Site description

The experiment was established in 1992 at the L’Acadie Experimental Farm (45°18′N, 73°21′W) of Agriculture and Agri-Food Canada in Quebec, Canada. Full details of the experiment were previously reported by Ziadi et al. (2014). Briefly, the soil is a poorly drained St. Blaise clay loam (364 g·kg⁻¹ clay, 204 g·kg⁻¹ sand in the Ap horizon) of the Orthic Humic Gleysol class (Soil Classification Working Group 1998) with flat topography and tile drains installed. The site was cropped to alfalfa (Medicago sativa L.) prior to 1992 and corn (Zea mays L.) from 1992 to 1994. The corn–soybean (Glycine max L.) rotation was initiated in 1995. At the initiation of the experiment, the site had 22 g SOC·kg⁻¹, 135 kg Mehlich-3 P·ha⁻¹ and a pH of 6.3 in the 0–15 cm depth. Poirier et al. (2009) reported 23 g SOC·kg⁻¹ in the 0–20 cm depth.

During the experimental period (1997–2016), growing season precipitation (GSP) varied from 440 mm in 2013 to 802 mm in 2005 (Fig. 1). The GSP was 28% and 12% above normal in 2005 and 2006, respectively, and 23%, 23%, and 30% below normal in 2004, 2012, and 2013, respectively, and near normal for the other years. Growing season air temperature ranged from 14.9 °C in 1997 to 17.3 °C in 2012 and was always within 10% of the climate normal (Fig. 1).

Fig. 1.

Growing season precipitation and mean air temperature during the study period.

Experimental design and treatments

The experiment was designed as a split plot with two tillage practices (MP and NT) as main plots and nine fertilizer combinations (0, 80, and 160 kg N·ha⁻¹ in a factorial with 0, 17.5, and 35 kg P·ha⁻¹) randomized as sub-plots, all replicated in four randomized blocks. Each sub-plot (experimental unit) was 25 m long × 4.5 m wide. The 160 kg N·ha⁻¹ and the 35 kg P·ha⁻¹ represented the recommended rates for grain corn in the region (CRAAQ 2010), but no N or P was added in the soybean phase. In this study, we only considered the three N rates in combination with the recommended P rate for grain corn for the two tillage practices, for a total of 24 experimental units.

Corn, grown for grain, was planted every year from 1992 to 1994 and in even years thereafter at a rate varying from 74 × 10³ seeds·ha⁻¹ in the first three years to 83 × 10³ seeds·ha⁻¹ in subsequent years. Soybean was sown at rate of 45 × 10⁴ seed·ha⁻¹ in odd years from 1995. Planting of both crops was done with a six-row, no-till corn planter with 76 cm row spacing. For NT, the plots were ridge-tilled from 1992 to 1997 and direct-seeded thereafter. Hence, our study focused on the period from 1997–2016 as the number of growing seasons were evenly distributed between the two crops from 1997. Nitrogen and P fertilizer in the form of urea and triple superphosphate, respectively, were side-banded (5 cm below and to the side of the seed row) at seeding at rates of 0, 40, and 40 kg N·ha⁻¹ and 35 kg P·ha⁻¹ followed by a side-dress, banded N application as ammonium nitrate at 0, 40, and 120 kg N·ha⁻¹ at the eight-leaf stage in corn with a disk opener (8–10 cm deep). All plots received 50 kg K₂O·ha⁻¹ side-banded at planting in 1992 and 2007 based on local soil test recommendations (CRAAQ 2010). The crop residues were left on the soil surface in all treatments after harvest. The MP operation was done every fall after crop harvest to 20 cm depth, followed by disking and harrowing to 10 cm in spring for seedbed preparation. Herbicides were applied based on provincial recommendations. Corn and soybean grain yield were determined at maturity (mid- to late-October) by harvesting the two inner rows of each plot on a 10 m long section as described by Ziadi et al. (2014).

