The field site was located on a commercial poultry-cash crop farm in Bayfield, Ontario, Canada (43°34’45.8”N, 81°39’52.2”W). The area was located 183 m above sea level with a 1.5% slope. Weather data (30-year mean) was obtained from a nearby weather station located in Dashwood (43°20’48.8”N, 81°38’07.6”W), Ontario with a mean annual temperature of 8.3°C and an average annual precipitation of 858 mm ( Figure 1 ). The soil was classified as a uniform calcareous Grey-Brown Luvisol with a sandy loam texture. In the years before biochar addition (prior to May 2016), commercial farming practices of the study site included an annual rotation of maize (Zea mays L.) and soybean (Glycine max Merr. L.). Poultry manure, based on switchgrass bedding (Panicum virgatum L.), was added in April at 600 g m^-2 and topped off with nitrogen fertilizer at 135 kg N ha^-1 in the years when maize was produced ( Table 1 ). The soil was cultivated in April of each field season using a disc harrow (0-20 cm) and weeds were controlled with N-phosphonomethyl glycine (Glyphosate).

The field experiment was established in May 2016 and included 600 g m^-2 poultry manure + 13.5 g N m^-2 fertilizer (MN); 300 g m^-2 poultry manure + 300 g m^-2 biochar (MB); and 300 g m^-2 poultry manure, 13.5 g N m^-2 fertilizer + 300 g m^-2 biochar (MNB). The experimental design was a randomized design, and each treatment was replicated three times. The plot size for each treatment replicate was 10 m x 10 m, with a 3 m border between plots. Biochar was added using a drop spreader and incorporated into the soil using a Salford RTS vertical tillage unit to ensure uniform distribution of the biochar. Commercial farm management operations, including herbicide additions and fertilizer application rates, used standard agronomic practices for this region of southwestern Ontario, Canada. The field study was conducted over three growing seasons, beginning in May 2016 with a maize crop and biochar addition, a soybean crop in 2017 and a maize crop in 2018. Biochar was added to the respective treatment replicates only in 2016, whereas manure and nitrogen fertilizer were added biennially when maize was produced. No soil amendments were added during the years soybeans were produced. The biochar was provided by Titan Carbon Smart Technologies (Saskatoon, Saskatchewan, Canada). The feedstock of the biochar was a 1:1 mix of pine (Pinus spp.) and spruce (Picea spp.) ( Table 1 ). Biochar was produced using slow pyrolysis (550°C, 15 min).

Prior to the initiation of the study, five soil samples (0-20 cm) were collected in April 2016 from each treatment replicate to determine baseline values. Further soil samples were collected from each treatment replicate after crop harvest in October 2016, 2017, and 2018. Fresh soil was used to quantify pH using a 1:1 soil and ultrapure water mixture using a BioKit AB15B (Houston, TX, USA) pH meter (24). The remaining soil was air dried and sieved to 2 mm. Carbonates were removed by washing 2 g of soil with 50 mL of 0.5 M HCl (25). The soil-acid solution was shaken three times over 24 h on a reciprocating shaker at 200 rpm (Heidolpj Unimax 1010 DT, Schwabach, Germany). Following a settling period of 30 min, the acid was removed using a pipette and discarded (25). The acid treated soil was washed by adding 50 mL ultrapure water and shaking the soil at 200 rpm for 15 minutes, after which the water was removed with a pipette and discarded (25). The washing procedure was repeated daily for four days after which the soil was dried at 30°C for two days, ground in a ball mill (Retsch^® ZM1, Haan, Germany) and analyzed for soil organic carbon and total nitrogen using a Costech 4010 (Valencia, USA) elemental analyzer.

