Machine Learning Driven Optimization of Carbon Sequestration in Intercropping Systems Using XGBoost Modeling and Partial Dependence Analysis

Machine Learning Driven Optimization of Carbon Sequestration in Intercropping Systems Using XGBoost Modeling and Partial Dependence Analysis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Machine Learning Driven Optimization of Carbon Sequestration in Intercropping Systems Using XGBoost Modeling and Partial Dependence Analysis Nallagatla Vinod Kumar, Gajanan L. Sawargaonkar, C. Sudha Rani, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7307700/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract This study integrates XGBoost modeling with Partial Dependence Plots to optimize carbon sequestration. The experiment was conducted using a split-split plot design replicated thrice with the main plot compressing tillage practices viz. , minimum tillage (M 1 ) and conventional tillage (M 2 ), subplot consists of row ratios viz. pigeonpea + maize (1:2 ratio) (R 1 ), pigeonpea + maize (1:3 ratio) (R 2 ), and sole pigeonpea (R 3 ) and sole maize (R 4 ), sub-subplot consists residue management practices viz. on farm produced biochar application (S 1 ), on farm produced residue application (S 2 ), and control with no biochar or residue application (S 3 ). Machine learning models of XGBoost accurately predicted CSQ while Artificial Neural Network performed best for SP and CF, confirming nonlinear relationships. PDPs showed available nitrogen increased productivity from 3.0 to 3.4 Mg ha⁻¹ y⁻¹ (150–300 kg ha⁻¹), while SOC reduced it from 3.8 to 3.2 Mg ha⁻¹ y⁻¹ (0.425–0.525%). Sensitivity analysis identified SOC, moisture, and enzyme activities as key drivers, with SOC enhancing sequestration by ~ 1.5 Mg C ha⁻¹ and improving footprint by ~ 1.2 Mg CO₂-Ce ha⁻¹. These insights enable targeted, efficient soil management for productivity and climate benefits. Machine learning-based Partial Dependence Plot (PDP) and sensitivity analysis (SA) revealed SOC, BD, and microbial parameters as key drivers of these outcomes. System productivity Carbon sequestration Carbon footprint Partial Dependence Plot and Sensitivity Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Introduction Climate change mitigation through agricultural carbon sequestration has emerged as a critical strategy for achieving global sustainability goals while maintaining food security (Lal, 2020 ; Smith et al., 2019 ). Intercropping systems, characterized by the simultaneous cultivation of multiple crop species, offer substantial potential for enhancing soil organic carbon (SOC) storage compared to monoculture practices (Li et al., 2020 ; Zhang et al., 2022 ). The integration of leguminous crops like pigeonpea with cereals such as maize creates synergistic effects that can significantly improve carbon sequestration rates through enhanced root biomass production, nitrogen fixation, and soil microbial activity (Kumar et al., 2021 ; Sharma et al., 2023 ). Traditional agricultural management practices, including tillage operations and residue management, profoundly influence carbon dynamics in intercropping systems (Chen et al., 2022 ). Minimum tillage practices have been shown to preserve soil structure and enhance carbon retention, while biochar application can provide long-term carbon storage benefits (Lehmann & Joseph, 2015 ; Zhou et al., 2021 ). However, the complex interactions between these management factors and their collective impact on carbon sequestration remain poorly understood, necessitating advanced analytical approaches. Machine learning techniques have revolutionized agricultural research by enabling the analysis of complex, non-linear relationships in ecological systems (Rashid et al., 2023 ; Wang et al., 2022 ). Extreme Gradient Boosting (XGBoost) has demonstrated superior performance in predicting agricultural outcomes due to its ability to handle feature interactions and non-linear patterns (Liu et al., 2023 ). Partial Dependence Plots (PDPs) complement machine learning models by providing interpretable insights into individual feature effects, making them particularly valuable for agricultural decision-making (Molnar, 2022 ; Thompson et al., 2023 ). Despite growing interest in machine learning applications for sustainable agriculture, limited research has integrated XGBoost modeling with PDP analysis to optimize carbon sequestration in intercropping systems. The complex interactions between tillage practices, cropping patterns, and residue management require sophisticated analytical frameworks to identify optimal management strategies. Furthermore, understanding the sensitivity