initial dose of N (r² = .08, n = 49, p = .05). Fields with linear yield response to N fertilizer application had stronger inverse association (r² = .65, n = 35, p < .001). These differing results suggest that fields with linear response had legitimate responses to the gradients in soil N availability and/or fertilizer N application, and that fields with non-linear response might have been constrained by additional factors other than N. Constraining factors may have included water stress, other nutrient stresses, or root-limiting texture or subsurface acidity issues (Morris et al., 2018).

3.3 Yield variation

Yield response to sidedress N application varied considerably among fields (Supplemental Table S2). Economically optimum N rate was significantly negatively associated with plant-available N (residual inorganic + mineralizable N) at low, medium, and high CVTs (r² = .21, p < .001; r² = .13, p < .001; r² = .09, p = .003, respectively). Large variability between yield response parameters and soil N supply properties was somewhat surprising, as this study employed a small-plot approach to try to narrow random variation. In a previous study, a diversity of field experimental approaches was investigated, including unreplicated strip trials with pseudoreplication along the length of strips, replicated strips, and replicated small-plot studies (Franzluebbers, 2018b). Similarly wide variation in yield response in the previous study was thought to be possibly attributable to the diversity of field approaches. Results of the current study with a uniform experimental approach suggest that wide variations in yield response were due to factors other than experimental approach. It is possible that within-field variations were simply greater than among experimental blocks employed here, and that larger experimental plots that are wholly harvested need to be considered (e.g., field-length strip trials).

To better understand the source of variation in yield response characteristics, coefficient of variation in various plant and soil properties was calculated for multiple tests on the same private farm or research station (Table 2). A total of 31 locations had multiple tests. Two and three tests on the same farm occurred 13 times each, while four tests occurred on three farms, and six and seven tests occurred on one farm each. Soil properties in the surface 10-cm depth had coefficient of variation among multiple tests on the same farm varying from 4 to 31% (18 ± 8% as mean ± standard deviation), while yield characteristics varied from 15 to 78% (46 ± 18% as mean ± standard deviation). As a comparison, median coefficient of variation in soil properties among the four replicates across the 111 fields and three depths (which expresses variation within a field as compared to variation among fields on the same farm) was 13 ± 6% for the same soil properties. Replicate variation in soil properties is shown in Figure 1. With only slightly greater variation in soil properties among fields compared to within fields, the two to three times greater variation in yield response characteristics than soil properties remains intriguing. Within-field variation in both soil and yield responses should be considered in future experiments to isolate profile-specific and/or landscape-specific causal factors. Future field trials might aim for reducing variation of EONR with replicated field-length strips (Rzewnicki et al., 1988).

Table 2. Mean coefficient of variation of yield characteristics and soil properties (0–10-cm depth) among multiple field tests conducted at the same private farm/research station

Yield characteristic	Coefficient of variation (%)	Soil property	Coefficient of variation (%)
Yield without sidedress N fertilizer, Mg ha^−1	29	Soil bulk density, Mg m^−3	5
Relative yield without N fertilizer (fraction)	15	Soil pH	4
Maximum yield with N fertilizer, Mg ha^−1	23	Cation exchange capacity, cmolc kg^−1	16
Yield response at initial dose of N, kg grain kg^−1 N	78	Soil-test P, g m^−3	31
EONR at low CVT, kg N ha^−1	52	Soil-test K, g m^−3	27
EONR at medium CVT	51	Residual inorganic N, mg kg^−1	28
EONR at high CVT	50	Total organic C, g kg^−1	19
NF at low CVT, kg N Mg^−1 grain	53	Soil microbial biomass C, mg kg^−1	20
NF at medium CVT	54	Soil-test biological activity, mg kg^−1	19
NF at high CVT	56	C mineralization, mg kg^−1 24 d^−1	18
		N mineralization, mg kg^−1 24 d^−1	18
		Plant-available N (inorganic + mineralizable), mg kg^−1	16

• Note: CVT, cost/value threshold; EONR, economically optimum N fertilizer rate.