Soil sampling and analysis

Composite soil samples (four cores per plot) were collected at the 0–5, 5–10, and 10–20 cm depths after corn harvest in fall of 2016 (25 yr after the initiation of the study). All visible plant materials were removed from the soil samples. A portion of the soil samples was air-dried and ground to pass a 2 mm sieve. Soil analyses on fresh soils were performed within 72 h. Soil MBC and MBN were determined on fresh samples (5 g oven-dry equivalent) using the chloroform fumigation-direct extraction method (Voroney et al. 2008). The C and N concentrations in the fumigated and non-fumigated extracts were quantified by combustion using an automated total organic C analyzer (SSM-5000A, Shimadzu, Kyoto, Japan). The MBC and MBN were calculated as the difference in C and N concentrations between fumigated and non-fumigated samples and corrected with an extraction efficiency factor of 0.45 for MBC (Wu et al. 1990) and 0.54 for MBN (Brookes et al. 1985). A 7 d aerobic incubation was used to measure soil respiration. A sub-sample of fresh soil (45 g oven-dry equivalent) was weighed into 250 mL Mason jars and moistened to initial field moisture content. The jars were loosely capped to minimize moisture loss and incubated in the dark at 20°C with soil moisture held constant throughout the incubation period. After each initial measurement (T₀), the jars were fitted with gas-tight lids with two ports for gas sampling. At T₀ and T₁ (24 h after), 20 mL of the jar headspace was taken using a syringe through the septum and transferred to pre-evacuated 12 mL glass vials. This first set of gas samplings (T₀ and T₁) were subsequently discarded and represented a pre-incubation period. After the T₀ gas sampling, the jars were left open for a few seconds for ambient air to circulate. During gas sampling at T₁, the valve of the other port was opened before removing the syringe to equalize the pressure in the syringe. The gas measurements were repeated for three other subsequent incubations (T₀ and T₂, T₀ and T_3, and T₀ and T₄) correcting the water content in each jar 24 h before the step T₀. The collected gas samples were immediately analyzed for CO₂ concentrations using a gas chromatograph (model 3800, Varian Inc., Walnut Creek, CA, USA) as described by Rochette and Bertrand (2008).

The POM was separated from whole soil as described by Gregorich et al. (2008), with slight modifications (St. Luce et al. 2014). Briefly, 25 g air-dried soil was dispersed by shaking on a reciprocating shaker at 180 excursions·min⁻¹ for 16 h with 100 mL of deionized water and 10 glass beads (6 mm diameter). The suspension was wet-sieved through a 53 μm sieve. The material retained on the sieve (POM + sand) was dried at 50°C until constant weight and ground using a mortar and pestle to pass a 250 μm sieve. The material that passed through the sieve, regarded as MAOM, was also dried and weighed. The average soil mass recovered after fractionation was 98.4%, with 22.7% and 75.7% recovered as POM and MAOM, respectively. Air-dried whole soil (<2 mm) was also ground to pass a 250 μm sieve. The C and N concentrations in POM and whole soils were determined by dry combustion using an Elementar CN Analyzer (Elementar Vario Macro Elementar Analyzer, Analysensysteme GmbH, Hanau, Germany). The C and N concentrations in POM and MAOM were calculated according to eq. 1:

(1)

where A = C or N concentration of POM and MAOM (g·kg⁻¹), respectively, C = SOC or STN concentration (g·kg⁻¹), F = C or N concentration of POM and MAOM (g·kg⁻¹), respectively, and T = sum of C or N concentration of both fractions. Total C and N content in whole soil, POM and MAOM were calculated for each layer by multiplying the concentrations by the corresponding bulk density. Although bulk density was not measured in our study, we used values reported by Li et al. (2017) that were taken from the same site two years prior to this study. Li et al. (2017) reported significant differences in bulk density only in the 0–5 cm layer between tillage systems; however, Poirier et al. (2009) found no significant tillage effect on bulk density in a study conducted at the same site in 2005. Nevertheless, to account for possible bulk density effects, SOC, STN, POMC, POMN, MAOMC, and MAOMN stocks (Mg C·ha⁻¹, Mg N·ha⁻¹) were expressed on an equivalent mass basis, using the average soil mass for the 0–5, 0–10, and 0–20 cm layers (Ellert et al. 2008). The fixed soil masses were 660, 1380, and 2830 Mg·ha⁻¹ for the 0–5, 0–10, and 0–20 cm soil layers, respectively.

