Volume 73, Issue 3 e13241
ORIGINAL ARTICLE
Open Access

Crude glycerol, a biodiesel byproduct, used as a soil amendment to temporarily immobilise and then release nitrogen

Mriganka De

Corresponding Author

Mriganka De

The Department of Agronomy, Iowa State University, Ames, Iowa, United States

The Department of Biological Sciences, Minnesota State University, Mankato, Mankato, Minnesota, United States

Correspondence

Mriganka De, The Department of Biological Sciences, Minnesota State University, Mankato, Minnesota 56001, United States.

Email: [email protected]

Contribution: Conceptualization (supporting), Formal analysis (lead), ​Investigation (lead), Methodology (lead), Project administration (supporting), Software (lead), Visualization (equal), Writing - original draft (lead)

Search for more papers by this author
John E. Sawyer

John E. Sawyer

The Department of Agronomy, Iowa State University, Ames, Iowa, United States

Contribution: Conceptualization (lead), Funding acquisition (lead), ​Investigation (supporting), Methodology (supporting), Project administration (lead), Visualization (supporting), Writing - review & editing (equal)

Search for more papers by this author
Marshall D. McDaniel

Marshall D. McDaniel

The Department of Agronomy, Iowa State University, Ames, Iowa, United States

Contribution: Conceptualization (lead), Funding acquisition (lead), ​Investigation (supporting), Methodology (supporting), Project administration (lead), Supervision (lead), Visualization (supporting), Writing - review & editing (equal)

Search for more papers by this author
First published: 20 April 2022
Citations: 2
Funding information Renewable Energy Group, Inc. (REG)

Abstract

Loss of nitrate-nitrogen (NO3–N) from Midwestern U.S. agricultural fields can impair water quality and be an economic loss to farmers. Winter cover crops have shown promise as a remedy, but low adoption illustrates the need for alternatives. Here, we tested whether adding a carbon (C)-rich soil amendment (i.e., crude glycerol, a biodiesel byproduct) can increase soil microbial biomass (MB) and promote N immobilisation under various conditions and then determined whether and when immobilised N would be released. We conducted a laboratory incubation with a full factorial combination of four glycerol rates (0, +117, +468 and +1872 mg C kg−1 soil), three supplemental NO3–N rates (0, +10 and +40 mg N kg−1) and two soils (Clarion clay loam and Sparta loamy sand). Soil inorganic N (NH4+–N and NO3–N) and MB were measured at seven and three time points, respectively, across the 98 days incubation period. Across all treatments, glycerol increased MBN in both short term (7 days; 4%–1137% compared to no glycerol addition) and long term (98 days; 10%–169%) and decreased NO3–N with increasing rate of glycerol. Adding glycerol caused net N immobilisation of 21%–61% (+117 mg C kg−1 addition) and ~100% (+468 and +1872 mg C kg−1 addition) compared to the control. Some of that immobilised inorganic N was likely released through MB turnover, but timing and rate of release depended on the soil and added N rate. Adding 40 mg N kg−1 with no glycerol showed nearly twice the net N mineralisation rate than with the low or no applied N – providing evidence for soil N priming. Overall, glycerol has the potential for use as a soil amendment to increase MB and temporarily immobilise NO3–N and then make some of that N crop available through MB turnover.

Highlights

  • Crude glycerol, a biodiesel by-product, was evaluated as a soil amendment to reduce soil nitrate.
  • Glycerol strongly increased soil microbial biomass and decreased nitrate under all conditions.
  • N immobilisation was temporary, and N was mineralised at a lower glycerol rate.
  • Glycerol rates, N rates and soil type affected N immobilisation-mineralisation dynamics.

1 INTRODUCTION

Intensified annual row crop production in the Midwest United States has made much progress with increasing grain yields but has come at the cost of impaired water quality. Loss of nitrate-nitrogen (NO3–N) and phosphorus is not only a primary cause of eutrophication of surface and ground waters but also an economic loss for producers (Anderson et al., 2002; Vitousek et al., 2009). Most NO3–N is lost from farms when fields are left fallow – especially early spring and late autumn in temperate agroecosystems (Di & Cameron, 2002; Dinnes et al., 2002). Recently, integration of winter cover crops has mitigated substantial NO3–N loss during these periods due to plant uptake or microbial net N immobilisation (Kaspar et al., 2012; Thapa et al., 2018; Waring et al., 2020).

Cover crop use in the Midwest United States has increased from 4.2 to 6.2 million ha between 2012 and 2017, an approximate 50% increase, but cover crop use in the state of Iowa and elsewhere remains low (<4% of total U.S. cropland or 160.4 million ha; USDA, 2019). Whether the reason for this low cover crop adoption rate is due to perceived or real cash crop yield declines (Marcillo & Miguez, 2017; Patel et al., 2019) or other socioeconomic factors (Bergtold et al., 2019; Roesch-McNally et al., 2018), it clearly illustrates other alternatives to reduce N leaching are needed. Thus, one alternative management strategy might be to add a carbon (C)-rich soil amendment to increase soil microbial biomass (MB) and promote immobilisation of inorganic N (NH4+–N and NO3–N), especially highly mobile NO3–N.

Unfortunately, little attention has been given to soil microbial N immobilisation as a management strategy because microorganisms and plants are in competition for N (Hodge et al., 2000; Kaye & Hart, 1997), and conventional wisdom in agriculture suggests that N immobilisation is ‘stolen’ N meant for crops. Immobilisation occurs in tandem with mineralisation, and immobilisation is often of lesser magnitude, resulting in net mineralisation. The N immobilisation can be a dominant process at times – resulting in net negative N mineralisation in agricultural soils (Ahmad et al., 1969; Kai et al., 1973; Romero et al., 2015). However, N immobilisation in agricultural soils only occurs when conditions are conducive, such as alkaline soil pH (Rochester et al., 1992), low concentrations of NH4+–N, strong nitrifier and heterotroph competition (Burger & Jackson, 2003) and probably most importantly, an abundant supply of labile C (Alotaibi & Schoenau, 2011; Qian et al., 2011; Redmile-Gordon et al., 2014; Romero et al., 2015). The addition of exogenous C appears to be a plausible means to immobilise N; however, the addition of C has shown inconsistent effects, and C rates appear to be important in regulating net N immobilisation (Cheng et al., 2017).