FIGURE 1

Coefficient of variation of soil properties from four field replicates across 111 site-years and three depths (0–10, 10–20, and 20–30 cm). Mid-point is median, box represents range of 25–75% of observations, and extension bars are 90th percentile

3.4 Soil and yield associations

Similar to results from an earlier study (Franzluebbers, 2018b), the N factor for economically optimum corn grain production at low CVT was negatively associated with plant-available N (r² = .11, n = 99, p < .001) and soil-test biological activity (r² = .07, n = 99, p = .007). When data were averaged over six consecutive fields to smooth large yield variations among fields, plant-available N was more closely associated with the N factor for economically optimum production at low CVT than at medium or high CVTs (Figure 2). Soil-test biological activity was also more closely associated with the N factor for economically optimum production at low CVT than at medium and high CVT, as well as having slightly better associations than with plant-available N. Plant-available N provides a direct estimate of soil N supply, but soil-test biological activity provides a rational, low-cost, easily determined approach to estimate potential soil N mineralization. The concept of basing N fertilizer requirement on some measure of soil-test biological activity is consistent with results found from 17 field studies in the Midwest(Yost et al., 2018). The approach of improving N efficiency of corn and other crops with site-specific N requirement via soil-test biological activity should have value for a wide array of stakeholders.

FIGURE 2

Nitrogen factor for economically optimum yield production as affected by plant-available N and soil-test biological activity at low cost/value threshold (CVT) (5 kg grain kg⁻¹ N), medium CVT (10 kg grain kg⁻¹ N; dashed line), and high CVT (20 kg grain kg⁻¹). Data points are means from six fields when sorted by ranked plant-available N and soil-test biological activity (upright open triangles are for low threshold, gray squares are for medium threshold, and inverted open triangles are for high threshold). ns is not significant at p ≤ .05. * Significant at p ≤ .05. ** Significant at p ≤ .01. *** Significant at p ≤ .001

The low, medium, and high CVTs in Figure 2 could serve as pseudo-boundaries for N fertilizer recommendations, given the fluctuating price conditions of fertilizer and commodity grain. The N factor boundaries at a given level of soil-test biological activity were somewhat greater than those reported earlier (Franzluebbers, 2018b), partly due to greater levels of pre-sidedress N applications in the current study (35 ± 41 kg N ha⁻¹). If pre-sidedress N application were discarded from calculations, then the relationships between studies would have been nearly identical. These results suggest that despite different farms and different years of evaluation, similar results were obtained. As an example, a soil with 100 mg kg⁻¹ soil-test biological activity would have N factor recommendation of 24.7, 18.5, and 12.8 kg N Mg⁻¹ grain at low, medium, and high CVTs, respectively. This would translate into N fertilizer rates of 247, 185, and 128 kg N ha⁻¹ to achieve a yield target of 10 Mg ha⁻¹ at low, medium, and high CVTs, respectively. A soil with 500 mg kg⁻¹ soil-test biological activity would have N factor recommendation of 8.3, 6.5, and 6.0 kg N Mg⁻¹ grain at low, medium, and high CVTs, respectively. To achieve a yield target of 12.5 Mg ha⁻¹, N fertilizer recommendations would be 104, 81, and 75 kg N ha⁻¹ at low, medium, and high CVTs, respectively. These scenarios illustrate the large implications of both prevailing economic conditions and soil biological conditions on adjustment of N fertilizer rates. They also highlight the much-reduced impact of cost/value fluctuations on N input adjustments when soil biological activity is at a high level. Therefore, these data provide evidence that improving soil health condition could lead to economic returns, if agricultural advisers and growers were to account for this condition.

The lower and upper limits described in Figure 2 serve to describe how sensitive profit margin of corn grain production would be to CVTs. If CVT were currently high (i.e., 20 kg grain kg⁻¹ N) and a grower had a 65-ha field with soil-test biological activity of 100 mg kg⁻¹, to which the grower would normally apply 265 kg N ha⁻¹ to achieve a yield goal of 12.5 Mg ha⁻¹ (i.e., target of 21.2 kg N Mg⁻¹ grain), then an expected deficit of $13,650 could have occurred as a result of large over-application of costly N fertilizer (i.e., $2.00 kg⁻¹). Alternatively, if CVT were currently low (i.e., 5 kg  grain kg⁻¹ N) and a grower had a 65-ha field with soil-test biological activity of 400 mg kg⁻¹, to which the grower would normally apply 225 kg N ha⁻¹ to achieve a yield goal of 12.5 Mg ha⁻¹ (i.e., target of 18 kg N Mg⁻¹ grain), then an expected deficit of $4,550 could have occurred as a result of moderate over-application of relatively cheap N fertilizer (i.e., $1.00 kg⁻¹). In both of these scenarios, if the grower had used the predicted N factor based on soil-test biological activity (as shown in Figure 2), but the actual yield response required 25% greater N fertilizer rate (i.e., 200 kg N ha⁻¹ required to optimize yield rather than 160 kg N ha⁻¹ predicted by soil-test biological activity in Scenario 1 and 194 kg N ha⁻¹ required to optimize yield rather than 155 kg N ha⁻¹ predicted by soil-test biological activity in Scenario 2), then an expected deficit of ∼$2,800 could have occurred [assuming missed yield response at near N saturation was one-third of the yield response at initial dose of N fertilizer of 33 kg grain kg⁻¹ N (average across sites)]. Both under-fertilization and over-fertilization cost growers lost revenue, and therefore, targeting the proper N dose will be of great importance to growers. Over-fertilization has ancillary costs to society in water quality cleanup (e.g., Chesapeake Bay pollution reduction efforts), drilling for new potable groundwater to avoid contamination by nitrate, and degradation of ecological conditions of farm landscapes, coastal estuaries, and indirect emission of nitrous oxide to the atmosphere. Improving N fertilizer recommendations based on site-specific soil conditions should be high priority for us to help solve many of these on-farm and off-farm issues.