of carbon sequestration to key soil parameters is essential for developing targeted interventions that maximize climate benefits while maintaining productivity. This study addresses these knowledge gaps by employing XGBoost modeling coupled with Partial Dependence Analysis to optimize carbon sequestration in intercropping systems. The research objectives were to: (1) evaluate the performance of machine learning models in predicting carbon sequestration, system productivity, and carbon footprint; (2) identify key drivers of carbon sequestration through sensitivity analysis; and (3) quantify the individual effects of soil parameters on system outcomes using Partial Dependence Plots. The findings contribute to evidence-based agricultural management strategies that enhance both productivity and climate mitigation potential. 2. Materials and methods 2.7 Machine Learning processes Machine learning (ML) models were adopted to study the dynamics of systems productivity, CSQ and CF based on soil parameters under pigeonpea + maize inter cropping system. All the data related to soil was analyzed as per mentioned protocols in (Supplementary file, ST-1) used in its initial form to build ML models in this study. Eight ML models were separately trained and evaluated to predict each CSQ, CF, and SP using soil parameters as input. These models included linear approaches viz., Ridge Regression, LASSO Regression, and Elastic Net, alongside nonlinear techniques viz., Random Forest, Support Vector Regression, Extreme Gradient Boosting, Artificial Neural Network, and Decision Tree. In order to avoid the overfitting issues, the soil parameters that have a considerable predictive effect on the respective response variable were selected using partial least square regression (PLSR) analysis. The soil parameters with a Variable Importance in Projection (VIP) score greater than 0.9 for a given response were selected to model the respective response (Supplementary file, ST-3 & SF- 1). To ensure robustness, 80% of the dataset was used for training, while the remaining 20% was reserved for validation. 2.7.1 Performance metrics used for the ML models Coefficient of determination (R 2 ) or Explained Variance score and other metrics of Errors viz ., Root Mean Squared Error, Mean Absolute Error, Mean Absolute Percentage Error, and Mean Absolute Bias Error. The best-performing model was identified based on the highest R² and the lowest error metrics for both training and validation datasets. The selection of the most suitable model also provided insights into the response behavior of SP, CSQ and CF for change in soil parameters. If a linear model such as Ridge, LASSO, or ENET demonstrated superior performance, it suggested a proportional effect of soil properties on the response variables across their entire range. Conversely, if a nonlinear model like RF, SVR, XGBoost, ANN, or DT exhibited the best predictive capability, it indicated complex interactions, where input variables influenced the response in a non-linear manner. To further interpret the model outputs, PDPs and SA were generated using the best-fitted model, providing a visual representation of how variations in specific soil parameters influenced CSQ, CF, and yield while keeping other variables constant. The PDP is generated as discussed below (Zhang et al., 2018), Let there are ‘k’ soil parameters and ‘n’ observations of each such that j=1, 2…n. Let the X i is the parameter of interest. The values of other (k-1) parameters are replaced with their average values for all the n observations. The response values are predicted through the best-fitted model using the new dataset. Plot the predicted values against the levels of the parameter of interest X i. These PDP facilitates a deeper understanding of the underlying relationships between soil health indicators and key agricultural sustainability metrics (Friedman, 2001). The sensitivity plots are generated by predicting CSQ, CF, and yield through the best-fitted model for each soil parameter while holding other parameters at the maximum and minimum values. The details of the metrics, source codes of which can be found in Scikit-learn are explained below: Root Mean Square Error (RMSE): $$\:\text{R}\text{M}\text{S}\text{E}=\:\sqrt{\frac{{\sum\:}_{i=1}^{N}\:{\left({Predicted}_{i}-{Actual}_{i}\right)}^{2}}{n}}$$ where, N is the total number of observations. Mean Absolute Error (MAE): $$\:\text{M}\text{A}\text{E}=\left(\frac{1}{n}\right)\text{*}\sum\:\left|{Actual}_{i}-{\:Predicted}_{i}\right|\:$$ where, n is the total number of observation and ‘i’ is the i th observation. Coefficient of determination or R 2 : $$\:{\text{R}}^{2}=\:\:\:\left(1-\frac{SSR}{SST}\right)$$ where, SSR is the sum of squared of the residual errors and SST is the total sum of the errors. Mean Absolute Percentage Error (MAPE): $$\:\:\:\:\text{M}\text{A}\text{P}\text{E}=\:\left(\frac{1}{n}\right)\:*\:\sum\:\left(\frac{\left|Actual-Predicted\right|}{\left|Actual\right|}\right)\:*100$$ where, n is the total number of observations. Mean Absolute Bias Error (MABE) = \(\:\frac{1}{n}\sum\:\left|Predicted-Actual\right|\) 2.8 Statistical analysis The carbon footprint (CF) was calculated based on the average data of carbon sources and sinks, along with their respective carbon equivalents. The ML models, PDP and sensitivity plots are generated using the R software version 4.4.3 (R Core Team, 2024) 3. Results 3.4. Machine Learning Model Analysis 3.4.1. Partial Dependence Plot Analysis System Productivity : Available nitrogen showed a strong positive relationship with productivity, which increased from 3.0 to 3.4 Mg ha⁻¹ y⁻¹ as nitrogen levels rose from 150 to 300 kg ha⁻¹ (Fig. 3a). Available potassium exhibited a positive but moderate influence, with productivity rising from 3.32 to 3.38 Mg ha⁻¹ y⁻¹ as potassium increased from 150 to 400 kg ha⁻¹. In contrast, phosphorus displayed an inverse relationship with productivity. SOC showed a negative effect on productivity when isolated, with yields decreasing from 3.8 to 3.2 Mg ha⁻¹ y⁻¹ as SOC rose from 0.425% to 0.525%. Smaller water-stable aggregates (<0.25 mm) enhanced productivity from 3.0 to 3.6 Mg ha⁻¹ y⁻¹ as their percentage increased from 4% to 12%, whereas larger aggregates (0.25-2 mm) reduced productivity from 3.5 to 3.3 Mg ha⁻¹ y⁻¹. Field capacity demonstrated a negative relationship with productivity, while bulk density (BD) showed a positive trend. Bacterial abundance was negatively correlated with productivity, while actinomycetes demonstrated a positive effect. Dehydrogenase activity exhibited a non-linear relationship with productivity. Carbon Sequestration : SOC exhibited clear threshold increases in sequestration capacity, rising from 2.5 to 4.0 Mg C ha⁻¹ as SOC increased from 0.425% to 0.525% (Fig. 3b). BD demonstrated a notable increase in sequestration from 3.2 to 3.9 Mg C ha⁻¹ when density reached 0.125 g cm⁻³. Bacterial abundance showed a positive effect on sequestration, which rose from 3.45 to 3.85 Mg C ha⁻¹ as populations increased from 30 to 65 CFU g⁻¹ soil. Actinomycetes exhibited a negative trend, with sequestration decreasing from 4.2 to 3.7 Mg C ha⁻¹ as their counts increased. Fungi showed stepwise increases in sequestration from 3.7 to 4.0 Mg C ha⁻¹ as their abundance rose. Enzyme activities showed varying relationships with sequestration. Dehydrogenase activity remained relatively stable until exceeding 15 μg TPF g⁻¹ soil 24 hr⁻¹, beyond which sequestration sharply increased. Microbial biomass carbon exhibited a dramatic threshold response, with sequestration sharply rising from 3.7 to 4.0 Mg C ha⁻¹ when biomass exceeded 500 μg C g⁻¹ soil. Carbon Footprint : The isolated effect of pH was substantial, with CF becoming less favorable (from -1.9 to -1.0 Mg CO₂-Ce ha⁻¹) as pH increased from 7.5 to 9.0 (Fig. 3c). SOC showed a strong pattern, with carbon footprint improving (more negative) from -0.8 to -2.0 Mg CO₂-Ce ha⁻¹ as SOC increased from 0.425% to 0.5%. BD independently improved the carbon footprint from 0 to -2.5 Mg CO₂-Ce ha⁻¹ as density increased from 0.11 to 0.14 g cm⁻³. Soil moisture showed a U-shaped curve with the lowest carbon footprint (-1.9 Mg CO₂-Ce ha⁻¹) at approximately 0.05% moisture content. Among microbial indicators, actinomycetes showed worsening carbon footprint as their abundance increased, while fungi displayed a U-shaped relationship. Urease activity was associated with an improved carbon footprint, decreasing from 0 to -2.0 Mg CO₂-Ce ha⁻¹ as activity increased. Microbial biomass carbon showed a strong inverse correlation with carbon footprint. 3.4.2. Sensitivity Analysis Sensitivity analysis revealed distinct patterns in how soil and microbial parameters influenced system productivity, carbon footprint, and carbon sequestration (Fig. 4). System productivity showed the greatest sensitivity to carbon footprint parameters, with maximum values reaching ~5.5 Mg ha⁻¹ y⁻¹ during carbon sequestration phases (Fig. 4a). Carbon sequestration was most significantly influenced by SOC content, with maximum values approaching 3.8 Mg C ha⁻¹ (Fig. 4b). Bulk density was associated with the lowest sequestration values (1.5 Mg C ha⁻¹). Narrow gaps between maximum and minimum values for acid phosphatase indicated a consistently strong effect on sequestration