The rapid growth in the US biodiesel production has led to large concurrent production of crude glycerol (IUPAC name: Propane-1,2,3-triol; chemical formula: C3H8O3). For example, using an estimated 7.2B L in biodiesel production (EIA, 2019), and glycerol production ratio of 79 g glycerol per 1 L of biodiesel (Thompson & He, 2006), approximately 570 M kg of crude glycerol was generated in the United States alone in 2018. Crude glycerol is generated through transesterification of oils and has other minor contaminants like salts or small concentrations of methanol (Fukuda et al., 2001; Leung et al., 2010). Some of the byproduct glycerol is purified of contaminants and used for commercial purposes (Dalias & Polycarpou, 2014; Lammers, Kerr, Honeyman, et al., 2008; Lammers, Kerr, Weber, et al., 2008; Mortensen et al., 2017), but the purification process is expensive. Moreover, the demand for commercial use of glycerol is small compared to the large supply (Ciriminna et al., 2014), so large quantities of glycerol are currently disposed of by incineration, releasing the CO2 to the atmosphere (Alotaibi & Schoenau, 2011). Hence, land application of crude glycerol is an appealing option for N immobilisation and one that mitigates the direct release of a greenhouse gas (GHG), bolstering the sustainability of the biodiesel supply chain (Alotaibi & Schoenau, 2013; Taheripour et al., 2010).

Given the surplus production of crude glycerol in the United States and potential for glycerol to immobilise N, it could serve as a viable alternative management strategy for farms where winter cover crops will not work. However, before glycerol addition is implemented at a field scale, it needs to be tested in a laboratory to determine the N immobilisation potential under a variety of soils and conditions. Several laboratory studies have investigated the application of glycerol as a soil amendment to either improve microbial activity and/or environmental quality (Alotaibi & Schoenau, 2011, 2013, 2016; Dalias & Polycarpou, 2014; Qian et al., 2011; Redmile-Gordon et al., 2014). While these studies are a first good attempt at assessing glycerol's ability to improve soil health, and perhaps reduce leachable N, four critical, unanswered questions remain: (i) would soil applied glycerol increase microbial biomass nitrogen (MBN) and reduce residual NO3–N (i.e., potentially reduce loss via leaching)? (ii) how long does immobilised N remain plant unavailable and when would immobilised N in MB be re-mineralised and become plant-available N? (iii) how does added fertiliser N affect N immobilisation and mineralisation? and (iv) do N immobilisation-mineralisation dynamics with glycerol addition differ between soils? We hypothesised that: (1) adding glycerol to soils would increase MBN and immobilise NO3–N, (2) immobilisation would be temporary and N would eventually be released after MB turnover, (3) adding supplemental NO3-N would increase the amount of N stored, especially in combination with greater glycerol rates and (4) a soil with greater soil organic matter (SOM) and finer texture would have greater N immobilisation potential due to a greater MB capacity than a low-SOM, coarse-textured soil. To test these hypotheses, we conducted a laboratory-based incubation study with two contrasting soils from fields under maize (Zea mays L.) – soybean (Glycine max (L.) Merr.) based crop rotation.

2 MATERIALS AND METHODS

2.1 Soil sampling and site description

The study included two contrasting soils from fields under decades of maize – soybean crop rotation. One soil was from the long-term Marsden Farm experiment near Boone, Iowa, USA (42o01′N; 93o47′W) − it is a clay loam, moderately well-drained, glacial till derived Typic Hapludoll (hereafter referred as Clarion series; USDA-NRCS, 1981) with 6.4% SOM and <2% slopes. The other soil was collected at a commercially operated farm in southeast Iowa near Conesville, Iowa, USA (41°36′N, 91°35′W) – it is classified as a loamy sand, very deep, excessively drained, mesic alluvial sediment derived Entic Hapludolls (hereafter referred as Sparta series; USDA-NRCS, 1989) with 2% SOM and 0%–2% slopes. Soil was collected from each site between 25 and 26 September 2017 when soybean was close to maturation. The climate at both sites is temperate continental, with an average annual rainfall (1988–2017) of 971 and 881 mm for the locations of the Clarion and Sparta soils, respectively (Iowa State University, 2019). The mean annual temperatures for each site were 9 and 10°C (Iowa State University, 2019).

At each sampling site, a hand probe with 2.54 cm diameter was used to collect 10 soil cores composited from the 0–15 cm depth. Composited field moist soil samples for each study location were stored in a cooler while transported to the laboratory for refrigeration at 4°C until analyses. Four subsamples were taken from each composited soil to determine gravimetric soil moisture content (Gardner, 1986). After soil moisture measurement, the remaining field moist soil subsample was passed through a 2-mm sieve. Some sieved soils were kept field moist for laboratory incubation and some were air-dried for initial soil analyses. Table 1 shows a description of the initial soil properties.

TABLE 1. Soil physical, chemical and biological characteristics for the Clarion and Sparta soil series used in this incubation (n = 4; mean ± standard error)
Parameters Clarion soil Sparta soil
Physical
Sand (%) 24.4 ± 0.3 86.5 ± 0.9
Silt (%) 42.8 ± 0.1 10.9 ± 0.9
Clay (%) 32.8 ± 0.2 2.6 ± 0.2
Texture Clay loam Loamy sand
Gravimetric moisture content (GMC, g g−1) 0.20 ± 0.001 0.08 ± 0.001
Maximum water holding capacity (MWHC, g g−1) 0.64 ± 0.01 0.39 ± 0.02
Chemical
pH 6.3 ± 0.05 5.7 ± 0.02
Electrical conductivity (EC, μS cm−1) 225.9 ± 9.2 161.6 ± 12.1
Soil organic matter (SOM, %) 6.4 ± 0.01 2.0 ± 0.01
Particulate organic matter (POM, %) 1.36 ± 0.02 0.60 ± 0.04
Total carbon (TC, g kg−1) 29.3 ± 0.52 8.32 ± 1.15
Total nitrogen (TN, g kg−1) 2.43 ± 0.01 0.83 ± 0.01
C:N 12.1 ± 0.2 10.0 ± 0.2
Ammoniacal nitrogen (NH4+–N, mg kg−1) 0.10 ± 0.04 0.20 ± 0.06
Nitrate nitrogen (NO3–N, mg kg−1) 5.82 ± 0.08 4.44 ± 0.05
Biological
Microbial biomass carbon (MBC, mg kg−1) 259 ± 4 77 ± 4
Microbial biomass nitrogen (MBN, mg kg−1) 32 ± 1 7 ± 1