Carbon and N inputs from corn and soybean during the study period (1992–2016) were determined using estimated dry matter yields with formulas proposed by Bolinder et al. (2007) since dry matter and plant tissue nutrient concentrations were not measured during the study. The harvest index was estimated using the relationship between grain yield and harvest index (Fan et al. 2017). We assumed that straw and root tissues for both crops contained 45% C (Bolinder et al. 2007). Furthermore, values used for straw and root N concentrations, respectively, were 9.4% and 7.6% for corn (Thiagarajan et al. 2018), and 6.6% (Thiagarajan et al. 2018) and 10% (Janzen et al. 2003) for soybean. The shoot:root ratios given by Thiagarajan et al. (2018) for the whole root profile for corn (4.0) and soybean (4.5) were used to estimate C and N input from roots. Carbon input through rhizodeposition represented 65% of the C input from roots (ref" rid="refg9">Bolinder et al. 2007); N input through rhizodeposition was not estimated. The cumulative C (C_C) and N (C_N) inputs for each crop were reported as the sum of the annual C and N inputs, respectively. Finally, the total cumulative C (C_CT) and N (C_NT) inputs were the sum of C_C and C_N across both crops.

Statistical analyses

Analysis of variance (ANOVA) on the effect of tillage, N rate, depth, and their two- and three-way interactions on the concentrations of the various soil parameters and the proportion of SOC and STN as POMC and MAOMC, and POMN and MAOMN, respectively was done using the Mixed procedure of SAS (SAS Institute 2020). Block and its interaction with tillage were treated as random effects. For SOC, STN, POMC, POMN, MAOMC, and MAOMN stocks in the 0–20 cm layer, estimated C_C input for each crop, tillage, and N rate were fixed effects in the analysis, with replicates and replicate interaction with tillage as random. Where necessary, data were transformed using the Box-Cox power transformation to meet the assumptions of normality and equality of variances. Treatment effects were deemed significant at p < 0.05. Means were separated with a post-hoc LSMEANS test with Tukey–Kramer adjustment at p < 0.05. A heat map of the Pearson correlations between the labile SOM fractions and SOC and total N was created, for the full data and separately by tillage system, using R software version 4.0.3 (R Core Team 2020) with the corrplot package (version 0.84).

Soil organic carbon and total nitrogen

Tillage, depth, and the tillage × depth interaction significantly affected SOC and STN concentrations (Table 1). We observed a decrease in SOC and STN concentrations with depth under NT; however, SOC and STN concentrations were evenly distributed under MP (Fig. 2). Moreover, SOC and STN concentrations were higher under NT than MP at all depths (Fig. 2), more so in the 0–5 cm layer (43% and 37% higher, respectively) compared with the 5–10 cm (28% and 26% higher, respectively) and 10–20 cm (11% and 16% higher, respectively) layers. We also found 22% and 20% higher SOC and STN stocks, respectively (Table 2), under NT compared with MP when the entire sampling depth (0–20 cm) was considered. There was no significant effect of N rate or tillage × N rate interaction on SOC and STN stocks at the 0–20 cm depth (Table 2).

Table 1.

Analysis of variance on the interactive effects of tillage, nitrogen fertilizer rate, and sampling depth on concentrations of the various soil parameters after 25 yr in a corn–soybean rotation.

Fig. 2.

Interactive effect of tillage and soil depth on soil organic carbon (SOC) and soil total nitrogen (STN) concentrations after 25 yr in a corn–soybean rotation. Error bars represent standard error of the mean for the interaction.

Table 2.

Effect of tillage and nitrogen fertilization on carbon and nitrogen stocks in whole soil and organic matter fractions in the 0–20 cm layer after 25 yr in a corn–soybean rotation.

The significantly higher SOC and STN stocks in the soil profile (0–20 cm) under NT than MP were expected given the greater SOC and STN concentrations under NT at all depths as well as similar bulk density below 5 cm (Li et al. 2017). Results from the previous sampling at this site in 2005 also found higher SOC stocks in the 0–20 cm layer under NT than MP (Poirier et al. 2009). Other studies also reported higher SOC and STN concentrations under NT than MP, with greater differences predominantly observed in the surface layer, and less apparent differences at lower depths (Angers and Eriksen-Hamel 2008; Zuber et al. 2015; Martínez et al. 2020). Our findings could be attributed to greater residue placement near the soil surface, reduced soil disturbance, enhanced microbial activity and macro-aggregate formation under NT (Balesdent et al. 2000; Six et al. 2000; Kan et al. 2020; Rahmati et al. 2020; Li et al. 2021). Dolan et al. (2006) reported 30% more SOC and STN stocks with NT than MP in the 0–20 cm layer in a silt loam soil in Minnesota, USA, but MP had twice as much SOC stock and about 30% more STN stock than NT in the 0–30 cm depth. Furthermore, unlike MP, we found significant stratification of SOC and STN under NT, which is agreement with previous findings (Blanco-Moure et al. 2013; Tivet et al. 2013; Dimassi et al. 2014). Mechanisms responsible for stratification under NT include accumulation of crop residues near the soil surface, leading to greater levels of aggregation and organic matter accumulation in the top 5 cm of soil (Balesdent et al. 2000).