regardless of external variability. Carbon footprint remained negative across most parameters, reflecting net sequestration potential in the system (Fig. 4c). Alkaline phosphatase had the most positive influence on carbon dynamics, while SOC and urease activity exhibited the most negative influence, with footprint values reaching as low as -2.5 Mg CO₂-Ce ha⁻¹. 4. Discussion 4.4. Machine Learning Insights into Soil-Plant-Microbial Interactions 4.4.1. Partial Dependence Relationships The analysis of soil-property interactions using partial dependence plots provided valuable insights into the complex relationships between soil properties and agricultural sustainability metrics. Soil physical properties, particularly water-stable aggregates, showed significant influence on key outcomes. Improved soil structure resulted in yield increases, enhanced carbon sequestration, and reduced carbon footprint, aligning with Bronick and Lal (2005), who emphasized that soil structure governs key processes such as water infiltration, aeration, and microbial habitat provision. The relationship between bulk density and agricultural outcomes suggested that moderate compaction can enhance root-soil contact while preserving adequate pore space, consistent with Reynolds et al. (2009), who highlighted optimal physical quality ranges for soil productivity. Field capacity and soil moisture dynamics underscored the importance of water management in influencing both crop performance and carbon cycling, supporting findings by Rawls et al. (2003). Among soil chemical properties, available nitrogen demonstrated a strong positive correlation with productivity, affirming Robertson and Vitousek's (2009) conclusion on its foundational role in crop systems. Soil pH emerged as a significant regulator of carbon footprint, aligning with Lal (2004), who noted the role of pH in moderating greenhouse gas emissions through microbial-mediated processes. SOC proved to be the most influential property, driving increases in yield, reductions in carbon footprint, and improvements in carbon sequestration. These results validate Lal's (2016) assertion that managing SOC offers a dual solution for boosting productivity while mitigating climate change impacts. The study also highlighted the pivotal role of soil biological processes, with microbial populations and enzyme activities showing distinct relationships with agricultural outcomes. These findings agree with Bardgett and van der Putten (2014), who emphasized the essential contributions of soil biota to ecosystem functions and Burns et al. (2013), who documented threshold effects in enzyme activities. 4.4.2. Sensitivity Analysis Implications Our sensitivity analysis revealed significant variations in agricultural ecosystem metrics in response to changes in soil and microbial parameters. System productivity demonstrated remarkable sensitivity to carbon dynamics, with up to 187% variation between minimum and maximum values when influenced by carbon footprint parameters. This significant range highlights the potential for climate-friendly management practices to deliver dual benefits: optimizing agronomic outputs while achieving environmental sustainability (Lal, 2020). Carbon sequestration potential showed a 98% increase in response to changes in SOC content, revealing powerful positive feedback mechanisms that can accelerate carbon capture in well-managed semi-arid soils. This self-reinforcing relationship provides tremendous opportunity for climate change mitigation through enhanced agricultural carbon storage (Paustian et al., 2016). The carbon footprint analysis revealed predominantly negative values, indicating the potential for agricultural systems to act as carbon sinks rather than sources. Even modest increases in SOC from 0.40% to 0.50% resulted in a 136% improvement in carbon footprint. This non-linear relationship emphasizes how minimal investments in enhancing soil carbon can yield disproportionately large environmental benefits, providing robust support for policies incentivizing soil carbon enhancement as a climate mitigation strategy (Smith et al., 2019). The identification of threshold effects, particularly step increases in carbon sequestration associated with specific parameter thresholds, suggests that targeted interventions could yield disproportionate environmental benefits. Increasing soil microbial biomass carbon beyond 500 μg C g⁻¹ soil triggered an 8% jump in sequestration, highlighting the potential value of management practices that enhance soil microbial abundance and activity. Conclusions This study successfully demonstrated the effectiveness of machine learning