Introduction

Accurate soil C estimates are critical to developing effective environmental policies and C trading initiatives. Dryland areas of the Inland Pacific Northwest (iPNW) offer significant opportunities for C sequestration and substantial economic potential for offsetting or reducing carbon dioxide (CO₂) emissions through sustainable agricultural practices1,2. However, reliable estimates are needed to support incentive programs and promote C sequestration efforts.

Winter wheat–summer fallow (WW–SF) dryland cropping systems have lost soil C stock due to extended fallow periods, frequent tillage, and the removal of crop residue3. These practices contribute to the rapid depletion of soil organic matter (SOM) and essential nutrients, compromising soil health and reducing crop yields4. These dryland systems are highly valuable given that around 17% of the nation’s wheat is produced in the iPNW5,6. Although the amount of organic C that can be stored in soils is finite, more than a century of wheat farming in this region has resulted in large C losses, which means that there is great potential for increasing C sequestration through improved management practices7. Opportunities to enhance soil organic C in the iPNW include eliminating fallow, reducing tillage intensity, and diversifying and intensifying crop rotations by incorporating legumes into annual cropping systems8,9,10. Improved management practices such as no-till (NT) and legume integration enhance microbial biomass C (MBC) and enzyme activity, contributing to substrate use efficiency and C storage11,12. Legume-based cropping systems also yield greater biomass production and plant residue, further aiding SOM retention13.

While C accumulation rates are often slow to respond to management changes in cold climates with low precipitation and limited plant productivity, some soils can show measurable improvements under the right conditions14. While dryland agricultural systems in the iPNW have shown a relatively quick response in soil C accrual after adopting conservation tillage, this behavior is not always observed15,16. The variability in reported C sequestration rates leads to inconsistencies, with some estimates potentially exacerbated, affecting assessment accuracy7. Furthermore, IPCC-based and modeling approaches often rely on generalized assumptions, failing to account for specific land use and management practices, which limits their accuracy17. These accumulation rate estimates are critical in correctly interpreting experimental and modeling data relevant to soil organic C in the iPNW.

The dynamics of SOM in this region is mainly influenced by aggregation and mineral interaction18,19. However, the extent to which agricultural conservation practices contribute to C stabilization in more persistent SOM pools, particularly mineral-associated organic matter (MAOM), remains uncertain. The distribution of C and nitrogen (N) across various labile and persistent SOM pools significantly affects their turnover rates and permanence, thereby influencing long-term storage and nutrient cycling20,21. Although no C sink is truly permanent, the MAOM fraction acts as a long-term C storage pool, helping stabilize SOM and slow the release of CO₂ into the atmosphere22. It is essential to conduct a comprehensive, site-specific assessment to accurately estimate soil C sequestration rates and MAOM formation, both of which are crucial for developing effective earth systems models and carbon trading markets.

Since the MAOM fraction is an emerging indicator, historical data on its formation rates is often limited or unavailable23. Additionally, estimating bulk soil accumulation rates remain challenging due to methodological inconsistencies over time and, in some cases, a lack of baseline data. Long-term experiments and archived soils offer invaluable opportunities to improve these estimates, refine agricultural strategies, and progress in this field. Our specific objectives are to: (1) use archived soils to document C sequestration rates in bulk soil and the MAOM fraction over time, (2) assess the potential for C stabilization and net CO2-equivalent (CO2e) sequestration resulting from cereal-legume incorporation into WW–SF cropping systems, and (3) evaluate the opportunities for local wheat growers in Eastern Oregon to participate in C offset programs for Walla Walla silt loam soils.

Materials and methods

Study areaThe study was carried out with data from several long-term experiments (LTEs) at the Columbia Basin Agricultural Research Center (CBARC) near Pendleton, OR (45º42’N, 118º36’W, 438 m elevation), some of which date back to 1931 (Fig. 1). The soil texture at a depth of 0–30 cm consists of 18% clay, 22% sand, and 60% silt, and is classified as Walla Walla silt loam (coarse-silty, mixed, superactive, mesic Typic Haploxeroll)24. The site has a Mediterranean climate characterized by average annual maximum and minimum temperatures of 17.4 and 3.06 °C, respectively, and annual precipitation of 421 mm. About 70% of the precipitation is received between September and March.

Fig. 1
figure 1

Location map of the study area. To estimate carbon (C) sequestration, a subbasin covering a total area of 1,414.7 km², including 14,758 ha of wheat lands, was defined. Landsat-8 Operational Land Imager (OLI) bands 4, 5, and 6 were used for the calculation of NDVI through random forest (RF) supervised classification processes to obtain detailed information on wheat-based cropping systems. The map was created in ArcGIS Pro using the SSURGO dataset as the primary source for identifying the dominant soil types.