The use of soil-test biological activity as a proxy for net N mineralization, and therefore, as a predictor of soil N supply has a strong foundation. Net N mineralization was highly associated with soil-test biological activity within each of the three depths sampled in this study (Figure 3). Intercept and slope were very similar across depth increments, despite the large difference in range of values among depth increments. In contrast, the association of residual inorganic N to soil-test biological activity was much weaker, but still significant at 10–20- and 20–30-cm depths. The weak, but significant association between residual inorganic N and soil-test biological activity was likely indirect from greater residual inorganic N from mineralized N in the field just prior to the time of sampling. Small inorganic N applications prior to sampling in some fields and none in others could have caused large variability in the association, as was observed.

FIGURE 3

Net N mineralization during 24 d of aerobic incubation and residual inorganic N in association with soil-test biological activity at depths of 0–10, 10–20, and 20–30 cm. ns, not significant. ** Significant at p < .01. ***Significant at p < .001

As the best estimate of soil N supply, plant-available N from the sum of residual inorganic N and mineralizable N during 24 d of aerobic incubation was also highly associated with soil-test biological activity (Figure 4). Although residual inorganic N is a key contributor to plant N uptake, it was only a fraction of potentially available N at the time of sampling in spring (0.19 ± 0.11 kg kg⁻¹ at depth of 0–10 cm, 0.24 ± 0.10 kg kg⁻¹ at depth of 10–20 cm, and 0.46 ± 0.13 kg kg⁻¹ at depth of 20–30 cm). Residual soil nitrate was 7 ± 11 kg ha⁻¹ at 0–10 cm, 10 ± 15 kg ha⁻¹ at 0–20 cm, and 14 ± 16 kg ha⁻¹ at 0–30 cm. The association of net N mineralization with soil-test biological activity was even stronger than from plant-available N, verifying that soil-test biological activity should be considered a rapid, robust, and reliable indicator of soil N supply.

FIGURE 4

Association of plant-available N with soil-test biological activity across three soil depths

Estimating soil N supply capacity via plant-available N (i.e., residual inorganic N + mineralizable N in 24 d) would be the most direct and scientific approach. However, considering expense and time as constraints for soil testing, substitution of soil-test biological activity for plant-available N would seem to be reasonable. Consistent relationships between the two soil variables (i.e., net N mineralization and soil-test biological activity), between these soil variables and the N factor for economically optimum production, and between N factor and soil-test biological activity from this and the previous evaluation (Franzluebbers, 2018b) give confidence for this substitution.

Large variation in estimating EONR has been expressed as a concern across many different experiments in the United States (Morris et al., 2018). Unless field-length, replicated strip trials can be conducted (Rzewnicki et al., 1988; Shapiro, Kranz, & Parkhurst, 1989), large variations in estimating EONR should be considered a reality and could be overcome with alternative approaches, such as conducting numerous individual experiments over a wide range of conditions to characterize a population of inference as in this study. Relative standard error (i.e., standard error divided by the mean) for the relationship between N factor for economically optimum production at low CVT and soil-test biological activity (Figure 2) was 10% for the intercept and 27% for the slope estimate. Relative standard error was 12 and 32%, respectively, at medium CVT, and 17 and 52%, respectively, at high CVT. If each individual field were used in the regression instead of the six-neighbor approach, relative standard errors of intercept and slope were of the same magnitude (13 and 36% at low CVT, 16 and 47% at medium CVT, and 18 and 59% at high CVT, respectively). These variations were within the range of coefficients of variation for N factor to meet economic production in the three physiographic regions of North Carolina (29–47%) (Rajkovich et al., 2015).