approaches, particularly XGBoost modeling coupled with Partial Dependence Analysis, for optimizing carbon sequestration in intercropping systems. The key findings provide valuable insights for sustainable agricultural management and climate change mitigation strategies. The XGBoost model achieved superior performance in predicting carbon sequestration (CSQ), while Artificial Neural Networks excelled in forecasting system productivity (SP) and carbon footprint (CF). These results confirm the presence of complex, non-linear relationships within intercropping systems that require advanced analytical techniques for accurate prediction. The high predictive accuracy of these models validates their potential for real-world agricultural decision support systems. Machine learning analyses revealed complex relationships between soil parameters and sustainability metrics, with SOC emerging as the key driver of both carbon sequestration and reduced carbon footprint. Sensitivity analysis identified threshold effects and quantified trade-offs, providing valuable guidance for precision agriculture approaches that optimize both productivity and environmental services. Future research should focus on validating these models across different geographical regions and extending the analysis to include economic considerations. Long-term studies examining the temporal dynamics of carbon sequestration under varying climate scenarios would further enhance the practical applicability of these findings. Additionally, integration of remote sensing data could expand the spatial scale of model applications. This research contributes significantly to the growing body of knowledge on sustainable agriculture and climate-smart farming practices. The demonstrated effectiveness of machine learning approaches for optimizing carbon sequestration provides a foundation for developing evidence-based policies and management recommendations that address both food security and climate change challenges. Declarations Funding No funding is available. Availability of data and materials: The data is available upon request CRediT authorship contribution statement Nallagatla Vinod Kumar : Conceptualization, Formal analysis, Investigation, Methodology, Data curation, Writing -original draft, Writing - review & editing. Gajanan Sawargaonkar, C. Sudharani : Conceptualization, Methodology, Resources, Supervision, Writing - review & editing. T. Ram Prakash, S. Triveni, Ch. Sarada : Methodology, Validation, Writing - review & editing. Ajith S , Peace Raising and M. Prabhakar : Formal analysis, Methodology, Visualization, Writing -review & editing. Acknowledgement We acknowledge the ICRISAT and PJTSAU for providing experimental field and laboratory facilities during the experimentation. Clinical trial registration: Not applicable. Ethics and Consent to Participate declarations: Not applicable. Conflict of interest The author declares that there is no confict of interests regarding the publication of this paper. Data availability Data will be made available on request. Consent to Participate and Consent to Publish Informed consent to participate and consent to publish, including the use of personal information and photographs (Fig. 1), were obtained from all participants. No participants under the age of 18 years were involved in this study. References Bardgett, R. D., & van der Putten, W. H. (2014). Belowground biodiversity and ecosystem functioning. Nature, 515, 505-511. Bronick, C. J., & Lal, R. (2005). Soil structure and management: A review. Geoderma, 124, 3-22. Burns, R. G., DeForest, J. L., Marxsen, J., Sinsabaugh, R. L., Stromberger, M. E., Wallenstein, M. D., ... & Zoppini, A. (2013). Soil enzymes in a changing environment: Current knowledge and future directions. Soil Biology and Biochemistry, 58, 216-234. Chen, X., Liu, Y., Wang, Z., & Zhang, H. (2022). Tillage effects on soil carbon dynamics in intercropping systems: A meta-analysis. Soil and Tillage Research, 218, 105-118. Chen, Y., Li, M., Zhang, H., Wang, L., & Liu, S. (2022). Effects of tillage and residue management on soil carbon dynamics in intercropping systems: A meta-analysis. Agriculture, Ecosystems & Environment, 327, 107825. Friedman, J.H., 2001. Greedy function approximation: a gradient boosting machine. Annals of statistics, pp.1189-1232. Kumar, A., Singh, P., Patel, R., & Gupta, S. (2021). Nitrogen fixation and carbon sequestration in legume-cereal intercropping systems. Agriculture, Ecosystems & Environment, 312, 107-122. Kumar, A., Singh, R., Patel, N., & Sharma, K. (2021). Nitrogen fixation and carbon sequestration in legume-cereal intercropping systems: Mechanisms