Full size image

Trial establishment and management. This study was conducted based on the ongoing winter wheat and spring pea (WW–SP) long-term experiment involving a 2-year rotation of wheat (Triticum aestivum L.) and spring pea (Pisum sativum L.), with both phases of the rotation present each year to facilitate yearly data collection. The experimental design is a randomized complete block with four replicates, each subplot measuring 7.3 m by 36.5 m. Winter wheat is planted in early October (18 cm row spacing) and harvested in late July, while spring peas are sown in late March or early April and harvested in June or July. For the past 20 years, wheat plots received 90 kg N ha − 1. This study investigated two tillage treatments within the WW–SP cropping system: (1) conventional tillage (CT), involving fall moldboard plowing since 1963, and (2) no-till (NT), which began as minimum tillage in 1963 and transitioned to NT in 1995. The CT treatment was chosen because previous studies have shown that fall plowing in the WW–SP system resulted in the most significant reduction of soil C levels25. The CT treatment included soil inversion, residue incorporation, and spring cultivation before pea planting, followed by plowing in pea vines and soil cultivation before planting wheat. The NT treatment was chosen because it was shown to partially restore native levels of soil C in the WW–SP3. Further details are available in the study by Machado et al. (2008). We compared soils from the WW–SP experiment to those from a wheat–summer fallow (WW–SF) system treated with 90 kg N ha⁻¹, as part of the Crop Residue (CR) LTE established in 1931 The CR treatment included fall moldboard plowing and spring cultivation with a cultivator and harrow, incorporating crop residues at a 20-cm depth during primary tillage in fallow years26.

Soil Sample Collection and Preparation from Archived Inventories. In this study, we define “present-day samples” and “archived samples.” Present-day samples were collected in 2015–2016 for the WW–SF system and in 2021 for the WW–SP system. The differences in sampling years arise from the fact that the most recent soil samples were typically collected every five years, whereas, prior to 2000, samples were taken every ten years. Additionally, not all LTEs are sampled in the same years due to financial constraints. Archived WW–SF samples include data from 1976 to 1977, 2005–2006, and 2010–2011, while archived WW–SP samples are from 1985, 1995, 2005, and 2010 as indicated in Table 1. Sampling depth intervals also varied over time, with early CR samples taken at 15 cm intervals and WW–SP samples at 0–8 cm, 8–15 cm, and 15–30 cm. Over the last 30 years, both experiments have used 10 cm intervals. Initially, hand probes were used, but since 1986, a 3.6 cm diameter truck-mounted Giddings Hydraulic Probe (Giddings Machine Company, Inc., Windsor, CO) was employed.

Table 1 Soil collection methods and sampling years used in the treatments of this study, including winter wheat–summer fallow (WW–SF) and winter wheat–spring pea (WW–SP) systems. *Averaged C and N values (from consecutive years).
Full size table

Laboratory analysis. Total C and N content were analyzed using the dry combustion method at 950 °C with a LECO CN628 analyzer (LECO Corp., St. Joseph, Michigan, USA). The experimental site has no inorganic C in the 0–30 cm upper soil layer. Consequently, C measurements were reliably assumed to be of soil organic C. Soil pH was measured in air-dried samples with a glass electrode in deionized water (1:2.5, w/v) with 0.01 M CaCl29. 

A size fractionation method was employed to analyze particulate organic matter (POM) and MAOM fractions in the present-day samples27,28. Initially, 5.5 g of air-dried and 2 mm sieved soil underwent agitation in a 30 mL solution containing 0.5% sodium hexametaphosphate for 18 h. The resulting suspensions were carefully poured over a 2 mm sieve nested on top of a 53 μm sieve into pre-weighed aluminum loaf pans. The POM that remained on the 53 μm sieve was carefully rinsed with deionized water, and the material passing through the sieve was considered MAOM. The collected fractions were then oven-dried at 60 ◦C and subsequently weighed. The dried POM and MAOM fractions were then finely ground with a ball mill for elemental analysis. The SOM fractions were sent to the Oregon State University’s soil health lab for total C and N analyses using an Elementar analyzer (Vario MAX cube, Elementar, Germany).