Century is a site-specific model based on the interactions between climate, crops, and soil. Parameterization of the model can be achieved by modifying climate, soil chemical, physical and biological characteristics, and agroecosystem management practices. Plant production is a sub-model in century that stimulates aboveground and belowground biomass productivity of forests, grasslands, savannas, or crops (26). The soil organic matter sub-model predicts changes in soil active, slow, and passive carbon fractions based on microbial decomposition of crop residues and resultant microbial products that are the foundation in the formation of humus (20). Of the total soil organic matter pool, approximately 2% to 4% is composed of the active fraction that includes soil microbes and microbial products with a turnover time of several months to a few years (27). The slow fraction composes 45% to 65% of the total soil organic matter pool with a turnover time of 20 to 50 years and includes resistant plant material acquired from stabilized soil microbial products and structural plant material (27). The passive fraction contains 30% to 40% of the total soil organic matter pool and is represented by chemically and physically protected mineral associated organic material (20). Soil texture controls the extent to which the decomposed compounds are stabilized (26). The Century model was originally developed to determine changes in soil organic carbon stocks for grassland ecosystems (26), but the model can also simulate changes in temperate and tropical environments that include various land management systems ranging from row cropping to complex agroecosystems (28, 29). To date, Century is one of the most widely used models because of its ability to simulate changes in soil organic carbon in diverse environments (20).

We used version 4.0 of the Century model to simulate changes in soil organic carbon and its associated fractions (0-20 cm) after a low dosage and one-time addition of biochar. The model was initiated for >10,000 years under a mixed deciduous-coniferous forest which is native to this region of southwestern Ontario, Canada to estimate equilibrium in plant productivity and soil organic carbon levels i.e., when soil organic carbon stabilizes and there are no further changes with time. After equilibrium was attained the first year of simulation began in 1882 with grass/hay for livestock production ( Table 2 ). Monthly average maximum and minimum temperature and monthly total precipitation (30-year average) were obtained from a nearby Environment and Climate Change Canada meteorological station managed in Dashwood, Ontario ( Figure 1 ). After initializing soil organic carbon levels, land management events were scheduled according to actual historical events prior to the addition of biochar in 2016 ( Table 2 ). Soil properties prior to the addition of biochar ( Table 3 ) were obtained from Jiang (30). The proportion of soil organic carbon initial values for the active fraction was 3%, 65% for the slow fraction and 32% for the passive fraction (31). The OMAD.100 file was changed and reflected the addition of 184 g C m^-2 manure for MN, or 91 g C m^-2 manure + 240 g C m^-2 biochar for MB and MNB ( Table 4 ). The C/N ratio of the manure and biochar were also adjusted accordingly ( Table 4 ). An additional file in FERT.100 was created to reflect the addition of 13.5 g m^-2 nitrogen fertilizer to MN and MNB. Crop parameters for maize and soybeans were modified to accommodate regional crop production values ( Table 4 ). Default values set by Century were used for atmospheric nitrogen deposition and nitrogen fixation rates. All other parameters used in this simulation were provided by Century and modified parameters are listed in Table 4 . We used a monthly time-step for simulation output to align predicted values with their corresponding month of soil sampling in the field.

Measured field data were compared with predicted data using the regression function in SPSS version 28 (33). The following metrics were used to determine the strength of the relationship between measured and predicted values: correlation coefficient (r), coefficient of determination (r²), adjusted r², root mean squared deviation (RMSD), root mean squared error (RMSE), the mean difference (M) between measured and predicted values of soil organic carbon, the coefficient of residual mass (CRM) to determine if the model over- or under estimated values of soil organic carbon, and modeling efficiency (EF) which compares predicted values to the average measured values (34, 35). The threshold probability level for determining goodness of fit was P < 0.05.