2.2 Soil and glycerol analyses

Soil pH and electrical conductivity (EC) were measured after equilibrating 10 ± 0.04 g of soil with 20 ml of deionised water (1:2, w:v) for 4 h with a digital HQ430D laboratory single input, multi-parameter meter (Hach Company, Loveland, CO, USA). Gravimetric soil moisture content was determined as described by Gardner (1986) and used to correct the soil mass to determine inorganic N (IN; NH4+–N + NO3–N) and MB. The maximum water holding capacity (MWHC) for each soil series was calculated as the difference in weight between a saturated soil that was allowed to drain for 6-h, and the weight after the soil was oven-dried for 48 h at 105°C (De et al., 2020; Haney & Haney, 2010). Particle size analysis (sand, silt and clay) was conducted using the pipette method (Gee & Bauder, 1986). A 5- to 10-g air-dried subsample was ball-milled to pass through a 60 mesh (0.25 mm) screen and was used for total C and N analyses. The total C and N content was measured using a high-temperature dry combustion method in a Leco CN analyser (Leco Corp., St. Joseph, MI, USA) or a Vario Max CN analyser (Elementar Americas, Mt. Laurel, NJ, USA) (Nelson & Sommers, 1996). The loss-on-ignition method was used to measure the SOM content (Cambardella et al., 2001). The difference between the soil mass before ignition and soil mass after 4–6 h in a muffle furnace at 450°C was estimated as SOM content. The particulate organic matter (POM) was measured following the procedure reported by Cambardella et al. (2001).

Crude glycerol was obtained from a biodiesel production company (Renewable Energy Group, Inc., Ames, Iowa, USA). The University of Missouri-Columbia Experiment Station Chemical Laboratories, Columbia, Missouri, USA, measured the glycerol chemical composition and characteristics. Characterisation of the crude glycerol and the methods of the analyses are listed in Table 2.

TABLE 2. Characterisation of the crude glycerol, a biodiesel byproduct, used in the incubation
Parametersb Valuea Analytical method
Total glycerol (%) 78.6 ± 0.28 Determined by differencec
Moisture (%) 14.2 ± 0.35 AOACd 984.20
Ash (%) 6.52 ± 0.01 AOACd 942.05
Methanol (%) 0.38 ± 0.12 Gas chromatography (proprietary method)
Chloride (%) 3.98 ± 0.02 AOACd 956.01
Sodium (%) 2.67 ± 0.03 AOACd 956.01
Potassium (%) 0.01 ± 0.0003 AOACd 9.15.01, 943.01
Total fatty acids (%) <0.01 AOCSe G 4.40 modified for glycerol
Crude fat (%) <0.01 AOACd 920.39 (A)
pH 5.25 ± 0.05 Orion 230A pH meter with 9107 BN probe
Colourf #1 AOCSe Cc 13a-43
  • a n = 4; mean ± standard error.
  • b Analysed by University of Missouri-Columbia Experiment Station Chemical Laboratories, Columbia, MO, USA.
  • c Determined by difference as: 100−%methanol−%total fatty acid−%moisture−%NaCl.
  • d AOAC (1995).
  • e AOCS (2000).
  • f Colour determined on a Lovibond 3000 comparator, FAC colour wheel.

2.3 Laboratory incubation study

2.3.1 Treatments and experimental design

To study the effect of glycerol on soil IN (NH4+–N + NO3–N) dynamics and MB, we used a laboratory incubation with the sieved (<2 mm), field-moist soils. To begin the incubation study, 5 ± 0.04 g of field moist soil was weighed from each site between 27 and 30 September 2017 and placed into a 50 ml centrifuge tube. We started our incubation on 2 October 2017 at ~25°C with the addition of four crude glycerol and three calcium nitrate [Ca(NO3)2 · 4H2O] solution rates, with the combinations diluted with water to correspond to 60% of the MWHC of each soil. Any moisture deficits across the incubation period were corrected by weight by adding deionised water back to 60% MWHC. To maintain aerobic conditions, all tubes were loosely capped and aerated daily for the first 2 weeks when respiration rates were very high and then weekly as the respiration rate decreased (De et al., 2020).

The incubation was a completely randomised design, included 12 (3 calcium nitrate N rates × 4 crude glycerol rates) treatments for each soil, with four replicates for each of seven sample dates (i.e., a total of 672 tubes for two soils). Three calcium nitrate N rates were 0, 10 and 40 mg N kg−1 (equivalent to 0, 19.5 and 78 kg N ha−1), approximating low (0N), medium (+10N) and high (+40N; above the critical value for the late-spring nitrate test for maize). Four crude glycerol rates were 0, 117, 468 and 1872 mg C kg−1 soil. Hereafter referred as 0C, +117C, +468C and +1872C. Assuming each rate would mix with 15 cm of topsoil, these would be roughly equivalent to 0, 228.2, 912.6 and 3650.4 kg C ha−1. Combined these rates (glycerol C and NO3–N) provide a range in C-to-N ratio of inputs from 2.9 to 187.2.

2.3.2 Inorganic nitrogen, net nitrogen mineralisation and microbial biomass

All incubated soil samples were extracted for IN (NH4+–N + NO3–N) analysis at 1, 3, 7, 14, 28, 63 and 98 days after adding the glycerol and N treatments. The soil IN was determined by extraction with 0.5 M potassium sulphate (K2SO4; 1:5 ratio, w:v), shaking for 1 h, filtering the supernatant through Whatman filter paper No. 1 (VWR International, LLC, Radnor, PA, USA) and storing the extract at −20°C until analysis. Soil extracts were analysed colorimetrically on a Synergy™ HTX Multi-Mode Microplate Reader (BioTek Instruments, Inc., Winooski, VT, USA) using the salicylate and ammonia cyanurate reagent packets (Hach Company, Loveland, CO, USA) at 595 nm for NH4+–N and using the single-reagent method (vanadium III, sulphanilamide and N-(1-naphthyl)-ethylenediamine dihydrochloride) at 540 nm for NO3–N (Doane & Horwáth, 2003).

Microbial biomass carbon (MBC) and MBN were determined on days 0 (initial), 7 (shortly after C/N addition) and 98 (final) using the chloroform-fumigation extraction method (Vance et al., 1987). Briefly, for each soil, one of the incubated 5 ± 0.04 g soil samples were placed in a 50 ml beaker, fumigated with chloroform for 24 h in a desiccator. Another 5 ± 0.04 g soil sample was not fumigated but placed in the fume hood next to the desiccator. After 24 h, the soils were extracted with 25 ml 0.5 M K2SO4 for 1 h on an orbital shaker, the supernatant filtered through Whatman #1, and stored at −20°C until analysis. Salt extractable C and N in the extracts were determined on a TOC-TN analyser (TOC-V-CPN, Shimadzu Scientific Instruments Inc., Columbia, MD, USA). The MBC was calculated as the difference in salt extractable C levels between fumigated and non-fumigated soil samples with a proportionality factor of 0.45 (Vance et al., 1987). The MBN was calculated as the difference in salt extractable N contents between fumigated and non-fumigated soils with a proportionality factor of 0.54 (Brookes et al., 1985).