Soil bulk density (BD) was used to convert C concentration (g kg−1) to C stock per area (t C ha−1). BD was estimated using a Giddings hydraulic probe with an internal diameter of 3.6 cm. Cores were collected near the locations where soil samples were taken. Each soil core was then divided into 10 cm increments, which were oven-dried at 105 °C and weighed29. The dry soil mass and sample volume (cm3)were recorded after drying. For years with missing information, pedotransfer functions (PTFs) were used to estimate BD in Mollisols within the suborder of Xerolls based on available input variables30.

$$BD = 1.713-0.193\: \times \:{OC}^{\frac{1}{2}}-0.015 \times \:WC$$
(1)

 

where \(\:OC\) is the soil organic C in percent (%) and \(\:WC\) is the gravimetric water content at −1,500 Kpa (w/w).

FTIR spectroscopy analysis. In this study, mid-infrared (MIR) spectra were obtained from air-dried bulk soil (35–37 °C), which was passed through a 2-mm sieve and ground to < 180 μm using a planetary ball mill (Pulverisette 7 Premium Line, Fritsch). The samples were scanned in the MIR spectral region (4,000–700 cm−1) using a Spectrum 3 Tri-Range FT-IR Spectrometer (PerkinElmer, USA). Each spectrum comprised 64 co-added scans at a resolution of 4 cm−1. Scans were conducted in duplicate for each sample, followed by standardization using Standard Normal Variate (SNV) and averaging.

Calibration models.Since various methods have been used for C and N analysis of LTE samples, ranging from the Walkley–Black method (pre-1976) to dry combustion analysis with different instruments (1976–present), a calibration approach and the reanalysis of archived samples using this approach were necessary to address methodological inconsistencies and improve the reliability of long-term soil data. Calibration models based on soil spectra were developed using a stratified sampling strategy, incorporating 160 samples. This approach included topsoil (0–10, 10–20, 20–30 cm) and subsoil at 30–40, 40–60, and 60–80 cm depths and combining data from the WW–SP and CR experiments. The calibration also included samples from the topsoil taken every and 2 and 5 cm. Machine learning analyses were implemented using the Caret package in the R software environment31.

Soil C stocks and accumulation rates. The total C and N stocks across locations at a 30 cm depth was calculated as a summation of total C and N stocks in the 0–10, 10–20, and 20–30 cm soil depth intervals.

$$\:{C}_{stock}=\frac{OC*BD*D}{10}$$
(2)

where \(\:{C}_{stock}\) is soil organic C stock (t C ha−1), \(\:OC\) is soil organic C concentration (g kg−1); BD is soil bulk density (g cm−3) and \(\:D\) soil thickness (cm).

The annual rate of change in carbon (\(\:\varDelta\:C)\) were calculated as the difference between C stock in the later time point (\(\:{C\:stock}_{{t}_{2})}\) and at the earlier time point \(\:{(C\:stock}_{{t}_{1}})\) divided by the number of years between the first \(\:{Y}_{{t}_{1}}\) and the last survey \(\:{Y}_{{t}_{2}}\), assuming a constant rate of C change over time. Increases in C sequestration rates under specific management practices were evaluated relative to a baseline condition, with positive values indicating a measurable improvement.

$$\:\varDelta\:C=\frac{{C\:stock}_{{t}_{2}}{-\:C\:stock}_{{t}_{1}}}{{Y}_{{t}_{2}}{-\:Y}_{{t}_{1}}}$$
(3)

As mentioned previously, the soil depth intervals at which carbon measurements were reported varied over the years. To better compare years and cropping systems, we standardized the observed C measurements to common depth intervals of 0–10, 10–20, and 20–30 cm. We followed the normalization method for soil organic C (\(\:OC\)) depth distribution as described by32,33.

$$\:{OC}_{d2}=\frac{1-{\beta\:}^{d2}}{1-{\beta\:}^{{d}_{0}}}{OC}_{{d}_{1}}$$
(4)

where \(\:{OC}_{d2}\) is the value of soil organic C (t C/ha) at a new depth (cm) interval \(\:{(d}_{2}\)) of the soil profile. The constant β is 0.978634, which is the rate of decrease in the C pool relative to soil depth. \(\:{d}_{1}\)is the original soil depth (cm) reported in the study.

CO2 -equivalents for trading on the CO2 market. Using a conversion factor of 3.67, corresponding to the atomic weight ratio of CO2 to C (44/12), the mean C accumulation rate was then converted to CO2 equivalents (\(\:{CO}_{2}e)\). This factor indicates that an increase of 1 t ha−1 in soil C corresponds to the removal of 3.67 tons of CO2from the atmosphere35. Given that 1 ton of soil C = 3.67 tons of CO2 (sequestered or emitted), the equivalent CO2 sink was estimated as follows:

$$\:{CO}_{2}e=3.67*\varDelta\:C$$
(5)