We found that land conversion from mixed deciduous-coniferous forest to grass and hay production for livestock forage, and subsequent conversion to row crop production led to a rapid decline in soil organic carbon stocks and associated carbon fractions. Such a decline in soil organic carbon stocks upon land conversion to agriculture have been observed on a global scale (36), and losses will likely be heightened in the future due to a changing climate (37) since warming enhances microbial activity and the transformation of soil organic carbon (1). Additionally, agriculture also causes a fundamental disruption in the soil’s steady state due to a drastic reduction in the quantity and quality of organic matter returned to the soil that leads to a depletion of soil organic carbon stocks (39, 40). For example, crop residue removal has increased from 0.5 Pg C y^-1 to 1.6 Pg C y^-1 between 1910 and 2005 due to heightened demands for food, fibre, and fuel (39, 41). Furthermore, Li et al. (42) determined that a 25% reduction in organic matter input caused a 1.6% decrease in soil organic carbon. In fact, we found an 51% decrease in soil organic carbon over 113 years, representing the time from initiation of agriculture in 1881 to 2015. Our results corroborate with those of Powlson et al. (43) who reviewed data from multiple long-term experiments in temperate and tropical agroecosystems. They determined that soil organic carbon stocks declined between 38% and 67% compared to pre-cultivation levels (43). Although our study showed a considerable increase (84%) in soil organic carbon stocks over 115 years once manure and (or) biochar were added, we found that pre-cultivation soil organic carbon stocks were not reached. Similarly, Powlson et al. (43) found that increases in soil organic carbon levels ranged from 60% to 70% of pre-cultivation values once conservation practices were initiated. But they concluded that soil organic carbon stocks will likely never approach pre-cultivation levels (43). This is because the rate of soil organic carbon loss and accumulation is influenced by land management practices including the quantity and quality of residue returned to the soil and the type and frequency of tillage (39, 44). Furthermore, there is an upper limit of carbon storage that depends on the soil’s inherent soil organic carbon level which is closely tied to soil texture (45). However, we also found that after 115 years of manure and (or) biochar addition, soil organic carbon equilibrium was not reached. This was likely due to an inherently low soil organic carbon level in the coarse textured soil (60% sand, 22% silt, 18% clay) used in our study. This suggested that soil amended with manure and (or) biochar has a greater carbon sequestration potential if the soil texture is coarse, since silt and clay content control the upper level (saturation) of soil carbon sequestration (46).

Although carbon input in MB and MNB (331 g C m^-2) was almost double compared to that in MN (184 g C m^-2), soil organic carbon accumulation was greater in MN after 115 years. This was due to the biennial addition of manure (i.e., in the years when maize was produced) since Century was scheduled to mirror commercial farming practices that took place at the field site. Conversely, Century was scheduled, paralleling field site management practices, only by a one-time addition of biochar to MB and MNB. Therefore, over the 115-year simulation carbon input from manure was cumulative and contributed to the greater soil organic carbon stock in MN. We therefore recommend that a more frequent low dosage of biochar should be applied in future field studies that can further increase soil organic carbon stocks. For example, Dil and Oelbermann (22) used Century and determined that an annual addition of biochar at 200 g C m^-2 increased soil organic carbon stock by 10,522 g m^-2 over 150 years in a light textured soil. However, the study by Dil and Oelbermann (22) did not integrate field data into their simulation. Although there is a wide variation in the ability of manure to increase soil organic carbon stocks over the long-term (>5 years), a recent meta-analysis by Gross and Glaser (47) determined that manure increased soil organic carbon stocks, regardless of environmental conditions. For example, they found a 40% increase in soil organic carbon after 20 years in temperate agroecosystems (47). However, our results were comparable to those by Jenkinson and Rayner (48) who predicted changes in soil organic carbon over a 120-years using field data from long-term research sites in England. Furthermore, nitrogen availability becomes unrestricted when a legume is integrated into the rotation (49). The greater nitrogen availability from the legume decreases the mineralization of old carbon and leads to the accumulation of soil organic carbon (50). This could also explain the difference in soil organic carbon stocks in our study (e.g., MN) compared to Gross and Glaser (47), who did not include legumes in their meta-analysis.