3.5 Farm type effects

In a previous investigation, net N mineralization was greater from soil collected on private farms than on state-sponsored research stations (Franzluebbers et al., 2018). This comparison was tested further with the current dataset having larger number of observations (Table 3). Several soil properties at a depth of 0–10 cm were greater on private farm fields than on research stations, including total organic C, residual inorganic N, soil-test biological activity, soil microbial biomass C, cumulative C mineralization during 24 d of aerobic incubation, and net N mineralization. Soil bulk density was lower on private farm fields than on research stations at a depth of 0–10 cm. Many other properties were not different between sources of studies, including clay and sand concentrations, cation exchange capacity, soil pH, soil-test P and K, and all of the soil organic C and N fractions below 10-cm depth. Of the 87 fields on private farms, 34 fields had an organic amendment applied. Even without those fields with organic amendment, soil-test biological activity and net N mineralization at a depth of 0–10 cm were either greater or tended to be greater on private farms than research stations (221 vs. 172 mg CO₂–C kg⁻¹ soil 3 d⁻¹, respectively, p = .02; 64 vs. 54 mg N kg⁻¹ soil 24 d⁻¹, respectively, p = .16).

Table 3. Soil and surface residue properties and associated corn grain yield characteristics as affected by testing on research stations versus private farms

Response variablea	Private farm	Research station	Pr > F
Soil properties (87 private fields and 24 research station fields)
BD at 0–10-cm depth	1.22	1.33	<.001
Clay at 0–10-cm	200	203	.90
Sand at 0–10-cm	491	556	.10
CEC at 0–10 cm	9.2	8.4	.33
STP at 0–10 cm	187	175	.73
STK at 0–10 cm	205	163	.11
RIN at 0–10 cm	18	12	.05
pH at 0–10 cm	6.13	6.06	.51
pH at 10–20 cm	5.87	5.93	.67
pH at 20–30 cm	5.66	5.91	.38
TOC at 0–10-cm	21.5	15.7	.004
SMBC at 0–10 cm	820	593	.008
CMIN at 0–10 cm	716	482	.001
NMIN at 0–10 cm	78	54	.006
NMIN at 10–20 cm	25	21	.23
NMIN at 20–30 cm	10	13	.20
STBA at 0–10 cm	259	172	<.001
STBA at 10–20 cm	83	69	.19
STBA at 20–30 cm	46	55	.38
Cover crop biomass (26 private fields and 5 research station fields in 2017 only)
Total C (kg ha^−1)	524	464	.72
Total N (kg ha^−1)	21	15	.33
Total surface residue cover (49 private fields and 17 research station fields in 2018 only)
Total C (kg ha^−1)	3018	1982	.06
Total N (kg ha^−1)	126	92	.15
Yield characteristics (76 private fields and 23 research station fields)
Relative yield (Mg ha^−1)	8.0	4.8	<.001
Maximum yield (Mg ha^−1)	10.3	7.5	.001
Relative yield (fraction)	0.77	0.64	.005
Initial yield response	30	36	.56
PSN	42	9	<.001
EONR at low CVT	155	142	.58
EONR at medium CVT	118	99	.38
EONR at high CVT	84	64	.26
NF at low CVT	17.3	24.3	.06
NF at medium CVT	13.7	16.7	.36
NF at high CVT	10.3	10.5	.92

• a BD, bulk density (Mg m⁻³); Clay and sand  (g kg⁻¹); CEC, cation exchange capacity  (cmol_c kg⁻¹); STP, soil-test phosphorus (g m⁻³); STK, soil-test potassium (g m⁻³); RIN, residual inorganic N (mg kg⁻¹); TOC, total organic C (g kg⁻¹); SMBC, soil microbial biomass C (mg kg⁻¹); CMIN, cumulative C mineralization (mg kg⁻¹ 24 d⁻¹); NMIN, net N mineralization (mg kg⁻¹ 24 d⁻¹); STBA, soil-test biological activity (mg kg⁻¹ 3 d⁻¹); PSN, pre-sidedress N (kg N ha⁻¹); EONR, economically optimum N fertilizer rate (kg N ha⁻¹); CVT, cost/value threshold; NF, nitrogen factor (kg N Mg⁻¹ grain).Initial yield response = yield response at initial dose of N (kg grain kg⁻¹ N)