To estimate the C credit revenue resulting from transitioning from a conventionally managed wheat system to NT farming, we multiplied the emissions by the current market value. The currency value for the C offset was estimated based on the rate used by the State and Trends of C Pricing36. The annual offset credit payments or revenue (USD/year) for agricultural soil C was estimated as follows,

$$\:C\:credit\:revenue\:={CO}_{2}e*market\:price\:per\:t{CO}_{2}e\:$$
(6)

Subbasin delineation and land cover identification. Our approach seeks to extend LTE scale results to a subbasin level by considering the diverse microclimates and subtle precipitation and soil composition variations across the dryland PNW. We delineated a subbasin to preserve the unique climate and soil conditions shaped by local topography, hydrology, and vegetation to achieve more accurate estimates (Fig. 1). These factors significantly determine crop management practices in the region. Two Landsat-8 Operational Land Imager (OLI) images were acquired on 25 July 2022 and 5 July 2023. In the study, support vector machine (SVM) classification algorithms were utilized. Three bands were included: Red (B4), Near-Infrared (NIR, B5), and Short-Wave Infrared (SWIR 1, B6). The Normalized Difference Vegetation Index (NDVI) was selected to enhance winter wheat’s weak vegetation signal characteristics. Data analysis and visualization were performed using Esri ArcGIS Pro software (version 3.1).

Statistical Analysis. Although there were missing archive samples at certain time intervals and soil depths, which prevented a fully consistent dataset across the same subplots and replicates throughout the study period, this challenge was managed by utilizing the available data. To ensure a robust analysis, we employed both one-way and two-way analysis of variance (ANOVA). Normality of the data was assessed using the Shapiro-Wilk test. When significant differences were identified, a post-hoc Tukey test was applied to determine pairwise differences among treatments. All statistical analyses were conducted at a 0.05 significance level using StatGraphics Centurion software. To assess temporal trends in total C, N, and MAOM C within each treatment, separate one-way ANOVAs were performed individually for each treatment group. Additionally, a two-way ANOVA was conducted to evaluate the effects of tillage (NT vs. CT in WW–SP) and sampling year (1985, 1995, 2005, 2010, and 2021) on total C and N in bulk soil as well as C in MAOM. The control treatment (CT WW–SF) was excluded from two-way ANOVA due to differences in sampling years compared to the legume-based system. By focusing on the WW–SP system, this analysis evaluates how long-term tillage practices and temporal changes influence soil C and N pools in legume-based cropping system, as well as the interactive effects of tillage and sampling year.

Predictive performance of PLS models developed using the mid-infrared (MIR) region was assessed through repeated five-fold cross-validation (repeated k-fold CV). The evaluation relied on two statistical metrics: root mean square error (RMSE) and the coefficient of determination (R-squared). RMSE quantifies the absolute difference between predicted and observed values, while R-squared represents the proportion of variance in the measured data explained by the model, indicating its overall predictive accuracy.

Results and discussion

C sequestration in bulk soil. The results from the PLS calibrations are presented in scatter plots, showing the measured and predicted values for total C and N in bulk soil as well as C in the MAOM and POM fractions, as indicated in Fig. 2. Total C, N and MAOM C were accurately predicted with R-squares of 0.93, 0.92, 0.95 and respectively. POM C prediction accuracy was reasonably good (R-square = 0.78; RMSE = 0.93), however, it was often overestimated for values between 0.0 and 3.0 g kg −1. Given that this level of accuracy is insufficient for applications requiring high precision, we did not use POM C predicted values for the purpose of this study.

Fig. 2
figure 2

Scatter plots of measured and predicted values of (a) total carbon (TC); (b) carbon content in mineral-associated organic carbon (C-MAOM); (c) total nitrogen (TN); carbon content in particulate organic matter (C-POM); resulting from cross-validation using MIR-PLS regression by using present-day samples in soil derived from the wheat and pea experiments at different soil depths. The 1:1 line is plotted in black in each plot, the regression line in red and the shaded area indicates 95% confidence interval. Repeated 5–fold cross–validation was used as the resampling method.

Full size image

Our results demonstrate that wheat-based systems in eastern Oregon can effectively sequester soil C through conservation-oriented practices such as reduced tillage and legume-based crop rotations. The annual C sequestration rate in wheat-based systems varied by treatment, as shown in Table 2. The C accumulation in treatments that included spring pea into rotation consistently exhibited modest yet positive sequestration rates throughout the soil profile over the 58-year period (1963–2021). The average annual C sequestration rates ranged from 0.15 t C ha⁻¹ yr⁻¹ in CT to 0.16 t C ha⁻¹ yr⁻¹ in NT legume-based rotation systems. In contrast, our findings indicate that the WW–SF systems resulted in substantial C stock losses over time, with a decline rate of −0.11 t C ha⁻¹ yr⁻¹ in the 0–30 cm soil layer (Table 2). Our results are consistent with previous studies37from LTEs in eastern Oregon, revealing the difficulties in maintaining soil C in WW–SF systems, where losses of up to 0.29 t C ha–1 yr–1has been observed in the 0–60 cm soil profile3.