2.4 Statistical analyses

The laboratory incubation was a completely randomised design with four replications (considered random), with the four glycerol rates, three N rates, two soils and time (d) as main fixed factors. We checked the data (NH4+–N, NO3–N, MBC and MBN) with the Shapiro–Wilk's normality test and Barlett's heterogeneity of variances test before analyses, and data were transformed to avoid anova assumption violations. Thus, NH4+–N and NO3–N were square root transformed and MBC was log transformed for anova. We used a four-way anova to determine the main and interactive effects of glycerol rates, N rates, soils and sample time on response variables (NH4+–N, NO3–N, MBC and MBN).

For NH4+–N and NO3–N, the means of glycerol treatment differences at each sampling day, soil and fertiliser N rate were separated using Fisher's least significant difference (LSD). Net N mineralisation was calculated using the slope of the IN (mg kg−1 day−1) once IN was released from MB turnover by taking a linear regression (all regressions had p values <0.05, and R2 > 0.93) between the last >1 mg N kg−1 concentration and succeeding points after that (Dou et al., 1996; Thangarajan et al., 2015). A two-way anova was used to determine the main and interactive effects of glycerol rate and N rate on mineralisable IN release rates within each soil. Furthermore, the mean separations for net N mineralisation rates among glycerol and N rates were performed using the least-squares method (LSMEANS procedure). A one-way anova was used for MBC, MBN and IN to determine statistically significant differences between glycerol treatments within each soil and at each N rate. All statistical analyses were done at the 5% significance level (α = 0.05) with the R statistical package (R Core Team, 2018).

3 RESULTS

3.1 Initial soil properties

The Clarion soil had larger values for nearly all soil properties than Sparta soil, except for NH4+–N (Table 1). The USDA soil textural classification showed that Clarion soil was characterised as clay loam, while Sparta soil was loamy sand. The gravimetric moisture content at the time of collection and MWHC of Clarion soil was 150% and 64% greater than Sparta soil. The Clarion soil had nearly three times greater total N concentration (2.43 g kg−1) than the Sparta (0.83 g kg−1). Most of the salt-extractable IN in both soils was present as NO3–N (96%–98%), followed by NH4+–N (2%–4%), but both were low (NH4+–N < 0.30 mg kg−1 and NO3–N < 6 mg kg−1). The MB in both soils was 0.88%–0.92% of total C and 0.88%–1.3% of total N. The Clarion soils had 3.4 times greater MBC and 4.4 times greater MBN than the Sparta soils.

3.2 Impact of glycerol and nitrogen application on inorganic nitrogen

Glycerol rate, N rate, soil, sample time and their interactions had significant main effects on NH4+–N concentrations (Table 3). Over the 98-day incubation and across all glycerol and N rates, the mean NH4+–N concentrations were low in both soils (<2.1 mg kg−1, Figure 1). In the Clarion soil under all treatments, NH4+–N concentrations gradually increased with time. In contrast, the Sparta soil showed markedly different dynamics, with rapid increase in NH4+–N within first 30 days and then began slowly declining with time. Greater glycerol rates (especially +468C and +1872C), regardless of other factors, tended to increase NH4+–N concentrations but timing and magnitude of peak NH4+–N differed quite dramatically (Figure 1), suggesting stimulated N mineralisation in the low-SOM, coarse-textured Sparta soils, although all concentrations were <3 mg kg−1.

TABLE 3. Summary of four-way anova* on the effect of adding glycerol and nitrogen rates to Clarion and Sparta soils across a 98-day incubation
anova factor df NH4+–N NO3–N df MBC MBN
F value p value F value p value F value p value F value p value
Glycerol rate (GR) 3 32.5 <0.001 14,096.5 <0.001 3 299.4 <0.001 42.5 <0.001
Nitrogen rate (NR) 2 4.13 <0.05 2006.2 <0.001 2 0.12 0.88 32.9 <0.001
Soil (S) 1 628.6 <0.001 125.4 <0.001 1 100.4 <0.001 484.2 <0.001
Time (T) 6 170.1 <0.001 1584.3 <0.001 1 123.7 <0.001 1.6 0.21
GR × NR 6 2.4 <0.05 649.5 <0.001 6 3.19 <0.01 12.8 <0.001
GR × S 3 41.0 <0.001 52.4 <0.001 3 75.1 <0.001 17.7 <0.001
NR × S 2 2.71 0.067 7.0 <0.01 2 0.03 0.97 11.4 <0.001
GR × T 18 17.3 <0.001 435.8 <0.001 3 20.1 <0.001 28.0 <0.001
NR × T 12 2.20 <0.05 15.7 <0.001 2 2.58 0.08 2.74 0.07
S × T 6 88.7 <0.001 5.92 <0.001 1 21.9 <0.001 3.37 0.07
GR × NR × S 6 3.23 <0.01 15.3 <0.001 6 1.91 0.08 0.50 0.80
GR × NR × T 36 2.11 <0.001 29.4 <0.001 6 6.54 <0.001 8.14 <0.001
GR × S × T 18 8.55 <0.001 20.1 <0.001 3 61.3 <0.001 11.5 <0.001
NR × S × T 12 1.94 <0.05 9.52 <0.001 2 5.75 <0.01 1.27 0.29
GR × NR × S × T 36 1.61 <0.05 9.98 <0.001 6 6.23 <0.001 2.39 <0.05
  • Abbreviations: df, the degree of freedom; MBC, microbial biomass carbon; MBN, microbial biomass nitrogen.
  • * The significance for all analyses was set at an alpha (α) of 0.05 (significant values are shown in bold).
Details are in the caption following the image
Ammonium (NH4+–N) concentrations in the Clarion and Sparta soils across the 98-day incubation. Means (n = 4) of treatments within glycerol and nitrogen rates. Vertical bars above the figures represent Fisher's least significant difference (LSD, p < 0.05)

The NO3–N concentrations were significantly influenced by all factorial combinations and their interactions (Table 3). The greatest NO3–N concentrations, within soil and N rate, always occurred in the 0C treatment (Figure 2), indicating that all treatments without added glycerol C resulted in net N mineralisation. Compared to 0C, and across N rates, +117C reduced NO3–N by 40%–61% and 21%–56% in the Clarion and Sparta soils by the end of the incubation, respectively. The highest rates of glycerol (+468C and +1872C) resulted in nearly 100% NO3–N reduction (<8 mg N kg−1) in both soils to the end of incubation. At 0N and +10N rates, the NO3–N concentrations with any rate of glycerol added were not significantly different from zero up to 28 days, and <3.5 mg N kg−1 across both soils, suggesting net N immobilisation during this period. However, the Sparta soil that received +468C with 0N had a significant difference compared with +1872C but released only 7 mg N kg−1 at 98 days (Figure 2). Conversely, the Clarion soil with +40N and +468C released 6 mg N kg−1 at 98 days, albeit not significantly different from +1872C. Overall, the two higher glycerol rates (+468C and +1872C) were not significantly different across the soils, N rates and sampling days.