and management strategies. Field Crops Research, 276, 108371. Lal, R. (2004). Soil carbon sequestration impacts on global climate change and food security. Science, 304, 1623-1627. Lal, R. (2016). Soil health and carbon management. Food and Energy Security, 5, 212-222. Lal, R. (2020). Soil carbon sequestration to mitigate climate change. Geoderma, 123, 1-22. Lal, R. (2020). Soil organic carbon sequestration impacts on global climate change and food security. Science, 304(5677), 1623-1627. Lehmann, J., & Joseph, S. (2015). Biochar for environmental management: Science, technology and implementation (2nd ed.). Routledge. Lehmann, J., & Joseph, S. (2015). Biochar for environmental management: Science, technology and implementation (2nd ed.). Routledge. Li, M., Chen, Q., Zhang, L., & Wang, J. (2020). Intercropping enhances soil organic carbon storage through improved root systems. Field Crops Research, 245, 107-115. Li, W., Zhang, P., Ye, J., Li, L., & Baker, P. A. (2020). Impact of two different types of rural land use change on soil organic carbon in a semi-arid region. Plant and Soil, 452(1), 149-166. Liu, H., Wang, S., Chen, M., & Li, X. (2023). XGBoost applications in agricultural prediction: A comprehensive review. Computers and Electronics in Agriculture, 198, 106-118. Liu, X., Wang, Y., Chen, M., Zhang, L., & Thompson, R. (2023). XGBoost applications in agricultural systems: A comprehensive review of predictive modeling approaches. Computers and Electronics in Agriculture, 205, 107634. Molnar, C. (2022). Interpretable machine learning: A guide for making black box models explainable (2nd ed.). Leanpub. Molnar, C. (2022). Interpretable machine learning: A guide for making black box models explainable (2nd ed.). Lean Publishing. Paustian, K., Lehmann, J., Ogle, S., Reay, D., Robertson, G. P., & Smith, P. (2016). Climate-smart soils. Nature, 532, 49-57. Rashid, M., Ahmed, K., Hassan, N., & Ali, S. (2023). Machine learning revolution in sustainable agriculture: Current trends and future prospects. Precision Agriculture, 24, 892-915. Rashid, M., Ahmed, S., Khan, A., & Ali, M. (2023). Machine learning approaches for sustainable agriculture: Applications, challenges, and future perspectives. Agricultural Systems, 208, 103652. Rawls, W. J., Pachepsky, Y. A., Ritchie, J. C., Sobecki, T. M., & Bloodworth, H. (2003). Effect of soil organic carbon on soil water retention. Geoderma, 116, 61-76. Reynolds, W. D., Bowman, B. T., Drury, C. F., Tan, C. S., & Lu, X. (2002). Indicators of good soil physical quality: Density and storage parameters. Geoderma, 110, 131-146. Robertson, G. P., & Vitousek, P. M. (2009). Nitrogen in agriculture: Balancing the cost of an essential resource. Annual Review of Environment and Resources, 34, 97-125. Sharma, P., Kumar, S., Singh, A., & Patel, R. (2023). Intercropping effects on soil microbial communities and carbon cycling: Insights from pigeonpea-maize systems. Soil Biology and Biochemistry, 179, 108986. Sharma, V., Patel, K., Reddy, M., & Singh, R. (2023). Synergistic effects of pigeonpea-maize intercropping on soil carbon dynamics. Agroforestry Systems, 97, 445-462. Smith, P., Adams, J., Beerling, D. J., Beringer, T., Calvin, K. V., Fuss, S., ... & Arneth, A. (2019). Land-management options for greenhouse gas removal and their impacts on ecosystem services and the sustainable development goals. Annual Review of Environment and Resources, 44, 255-286. Thompson, K., Martinez, L., Brown, J., & Wilson, D. (2023). Partial dependence plots in agricultural machine learning: Interpretation and applications. Precision Agriculture, 24(2), 456-478. Thompson, R., Davis, L., & Miller, K. (2023). Interpretable machine learning for agricultural decision support systems. Agricultural Systems, 208, 103-117. Wang, H., Liu, Y., Chen, X., Zhang, S., & Lee, K. (2022). Machine learning in precision agriculture: Current applications and future prospects. Smart Agricultural Technology, 3, 100098. Wang, T., Liu, G., Zhang, F., & Zhou, Y. (2022). Deep learning applications in crop yield prediction and optimization. Journal of Agricultural Science, 160, 234-248. Zhang, Q., Liu, H., Wang, J., Chen, F., & Li, Y. (2022). Carbon sequestration potential of intercropping systems: A global synthesis. Global Change Biology, 28(12), 3652-3668. Zhang, W., Liu, K., Wang, H., & Chen, Y. (2022). Carbon sequestration potential of intercropping systems: A global synthesis. Global Change Biology, 28, 1245-1261. Zhang, Z., Beck, M. W., Winkler, D. A., Huang, B., Sibanda, W., Goyal, H., 2018. Opening the black box of neural networks: methods for interpreting neural network models in clinical applications. