The field managed with annual chisel plowing (Chisel), had two similarly textured Barnes-Aastad complexes. Using USDA-SCS soil maps [24] plots were arranged to block soil variability within replications, therefore three replications were on Barnes soil (fine-loamy, mixed, superactive, frigid Calcic Hapludoll) with the fourth replication on Aastad (fine-loamy, mixed, superactive, frigid Pachic Hapludoll). Both the NT2005 and NT1995 fields were established on an area mapped nearly exclusively as Barnes. All fields were managed with a corn/soybean rotation, with both crops present each year.

Prior to 2005, the Chisel field had been moldboard plowed (>20 cm) in the fall after harvest with one or two disking (~10–15 cm depth) operations prior to planting. Beginning in 2005, the moldboard plow was replaced with a chisel plow (~20 cm) but seedbed preparation remained the same. For at least ten years prior to establishing this stover study, the NT1995 had been managed without tillage. During the previous years, both NT1995 and Chisel were planted to continuous corn (10 years) or in a corn-soybean rotation (four years). The pre-experimental cropping history for NT2005 was a corn, soybean, wheat (Triticum aestivum L.) rotation with moldboard plow tillage (>20 cm) each fall after harvest for at least 10 years. This field was last moldboard plowed in 2004.

Corn stover return treatments were initiated in 2005 on Chisel and NT1995 and in 2006 on NT2005. In the Full Return treatment, only grain was harvested with all stover including cob material retained in the field. The amount of stover returned in Full Return was assumed to be the same as the stover yield measured at physiological maturity (described in section 2.3). The other two treatments correspond to (1) a Moderate Return rate intended to balance stover harvest with maintaining soil quality, and (2) a Low Return rate intended to maximize stover harvested. The methods used for implementing the stover return treatments were based on equipment availability. Between 2005 and 2008 after grain harvest, a two-row, flail-knife forage harvester, which cut the stalks at or below 10 cm from the soil surface, was used in four of the eight rows (Moderate Return) or all (Low Return) rows in the plot area to remove corn stover. Between 2009 and 2011, a prototype one-pass combine was used at two cutting-heights; just below the ear (about 65 to 70 cm) (Moderate Return) and about 10 cm above soil surface (Low Return). The amount of stover returned was determined annually by collecting standing stalk, plus all material on the ground in a 0.762 m² area.

All fields were planted with glyphosate-tolerant corn (Croplan 296TS) and soybean (508-M8) with both crop phases present each year. Planting density was 78,000 (Chisel) and 81,500 plants ha⁻¹ (NT19995 and NT2005) for corn and 247,400 plants ha⁻¹ for soybeans in all fields. Listing even-year crop first, the rotations were designated corn-soybean or soybean-corn. Corn plots received 10 and 15 kg N and P of starter fertilizer in 2005 through 2008. Prior to planting in 2009 to 2011 all plots in all fields received knife-injected (5–7 cm) 21, 34, 65 and 52 kg ha⁻¹ of N, P, K and S, respectively; plus, an additional 4 and 6 kg ha⁻¹ of N and P was applied when planting corn. Nitrogen fertilizer was applied annually during the corn phase as side-dressed anhydrous ammonia with average annual application rates of 136, 139 and 148 kg N ha⁻¹ applied in Chisel, NT2005 and NT1995, respectively. Side-dress rates were based on spring soil tests (0–30 cm) and the ARS Nitrogen-decision aid [25]. Weeds were controlled in both crops with two glyphosate applications annually; insecticides were only applied to soybean.

Corn and soybean grain yield at standard grain water contents of 15.5 and 13 g kg⁻¹, respectively were based on harvest with a two-row plot scale combine. Soybean straw production was determined on plants collected from a 0.76 m² area prior to leaf-drop R6 [26]. At physiological maturity, corn was collected from 1.5 m² to determine stover yield. Corn stover and soybean straw were reported as dry mass per area, based on oven-dried (60 °C) mass. Harvest index [27] was calculated as dry grain divided by dry grain plus stover.