Details are in the caption following the image
Nitrate (NO3–N) concentrations in Clarion and Sparta soils across the 98-day incubation. Means (n = 4) of treatments within glycerol and nitrogen rates. Vertical bars above the figures represent Fisher's least significant difference (LSD, p < 0.05)

We calculated net N mineralisation from the slope of IN for those soil/treatment combinations that eventually released IN (Table 4). The lowest glycerol rate (+117C) and in a few instances the moderate rate (+468C), completely immobilised N but released it later in the incubation (Figure 2; Tables 4 and 5). The more N added, the quicker the release and the greater net N mineralisation rate across both soils (Tables 4 and 5). Adding glycerol C, however, changed these dynamics. At +10N with +117C, Sparta soil released IN earlier than Clarion soil, 14–28 versus 28–63 days, but there was no difference in net N mineralisation rates (0.57–0.74 mg N kg−1 day−1, Table 4). Both soils also released IN earlier (7–14 days) when +40N was added, but only with the lower C rate (+117C). Adding the +117C increased net N mineralisation across all N rates by 102% and 96% in the Clarion and Sparta soils, respectively. We also observed that when no glycerol C (0C) was added, +40N increased net N mineralisation rates by 158% and 175% in the Clarion and Sparta soils compared to 0N and +10N.

TABLE 4. Net nitrogen (N) mineralisationa after release at different time points during the 98-day incubation (mean ± standard error)*
N rates (mg N kg−1 soil) Glycerol rates (mg C kg−1 soil) Clarion soil Sparta soil
Release period (days) Net N mineralisation rate (mg kg−1 day−1) Release period (days) Net N mineralisation rate (mg kg−1 day−1)
0 0 0 0.44 ± 0.03 bB 0 0.49 ± 0.02 bB
117 28–63 0.59 ± 0.07 aB 14–28 0.58 ± 0.02 aB
10 0 0 0.46 ± 0.03 bB 0 0.47 ± 0.02 bB
117 28–63 0.74 ± 0.07 aB 14–28 0.57 ± 0.05 aB
40 0 0 1.16 ± 0.04 aA 0 1.32 ± 0.03 aA
117 7–14 0.64 ± 0.01 bB 7–14 0.75 ± 0.04 bA
anova factorb df F value p value F value p value
Glycerol rate (GR) 1 0.39 0.54 25.9 <0.001
N rate (NR) 2 35.7 <0.001 177.3 <0.001
GR × NR 2 39.7 <0.001 75.6 <0.001
  • a Net N mineralisation calculated as slope of inorganic N (NH4+–N + NO3–N) during incubation.
  • b Summary of two-way anova on the effect of adding glycerol and nitrogen rates in Clarion and Sparta soils. The significant values (p < 0.05) are shown in bold.
  • * Significant difference (p < 0.05) among N rates within each glycerol rates indicated with capital letters, among glycerol rates within each N rates with lower case letters.
TABLE 5. Surplus/deficient nitrogen (mean ± standard error) in microbial biomass at 7 and 98 days
Nitrogen rates (mg N kg−1 soil) Glycerol rates (mg C kg−1 soil) Clarion soil Sparta soil
Surplus/deficient nitrogen (mg N kg−1 soil)a Surplus/deficient nitrogen (mg N kg−1 soil)a
C:Nb At 7 days At 98 days C:N At 7 days At 98 days
0 117 19.8 −5.05 ± 2.24 a 6.69 ± 6.71 b 25.2 4.52 ± 0.86 aB 1.55 ± 1.35 aA
468 79.1 −7.33 ± 2.80 a 14.5 ± 5.07 a 100.9 3.02 ± 1.07 aB 1.10 ± 0.91 aB
1872 316.2 −9.78 ± 2.54 a 2.59 ± 2.67 b 403.4 3.19 ± 0.81 aC 5.25 ± 0.77 aB
10 117 7.35 3.52 ± 1.00 a −1.29 ± 5.78 c 8.0 5.13 ± 1.05 aB −0.93 ± 2.45 bA
468 29.4 0.94 ± 0.83 a 20.4 ± 7.42 a 32.0 4.84 ± 1.14 aB 4.18 ± 1.73 aB
1872 117.6 −1.94 ± 0.80 b 8.26 ± 6.77 b 127.9 7.83 ± 0.92 aB 9.04 ± 1.32 aB
40 117 2.55 22.7 ± 1.05 a −5.87 ± 4.01 b 2.6 16.5 ± 1.38 bA 1.14 ± 1.85 cA
468 10.2 16.1 ± 0.75 b 18.1 ± 8.09 a 10.5 13.3 ± 0.88 bA 8.49 ± 1.27 bA
1872 40.8 11.8 ± 0.55 b 12.8 ± 4.21 a 41.9 21.4 ± 0.63 aA 19.7 ± 3.83 aA
anova factorc df p value p value p value p value
Glycerol rate (GR) 2 <0.001 <0.01 <0.001 <0.001
N rate (NR) 2 <0.001 0.97 <0.001 <0.001
GR × NR 4 0.36 0.39 <0.01 <0.05
  • a Surplus/deficient nitrogen (N) calculated as the microbial biomass N difference between glycerol carbon (C) addition and control (0 glycerol) at each N rate.
  • b C:N ratio = (Added glycerol C)/(Added N rate + Background soil inorganic N).
  • c Summary of two-way anova on the effect of adding glycerol and nitrogen rates in Clarion and Sparta soils at 7 and 98 days. The significant values (p< 0.05) are shown in bold. Significant difference (p < 0.05) among N rates within each glycerol rates indicated with capital letters, among glycerol rates within each N rates with lower case letters.