Nitrogen fertilization may impact SOC and STN due to increased crop biomass production and return to the soil, as well as by enhancing residue decomposition rates (Stewart et al. 2017; Chen et al. 2018). Corn grain yield during the study period increased with N fertilization ( Supplementary Fig. S1 (cjss-2021-0129suppla.pdf) ¹ 1), but soybean grain yield was not affected by N rate applied in the corn phase ( Supplementary Fig. S2 (cjss-2021-0129suppla.pdf) ¹). In addition, corn and soybean grain yields were similar between MP and NT [6.2 Mg·ha⁻¹ and 5.3 Mg·ha⁻¹ (corn;  Supplementary Fig. S1 (cjss-2021-0129suppla.pdf) ¹), and 2.6 Mg·ha⁻¹ and 2.2 Mg·ha⁻¹ (soybean;  Supplementary Fig. S2 (cjss-2021-0129suppla.pdf) ¹) for MP and NT, respectively]. However, the C_C and C_N inputs were significantly higher for MP than NT for corn but similar between tillage systems for soybean (Table 3). With respect to N fertilization, C_C and C_N increased with N fertilizer rate for corn, while for soybean, they were higher for the unfertilized N than the plots that received 80 kg N·ha⁻¹ (N applied only in the corn phase). Hence, the effect of N fertilization on C_C and C_N inputs from corn was in agreement with corn grain yield ( Supplementary Fig. S1 (cjss-2021-0129suppla.pdf) ¹). However, the lower C_C and C_N inputs for soybean from the plots that received 80 kg N·ha⁻¹ compared with those that received 0 kg N·ha⁻¹ in the corn phase were probably due in part to slightly lower soybean grain yield ( Supplementary Fig. S2 (cjss-2021-0129suppla.pdf) ¹) and harvest index for the 80 kg N·ha⁻¹ plots (data not shown).

Table 3.

Effect of tillage and nitrogen fertilization on plant carbon and nitrogen inputs after 25 yr in a corn–soybean rotation.

For the period 1992–2005 at the same experimental site, Poirier et al. (2009) also reported similar corn and soybean grain yields. Certainly, environmental conditions have profound impacts on crop yield; however, average corn and soybean grain yields were quite similar between the two studies. The fact that tillage system, as a main factor, had no impact on corn and soybean grain yields suggests that increases in SOC and STN observed under NT were not reflective in corn and soybean grain production and stability under these conditions. The coefficient of variation in grain yield over the study period was 39% and 47% for corn and 39% and 44% for soybean under MP and NT, respectively (data not shown). In years where grain yields differed between the tillage systems, they were always higher for MP than NT ( Supplementary Figs. S1 (cjss-2021-0129suppla.pdf) and  S2 (cjss-2021-0129suppla.pdf) ¹). In contrast, Congreves et al. (2017) showed that corn grain yield stability can be improved by increasing SOC content. In our study, it was notable that soybean grain yield in the unfertilized N plot was similar to the fertilized plots, indicating that soybean grain yields were not influenced by N rates applied to corn. It is likely that soybean nodulation and biological N fixation were reduced in the plots fertilized in the corn phase and probably contributed to lower harvest index, and reduced C and N accumulation in soybean biomass and rhizodeposition in these plots (Fustec et al. 2010; Wu et al. 2017). Hence, the soybean C and N inputs without N fertilization partly offset the gain in corn C and N inputs with N fertilization. This suggests that to maximize C and N return to the soil in this system, corn should receive N fertilization but not soybean, confirming local recommendations.