Compared to the pre-2016 land management practice, amendments with biochar also increased soil organic carbon stocks, over the 115-year simulation, by ≥70%. This is because biochar, due to incomplete combustion, is the only plant-derived organic matter that as a higher intrinsic stability than other sources of organic matter (51). Although biochar is stable over the long-term, it is also affected by microbial degradation depending on the pyrolysis conditions under which it was produced (52). For example, a high pyrolysis temperature, carbon to nitrogen ratio and a pH >8.1 caused an increase in soil organic carbon (52). However, biochar produced with a low pyrolysis temperature and narrow carbon to nitrogen ratio, like that used in our study, is more readily available to the microbial community (52). This could also explain the lower soil organic carbon stock predicted by Century in our study in MB and MNB compared to MN. Furthermore, the soil used in our study was sandy, which does not provide a physical and chemical protection mechanism for the biochar, allowing the readily decomposable fractions to be more accessible to the microbial community (52).

The greater accumulation of soil organic carbon in MNB, compared to MB, over the 115-year simulation was likely due to the presence of nitrogen fertilizer combined with manure and biochar. Previous studies showed that an effective way to promote efficient soil organic matter formation includes the input of a diverse source of organic materials (1). This is because diverse sources of organic matter (i.e., molecular diversity) causes an energy limitation on microbes, decreasing soil organic matter mineralization (53). Additionally, van Groenigen et al. (54) found that an increasing input of nitrogen leads to an accumulation of soil organic carbon which, in our study, likely contributed to the greater soil organic carbon stock in MNB. In a 15-week mesocosm study, Ibrahim et al. (55) determined that biochar blended with fertilizer regulated carbon mineralization because more nitrogen substrates were available to the microbial community that depended on nitrogen. Although blending biochar with manure and a source of mineralizable nitrogen is contradictory with respect to potentially greater greenhouse gas emissions, results from a previous study at the same field site as our study showed no significant difference in CO₂ and N₂O emissions among amendment types (16). We also determined that model performance for MNB was poor, compared to MB, and it did not show a significant relationship between measured and predicted values. We recommend that more reliable simulation outcomes should be based on a greater number of data points (e.g., number of years) from field measurements to provide a more robust comparison with predicted values.

Soil organic carbon fractions decreased upon initiation of agricultural activities in 1882, with little change in the passive fraction. However, all three fractions increased after 2016, once manure and (or) biochar were added, where the greatest rise was observed in the active and slow fractions. This suggested that the active and slow carbon fractions were controlled by land-use change (i.e., change from the pre-cultivation forest ecosystem to cultivated agroecosystems) and different agroecosystem management practices (i.e., change from grass/hay production to cereal and maize crop rotations). Land conversion and changes in land-use management decreased primary production, soil nutrient status and the quantity and quality of organic matter returned to the soil compared to pre-cultivation levels (39). This likely led to the mineralization of old carbon sources, causing a decline in the various carbon fractions (56). However, a sufficient input of organic matter, and therefore an increase in each of the carbon fractions, occurred once manure and (or) biochar were applied to the soil. For example, Gross and Glaser (47) determined that soil with a low initial soil organic carbon stock will have a large response to the addition of organic matter and show no depletion with time. Also, the addition of organic matter increased nitrogen availability, which affected microbial growth and the mineralization of new carbon derived from the amendments (57). This suggested that carbon stabilization is controlled by the quantity and quality of the organic residues entering the soil ecosystem.

The greater accumulation of carbon in the active and slow fractions in MN, compared to MB and MNB, was due to the biennial addition of manure (47). However, the one-time addition of biochar, when blended with manure and nitrogen fertilizer (MNB), led to a similar accumulation of carbon in the active and slow fractions as MN. This suggested that biochar increased soil organic carbon stabilization within the organo-mineral fraction (60). Additionally, our results showed that the active and slow carbon fractions did not reach equilibrium during the 115-year simulation period. Gross and Glaser (47) suggested that soil carbon saturation not only depends on the properties of the soil and of the organic matter added to the soil, but also on the inherent soil organic carbon content. They found that soil with an inherently low soil organic carbon content requires a longer time to reach equilibrium compared to soil with an inherently high soil organic carbon content (47).