3.3 Impact of glycerol and nitrogen application on MB

A complex interaction of all factors significantly influenced the soil MBC (Table 3). For each soil, N rate and sampling day (at 7 and 98 days only), the mean MBC generally increased with increasing rate of glycerol (Figures 3 and 4). Glycerol greatly increased MBC at 7 days, especially with +468C and +1872C treatments regardless of soil or N rate. At 7 days, +1872C glycerol with or without N showed the greatest mean MBC (809–2359 mg C kg−1), and MBC at different glycerol rates was approximately 66%–545% and 93%–1323% greater than 0C in the Clarion and Sparta soils, respectively. MBC declined by 98 days, but glycerol still had mostly positive effects on MBC at moderate (+468C) and high (+1872C) rates of glycerol with 41% to 8179% increases across N rates and soils (Figures 3 and 4). However, in the Sparta soil that received +1872C, there was an unusually high concentration of MBC at 98 days, ranging from 1009 to 7286 mg C kg−1 (Figure 4).

Details are in the caption following the image
Microbial biomass carbon (MBC) in the Clarion soil at 0 (background or no glycerol added), 7, and 98 days of the incubation. Means and standard error bars are shown (n = 4). Significance between glycerin rates indicated by lowercase letters (p < 0.05)
Details are in the caption following the image
Microbial biomass carbon (MBC) in the Sparta soil at 0 (background or no glycerol added), 7 and 98 days of the incubation. Means and standard error bars are shown (n = 4). Significance among glycerin rates, within N rates, indicated by lowercase letters (p < 0.05)

The proportion of N in MB versus salt-extractable IN was very dynamic (Figures 5 and 6), as indicated by significant interactions among all factors (Table 3). In the Clarion soil, early in the incubation (7 days), +117C increased MBN by 16%–50% but only with addition of N, however, glycerol additions above +117C decreased MBN by 9%–30% compared to lower additions (Figure 5). The MBN response to glycerol in the Sparta soil was more straightforward, with 58%–1137% increases in MBN compared to 0C across glycerol rates (Figure 6).

Details are in the caption following the image
Microbial biomass nitrogen (MBN; solid bars) and inorganic nitrogen (IN; NH4+-N + NO3-N; hatched bars) in the Clarion soil at 0 (background or NO glycerol added), 7 and 98 days of the incubation. The significant difference among glycerol rates indicated by capital letters for MBN and lowercase letters for IN at p < 0.05
Details are in the caption following the image
Microbial biomass nitrogen (MBN; solid bars) and inorganic nitrogen (IN; NH4+-N + NO3-N; hatched bars) in the Sparta soil at 0 (background or no glycerol added), 7 and 98 days of the incubation. The significant difference among glycerol rates indicated by capital letters for MBN and lowercase letters for IN at p < 0.05

Later in the incubation, at 98 days, IN was likely released from both MBN and mineralised SOM turnover. The evidence for this comes from the increase in total soil N pools across treatments but also declines in MBN pool from 7 to 98 days (Figures 5 and 6). The proportion of IN relative to MBN decreased with increasing glycerol rates in both soils. At 98 days, the +117C rate decreased IN approximately 40%–63% and 21%–56% in Clarion and Sparta soils, respectively, compared to 0C. Adding N increased MBN at 98 days by 10%–166% and 14%–167% in +468C and +1872C, respectively, across N rates and soils. With +468C and +1872C, however, there was less IN compared to 0C, and most of that N was probably still in MBN (65%–99% on average across N rates and soils). We also observed that there is a relationship between the MBN drop (change in MBN between 7 and 98 days) and cumulative net N mineralisation, but more in Clarion soil than the Sparta (data not shown). Across both soils, for every 1 mg N kg−1 increase in MBN between 7 and 98 days cumulative net N mineralisation increased by 25 mg N kg−1 (data not shown).

4 DISCUSSION

In agricultural soils, application of moderate to high rates of N fertiliser at the wrong time can result in large N losses via runoff, leaching, tile drainage and denitrification (Charles et al., 2017; Dinnes et al., 2002). Therefore, the challenge is to reduce N losses while enhancing crop fertiliser N use efficiency. Our laboratory incubation showed that adding a labile C soil amendment, in the form of glycerol, can enhance microbial N immobilisation and decrease soil NO3–N concentrations. This N immobilisation was temporary, and timing and magnitude of glycerol's effect on N dynamics were highly dependent on glycerol rates, soils and added N.

4.1 Nitrogen immobilisation with crude glycerol

Supply and demand for soil N by microorganisms is largely regulated by C availability. Labile C limits soil MB, which represents only a small fraction of the total soil C and N (<6%), but is highly dynamic (Jenkinson & Ladd, 1981; Vance & Chapin III, 2001). A recent meta-analysis reported that adding simple C substrates (e.g., glucose and acetate) at rates ≥500 mg C kg−1 can enhance NO3–N immobilisation, but lower rates were insufficient to stimulate soil microbial NO3–N immobilisation (Cheng et al., 2017). Since glycerol is a highly labile hydrocarbon, adding glycerol alone or in combination with N immediately increased MBC and MBN in both soils (Figures 3-6), supporting our first hypothesis that adding glycerol to soils will increase MB, but the increase and length of effect may vary between the soils (as discussed in Section 4.2). Similarly, other lab-based and field studies show MB increases when soils are amended with glycerol (Alotaibi & Schoenau, 2011, 2016; Redmile-Gordon et al., 2014), other carbohydrates (Recous et al., 1990; Sparling et al., 1981; Wesselsperelo et al., 2006), oily food wastes (Plante & Voroney, 1998; Rashid & Voroney, 2003) and plant residues (Baldwin-Kordick et al., ; Koranda et al., 2014; Li et al., 2020). In the Clarion soil, MBC declined by 98 days, but moderate (+468C) and high (+1872C) rates of glycerol still had positive effects on MBC across N rates (Figure 3). However, in the Sparta soil that received a high (+1872C) rate of glycerol, MBC was an unusually high (1009–7286 mg C kg−1) by 98 days (Figure 4). We have double-checked this for analytical or experimental error, and we have found no explanation for this other than perhaps MB is C saturated.

In our study, glycerol additions resulted in a high concentration of MBC lasting up to 98 days (Figures 3 and 4) but not always coupled with a consistent increase in MBN (Figures 5 and 6). The reason for this is likely due to supply and demand for N resulting in a flexible microbial C and N stoichiometry. Carbon is more plastic and can have a wider range, but N is more restricted with these levels of C addition. We found that N became limiting when labile C supply was high, in other words, MBN was increased by additional N. In a similar lab-based incubation study, glycerol additions increased MBC and MBN by 300%–700% and 200%–1050%, respectively, compared to a control just 10 days after adding glycerol, and effects were largely dependent on glycerol rates (20, 200 and 2000 mg C kg−1) alone or combined with 150 mg N kg−1 as urea (Alotaibi & Schoenau, 2011). They also suggested that microbial growth was not limited by C at the higher glycerol rates but was limited by N (and our data show this).