Table 2 Annual rates of soil carbon (C) sequestration (t C Ha yr−1) in bulk soil and mineral-associated organic matter (MAOM) and CO2 equivalent (CO2e) emission in the 0–30 depth assessed in conventional till (CT) winter wheat-summer fallow (WW–SF), CT winter wheat- spring pea (WW–SP) and no-till (NT) WW–SP cropping systems. The WW–SF rates were estimated for the period 1976–2015, while the WW–SP plots were analyzed from 1985 to 2021.
Full size table

Among other practices, conservation tillage is an active area of research in dryland wheat-based cropping systems. From our most recent sampling, we found that adding spring peas to crop rotations under NT significantly (p = 0.000) increased C stocks in bulk soil, nearly doubling those observed in WW–SF systems at the 0–10 cm depth (Fig. 3a). Specifically, we observed 23.3 ± 1.7 for NT WW–SP and 12.7 ± 1.3 in the traditional WW–SF system. However, below 10 cm, the net effect of NT WW–SP on C stocks was only significantly different (p = 0.015) from WW–SF, but not significantly different from CT WW–SP (Fig. 3a). The decline in C stocks in the WW–SF systems was particularly noticeable between 1976 and 2015, with a significant decrease observed in the first 10 cm (Fig. 4a). In contrast, C stocks in the CT WW–SP system (Fig. 4b) remained stable or showed only minor variations at the 0–10 cm depth compared to cropping systems utilizing summer fallow management (Fig. 4a). The near identical C sequestration rates in legume-based rotations under CT and NT at the 0–30 cm depth indicate that tillage was less detrimental to C sequestration than WW–SF (Table 2), suggesting that cropping intensification, regardless of tillage practices, contributes significantly to C sequestration. Similar results were found when comparing continuous CT and NT wheat and barley cropping systems13.

Fig. 3
figure 3

Changes in total Carbon (C) and nitrogen (N) stocks (t ha⁻¹) across different soil depths. (A) Total C in bulk soil, (B) Total N in bulk soil, and (C) Total C in mineral-associated organic C fraction (MAOM C) per 10 cm depth increments across three cropping systems: conventional tillage (CT) winter wheat–summer fallow systems (WW–SF) (samples collected in 2015, maintained for 84 years), and winter wheat–spring pea systems (WW–SP) (samples collected in 2021, maintained for 58 years), under CT and no-till (NT). Statistical analysis (± SD (N = 4) was conducted using one-way ANOVA followed by Tukey’s post-hoc test (p < 0.05).

Full size image
Fig. 4
figure 4

Total carbon (C) stocks (t C ha⁻¹) across 10 cm depth intervals measured at different sampling years in wheat-based cropping systems under (A) conventional tillage (CT) winter wheat-summer fallow (WW–SF), (B) CT winter wheat-spring pea (WW–SP), and (C) no-till (NT) WW–SP. Most of the data were reconstructed from archived soil samples using mid-infrared spectroscopy and partial least squares (PLS) models.

Full size image

Adoption of NT leads to greater C stratification; NT systems may result in higher rates of C sequestration over time, although the effect is primarily confined to surface layers13,41. Similarly,, our results show that untilled soil in WW–SP systems accumulated more C in the surface layer (< 10 cm depth) but sharply diminished with increasing soil depth (Figs. 3a and 4c). A similar trend was observed for total N stock at the 0–10 cm depth, with values significantly (p = 0.000) higher in NT WW-SP systems reaching 2.01 t N ha⁻¹. However, N stock below the 10 cm layer dropped to 1.23 t N ha⁻¹ in NT WW–SP, showing no significant difference (p = 0.039) from the levels observed in WW–SF systems (1.31 t N ha⁻¹) (Fig. 3b). In the legume-based CT rotation system, C and N stock show a more uniform distribution throughout the soil profile, which is expected as tillage redistributes topsoil into deeper layers (Fig. 3a and b). The increase of C stock in the upper topsoil (0–10 cm) under NT is consistent with studies showing that positive effects on C accumulation are mainly limited to surface soil layers13,42,43.

C sequestration in the MAOM fraction. The inclusion of legumes in wheat-based cropping systems helps offset N losses and support enhanced SOM stability44. Based on the most recent sampling period, the inclusion of legumes significantly (p = 0.000) increased MAOM C fraction, even under CT management, with this effect consistently observed across all 10 cm depth intervals (Fig. 3c). During the entire period of this study, our results clearly demonstrate that MAOM decomposes more slowly than bulk soil C yet remains influenced by tillage and cropping systems. Notably, MAOM in WW–SF systems exhibited increased C turnover (Fig. 5a), while C stocks in the MAOM fraction remained relatively stable in WW–SP systems under CT management (Fig. 5b). Traditional WW–SF systems led to a decline C storage in the MAOM fraction from 1976 to 2015 at a rate of −0.28 t C ha⁻¹ yr⁻¹ (Fig. 5a; Table 2). In contrast, incorporating legumes into crop rotations, particularly within NT systems, mitigated C losses in the MAOM fraction, particularly the 0–10 cm soil layer (Fig. 5c). At 30 cm depth (Fig. 6), the most recent data for WW–SF and WW–SP systems show that MAOM C stocks were higher in untilled soils under WW–SP (47.3 ± 1.1 t C ha⁻¹) than CT WW–SP soils (42.8 ± 1.9 t C ha⁻¹). Even under conventional tillage, WW–SP systems maintained substantially higher C stocks compared to WW–SF. Furthermore, our findings indicate that soil N stocks in WW–SP systems are greater than those in traditional WW–SF systems at 30 cm depth, with an observed increase of 11% under CT (4.1 ± 0.3 t N ha⁻¹) and 19% under NT conditions (4.4 ± 0 0.4 t N ha⁻¹) compared to 3.7 ± 0.3 t N ha⁻¹ in WW–SF systems (Fig. 6).

Fig. 5
figure 5

Total carbon (C) stock in mineral-associated organic matter fraction (MAOM C) stock (t C ha −1) across 10 cm depth intervals measured at different sampling years in wheat-based cropping systems under (A) conventional tillage (CT) winter wheat-summer fallow (WW–SF), (B) CT winter wheat-spring pea (WW–SP), and (C) no-till (NT) WW–SP. Most of the data were reconstructed from archived soil samples using mid-infrared spectroscopy and partial least squares (PLS) models.

Full size image
Fig. 6
figure 6

Changes in carbon (C) and nitrogen (N) stocks (t ha⁻¹) at a 0–30 cm depth over time in bulk soil and the mineral-associated organic carbon fraction (MAOM C) across three treatments: (A) tilled winter wheat–summer fallow systems (WW–SF), and winter wheat–spring pea systems (WW–SP), under (B) conventional tillage (CT) and (C) no-till (NT). Each bar represents the mean ± standard deviation of carbon (C) in bulk soil and MAOM, while the black circles represent nitrogen (N). Mean values sharing the same letter indicate no significant difference mong sampling times within each treatment. Green letters denote significant differences in bulk soil C, while gray capital letters correspond to significant differences in MAOM C, as determined by the Tukey post-hoc test (p ≤ 0.05).

Full size image

A two-way ANOVA assessed the effects of NT and CT in wheat legume-based systems and time on C and N in bulk soil and MAOM C (Table 3). Tillage had a significant effect on C in bulk soil (p = 0.020) and MAOM fraction (p = 0.004), while year (p = 0.110) and the tillage × year interaction (p = 0.123) were not significant, indicating temporal trends and interaction effects were minimal. For N, the tillage showed a marginal effect (p = 0.052), but year (p = 0.431) and the interaction were not significant. A one-way ANOVA revealed that MAOM remained stable over time in NT and CT, whereas the control CT WW-SF exhibited significant temporal fluctuations (Fig. 6), likely due to alternating fallow periods. These findings are consistent with prior research45,46suggesting that legume-integrated agricultural systems, which provide higher-quality inputs (i.e., lower C/N ratio) and straw incorporation, might reduce C decomposition and facilitate the formation of stable SOM by stimulating soil microbial activity. This, in turn, enhances C use efficiency (CUE) and increases the likelihood of C stabilization in MAOM47,48. However, no consensus remains regarding the long-term impacts of tillage practices and legume inclusion on the persistence of soil organic C in Mollisols such as Walla Walla silty loam, with varying results observed across different studies and climatic conditions49,50.

Table 3 Effects of no-till (NT) and conventional tillage (CT) over time on carbon (C) and nitrogen (N) stocks in bulk soil, as well as carbon in mineral-associated organic matter (MAOM), within a winter wheat–spring pea (WW-SP) system at a 30 cm soil depth. Mean values ± standard deviation (SD) (n = 4) are presented. Statistical analysis was conducted using a two-way ANOVA. A post hoc Tukey HSD test was performed for multiple comparisons, with different lowercase letters indicating significant differences (p < 0.05) among soil variables within the cropping system. The P × T interaction represents the combined effect of tillage (T) practices and time (Y).
Full size table

C sequestration upon conversion from WW–SF to WW–SP.Soils that have lost a significant portion of their original soil C pool can still sequester C when converted to restorative land uses and managed with recommended practices51. While not all soils can be effectively managed for C sequestration, our results demonstrate that Walla Walla silt loam soils can be either C sinks or sources of C. The potential for C sequestration during the transition from WW–SF to WW–SP cropping system was estimated as the difference in C stocks in recent years, normalized by 58 years of legume incorporation, with WW–SF values assumed as a baseline (Fig. 7). Specifically, Fig. 7a illustrates this potential, where \(\:P1\) and \(\:P2\) represent the attainable C sequestration potential at a 30 cm depth under CT NT practices, respectively. We infer that rapid mineralization of soil C occurred in the traditional WW–SF cropping system, impacting C in bulk soil as well as in the MAOM C fraction. This led to a significant reduction in soil C levels, which eventually stabilized at a new equilibrium of 39.0 t C ha⁻¹ (Fig. 6). After 80 years of consistent management (1931–2015), nearly constant C losses were expected and observed within the top 30 cm. (Table 2). According to the IPCC guidelines, C stocks, following all management changes—including the initial cultivation of native soil—are expected to stabilize at a new steady state after 20 years40,52. After 58 years or less, soil C stocks reached 48.0 t C ha[− 1under CT and 53.4 t C ha− 1 under NT practices in WW–SP cropping systems (Fig. 6). The transition from the depleted steady-state (Phase III) in the WW–SF to the new C stock levels in the WW–SP cropping system resulted in an approximately 38% increase in bulk soil C and a remarkable 60% increase in C in MAOM. Specifically, our findings indicate that C sequestration per unit area reached at least 0.16 t C ha⁻¹ yr⁻¹ under CT (\(\:P1\)) and 0.25 t C ha⁻¹ yr⁻¹ under NT (\(\:P2\)) in the 30 cm depth profile (Fig. 7b), with these rates likely being higher during the initial period of transition, suggesting a new equilibrium has been established (phase IV; present). Previous studies have documented substantial C sequestration within the first five years of adopting conservation practices, followed by a gradual decline in sequestration rates until a new equilibrium is reached53. Our estimated C storage potential aligns with previous findings from national croplands2, with values ranging from 0.23 ± 0.2 to 0.32 ± 0.2 t C ha⁻¹ yr⁻¹ to a depth of 30 cm. Although this region generally exhibits lower average sequestration rates, which are typically observed in colder northern and arid western areas of the US54, substantial differences across studies remain. For example, another study reported annual sequestration rates of 0.71 ± 0.63 t C ha⁻¹ yr⁻¹ over 14 years and 0.21 t C ha⁻¹ yr⁻¹ over a 10-year period, both following the transition from CT to untilled systems at a depth of 20 cm7.

Fig. 7
figure 7

Carbon (C) sequestration potential (t C ha⁻¹ yr⁻¹) in wheat cropping systems converted from high-intensity tillage (winter wheat–summer fallow, WW–SF) to a legume-cereal crop rotation (winter wheat–spring pea, WW–SP). (A) Schematic representation illustrating how C sequestration rates were estimated, beginning with a WW–SF system depleted in C as the baseline, followed by a C reaccumulation phase after adopting WW–SP system. The attainable sequestration potential, denoted as \(\:P1\) and \(\:P2\), is assessed under conventional tillage (CT) and no-till (NT) at a 30 cm depth. (B) C sequestration rates in bulk soil and mineral-associated organic matter (MAOM), along with annual CO2 reduction in carbon dioxide equivalents (CO2e) estimated in \(\:P1\) and \(\:P2\).

Full size image

While our current research methodologies have some uncertainties and limitations, this study provides valuable empirical data to enhance our understanding of C dynamics. We offer a solid foundation for refining C simulation models by providing these annual rates of change in C. We also recognize the nonlinear trajectory of soil C dynamics, which are inherently complex and evolve in ways that cannot be captured by simple linear models38. In the initial phase of this pattern, more pronounced losses or gains are often observed, which gradually slow as the system approaches equilibrium, presenting an important opportunity for further exploration. Previous research suggests that soil organic C is generally lost more rapidly than it is accumulated39,40, which may alter our results. For example, if baseline conditions in \(\:{Y}_{{t}_{1}}\)continue to show C losses or gains, the annual C accumulation rates may be higher or lower than expected for any given response year (\(\:{Y}_{{t}_{2}}\)) than estimates based on a steady-state baseline scenario. However, we assume our cropping systems to be in a steady-state condition, as the samples were taken from long-established systems. By the time samples were collected in 1976, the WW–SF system had been maintained for 45 years (1931–1976), while the WW–SP system had been in place for 58 years (1963–1985). Although data from the initial phases are unavailable, our findings offer significant understanding of the the impact of management practices on C sequestration in dryland wheat systems in the inland Pacific Northwest (iPNW), as well as the minimum amount of C sequestration we can expect in these soils, as the baseline year was assumed to be in a steady-state condition.

