Interactions of Stover and Nitrogen Management on Soil Microbial Community and Labile Carbon under Irrigated No-Till Corn

Soil Sci. Soc. Am. J. 82:323–331 doi:10.2136/sssaj2017.07.0229 Received 12 July 2017. Accepted 2 Jan. 2018. *Corresponding author (catherine.stewart@ars.usda.gov). © Soil Science Society of America. This is an open access article distributed under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Interactions of Stover and Nitrogen Management on Soil Microbial Community and Labile Carbon under Irrigated No-Till Corn Soil Biology & Biochemistry

effects on corn root production; at low rates stimulating root production and at high rates reducing root production (Eghball and Maranville, 1993;Zhu et al., 2016).These effects can also be mediated by edaphic factors including water availability, soil texture, and N placement.Root exudation increases with N fertilizer rate influencing microbial composition (Zhu et al., 2016), potentially altering SOC storage.Soil C inputs as root biomass or exudation and their interaction with the soil microbial community will determine the overall effect of N and residue removal on SOC pools.
Nitrogen availability can also alter SOC stability depending on residue chemistry and soil nutrient status through stimulation of microbial residue decomposition or SOC priming (Stewart et al., 2015a).Residue decomposition rate is positively related to residue N content and N fertilizer stimulates initial residue cellulose degradation (Hobbie et al., 2012).However, in later stages of decomposition, N fertilizer decreases lignolytic enzyme activity, resulting in a larger, more slowly decomposing residue pool (Hobbie et al., 2012).Residue chemistry determines the proportion of residue or soil-derived CO 2 loss from soil incubations, suggesting microbial community composition was responsible for either residue or soil decomposition (Stewart et al., 2015a).In addition, N fertilizer can enhance residue and SOC decomposition.(Halvorson and Stewart, 2015;Russell et al., 2009;Stewart et al., 2015a).Treatments with greater residue N may decompose more rapidly, forming particulate organic matter (POM), which is comprised primarily of partially decomposed plant material.These competing drivers of C input and loss may explain why some studies find no significant effect of N fertilizer on POM-C stocks despite predicted increases in POM with N fertilizer based on aboveground production (Brown et al., 2014;Jin et al., 2015;Stewart et al., 2017).
Residue removal impacts on soil C pools will be mediated by nutrient cycling effects on microbial biomass and community composition.Despite intensive study, long-term N fertilizer effects have been predominately evaluated in undisturbed systems, with less known about its effects in agroecosystems (Geisseler and Scow, 2014).In this study, we evaluated residue removal and N rate (0, 67, and 202 kg N ha -1 ) effects on labile soil C (i.e., POM-C) and microbial biomass and composition (phospholipid fatty-acid carbon, PLFA-C) in a long-term, NT irrigated corn study after 7 yr of stover removal.Long-term residue removal should reduce POM-C and microbial biomass due to SOC loss (Halvorson and Stewart 2015) and alter microbial community toward a more bacterial community.Nitrogen addition could mediate these effects due to greater root input, but also could stimulate residue and SOC decomposition, leading to lower POM-C stocks.

Experimental Site and Treatments
The study site is located on the Agricultural Research Development and Education Center (ARDEC) near Fort Collins, CO (40°39¢6² N, 104°59¢57² W; 1535 m above sea level).Soils are classified as clay loam soil (fine-loamy, mixed, mesic Aridic Haplustalf ) with a 1 to 2% slope.The full experimental design is described in detail in Halvorson and Stewart (2015).The study is a split-plot randomized complete block experimental design with three replications.Plots were located in an irrigated NT field established in 2008 that had been cropped in continuous corn.The field consisted of corn directly planted into the previous year's corn residue and each plot was 10.7 m by 21.3 m for main plot N treatments and 5.4 m by 21.3 m for residue treatment subplots.Planting was followed by application of herbicides for weed control and triple superphosphate (0-46-0) was applied to avoid P deficiency in the corn.A linear-move sprinkler irrigation system was used to apply water as needed during the growing season.
For the partial retention of residue as a split-plot treatment, stover from the previous corn crop was removed from half of each N treatment in the spring before corn planting.Over the study, residue removal averaged 66% of dry stover mass.Full retention of stover from the previous corn crop remained for the other half of each N treatment.Subsamples of the harvested residue were collected to determine water content and to calculate the percentage of stover removal from each N treatment.
Each year at emergence three N fertilizer treatments (0, 67, and 202 kg N ha -1 ) were surface-band applied to the corn.The N source consisted either of polymer-coated urea (2008 to 2011) or SuperU (2012 to 2014).