Differences in soil organic C and N fractions in the surface 10 cm were associated with changes in yield response characteristics (Table 3). Both maximum yield and yield without sidedress N application were greater from fields on private farms than from research stations, as well as relative yield from the ratio of these two parameters. Part of the greater yield without sidedress N application was from greater basal rate of N fertilizer applied preplant or at planting on private farms than on research stations (42 vs. 9 kg N ha⁻¹, respectively; p < .001). Grain yield response at initial dose of sidedress N fertilizer was not different between sources of studies, nor was economically optimum N rate at low, medium, or high CVTs. However, the N factor scaled to production level tended to be lower on fields from private farms than on research stations at low CVT (17.3 vs.  24.3 kg N Mg⁻¹ grain, respectively; p = .06), but not at medium and high CVTs. These results suggest that fields from private farms were generally more enriched in surface soil organic C and N fractions and that enrichment led to lower requirement for supplemental N fertilizer. Nitrogen invested on the farm was being preserved with no-tillage management, diverse crop rotations, use of cover crops, and through animal manure applications. A possible impact of that investment could be realized with lower need for annual N fertilizer inputs. This analysis does not suggest that all private farms in the region have better surface soil conditions than research stations, but only that many of the fields sampled on private farms had better conditions. It is duly noted that more progressive private farmers than average were likely included in this sampling. It is also not to say that research stations were all below average, e.g. fields at Cherry Research Farm managed with no-tillage following sod, a field at Piedmont Research Station managed with no-tillage and cover crops, and a field at Mountain Research Station managed with no-tillage following sod had soil-test biological activity of 297 ± 75 mg kg⁻¹ d⁻¹, which was no different from the average on fields from private farms.

Given the large number of fields under investigation and as a means to show clearer the relationship of key soil properties to corn grain yield responses, data were sorted into ascending order of soil properties and means calculated for six consecutive fields for private farms and research stations separately. This reduced the number of observations from 111 to 19. Both grain yield response to initial dose of N fertilizer and N factor for economically optimum yield production at low CVT were negatively related with plant-available N (r² = .57 and .47, respectively; p ≤ .001). Soil-test biological activity also had significant negative association with these yield responses (Figure 5). Observations from private farms and research stations followed the same relationship for N requirement as affected by either plant-available N or soil-test biological activity. These results further support the attribution of biologically active soil C and N fractions as a key determinant for differentiating N factor for economically optimum yield production, and not whether fields were managed privately or not (Table 3). These results also validate the concept for using soil-test biological activity as a rapid, reliable, and robust indicator of soil N supply to meet the demands of corn grain in the Mid-Atlantic and southeastern U.S. region, as suggested in a previous study (Franzluebbers, 2018b).

FIGURE 5

Nitrogen factor for economically optimum yield production (at low cost/value threshold) and grain yield response to initial dose of N fertilizer as a function of soil-test biological activity. No differences in association occurred between private farms and research stations, and therefore, a common regression was plotted

3.6 Tillage effects

The majority of fields were managed with no tillage (n = 88). Other fields were managed with disk tillage (n = 10), strip tillage (n = 9), and plow tillage (n = 4). In the following, comparisons were made between minimal disturbance (no-till + strip-till) and severe disturbance (disk-till + plow-till). As expected, surface residue C and N contents were greater under minimal than under severe soil disturbance (3037 vs. 675 kg C ha⁻¹, p = .001; 129 vs. 29 kg N ha⁻¹, p = .001). Total organic C at 0–10-cm depth was greater under minimal than severe soil disturbance (20.9 vs. 15.8 g kg⁻¹, p = .04). Soil-test biological activity and net N mineralization at 0–10-cm depth were even more strongly differentiated between soil disturbance levels (254 vs.  142 mg CO₂–C kg⁻¹ soil 3 d⁻¹, p < .001; 78 vs.  41 mg N kg⁻¹ soil 24 d⁻¹, p < .001). Stratification ratio of total organic C (i.e., concentration at 0–10-cm depth divided by concentration at 10–20-cm depth) was greater under minimal than under severe soil disturbance (2.2 vs.  1.5 kg kg⁻¹, p < 0.001). Corn grain yield was greater with minimal than with severe soil disturbance (9.9 vs.  7.0 Mg Mg⁻¹, p = .01). Even with this greater yield achievement, fields with minimal soil disturbance had the same level of EONR as with severe soil disturbance (i.e., 152 kg N ha⁻¹). Therefore, the N factor at low CVT tended to be lower with minimal soil disturbance than with severe soil disturbance (18.1 vs.  26.2 kg N Mg⁻¹  grain, respectively; p = .12). Better soil-surface biological conditions having greater potential N mineralization with minimal soil disturbance provided a reserve of organic N supply that lowered external N input requirements. Greater soil organic C and N fractions with long-term minimal tillage systems were more directly expressed in greater relative yield without sidedress N application compared with severe soil disturbance (0.76 vs.  0.62 Mg Mg⁻¹, p = .05).