In the fall of 2005, baseline soil samples were collected for all fields using a Giddings (Giddings Machine Company, Windsor, CO, USA) hydraulic probe (5.33 cm i.d.) to 100 cm as recommended by Liebig et al. [28]. Three soil cores taken in each plot were divided into six depth increments (0–5, 5–10, 10–20, 20–30, 30–60, and 60–100 cm). Soil from the 0–5 and 5–10 cm depth increments was passed through a 2 mm sieve. A subsample (~50-g) was air-dried for POM isolation while soil was oven-dried at 37 °C for chemical analyses. Soil from one core was used for determining soil bulk density at depth intervals below 10 cm. Cores were assumed to be uncompressed when the core length equaled the whole depth [29]. In the surface 0–5 and 5–10 cm increments, a hand-held soil probe (5 cm i.d.) was used to collect a sample for bulk density. Soil texture was determined using the hydrometer method [30,31]. Soil pH (1:1 CaCl₂) [32], total C and N (LECO TRU-SPEC CN analyzer; LECO Corporation, St. Joseph, MI), inorganic C [33], Olsen P and extractable K [34] were determined. Organic C was calculated as the difference between total combustible C and inorganic C as these are calcareous soils. Particulate organic matter was isolated from air-dried soil as described by Cambardella and Elliot [16]. The mass of POM independent of sand [35] was determined by weight loss on ignition using a method by Schulte [36] as reported by Cambardella et al. [17]. In 2009, to obtain an early prediction of changes in soil organic matter due stover return rate, POM was isolated and bulk density determined from soil collected at the surface 0–5 and 5–10 cm as described above. As previously described for baseline samples, in the fall of 2010 (following harvest and preceding tillage in Chisel), soil samples were collected to 100 cm depth and assessed for bulk density, pH, total C and N by combustion, inorganic C, Olsen P and extractable K.

Percent soil coverage and dry aggregate size distribution were used as indicators for treatment induced changes in soil erosion potential. Soil coverage was assessed annually by the transect method [37,38]. The transect method, which does not require a minimum residue size, counts the number of times a 15 m tape with 100 equidistant marks intersects visible residue. Soil for dry aggregate size distribution was collected in the summer of 2011 within about 5 cm of the soil surface after removing visible surface residue. Dry aggregate size distribution using the method described by Chepil [39] and Pikul et al. [40] was determined using a rotary sieve operating at 6-rpm to separate air-dried soil into six aggregate size groupings: 0–0.5, 0.5–1.0, 1–2, 2–3, 3–5, 5–9, and 9–20 mm. Total soil mass in aggregates <20 mm and mass of each aggregate size was determined.

Biological parameters (microbial biomass and composition and enzyme activities) were evaluated on soil samples collected in 2008 prior to spring agronomic operations from Full and Low Return in all fields using a hand-held probe. Soil cores were divided into 0–5 and 5–10 cm depth increments. Soil MBC and MBN were determined on the field-moist samples, with a dry equivalent mass of 15-g, by the chloroform-fumigation-extraction method using 0.5 M K₂SO₄ as an extractant [41,42]. Organic C and N were quantified using a CN analyzer (Shimadzu Model TOC-V/CPH-TN; Shimadzu Scientific Instruments, Columbia, MD, USA). Microbial biomass C was calculated as (organic C extracted from fumigated soil minus organic C extracted from non-fumigated soil) divided by a k_EC factor of 0.45 [43]. The soil MBN was calculated similarly, but used a k_EN factor of 0.54 [44]. These factors correct for biomass solubilized during the fumigation-extraction [45]. Each sample had duplicate analyses and results are expressed on a dry-weight basis. Soil water content was determined after drying soil at 105 °C for 48 h.