Soil Sampling
Using a GPS to relocate the field sampling location, a soil core of 5 cm diameter was collected within each stover and N plot following harvest in 2014.Soils were collected in increments of 0 to 7.5, 7.5 to 15, and 15 to 30 cm.Soils were tested to determine total soil C, total soil N, soil inorganic C, and soil organic C.

Soil Preparation and Analyses
Soils samples were sieved to 8 mm and all large roots and non-soil materials removed prior to soil characterization and microbial analysis.Soil subsamples for microbial PLFA analysis were handpicked to remove all identifiable plant material, frozen at −22°C, then freeze-dried (Labconco FreeZone 77530, Kansas City, MO) and stored at room temperature until lipid extraction.

Soil analyses
After initial sieving and large root removal, soils were air dried and ground to pass through a 150 mm screen for C and N analysis.Plant C and N as well as soil C and N content were determined (Elementar vario Macro C-N analyzer, Elementar Americas, Inc., Mt. Laurel, NJ).Inorganic C was determined using the pressure-calcimeter method for both the whole soil and for the POM fraction (Sherrod et al., 2002).Soil organic C and POM organic C was calculated as the difference between total C and inorganic C.
Particulate organic matter carbon (POM-C) and nitrogen (POM-N) were determined using the method of Gregorich and Ellert (1993).Bulk density was used to express soil on a mass base.There were no significant effects of residue or N rate on bulk density, but it did increase with depth (1.37 g cm -3 at 0 to 7.5 cm; 1.56 g cm -3 at 7.5 to 15 cm; 1.56 g cm -3 at 15 to 30 cm).
Soil pH was determined in a 1:1 soil/water ratio and averaged 7.98 with no treatment effects for the entire study.

PLFA Extraction and Identification
PLFA subsamples were analyzed at Ward Laboratories, Inc. (Lincoln, NE).Methods for extraction were adapted and modified from Clapperton et al. (2005).Briefly, 2-g fractions of soil were extracted using dichloromethane/methanol/citrate buffer in a 1:2:0.8v/v ratio.Following a rigorous 1-h shaking at 240 rpm, 2.5 mL of dichloromethane (DCM) and 10 mL of saturated potassium chloride (KCl) was added to each sample tube and shaken for 5 min.Tubes were then centrifuged at 3000 rev min -1 for 10 min and the organic fraction was transferred into clean vials.After drying under a flow of N 2 at 37°C, samples were dissolved in 2 mL of DCM and stored at −20°C less than 2 wk until lipid-class separation could be run.
Samples were loaded onto silica gel columns for lipid-class separation.Dichloromethane, acetone, and methanol were used as eluents and the respective neutral, glycol-, and phospholipid fractions were collected in separate vials.The neutral and glycolipid fractions were discarded while the phospholipid fraction was dried under N 2 at 37°C in the fume hood.The dried fractions were then dissolved in a few milliliters of MeOH and stored at −20°C.
Fatty acid methyl esters (FAMEs) were formed from the phospholipid fractions by mild acid methanolysis.As soon as samples were dry using a flow of N 2 at 37°C in a fume hood, half a Pasteur pipette full of MeOH/H 2 SO 4 (25:1 v/v) was added and the vials were then placed in an 80°C oven for 10 min.Once cooled to room temperature, 2 mL of hexane was added to each sample, vials were vortexed for 30 s and left to settle for 5 min, and the lower fraction was discarded.Finally, samples were again dried under a flow of N 2 at 37°C in a fume hood and rinsed with 50 mL of hexane into individual 100-mL tapered glass inserts for gas chromatograph vials.
Samples were analyzed to identify and quantify individual PLFA biomarkers with an Agilent 7890A gas chromatograph equipped with a 7693 autosampler and a flame ionization detector with hydrogen as the carrier gas and a 50-m Varian Capillary Select FAME column #cp7420.The temperature program started at 190°C held for 5 min, raised to 210°C at a rate of 2°C min -1 , then raised to 250°C at a rate of 5°C min -1 , and held for 12 min.
Peaks were identified as specific PLFAs based on comparison of retention times to known standards.Before calculations, the abundance of PLFAs was expressed as micrograms of PLFA per gram of dry soil, derived from the relative peak area under specific peaks, as compared with the 19:0 peak values.A calibration curve was made from a range of concentrations of the 19:0 FAME standard dissolved in hexane.
Individual PLFAs were expressed as C mass (ng PLFA-C g soil -1 ) and were combined as a proxy for microbial biomass.Relative PLFA abundance was expressed as molar C percentage (mol%) of each biomarker: where (PLFA-C) i is the concentration of an individual biomarker PLFA-C in solution (mol L -1 ) and n is the total number of identified biomarkers.Individual biomarker relative abundance values were summed for each microbial functional group and used to assess changes in microbial composition.