Over the study period in this corn–soybean system, tillage, N fertilizer rate, and tillage × N fertilizer rate interaction had significant impacts on C_CT, whereas, tillage and N fertilizer rate significantly influenced C_NT (Table 3). In plots fertilized at 160 kg N·ha⁻¹ (N was applied only in the corn phase), C_CT input was 11.7% higher under MP than NT (Fig. 3). In addition, C_CT input increased with N rate under MP and was significantly lower for the control than N fertilized plots under NT (Fig. 3). Overall, C_CT and C_NT inputs were 6.9 Mg C·ha⁻¹ and 1.1 Mg N·ha⁻¹ higher, respectively, for MP than NT (Table 3). In addition, C_CT and C_NT inputs increased with N fertilization. The C input was 5.8 Mg C·ha⁻¹ higher for MP than NT for the 1992–2005 period (Poirier et al. 2009). We note that Poirier et al. (2009) used a single harvest index value per crop across all treatment combinations, while in our study, the harvest index was estimated based on the relationship between grain yield and harvest index, which better accounts for variability due to management practice, such as N fertilization (Fan et al. 2017). Although N rate and year × N rate, but not tillage × N rate, significantly affected soybean yield, all these factors significantly influenced corn grain yield in our study ( Supplementary Figs. S1 (cjss-2021-0129suppla.pdf) and  S2 (cjss-2021-0129suppla.pdf) ¹). Taken together, this suggests that the difference in C_CT and C_NT between MP and NT may continue to increase over time due to higher corn grain yield under MP than NT at 160 kg N·ha⁻¹ ( Supplementary Fig. S1 (cjss-2021-0129suppla.pdf) ¹). Other studies also found higher plant C inputs under MP than NT (Allmaras et al. 2004; Poirier et al. 2009) and an increase in input with N fertilization (Allmaras et al. 2004; Poirier et al. 2009; Congreves et al. 2017).

Fig. 3.

Interactive effect of tillage and nitrogen fertilization on total cumulative plant carbon input after 25 yr in a corn–soybean rotation. Error bars represent standard error of the mean for the interaction MP, moldboard plow; NT, no-till.

In spite of the results obtained for C_CT and C_NT inputs, N rate or its interaction with tillage and depth had no significant influence on SOC and STN concentration and stocks in our study, which corroborates findings reported in other studies (Neff et al. 2002; Dolan et al. 2006; Blanco-Canqui et al. 2014; Chahal et al. 2021), including a previous study at this site (Poirier et al. 2009). Dolan et al. (2006) reported no effect of N fertilization on SOC and STN storage from a 23 yr field experiment. The lack of an observed N rate effect or interactions with tillage on SOC and STN in our study was probably due to a dilution effect of tillage, which affected the soil condition and hence microbial access and turnover rates. While C_CT and C_NT inputs were higher for MP than NT (Table 3), the greater aeration of the soil matrix and greater disruption of soil aggregates coupled with mechanical breakdown of crop residues under MP are known to accelerate residue decomposition and mineralization of SOM, including previously protected SOM (Balesdent et al. 2000; Six et al. 2000). This illustrates that tillage and not N fertilization primarily controlled SOC and STN accumulation in our study. Nitrogen availability is necessary for microbes to decompose crop residues returned to the soil. Therefore, it is reasonable to assume that N fertilization would differently impact SOC and STN stocks under the two tillage systems due to variations in crop residue location and level of soil disturbance. In contrast to our findings, other studies on silt loam and clay loam soils and similar environmental conditions reported significant N fertilization effects and (or) interactions between tillage and N fertilization on SOC storage in the top 20 cm, with N fertilization increasing SOC under NT compared with MP or CT (Gregorich et al. 1996; Allmaras et al. 2004; Halvorson and Jantalia 2011).

The change in SOC stock from the establishment of the study to 2016 can be reasonably estimated based on our results and the experimental site information. At establishment in 1992, the experimental site contained 23 g SOC·kg⁻¹ in the 0–20 cm layer (Poirier et al. 2009). Based on our SOC stock calculations at equivalent mass, this was equivalent to 63.8 Mg C·ha⁻¹. Although soybean was not grown during the first three years, we can reasonably assume that the SOC stocks were more or less unchanged for the first three years, even considering the different N rates that were applied to corn. Within the 0–20 cm layer sampled in this study, we estimate that NT increased SOC stocks by 2.8%, but MP decreased it by 15.8%. These estimates support the adoption of NT as a best management practice to enhance SOC within the root zone of agricultural soils under humid temperate conditions (Poirier et al. 2009; Halpern et al. 2010).