Soil microbial community structure was characterized using fatty acid methyl esters (FAME) analysis on field-moist soil samples using the Microbial Identification System (MIS, Microbial ID, Inc., Newark, DE, USA) procedure as applied for soil analyses [46,47]. Briefly, the method consists of four steps: (1) saponification of fatty acids in 3 g field-moist soil with 3 mL 3.75 M NaOH (methanol:water, 1:1) solution under heat (100 °C) for 30 min; (2) methylation of fatty acids by adding 6 mL of 6 M HCl in aqueous methanol (1:0.85) under heat (80 °C) for 10 min; (3) extraction of the FAME with 3 mL of 1:1 hexane:methyl-tert butyl-ether solution and rotating the samples end-over-end for 10 min; and (4) washing the organic phases with 1.2 % diluted NaOH by rotating the tubes end-over-end for 5 min. The organic phase was analyzed in a 6890 GC Series II (Hewlett Packard, Wilmington, DE) equipped with a flame ionization detector and 25m × 0.2 mm fused silica capillary column using ultra high purity hydrogen as the carrier gas. The temperature program was ramped from 170 °C to 250 °C at 5 °C min⁻¹. Fatty acids were identified and their relative peak areas (percent) were determined with respect to the other fatty acids in a sample using the MIS Aerobe method of the MIDI system (Microbial ID, Inc., Newark, DE, USA).

Enzyme activities, β-glucosidase, β-glucosaminidase, and acid phosphatase were assayed as indicators of C, C and N, and P biogeochemical cycling potential, respectively. These enzyme activities were assayed using 1g of air-dried soil with their appropriate substrate and incubated for 1 h (37 °C) at their optimal pH as described by Tabatabai [48] and Parham and Deng [49]. Enzyme activities were assayed in duplicate with one control, to which substrate was added after incubation and subtracted from the sample value.

The impact of corn stover return was assessed on three fields differing in tillage management, but these experiments were not designed to make direct statistical comparisons among tillage management effects. Data on soil properties and crop yield within each field were analyzed using PROC MIXED of SAS version 9.2 [50] with residue return and crop phase considered as fixed effects and replication as a random effect. Crop yield and percentage soil-coverage data were evaluated using PROC MIXED, but including years as a random effect. Mean separation and multiple comparisons were obtained with a LSMEANS statement within PROC MIXED [50]. Additional analysis of the FAME profiles was completed using principal component analysis (PCA) plots of the correlation matrix for each soil depth with a 95% confidence ellipse to determine separation due to stover return (Full Return vs. Low Return) on the microbial community structure using R statistical software [51] with the Vegan package for PCA [52]. The R software was also used to evaluate the three enzyme activities together as affected by stover return according to PCA plots. Significance for all data is reported at P ≤ 0.05.

The mass of stover returned in each field is summarized in Table 1. In Chisel, the Moderate Return had 51% and Low Return had 22% stover returned compared to the Full Return. In NT2005, Moderate Return and Low Return had 46% and 15% stover returned, respectively, compared to Full Return. Similarly in NT1995, the Moderate Return had 50% stover and the Low Return had 21% stover returned compared to the Full Return. Corn was rotated with soybeans so the average total residue (Corn stover plus soybean straw) returned reflects the soybean contribution.

Within any year, stover return rate did not affect grain or residue yield for either crop. Averaged over six-years (2006–2011 for Chisel and NT1995), or over five-years (2007-2011 for NT2005), stover return rate only significantly altered yield in NT1995 (Table 2). In NT1995, a soybean grain average was similar between Full Return and Moderate Return, which averaged 8% more than Low Return. The corn harvest index was similar among treatments in all fields: Chisel (0.52), NT1995 (0.51), and NT2005 (0.50) fields.