Statistics
We used a general linear mixed effects model to analyze fixed main effects of residue management, N fertilizer rate, and depth with block as a random effect using SAS v. 9.3 (SAS Institute, 2011).Main variables were PLFA-C (proxy for SMB), fungi to bacteria ratio, gram-positive to gram-negative ratio, POM-C, POM-N, the POM-C to POM-N ratio, and whole-soil C and N.Where necessary, data were log transformed to meet assumptions of normality and equal variance.P-values are noted in the text after Bonferroni adjustment and significant differences were considered at P < 0.05.
The PLFA relative abundance (mol%) biomarker data were analyzed by biomarker groups using distance-based redundancy analysis (dbRDA).Bray's distance was used to examine microbial compositional differences between cultivar soils (Legendre and Anderson, 1999).A principal coordinate analysis (PCoA) was performed on the distance matrix, from which the eigenvalues (obtained in the PCoA) were used within a redundancy analysis.Permutation-based analysis of variance (ANOVA) was performed on all dbRDA models to determine significance among group differences.Ellipsoids represent 95% confidence intervals around the multivariate-group centroids.

Soil Properties
After 7 yr, long-term partial stover retention in this irrigated NT site had 16% lower SOC stocks (P < 0.002) and 8% lower total soil N stocks compared with full residue retention in the 0to 7.5-cm depth (P = 0.017), but not in the 7.5-to 30-cm depth resulting in a residue × depth interaction (P = 0.012; Table 1).There was no overall effect of N rate, but SOC stocks decreased significantly with depth (P < 0.0001).These results confirm other studies that document primarily surface decreases in SOC stocks with residue removal in both rainfed and irrigated NT systems (Halvorson and Stewart, 2015;Osborne et al., 2014;Sindelar et al., 2014;Stewart et al., 2015b).The lack of N fertilizer effects has previously been observed at this site (Follett et al., 2013) and in other residue-removal studies under continuous corn (Dolan et al., 2006;Wilts et al., 2004).