Globally, yield with no-till management has been shown to be reduced or similar compared with more conventional inversion tillage systems (Pittelkow et al., 2015). However, under the long, hot summertime conditions in the southeastern United States, significant yield improvement with adoption of no-till systems can be attained (Endale et al., 2002; Franzluebbers, 2005). The reason for this yield improvement is often linked to less plant water stress from a protected soil surface with a blanket of crop and cover crop residues that reduces evaporation losses (Klocke, Currie, & Aiken, 2009). It may also be that greater water infiltration and less water runoff occurs with improved soil-surface organic matter conditions (Raczkowski, Reyes, Reddy, Busscher, & Bauer, 2009; Sullivan, Truman, Schomberg, Endale, & Franklin, 2007). This current survey of on-farm results suggests that investment in conservation tillage equipment and a farmer's dedication to its approach over time leads to improved surface soil condition, consistent with results of many research investigations (Angers & Eriksen-Hamel, 2008; Franzluebbers, 2010a; Haddaway et al., 2017). This survey also illustrates the possibility to: (a) reduce nutrient runoff losses (Franzluebbers, 2010b), (b) lower cost of field operations (Allmaras & Dowdy, 1985); and (c) reduce N fertilizer requirements to further improve profitability. The observation that N factor could be reduced with minimal soil disturbance compared with conventional tillage systems is a key result that counters a recommendation for greater N inputs on fields converted to conservation tillage (particularly in the first few years) (Derpsch et al., 2014). In short-term tillage studies, crop yields without N fertilizer have often been reported lower with no tillage than with conventional tillage, such as in Nebraska during 4 yr of corn (Eghball & Power, 1999) and in Colorado during 5 yr of irrigated corn (Halvorson et al., 2006). However, the results in the current study are consistent with other observations in the southeastern United States, in which greater yields were implied with time under no-till management, as well as reduced N fertilizer requirements due to accumulation of organic-enriched surface soil (Franzluebbers, 2005). Further, on a silt loam soil in Kentucky, corn grain yield without N fertilizer was significantly lower with no tillage than with conventional tillage early in the study, but over time became greater with no tillage than with conventional tillage (Ismail, Blevins, & Frye, 1994). Based on regression over time in this Kentucky study, equivalent yield between tillage systems occurred at 14 yr from adoption of no tillage. Soil organic C and N in the surface 5 cm at the end of 20 yr was 66 ± 13% greater under no tillage than under conventional tillage. Greater yield, improved net N mineralization, and lower N fertilizer requirements with minimal than with severe soil disturbance in this study are supported by long-term systems research. Investigation of multiple, on-farm trials with growers that have adopted such systems provides one approach to validate key sustainability concepts.

3.7 Crop rotation effects

Soil properties and yield responses of corn following soybean [Glycine max (L.) Merr.] and peanut (Arachis hypogaea L.) (n = 59) were compared with properties and responses following non-legumes (n = 52; barley [Hordeum vulgare L.], corn, cotton [Gossypium hirsutum L.], hay, sunflower [Helianthus annuus L.], tobacco [Nicotiana tabacum L.], tomato [Solanum lycopersicum L.], and wheat [Triticum aestivum L.]). Grain yield tended to be greater following legume grain crops (10.2 Mg ha⁻¹) than following other crops (9.0 Mg ha⁻¹) (p = .09). Relative corn grain yield without N fertilizer was not different between previous crop groups. Total organic C, soil-test biological activity, and net N mineralization at depth of 0−10 cm did not differ between these groups. Yield response at initial dose of N fertilizer application was also not different between these two previous crop groups, nor was the N factor at either low, medium, or high CVT. Effects between corn and soybean as previous crop were not significant either. Rotation with legume grain or forage crops has been shown to enhance corn grain production (Baldock, Higgs, Paulson, Jackobs, & Shrader, 1981; Jawson, Franzluebbers, Galusha, & Aiken, 1994). Research in Nebraska showed that corn grain yield with optimum N input was not affected by rotation in some years (Varvel & Peterson, 1990), but was improved with rotation in other years (Peterson & Varvel, 1989). Corn production in North Carolina was greater following legume cover crops than without a cover crop, irrespective of N input level (Wagger, 1989), a result similar in response to that in the current study. Lack of difference in yield response to applied N between legume and non-legume previous crop in this study was consistent with a similar need for N to optimize yield in corn following corn, corn following cotton, and corn following soybean in Tennessee (Boyer et al., 2013).

These results suggest that crop rotation is not a clearly distinguishable characteristic that affects yield response to N fertilizer application. However, yield improvement with diverse crop rotations helps elevate system-wide decisions by farmers to balance both short-term and long-term considerations of agronomy and economics.

3.8 Cover crop effects

Another conservation management decision that was assessed for its differentiation in soil and yield responses was the use of cover crops. A total of 48 fields had multi-species cover crops prior to corn production, while 19 fields had single-species cover crops, and 37 fields had no cover crop. Multi-species cover crops were used on 37% of fields in the Coastal Plain, 51% of fields in the Piedmont, none of the fields in the Blue Ridge, and 75% of fields in the Great Valley. Corn grain yield was greater when preceded by single-species (11.0 Mg ha⁻¹) and multi-species cover crops (10.4 Mg ha⁻¹) than without cover crop (8.3 Mg ha⁻¹) (p = .009). Relative yield without N fertilizer was greater following multi-species cover cropping (0.80) than following no cover (0.68) or single-species cover (0.67) (p = .02). At soil depth of 0–10 cm, total organic C was 21.9 ± 1.0 g kg⁻¹ (mean ± standard error) following multi-species cover crops, 20.4 ± 2.5 g kg⁻¹ following single-species cover crops, and 17.7 ± 1.6 g kg⁻¹ following no cover crop. Values for net N mineralization [multi-species cover (86 mg kg⁻¹ 24 d⁻¹) > no cover (59 mg kg⁻¹ 24 d⁻¹) = single-species cover (65 mg kg⁻¹ 24 d⁻¹)] and soil-test biological activity [multi-species cover (297 mg kg⁻¹ 3 d⁻¹) > single-species cover (205 mg kg⁻¹ 3 d⁻¹) = no cover (180 mg kg⁻¹ 3 d⁻¹)] followed the same trend and were significant at p ≤ 0.05. Fields with multi-species cover cropping had lower N factor at low CVT (15.8 kg N Mg⁻¹ grain) than fields with no cover crop (25.4 kg N Mg⁻¹ grain) (p = .02). Fields with single-species cover crops had an intermediate N factor (17.2 kg N Mg⁻¹ grain). Yield response at initial dose of N fertilizer was greater with single-species cover crop (51 kg grain kg⁻¹ N) than with multi-species cover crop (25 kg grain kg⁻¹ N) (p = .04). Yield response without cover crop was intermediate (33 kg grain kg⁻¹ N). Although these data were derived from a survey approach (i.e., no control on matching practices on the same types of soils and landscape conditions), they do suggest that farmers using multi-species cover crops over a number of years help transform surface soil into a more biologically active condition. Improvement in soil health condition reasonably led to reduction in need for exogenous fertilizer N inputs.

Short-term negative effects of cover crops on corn grain yield have occurred in some studies due to net N immobilization with the decomposition of low-N containing small grain biomass (Wagger & Mengel, 1993) and due to excessive extraction of soil water affecting early-season corn growth (Munawar, Blevins, Frye, & Saul, 1990). In other studies, significant corn yield enhancement has occurred with cover cropping due to a variety of factors linked to greater N supply with legumes, soil water conservation with surface residue cover, greater water retention with improved soil organic matter, reduced soil temperature during daytime (and/or increased soil temperature during nighttime), greater soil biological diversity that enhances nutrient cycling, and/or improved plant health via release of allelochemicals and other biochemical stimulants (Kaye & Quemada, 2017; Masiunas, 1998; Thorup-Kristensen, Magid, & Jensen, 2003). Miguez and Bollero (2005) conducted a meta-analysis of corn yield responses to winter cover cropping and found that grass–legume cover crops increased corn yield by 21% in a limited number of studies, grass cover crops had little net effect on corn yield across studies, and legume cover crops increased corn yield when N fertilizer inputs were limited.