CO2-equivalent emissions. Croplands that transitioned from native conditions to intensive tillage prior to adopting NT practices exhibit a significant C sequestration capability. This transition results in a win-win outcome, increasing C storage and reducing CO2e emissions, following years of conventional tillage, fallow, and soil C depletion. Our results indicate that legume-based rotations could probably offset an average of 0.58–0.94 t CO₂e per hectare per year in Walla Walla silt loam soils (Fig. 7b). Based on the calculated C sequestration rates, we estimate that the current 14,758 ha included in the study (Fig. 1) under CT management would offset 13,770 tons of CO₂e emissions annually with the transition to WW–SP cropping systems. The study area represents only 0.8% of Oregon’s total cropland, which spans 1,912,588 ha55. The study area shown in Fig. 1could potentially offset 0.20% of Oregon’s agricultural emissions, which are estimated at 7 million metric tons annually56. In the US, emerging C crediting organizations have announced the price of C credits at $20–30 per ton of CO₂e. By 2030, across major economies, the High-Level Commission on C Prices —aimed at limiting global temperature rise to well below 2ºC— foresees these prices rising to $63–127 per ton of CO₂e36. Consequently, in this area, annual farm revenues from C offsets could total approximately USD $29 or $93 per ha under 2030 standards. Oregon’s average farm size is 170 ha (~ 400 acres); by adopting cereal-legume rotations, each farm could potentially earn an average of USD $4,780 annually.

In annual cereal-fallow systems, tillage plays a crucial role in C losses by influencing soil structure and long-term C retention57. Consistent with other studies, its impact is mitigated when cropping intensification occurs, such as extended rotations or spring crops, particularly in areas where moisture is less limiting than in other agroecological zones of the inland PNW58. Beyond agronomic factors, a key challenge is that farmers are more likely to adopt management practices when they provide clear benefits, such as increased profitability, long-term sustainability, and market diversification59. Wheat production costs in this region USDA ERS (Base Survey 2023)60 are estimated at approximately $195 per acre ($480 per ha), covering variable expenses like seed, fertilizer, labor, and machinery operation (excluding capital expenditure and land ownership costs). As mentioned earlier, annual revenue from carbon offsets in this region is estimated at around $29 per ha, or $93 per ha under 2030 standards. This indicates that returns from voluntary carbon markets may not be sufficient to incentivize widespread adoption of these practices. Even under the most optimistic market conditions, these incentives fail to cover operational costs. These findings emphasize the need for stronger financial incentives to enhance the economic viability of carbon offset programs, ensuring greater farmer participation. Rather than focusing solely on economic returns, the benefits of these management strategies are more closely linked to improvements in soil health, reductions in greenhouse gas emissions, and the promotion of long-term agricultural sustainability.

Conclusion

Given the need for credible C flux estimates in the dryland iPNW, this study explores the potential for C sequestration through the adoption of conservation practices in wheat-based systems of eastern Oregon. Although we found relatively low C sequestration rates, greater opportunities exist to build C into annual cropping systems. Our findings demonstrate that eliminating fallow and incorporating legumes into WW–SF cropping systems can significantly enhance the formation of more persistent C over time. This is particularly important in regions with low precipitation and frequent fallow periods, where efforts to increase soil C levels face persistent limitations. While the increase in C inputs and improvement in N cycling provides only a modest contribution to C sequestration in NT legume-based cropping systems, its impact remains tangible and beneficial. Our study emphasizes the importance of archived soils and long-term experiments in establishing realistic C recovery goals. This is especially important given the ongoing challenges in estimating C sequestration potential, as uncertainties persist due to variations in analytical methods across laboratories over time. Soil C sequestration efforts not only support GHG emission offsets but also enhance overall soil health, making the region well-positioned for participation in voluntary carbon markets. However, the immediate agronomic benefits may take priority over climate mitigation, particularly due to the low market value of C. This study emphasizes the need for financial incentives that promote soil health, reducing GHG emissions and ensuring long-term sustainability for farmers.