Soil Microbial Biomass
Similar to the bulk soil, soil microbial biomass (i.e., PLFA-C) was lower (32%) on average following partial residue retention (P = 0.002), decreased with depth (P > 0.001), and did not respond to increasing N rate (Fig. 1).Averaged over depth, residue effects were only observed in the 0 and 67 kg N ha -1 additions, with no residue effect at 202 kg N ha -1 (N rate × residue interaction, P = 0.016).Partial residue retention decreased SMB only at low rates: by 40% under 0 kg N ha -1 and by 42% under 67 kg N ha -1 (through 0 to 30 cm), but not at 202 kg N ha -1 .There was also a residue × depth interaction with partial residue retention having lower microbial biomass compared with full residue retention in the 0-to 7.5-cm and 7.5-to 15cm depths under 67 kg N ha -1 and the 0-to 30-cm depth under 0 kg N ha -1 .
Residue removal directly influences microbial dynamics by increasing surface temperature and evaporation, or indirectly, through changes in plant rooting dynamics and soil C stocks.Residue cover in NT systems promotes surface rooting due to water and nutrient availability (Newell and Wilhelm, 1987;Qin et al., 2006).Loss of this protective cover reduced surface root-C by 37% and concomitantly, SMB by 50% after 9 yr under rainfed NT corn (Stewart et al., 2015b;Stewart et al., 2016).However, other studies in rainfed NT corn systems have not observed decreased SMB with residue removal (Ahlschwede, 2013;Johnson et al., 2013;Lehman et al., 2014;Spedding et al., 2004).Some of the differences between studies may be due to differences in SMB determination method (fumigation-incubation, PLFA, EL-FAME) which differentially emphasizes inherent SOC content or field conditions such as soil moisture and rhizodeposited C. For example, Spedding et al. (2004) found that 10 yr of residue removal decreased SMB in the 0-to 10-cm depth when measured by chloroform fumigation-  incubation, but this difference was not detectable via PLFA analysis during a very dry year.
Nitrogen fertilizer effects on microbial communities are mediated through their effects on soil C, plant biomass and root exudate production (Zhu et al., 2016), competition between plants and microbes for available N, and toxicity at high application rates.In a meta-analysis of 64 long-term trials, Geisseler and Scow (2014) found that increased microbial biomass with fertilization was likely due to indirect effects of increased soil C from long-term N fertilization.Low rates of N fertilizer application can promote root growth by stimulating plant N demand (Garten et al., 2010;Stewart et al., 2016), and consequently increasing SMB (Stewart et al., 2015b).We observed that trend here, where greater SMB was observed at the 67 kg N ha -1 rates with residue retained, but not at the highest N rate of 202 kg N ha -1 .
In long-term agricultural systems, fertilization generally increases SMB from 8 to 15% (Geisseler and Scow, 2014;Kallenbach and Grandy, 2011), in contrast to undisturbed systems, where N fertilization generally decreases SMB.However, others have also found a limited response of soil microbial biomass to increasing N rate in barley and corn (Lupwayi et al., 2011;Lupwayi et al., 2012).Lack of N effects may be due to the calcareous nature of the soil and buffering of N-induced pH effects.Studies that observed decreased SMB under increasing rates of N fertilizer, also observe a significant decrease in soil pH with N fertilizer (Biederbeck et al., 1996;Liebig et al., 2002;Stewart et al., 2015b).