3.9 Organic amendment effects

A total of 34 fields had recent history of organic amendment (beef or dairy bedpack, dairy slurry, swine lagoon waste, poultry litter, or municipal biosolids), while 78 fields were without organic amendment. Fields with organic amendment had enriched soil-test biological activity (317 vs.  206 mg CO₂−C kg⁻¹ soil 3 d⁻¹, respectively; p < .001), total organic C (26.4 vs.  17.6 g C kg⁻¹ soil, respectively; p < .001), and net N mineralization (101 vs.  61 mg N kg⁻¹ soil, respectively; p < .001). Residual inorganic N was not different between fields with and without history of organic amendment at 0–10-cm depth (23 and 19 kg N ha⁻¹, respectively; p = .21), at 0–20-cm depth (35 and 28 kg N ha⁻¹, respectively; p = .09), and at 0–30-cm depth (56 and 48 kg N ha⁻¹, respectively; p = .36). On average, maximum yield was not different with and without organic amendment (9.2 and 9.8 Mg ha⁻¹, respectively; p = .49). However, yield response to the initial dose of N fertilizer was lower with rather than without organic amendment (12 and 39 kg grain kg⁻¹ N, respectively; p = .001). Additionally, N factor at low CVT (11.5 and 21.7 kg N Mg⁻¹ grain, respectively; p = .003), medium CVT (8.2 and 16.7 kg N Mg⁻¹ grain, respectively; p = .006), and high CVT (6.7 and 11.7 kg N Mg⁻¹ grain, respectively; p = .03) was lower with than without organic amendment.

Reduction in the need for N fertilizer inputs with application of animal manures has been oft observed due to significant N mineralization (Eghball, 2000; Jokela, 1992; Mitchell & Tu, 2005; Tewolde, Sistani, Feng, & Menkir, 2019). Determining the appropriate level of N fertilizer reduction with animal manure application is complicated by source of manure, variation in nutrient concentration of manures from the same source, handling of manure prior to application, and weather conditions following application that affect decomposition and release of nutrients (Eghball, Wienhold, Gilley, & Eigenberg, 2002; Larney & Hao, 2007; Zavattaro et al., 2017). Summarizing this comparison, application of organic amendment was manifested in greater soil organic C and N fractions (particularly net N mineralization and soil-test biological activity) that contributed to effective nutrient cycling and reduced the need for synthetic N inputs.

4 CONCLUSIONS

Magnitude of corn grain yield response to sidedress N application was inversely related to net N mineralization, plant-available N (residual inorganic + mineralizable N), and soil-test biological activity. These consistent relationships occurred across a diversity of conditions throughout North Carolina, South Carolina, and Virginia, including: (a) private farms and research stations, (b) Coastal Plain, Piedmont, and Blue Ridge physiographic regions; and (c) various management practices of tillage, crop rotation, cover cropping, and organic amendments. These data validated similar responses observed in a previous study in the region. Strong association occurred between soil-test biological activity and net N mineralization within and across individual depths. Soil-test biological activity of the surface 10 cm was an important attribute that is simple, rapid, and reliable to better define soil N availability and fine-tune N fertilizer recommendations on a site-specific basis. Management systems that promote soil health with greater soil-test biological activity are able to provide a greater supply of N that can be used to lower N fertilizer costs and overcome potential N deficits during adverse weather conditions. Adjustment of N fertilizer rates based on CVT (i.e., fertilizer cost to grain value) should be an important component of any fertilizer recommendation system. This study validated the concept of N fertilizer adjustment on a site-specific basis with knowledge of soil-test biological activity.

ACKNOWLEDGEMENTS

Hannah Frank, Ellen Leonard, Erin Silva, and Ashley Turner provided sound technical support in the laboratory. Phil Bauer, Jeff Cline, Bhupinder Farmaha, Brad Graham, Keith Heller, Ivy Lanier, Alec Lipscomb, Kyle Miller, Robert Shoemaker, and Morgan Welch provided key logistical support at field locations. Sincere appreciation is extended to all research station managers, including Joe Hampton, Teresa Herman, Cathy Herring, Andy Meier, Will Morrow, Kaleb Rathbone, Fred Smith, Matt Smith, Ariel Szogi, and Tracy Taylor. Collaborating farmers deserve special mention for their dedication, insights, and trust – thanks to Allan Baucom, Jeff Bender, Brian Chatham, Carl Coleman, James Cooley, Jason Davis, Paul Davis, Jimmy Crosby, Curtis Furr, Gary Hendrix, Kenny Haines, Russell Hedrick, Louis and Quinn Howard, Joseph Johnson, Josh Johnson, Matthew Knowles, Franklin Lee, Jeff Lewis, Adrian Locklear, Nathan Lowder, Kay McGirt, Wesley Moore, Trent Pendarvis, Brian Ragan, Dale Reeves, Sam Reid, Sid Rogers, Pat Shooter, Ray Showalter, Keith Sink, Reid Smith, Ray and Mark Styer, Robert Waring, Dwight Wenger, Jeff Westbury, William Wollett, Zeb Winslow, and Don York.