These clay loam soils were near neutral to slightly acidic in the 0-10 cm depth (Table 3), overlaying a calcareous, alkaline subsoil >30 cm (data not shown). Bulk density ranged from 1.24 to 1.37 g cm⁻³ in the 0–5 cm depth and ranged from 1.29 to 1.41 g cm⁻³ in the 5 to 10 cm depth increments, increasing to >1.5 g cm⁻³ below 30 cm. Total soil C ranged from 16 to 29 g kg⁻¹, remaining fairly constant throughout the soil profile, but below 30 cm most of soil C was inorganic C (data not shown). Soil P and K availability were considered high to very high for the region [53]. Soil organic C, total N and POM content declined from the surface 0–5 to 5–10 cm depth in all three fields.

A reduction POM as a result of reducing stover return was detected in NT1995 at 0–5 cm soil depth (Table 4). In NT1995 at 0–5 cm depth, POM was similar for Full and Moderate Return (14.0 g kg⁻¹ soil), but reduced in Low Return (11.3 g kg⁻¹ soil). In NT1995, although not significant, patterns in total N and soil organic C mirrored POM results. Total soil N concentrations of 2.5, 2.4, and 2.2 g N kg⁻¹ soil and soil organic C concentrations of 28.8, 27.0 and 26.2 g C kg⁻¹ soil were detected for the Full Return, Moderate Return and Low Return, respectively. In Chisel and NT2005, no discernible patterns for POM, SOC or total soil N were detected as a result of stover return rate. In all three fields, P, K, pH and bulk density were similar to baseline sampling and among stover treatments at all depth increment to 100 cm (data not shown).

Both the percentage of soil coverage and dry aggregate size distribution revealed potential for increased soil erosion risk if stover were removed. For example, the percentage of soil coverage declined proportionally to rate of residue returned (Full > Moderate > Low Return) (data not shown). Additionally, intensive tillage at the Chisel field showed <20% of the soil covered for all stover treatments, including Full Return, where all residues (stover and soybean stubble) were returned; whereas, NT2005 and NT1995 had at least 45% of the soil covered even in Low Return. In NT2005, significant increases in aggregates <1 mm and significant decreases in aggregates 5–9 mm were measured in Low Return compared to Full Return (Figure 1). Low Return had 15% and 60% more aggregates in the 0–0.5 and 0.5–1 mm classes, respectively, compared to Full Return, but Full Return had 14% more 5–9 mm aggregates compared to Low Return, with Moderate Return intermediate. In Chisel and NT1995, although means of aggregate distribution displayed a similar trend to the NT2005, no statistically significant increase in the frequency of aggregates <1 mm was detected.

In all fields, biological parameters such as MBC, MBN, and the activities of acid phosphatase, β-glucosaminidase or β-glucosidase were not altered by stover return rate (Table 5), except for acid phosphatase activity, which was reduced about 20% under Low Return compared to Full Return at the 0–5 cm depth in NT2005. The stover return rate treatments affected FAME profiles only in Chisel. In this field at the 0–5 cm depth, the fungal indicator 16:1ω5c had a greater abundance in Full Return compared to Low Return. At the 5–10 cm depth, the bacterial indicator i17:0 and fungal indicator 18:3ω6c had a greater abundance in Low Return compared to Full Return. The PCA plot to evaluate indicators of microbial community structure and function did not separate between Low and Full Return.

This study showed that corn grain and stover yields were not impacted on any field as a result of stover harvest. These crop responses contrast the dramatic corn and soybean yield reductions observed for dry land corn in Nebraska [54]. However, corn yield was not reduced in South Dakota [18], Iowa [7] nor in Minnesota even after 29 years of silage harvest [55]. The dramatic short-term responses in Nebraska were attributed to reduced water availability and increased soil temperatures. The lack of short-term response in our current studies suggests stover treatments did not result in yield-reducing changes in the microclimate. This observation is encouraging from both a bioenergy and agronomic perspective because, at least in the short-term, yields were not compromised by reducing corn stover returned.