The rapid increase in MBC paired with NH4+–N and NO3–N dynamics (Figures 1 and 2) point to enhanced co-metabolism of labile SOM and subsequently either anabolism of N into microbial bodies or nitrification to NO3–N (Qiu et al., 2008). A rapid increase in MB depletes soil solution N (via immobilisation) but also forces microbes to mineralise native SOM for N as evidenced by slow and then rapid increases in NH4+–N concentrations (Figure 1), albeit still low compared to NO3–N concentrations. This newly mineralised N may also be stored in microbial bodies (Figures 5 and 6; Table 5). Even at our lowest glycerol rate, there was hardly any measurable NO3–N for up to 14 days (Figure 2). In a similar lab-based incubation study, glycerol application (1500 mg C kg−1) alone or combined with 50 mg N kg−1 as NH4NO3 increased MBN by 144%–148% and lowered soil NO3–N concentrations by 53%–94% compared to a control within 4 days of glycerol application (Redmile-Gordon et al., 2014). Apart from glycerol, the addition of other labile C-compounds (e.g., glucose and acetate) has also been effective in lowering soil NO3–N concentrations (Azam et al., 1985, 1993; Kai et al., 1973; Ladd et al., 1977; Magill & Aber, 2000; Recous et al., 1990).

Timing of NO3–N release and subsequent net N mineralisation rates were dependent on glycerol and N rates (Table 4). The lowest glycerol rate (+117C) delayed initial NO3–N release compared to the control but increased net N mineralisation thereafter. The greatest two glycerol treatments, +468C and +1872C, never released NO3–N even to 98 days. Some of this N appeared to still be stored in MB (Table 5; Figures 5 and 6) and would likely be mineralised and released later if the incubation were continued longer than 98 days as discussed in the next section. These results suggest that synchronising glycerol rates and application timing could sustain N retention and delay N release to crop demand – something important to consider with field application of glycerol in early spring and late autumn (when NO3 leaching is most prevalent).

4.2 Release of plant available nitrogen and its fate

With current agricultural practices in many parts of the world, the amount of fertiliser N applied exceeds that of crop N uptake (Cassman et al., 2002) and some of this excess N can be lost from the agroecosystem (Charles et al., 2017; Dinnes et al., 2002; Jin et al., 2014). Therefore, the challenge is to synchronise soil N availability with peak crop N demand. Immobilising soil NO3–N with labile C amendment is ideal for immediately preventing losses and improving environmental quality but may come at an economic risk if this N is not available to crops. Thus, our second question is, would any observed N immobilisation be temporary? The answer is yes, supporting our second hypothesis that soil MB would temporarily immobilise NO3–N and then some of that N could be found in MB (Table 5) and would become available through MB turnover (Table 4; Figure 2). This turnover of N was most likely due to depletion of labile C reserves, resulting in mineralisation of SOM, or just natural cell death and MB turnover without the need for stimulating a positive priming effect through glycerol C addition (Redmile-Gordon et al., 2014). However, these N immobilisation-mineralisation dynamics were highly dependent on added N rate, glycerol rate and soil (Tables 4 and 5; Figure 2).

In our study, after release, net N mineralisation rates were 20%–64% greater with the low glycerol rate (+117C) compared to the control (0C) across both soils and lowest N rates (0N and +10N; Table 4). In a similar lab-based incubation study, Redmile-Gordon et al. (2014) reported that 4 days after substrate addition, glycerol alone or combined with 50 mg N kg−1 (1.6–1.8 mg N kg−1 day−1 mineralisation rate) almost doubled net N mineralisation rate of the control (0.8 mg N kg−1 day−1). Azam et al. (1985) showed in an 84-day incubation that N immobilisation-mineralisation dynamics occurred most rapidly with glucose addition (within 3 days) compared to cellulose, wheat straw or maize stalks – which were not re-mineralised until 21 to 28 days. Thus, glycerol may provide a quicker release and faster N mineralisation rate from MBN and labile SOM than more recalcitrant materials like crop residues.

One of the most interesting findings from our study was the large, overwhelming influence soil had on N dynamics (Table 4). Some trends are salient across both soils, like similar net N mineralisation trends (Table 4; Figure 2), but the differences between the soils might be even more interesting. Even though the soils were collected shortly before soybean harvest and under similar maize-soybean management, the response to glycerol and N rate varied (Figure 1 and Table 4). In partial support of our third hypothesis, the Clarion soil with greater SOM did retain N longer than the loamy sand Sparta soil and released NO3–N at slower rates (Table 4; Figures 2, 5 and 6). However, we cannot confirm that this is due to differences in SOM because the soils also differed in many other aspects (Table 1). It could be due to a greater C:N ratio in the Clarion (12:1) compared to Sparta (10:1) soil as well. Greater endogenous C relative to N in SOM may provide more immobilisation potential, and slower N release. While we do not know the exact mechanisms causing the differing response to glycerol and N inputs, it highlights the importance of soil type in regulating their interaction on soil N dynamics. Our finding aligns with another incubation study, using a 15N-enriched source, that found decomposition and net immobilisation and mineralisation of N in a sandy loam soil were more rapid than in a clayey soil (Van Veen et al., 1985). They attributed this to the clay particles having less habitable pore space for larger predators (such as nematodes) and protected bacteria from predation and desiccation.

Careful management of a labile C additive like glycerol is needed to synchronise N immobilisation and release with plant N demand. Both timing and balance of labile C with N should be considered with this regard (Tables 4 and 5, Figure 2). In some support for our fourth hypothesis, adding more N meant more N stored in MB, although this was a non-linear relationship and featured complex interactions (Figures 5 and 6). Increased N rates with glycerol rates increased total N storage in MB – especially later in the incubation and at the highest N rate (Table 5). Thus, getting the N immobilisation-mineralisation timing and balance correct is critical if labile C amendments are going to work at the field-scale in agroecosystems.