Microbial Community Composition
Microbial community composition was remarkably similar under the residue treatments, with no difference between relative abundance of PLFA biomarkers (RDBa, P = 0.749), fungi to bacteria ratios, or gram-positive to gram-negative ratios (Fig. 2; Table 2).Few other studies examine soil microbial community composition under residue treatments, but those that have conducted such measurements, have found mixed results.Spedding et al. (2004) found no difference between residue-harvested and residue-retained treatments under NT and CT corn in Ottawa, Canada, using PLFA biomarkers.Studies using EL-FAME or total FAME analysis also showed no overall microbial community changes in Nebraska (Ahlschwede, 2013) and Minnesota (Johnson et al., 2013), respectively.Minimal response of fungi Table 2. Microbial biomass (ng PLFA-C g soil -1 ), fungi to bacteria ratio, and gram-positive (G+) to gram-negative (G-) ratio for N rate (0, 67, and 202 kg N ha -1 ) and residue (partial vs. full retention of corn stover) treatments after 7 yr.to bacteria ratios using DNA and EL-FAMEs was also observed by Lehman et al. (2014) where three out of four NT sites under residue removal showed no significant differences between residue removal treatments.Although fungi to bacteria ratios might be expected to decrease with residue removal, NT may moderate this effect through reduced disturbance and a high proportion of root to aboveground biomass C input.Root biomass and rhizodeposition contributes more than aboveground biomass to soil C (Johnson et al., 2006;Wilts et al., 2004).
There was a strong effect of depth on microbial composition (P = 0.001; Fig. 2) and gram-positive to gram-negative ratio increased with depth (P = 0.001).RDBa of relative PLFA abundance explained 48% of the variability along Axis 1 and 6% along Axis 2 (Fig. 2).Surface soil (0-7.5 cm) had a greater relative abundance of gram-negative and fungi compared with the 7.5-to 15and 15-to 30-cm depths, which separated along gram-positive, actinomycetes and AMF.Gram-negative communities are associated with plant roots and take up rhizodeposit C (Denef et al., 2007;Roosendaal et al., 2016).Under NT, fungal communities should be associated with the surface soil where new litter-C is being added.As labile C becomes less available with depth, microbial communities shift to gram-positive and actinomycetes (Fierer et al., 2003;Roosendaal et al., 2016).
Residue removal would be expected to decrease fungal growth, and could impact plant-microbial associations that impact nutrient acquisition, such as AMF fungi.On a mass base, AMF showed a strong decline in the surface soil with residue removal (P = 0.026; Fig. 3) and decrease with depth (P < 0.0001).Other studies that quantified AFM markers via FAME found no significant effect of residue removal on AMF, mirroring no significant differences in overall biomass (Ahlschwede, 2013;Johnson et al., 2013).

Particulate Organic Matter
In contrast to the whole soil and microbial biomass, partial stover retention increased soil POM-C by 11% on average (P = 0.003) compared with full stover retention and decreased with depth (P < 0.0001; Fig. 4a).The increase under partial stover retention was in the 7.5-to 15-and 15-to 30-cm depths, resulting in a depth × residue treatment interaction (P < 0.0001; Fig. 4a).The POM-C and POM-N decreased with depth (P = 0.013) and tended to be greater in the 0-to 7.5-cm depth under full stover retention compared with partial stover retention (Fig. 4a, 4b).The C to N ratios were higher for the partial retention treatment compared with full stover retention (P < 0.0001) in the deeper depths: 7.5 to 15 cm (12.7 vs. 9.4) and 15 to 30 cm (11.9 vs. 7.4, residue × depth interaction, P = 0.0001; Fig. 4c).There was no effect of N rate for POM-C or the POM C to N ratio.
Greater POM-C under partial stover retention is surprising and contrasts other studies under NT that show decreased POM-C with residue removal ( Jin et al., 2015;Osborne et al., 2014;Sindelar et al., 2014;Stewart et al., 2016).The higher POM C to N ratio at depth under partial stover retention compared with full retention reflects stover C/N differences between the two treatments (93 vs. 88) (Halvorson and Stewart, 2015) and suggests greater root-C input and/or reduced POM-C decomposition under partial stover retention.Corn root growth is sensitive to early season soil temperatures and partial stover retention increased soil temperature at our site 2 to 3°C compared with full stover retention (Halvorson and Stewart, 2015), potentially stimulating more root growth deeper into the profile under partial stover retention (Barber et al., 1988;Qin et al., 2006).Agricultural management effects on root production and root chemistry can have a strong impact under NT systems, where C input in the deeper layers is root-derived (Stewart et al., 2016).