Our studies may suggest yield response to corn stover return rates may depend upon tillage. For example, in the field under longer term no-till management (NT1995), a small impact on soybean grain yield was observed when an average of 1.6 Mg ha⁻¹ yr⁻¹ stover was returned. Karlen et al. [7], reported decreases in soybean yield following stover harvest that was attributed to K deficiency. The K removed in the harvested stover was sufficient to decrease yield [7]. While it is possible K availability is contributing to the soybean response in NT1995, it is unlikely as K was added annually since 2009. However, the slight decline in soybean grain yield occurred in NT1995, which also had a decline in POM. This suggests that tillage management may be an important factor for maintaining productivity in biofuel cropping systems. Although, it may be speculative at this time, it warrants reassessing POM and other soil quality parameters in the future on these fields.

It was anticipated that returning less stover would reduce the percentage of soil covered and thus, soil erodibility potential. Implementing stover return rate treatments with a forage harvester or a one-pass combine provided sufficient coverage against early growing-season erosive forces in NT1995 and NT2005, but in the Chisel even Full Return failed to provide at least 30% soil coverage. The other parameter used to assess soil erosion risk was dry aggregate size distribution. The increased fraction of aggregates <1 mm under Low Return indicates that stover removal reduces aggregate size and stability. A similar shift, in dry aggregate size distribution toward fewer large aggregates and more small aggregates and a decline in wet aggregate stability was reported after four stover removal cycles in a corn-soybean rotation [18]. Even though the differences were not significant in the Chisel and NT1995 fields, the overall pattern for dry aggregate distribution implies that the fraction of small aggregates may be increasing and larger (more stable) aggregates decreasing. These observations suggest repeated Low Return associated with stover harvest has the potential to slowly decrease soil aggregate stability leaving it more prone to the degrading forces of wind and water. Thus, dry aggregate size distribution and wet aggregate stability should be reassessed after additional treatment cycles to determine the long-term sustainability of the reducing stover return.

Whole soil POM has been established as an indicator of soil organic matter change [16,35,40]. Soil organic matter was predicted to decline if stover return rate provided insufficient raw inputs [5,56,57]. In NT1995 after two treatment cycles, decreased total POM concentration in the 0–5 cm depth increment suggested insufficient residue was provided at Low Return to sustain soil organic matter over time. Similar decreases in POM were also reported after four stover removal cycles in another corn-soybean rotation [18]. Total residue (including soybean straw) returned in the Low Return for all fields was ≤3.0 Mg residue ha yr⁻¹. In a South Dakota corn-soybean rotation, 3.6 Mg residue ha⁻¹ yr⁻¹ was needed to maintain soil organic C [58]. Our findings are consistent in that more stover needs to be returned during the corn phase to compensate for the lack of residue during the soybean phase. Furthermore, Huggins et al. [59] found that about twice the amount of soybean derived C compared to corn derived C was needed to maintain soil organic C because soybean in the rotation accelerated the turnover of soil organic matter.

The clay loam soils in these studies had relatively high soil organic carbon, POM, and microbial communities compared to semiarid sandy soils [60], which may have prevented changes in microbial biomass, community structure or enzyme activities from being detectable. In contrast, decreases in MBC and MBN in response to annual residue harvest on a sandy loam soil were detected in as few as two years by Kashwaha et al. [19]. Stetson et al. [61] reported mixed results to stover harvest in a corn/soybean rotation as changes in labile C pools or glomalin—a soil protein related to mycorrhizal populations—were not detected while total microbial activity measured by fluoresceine diacetate was reduced. Perhaps our study shows the benefits of a system where the plant biomass is removed every other year for biofuel production rather than intensively removing it every year. Additionally, although these fields with different tillage managements were not statistically compared, the higher fungal indicator 16:1ω5c (an arbuscular mycorrhizal indicator) in Full Return compared to Low Return in the Chisel field may suggest more residue is needed in tilled fields to encourage soil conditions conducive to maintaining mycorrhizal populations. As changes in other microbial FAMEs were not detected, long-term impacts of residue return rates on the soil microbial community remain to be determined.