Input from the various carbon sources such as manure or manure blended with biochar likely stimulated the microbial community to become more active, which may have preserved the active carbon pool despite disturbance from tillage. For example, Ouyang et al. (61) also found that the addition of manure counteracted the impact of tillage on the active soil organic carbon fraction in northeastern China. Lorenz and Lal (52) also determined that when biochar was added to a coarse-textured soil, it caused an increase in soil-specific surface area that enhanced soil nutrient cycling and microbial activity. Furthermore, our modeling results determined that carbon associated with the slow fraction was the key driver in soil organic carbon accumulation once manure and (or) biochar were added to the soil. We also found that the slow fraction resisted the impact of disturbance from cultivation with a disc harrow (0-20 cm depth) and soil organic carbon was maintained by input of manure and (or) biochar. This is because the slow fraction is physically protected and has a more chemically complex structure, which makes it more resistant to decomposition than the active fraction (20). Additionally, Carvalho-Leite et al. (62) determined that carbon accumulation in the slow fraction is related to the quantity carbon and nutrients added to the soil. Although Cong et al. (63) found that nitrogen fertilizer decreased carbon in the passive fraction, our simulation results showed a 29% increase in the passive fraction in MNB, compared to a 18% in crease in MN and MB. Pulcher et al. (21) suggested that biochar carbon may not be directly mineralized to CO₂ but may transfer to carbon pools with different degradation rates. This is because the biochar used in our study has a hydrogen-to-organic carbon ratio lower than 0.4, which is considered highly stable with carbon remaining in soil for more than 100 years (64). Accordingly, results from our study predicted that a one-time addition of manure blended with biochar and nitrogen fertilizer (MNB) can stabilize a greater proportion of the carbon in the passive fraction than when blending manure with nitrogen fertilizer (MN) or when bending manure with biochar (MB).

Differences between measured and predicted values in our study ranged from +9% to -1.3% and was within range of other studies. For example, Alves Primo et al. (29) determined that Century over- and underestimated soil organic carbon stocks up to 15% in semiarid slash and burn and agroforestry systems in Brazil. Similarly, in temperate agroecosystems soil organic carbon stocks deviated between +10% to -15% between measured and predicted values (38, 65). However, our study was limited by the availability of only three years of field data for model validation. Discrepancies between measured and predicted values in our study were also due to Century’s failure to accommodate changes in bulk density with increasing soil organic carbon stocks. Instead, Century assumes bulk density as a constant parameter throughout the simulation (58). With increasing soil organic carbon stocks, bulk density decreases with time, influencing the quantity of carbon sequestered (47). For example, Lorenz and Lal (52) noted that biochar could reduce soil bulk density by 3 to 31%. Various environmental factors including fluctuating moisture and temperature conditions, freeze-thaw events and microbial activity control the dissolution, oxidation and fragmentation of biochar that can change its physicochemical characteristics (59). For example, Wang et al. (32) found that soil minerals accumulated on the surface of the aged biochar and suggested that this improved soil physical stability which could lead to enhanced carbon sequestration. Therefore, the soil organic carbon stock in amendments with biochar may be an underestimation in our study since Century does not account for biochar ageing. However, it is also crucial to consider how soil physicochemical properties and soil amendments (e.g., manure, biochar) are affected by future climate change scenarios which could affect long-term soil organic carbon stocks. For example, Jiang et al. (17) determined that short-term climate effects of elevated CO₂, warming or the combined effect of elevated CO₂ and warming did not influence biochar. But they found that climate effects influenced how carbon was accessed by the microbial community in amendments with manure and biochar (17). Furthermore, leaching and bioturbation cause the downward translocation and stabilization of biochar within the soil profile (52). However, translocation of carbon (i.e., biochar) to deeper soil horizons is not accounted for in Century since the model only predicts soil organic carbon stocks to a 20 cm depth.