Organic matter fractions

There was a significant effect of tillage, depth, and tillage × depth interaction on POMC, POMN, MAOMC, MAOMN, MBC, and MBN concentrations (Table 1). These organic fractions were more uniformly distributed within the 0–20 cm layer under MP, but were more concentrated within the 0–5 cm layer under NT, a clear demonstration of stratification, especially for MBC and MBN (Fig. 4). We found higher POMC, POMN, and MAOMC concentrations under NT than MP at the 0–5 and 5–10 cm depths, while MAOMN concentrations were higher under NT and MP at all depths (Figs. 4a–4d). The MBC and MBN followed a similar trend as POM and MAOM fractions, in that they were higher within the top 5 cm under NT than MP, with MBN also being higher under NT than MP at the 5–10 cm depth (Figs. 4e, 4f). Shi et al. (2012) examined the impact of tillage and P rates at the same experimental site and reported higher MBN concentration at the 0–15 cm depth under NT than MP, while MBC was similar between tillage systems at both the 0–15 and 15–30 cm depths. Furthermore, Shi et al. (2012) found higher MBC and MBN concentrations at the 0–15 cm than 15–30 cm depth under NT but not under MP. Nonetheless, higher MBC and MBN may not always indicate increased microbial activity (Gregorich et al. 2006b; Joergensen and Wichern 2018). The soil microbes under NT, at least within the top 10 cm, seemed to have a better environment for nutrient cycling through the presence of C and N sources and possibly adequate soil moisture and temperature (Nunes et al. 2020). The fact that N fertilizer rates had no impact on MBC and MBN in our study suggests that the soil microbial population was more influenced by tillage, which can affect soil moisture and aeration and crop residue distribution and size, than N availability. The higher amounts of crop residues left at the soil surface coupled with lower soil disturbance under NT could explain the increased concentrations of these SOM fractions as compared with MP (Franzluebbers and Arshad 1997; Zhang et al. 2016; Jilling et al. 2020; Li et al. 2021).

Fig. 4.

Interactive effect of tillage and soil depth on (a) particulate organic matter carbon (POMC), (b) mineral-associated organic matter carbon (MAOMC), (c) particulate organic matter nitrogen (POMN), (d) mineral-associated organic matter nitrogen (MAOMN), (e) microbial biomass carbon (MBC), and (f) microbial biomass nitrogen (MBN). Error bars represent standard error of the mean for the interaction.

Unlike in our study, differences in POMC and POMN were only detected in the 0–5 cm depth between NT and CT a long-term study in the Argentine Pampas (Martínez et al. 2020). We also found higher POMC, POMN, MAOMC, and MAOMN stocks under NT than MP (Table 2), suggesting that the higher SOC and STN stocks under NT than MP were reflected in POM and MAOM. This was not surprising since these labile SOM fractions are more sensitive to management practices than whole SOC and STN (Jilling et al. 2020). Higher accumulation of crop residues at or near the soil surface and reduced soil disturbance could partly account for the higher POM and MAOM concentrations under NT than MP. On the other hand, decomposition of crop residues, which could be accelerated by tillage (Balesdent et al. 2000), could also lead to mineralization of POM, thereby significantly reducing POMC and POMN under MP. Angers (1998) concluded that accumulation of POM in a silty-clay soil was enhanced by protection through macroaggregate formation. It has been proposed that mineralization of POM leads to the formation or replenishment of MAOM (Cotrufo et al. 2013; Lehmann and Kleber 2015). However, Haddix et al. (2020) found no consistent evidence that the formation of MAOM occurred from the decomposition of POM, suggesting that these SOM fractions may be formed by two separate pathways. In addition, Samson et al. (2020) concluded that POMC was closely related to the mean-weight diameter of water-stable aggregates and plant-derived carbohydrates, whereas MAOMC was closely related to microbial amino-sugars. Our results do not provide sufficient evidence to support or refute these findings.