Although it was not our original intent, here we found evidence for what is known as the N priming effect (Jenkinson et al., 1985; Kuzyakov et al., 2000). Adding N alone, without any glycerol, increased net N mineralisation rates from +5% (with +10N) to +169% (with +40N), but depended strongly on the soil and whether or not glycerol was added (Tables 3 and 4). This is in stark contrast to plethora of studies showing that adding synthetic N fertiliser decreases MB (Geisseler et al., 2016; Geisseler & Scow, 2014), enzyme activities (Liao et al., 2017; Zhang et al., 2015) and even mineralisation (Mahal et al., 2019). There is also limited, though significant, evidence for enhanced net N mineralisation when synthetic N fertiliser is added (Chen et al., 2018; Fernández et al., 2017; Lu et al., 2011). These inconsistent findings are likely due to specific fertiliser source and C and N stoichiometry of microorganisms (Craine et al., 2007; Mahal et al., 2019; Moorhead & Sinsabaugh, 2006). In other words, whether fertiliser N suppresses or enhances net N mineralisation may depend on quality of native labile SOM (C:N ratio), soil texture, amount of added N and whether the N addition is balanced with labile C. The consequences of a positive N priming effect are reduced SOM stocks (Mulvaney et al., 2009; Poffenbarger et al., 2017; Russell et al., 2009), increased soil NO3–N leaching and gaseous N losses via nitrification and denitrification (Lu et al., 2011; Ma et al., 1999; Schleusner et al., 2018). The contribution of SOM-N is loosely estimated or ignored when N fertilisers are added. Better understanding of this N priming effect, and role of labile C, could improve fertiliser N rate recommendations.

The N immobilised in MB from glycerol did not account for the entire N released compared to the control (0C, Table 5). The fate of this N is unknown but likely ended up in one or two places: (i) it was lost as N gases (e.g., NO, N2O or N2) through nitrification and denitrification or (ii) it was rapidly stabilised as organic N to soil colloids. There is limited work published on the effects of crude glycerol application on the evolution of gaseous N losses (Alotaibi & Schoenau, 2013). The N2O emissions are more worrisome since they are a potent GHG, and some studies show that labile C tends to exacerbate N2O emissions (Azam et al., 2002; Morley & Baggs, 2010; Weier et al., 1993). Further research is needed to quantify gaseous N losses after glycerol addition to understand if we are trading an air quality problem for a water quality problem.

The second fate of N with glycerol addition could have been rapidly transformed into organic, mineral-associated N (not extractable with 0.5 M K2SO4) by both biotic and abiotic mechanisms (Castellano et al., 2012; Johnson et al., 2000; Lewis & Kaye, 2012; Moritsuka et al., 2004). We can make some estimates based on assumptions and simple subtraction. For example, MBN concentrations at 98 days with the highest glycerol rates were 1.2–2.3 (+468C) and 1.1–2.7 (+1872C) times greater than that of control (0C) across N rates, and therefore, about 1–21 (+468C) and 2.6–20 (+1872C) mg N kg−1 would likely be released if the incubation were continued longer than 98 days (Figures 5 and 6; Table 5). However, specific results depended on the soil.

4.3 Further considerations for glycerol application to fields

Although it is difficult to extrapolate from a laboratory incubation to the field, we do know a few things about adding glycerol to soils: (i) NO3–N was rapidly immobilised with higher rates of glycerol addition, (ii) at the end of the 98 day incubation, MB still held a surplus of N compared to control (0C; Table 5) that could potentially be supplied to crops and (iii) range of soil temperature in Midwest U.S. winters are much lower (<0 to 5°C) than our incubation temperature (25°C) suggesting that the storage and release could be further delayed. Meanwhile, there are other sources of considerable variation in field studies that are not encompassed by our incubation study. The depth at which glycerol C will penetrate the soil surface will depend on rainfall during autumn months and winter snowmelt. No study, to our knowledge, has determined the depth of glycerol C penetration to depth. Furthermore, N release timing and mineralisation rates will be delayed and slowed in much cooler soils in late autumn to winter transition in the U.S. Midwest, especially compared to optimal conditions used in our laboratory incubation (25°C and 60% MWHC). Hence, a deeper mixing with the soil profile, delayed release time, and slower release rate will all change the effective depth and concentration of glycerol C that will be available to soil microbes and alter soil N dynamics. Altogether these factors will regulate the overwinter, cumulative NO3–N leaching from Midwestern U.S. farm fields, but a field study is needed to evaluate this.

5 CONCLUSIONS

Crude glycerol, a natural byproduct of biodiesel production, was effective at immobilising NO3–N and increasing MB. The soil type, to some extent, and additional NO3–N changed the glycerol effect on soil N dynamics. Net N immobilisation was temporary, and some of that immobilised N was eventually released from the lowest glycerol C rate through turnover of MB – an important consideration for growers concerned about microbes ‘stealing’ N from crops. Hence, a significant proportion of re-mineralised N may become available to plants, depending on the glycerol rate and timing between glycerol application and crop planting. Based on the timing of N release and net N mineralisation, we suspect application in late autumn after harvest in Midwestern U.S. agroecosystems would be preferable to springtime – but recommendation on glycerol timing and rates remains uncertain and needs to be tested in the field. This laboratory incubation was a proof of concept that shows promise of using glycerol as a soil amendment to immobilise NO3–N and may increase the sustainability of the biodiesel production supply chain. However, future incubation studies should be conducted using 15N tracing technique to calculate gross N immobilisation rate and MB15N. Further field trials are needed to evaluate the broader and long-term impacts of glycerol application on crop yield, soil contamination especially those concerning accumulation of minor contaminants like salts or small concentrations of methanol, N leaching and GHG emissions from maize-soybean rotations of the Midwestern United States.

ACKNOWLEDGEMENTS

The authors would like to thank the Renewable Energy Group, Inc. (REG) for providing glycerol and financial support. Special thanks are given to Jon Scharingson, Nick Schurr and Derek Huser from REG for their continuous help and support to carry out this study. The authors thank Philip M. Dixon and Katherine Goode, statistics consultants, Department of Statistics, Iowa State University for their co-operation and assistance with statistical analysis using R statistical software. We would also like to thank Mitchell T. Roush, Theresa Brehm, Christopher Kim and Stephen Potter for field and laboratory assistance. Open access funding provided by the Iowa State University Library.

    CONFLICT OF INTEREST

    The authors declare no conflicts of interest.

    AUTHOR CONTRIBUTIONS

    Mriganka De: Conceptualization (supporting); formal analysis (lead); investigation (lead); methodology (lead); project administration (supporting); software (lead); visualization (equal); writing – original draft (lead). John E. Sawyer: Conceptualization (lead); funding acquisition (lead); investigation (supporting); methodology (supporting); project administration (lead); visualization (supporting); writing – review and editing (equal). Marshall D. McDaniel: Conceptualization (lead); funding acquisition (lead); investigation (supporting); methodology (supporting); project administration (lead); supervision (lead); visualization (supporting); writing – review and editing (equal).

    DATA AVAILABILITY STATEMENT

    The original contributions presented in the study are included in the article. The data that support the findings of this study can be available from the corresponding author upon reasonable request.