N and Microbial Impacts on Labile Soil C Pools
In contrast to many studies, we found that SMB did not correlate well to POM, suggesting alternative C sources for the microbial community.Traditional decomposition models suggest that plant residues are decomposed via microbes and C stabilized on mineral surfaces or by aggregation (Six et al., 2002).However, new models emphasize the importance of labile C sources in creating stable soil organic matter via microbial processing (Cotrufo et al., 2015).Root exudates and dissolved organic C are relatively low-molecular weight compounds, easily metabolized by SMB with a relatively efficient conversion to SOM compared with residue with more recalcitrant structural lignins (Cotrufo et al., 2013).Soil C storage in these irrigated soils appears to follow a more microbial pathway, since full stover retention had significantly greater SOC, but not POM-C, compared with partial stover retention in the high N rate.
Residue removal effects on SMB were mediated by nutrient limitations.In partial retention treatments under low N rates, SMB was clearly C limited, and increased with additional residue C and bulk SOC content of the full retention treatment.However, at the highest N rate, there was no effect of additional residue on SMB compared with partial stover retention.This suggests that at high N fertilizer rates, the microbial C source was not primarily residue-derived.Instead, microbes in the high-N treatment may have received more C from root exudates, which have been shown to increase with N fertilizer addition rates (Zhu et al., 2016).In addition, soil microbes change their C source depending on residue N content (Stewart et al., 2015a).At low residue N contents, SMB preferentially decomposed shoot and root structural C, whereas at high residue N contents, microbes preferentially degraded soil C.
Although we anticipated that N-fertilizer treatments would decompose residue more rapidly, forming larger pools of POM-C, this was not the case.Instead, the high N content of residues likely promoted their rapid decomposition and incorporation into aggregate and mineral-associated C pools.Nitrogen fertilizer substantially decreased residue C to N ratio from 106, 96, to 70 in the 0, 67, and 202 kg N ha -1 treatments, respectively.Residue decomposition rate is positively related to residue N content (Adair et al., 2008) and N fertilizer stimulates initial residue cellulose degradation (Hobbie et al., 2012) and cellulose degrading enzyme activity (Geisseler and Scow, 2014).More rapid decomposition of higher N content residues may also explain no N effects on POM-C in other studies (Brown et al., 2014;Stewart et al., 2017)

C Cycling in Semiarid Irrigated Systems
In this irrigated no-till system, SOC loss may be more strongly influenced by increased decomposition than reduced C inputs under partial stover retention (Denef et al., 2008;Stewart et al., 2017).Residue removal studies that use stable isotopes to trace recent corn-derived contributions to SOM show decreased corn contribution to SOC, and increased turnover of older C3derived C, with residue removal (Clapp et al., 2000).Frequent irrigation promotes wet-dry cycles that form and destroy soil aggregates (Gillabel et al., 2007) and connects previously physically isolated microbial biomass and labile, dissolved organic C sources (Schimel et al., 2011), promoting large losses through respiration.The majority of soil C at this site is mineral-associated, and thus SOC loss through aggregate turnover and decomposition is likely significant.Other researchers have found that CO 2 pulses after rewetting are a significant source of SOC loss (Huxman et al., 2004;Munson et al., 2010), even from soils deep in the soil profile (Schimel et al., 2011).
Residue requirements for maintaining SOC stocks have been estimated at 2.5 ± 1.7 Mg C ha -1 ( Johnson et al., 2010) for predominately tilled soils in the Corn Belt.Stover returned to the soil under partial stover retention at our site was 0.9 to 1.24 Mg C ha -1 , well below sustainable removal rates with continuous-corn rotations, and clearly unable to maintain SOC stocks.Full residue retention contributed 2.74 to 3.82 Mg C ha -1 and maintained SOC stocks.No-tillage management was unable to compensate for reduced C inputs through residue removal and increased soil decomposition caused by increased early-season temperatures.These results emphasize the importance of residues not only as a source of C input, but in NT systems in mitigating enhanced soil C decomposition by reducing soil temperature.

Table 1 . Soil organic C (SOC) stock, total soil N stock, and the C to N ratio under N rate (0, 67, and 202 kg N ha -1 ) and residue (partial vs. full retention of corn stover residue) treatments after 7 yr. 0 kg N ha -1 67 kg N ha -1 202 kg N ha -1 All N rates
§ Lowercase letters in italics indicate significant differences among sampling depths.

0 kg N ha -1 67 kg N ha -1 202 kg N ha -1 All N rates
‡ Lowercase letters indicate significant differences among N rates and sampling depths.