The proportion of SOC as POMC and MAOMC was significantly influenced by tillage (p = 0.021 and 0.028), depth (p < 0.001 for both), and tillage × depth (p = 0.003 and 0.006), while the proportion of STN as POMN and MAOMN was significantly affected by depth (p < 0.001 for both) and tillage × depth (p = 0.006 for both; data not shown). Under MP, the proportions of SOC and STN as POM and MAOM were uniformly distributed within the top 20 cm layer (Fig. 5). Under NT, the proportion of SOC as POMC was 75% and 43% higher at the 0–5 and 5–10 cm layers than the 10–20 cm layer (Fig. 5a), while the proportion of STN as POMN was 77% and 53% higher at the 0–5 and 5–10 cm layers than 10–20 cm layer (Fig. 5b). Interestingly, the opposite trend was observed for the proportion of SOC as MAOMC and STN as MOAMN: under NT, the proportion of SOC as MAOMC was 11.6% and 6.3% higher at the 10–20 cm layer than at 0–5 and 5–10 cm layers, respectively (Fig. 5c), while the proportion of STN as MAOMN was 8.3% and 5.8% higher at the 10–20 cm layer than at 0–5 and 5–10 cm layers, respectively (Fig. 5d). The lower proportions as MAOM in the upper layers under NT were probably due to a higher concentration of recently returned plant materials present as large fragments (POM) at the soil surface. Further processing through microbial activity likely reduced their size and molecular weight or solubilized part of the decomposed materials (St. Luce et al. 2014), resulting in downward transfer through macropores under NT. The MAOM is regarded as a longer-term sink for C and N compared with POM, which has a faster turnover rate (Kögel-Knabner et al. 2008; Jilling et al. 2020). Within the top 10 cm soil layer, up to seven times more C and eight times for N were stored in the MAOM fraction than POM under MP, compared with four times more C and six times more N under NT. Other studies also found a greater proportion of SOC and STN in the MAOM fraction compared with POM (Franzluebbers and Stuedemann 2002; Denef et al. 2013; Jilling et al. 2020). However, it is important to note that soil texture may play a significant role in influencing SOC and STN distributions between POM and MAOM (Jilling et al. 2020), but in some cases land use and not soil texture may be the dominant factor (DeGryze et al. 2004). Our results point to a slightly higher level of MAOM formation under MP in the top 10 cm due to greater turnover rate of crop residues by mechanical breakdown and incorporation, creating increased contact between crop residues and soil mineral particles. The incorporation of soybean residues by MP may have further favored MAOM formation by enhancing the contact between labile crop residues with the soil and microorganisms (Cotrufo et al. 2013). However, it is likely that MAOM may be more stabilized under NT due to the lower soil disturbance.

Fig. 5.

Interactive effect of tillage and soil depth on (a, c) proportion of soil organic carbon (SOC) as particulate organic matter carbon (POMC) and mineral-associated organic matter carbon (MAOMC), and (b, d) proportion of soil total nitrogen (STN) as particulate organic matter nitrogen (POMN) and mineral-associated organic matter nitrogen (MAOMN). Error bars represent standard error of the mean for the interaction.

Soil respiration

Respiration was significantly influenced by tillage, depth, tillage × N rate, tillage × depth, and N rate × depth (Table 1). While respiration rate was similar among soil depths under MP, it decreased with depth under NT (Fig. 6a), especially in plots that received 80 and 160 kg N·ha⁻¹. Overall, respiration was higher under NT than MP at the 0–5 and 5–10 cm depths. The decrease in respiration with depth under NT was in agreement with previous findings (Nunes et al. 2020) and could be explained by reduced available C and microbial biomass at lower depths under NT (Fig. 4). Similarly, higher respiration under NT than MP in the top 10 cm of soil could be attributed to lower soil disturbance and greater accumulation of crop residues at the soil surface under NT (Nunes et al. 2018). There was a slight trend toward higher respiration at the 10–20 cm layer under MP compared with the upper layers. This was probably due to mechanical mixing under MP, leading to the deposition of crop residues and increased SOC at or near the bottom of the plow layer (Poirier et al. 2009). In contrast to our findings, Martínez et al. (2020) found higher respiration at the 5–10 cm depth under CT than NT in a long-term study, with no differences at the 0–5 and 10–20 cm depths, which was attributed to greater access to labile SOM, increased aeration and mineralization due to plowing. A meta-analysis by Zuber and Villamil (2016) found higher MBC, MBN, and enzyme activities under NT than tilled systems, but metabolic quotient, an indicator of microbial activity, was the reverse. However, in long-term field experiments, metabolic quotient was similar between systems, indicating that microbes under NT systems may eventually become as active as those in tilled systems since they are more closely driven by the labile C pools that have accumulated in the soil.

Fig. 6.

Soil respiration as influenced by (a) the interactive effect of tillage and soil depth, (b) the interactive effect of tillage and nitrogen fertilization, and (c) the interactive effect of nitrogen fertilization soil depth after 25 yr in a corn–soybean rotation. Error bars represent standard error of the mean for the interaction.

Under both MP and NT, soil respiration was similar across N rates, but there was a trend toward lower respiration in plots fertilized at 80 kg N·ha⁻¹ under NT (Fig. 6b). In addition, within N rates, respiration was higher under NT than MP in plots fertilized at 0 kg N·ha⁻¹ and 160 kg N·ha⁻¹, respectively (Fig. 6b), and was lowest at the 10–20 cm depth for the 80 and 160 kg N·ha⁻¹ rates (Fig. 6c). While respiration was similar across N rates in the 0–5 and the 5–10 cm depths, respectively, it was lower for 80 kg N·ha⁻¹ than 0 kg N·ha⁻¹ at the 10–20 cm depth (Fig. 6c). Although N rate, as a main factor, had no influence on respiration in the present study, others reported an increase (Sun et al. 2018; Wang et al. 2019; Bean et al. 2020), decrease (Ramirez et al. 2012; Bean et al. 2020), or no change (Rochette and Gregorich 1998; He et al. 2019) due to N addition. This highlights the greater dependence of respiration on resource accessibility and (or) availability including C (He et al. 2019; Wang et al. 2019; Nunes et al. 2020) and soil moisture and temperature (Gregorich et al. 2006b) than N addition.

Relationships between soil organic carbon, soil total nitrogen, and SOM C and N fractions

There were strong positive relationships (r > 0.85) between SOC, STN, the labile SOM fractions, and respiration for the full dataset (Fig. 7a). Interestingly, the correlation matrix for NT (Fig. 7b) somewhat resembled that of the full dataset with mostly strong positive relationships, whereas these two correlation matrices drastically differed from that of MP (Fig. 7c). Moreover, although not significant in all cases, there were negative relationships under MP (Fig. 7c). For example, POMC and POMN were negatively (p < 0.01) related to MAOMC and MAOMN under MP, respectively, a possible indication of turnover of POM, leading to MAOM formation (Cotrufo et al. 2013). Our results are in agreement with other studies that demonstrated the positive impact of NT on labile SOM fractions and soil biological activity (Nunes et al. 2020) and strong positive correlations among these parameters (Nunes et al. 2018). The strong positive relationships under NT, considering only the top 20 cm of soil, indicate that NT provides a localized or concentrated area for plant litter inputs and an ideal condition for accumulating SOM and improving soil fertility (greater microbial biomass and respiration) as compared with MP.

Fig. 7.

Heat map of Pearson correlations between the labile soil organic matter fractions and soil organic carbon and soil total nitrogen for (a) All (b) under no-till (NT) and (c) moldboard plow (MP). SOC, soil organic carbon; STN, soil total nitrogen; POMC, particulate organic matter carbon; POMN, particulate organic matter nitrogen; MAOMC, mineral-associated organic matter carbon; MAOMN, mineral-associated organic matter nitrogen; MBC, microbial biomass carbon; MBN, microbial biomass nitrogen; *, **, ***, significant at 0.05, 0.01, and 0.001 probability levels, respectively. [Colour online.]

The SOC and STN were more strongly related to each other for the entire dataset (r = 0.99) and under NT (r = 0.99) than MP (r = 0.89). Under MP, there was no significant positive relationship between soil respiration and any of the parameters; however, respiration was negatively related to POMN (r = −0.70; p = 0.035). This lack of significant positive relationship between respiration with other parameters under MP could be due to the recurrent disruption of microbial habitats by tillage and placement of C and N sources at depth where soil temperature and oxygen availability are reduced (MacDonald et al. 2010; Wang et al. 2019; Nunes et al. 2020). In eastern Canada, MacDonald et al. (2010) found significantly reduced CO₂ emissions under MP in managed grasslands due to reduced mineralization of plant residues at depth under the cold and humid climate in the region. In contrast, respiration was positively related to all parameters except POMN under NT, suggesting a greater availability of substrates for microbes and an overall more conducive habitat under NT (Nunes et al. 2020). Together, our findings further support the use of these SOM fractions as indicators of management-induced changes and highlight the profound impact of tillage practices on soil processes, with implications for soil health and productivity.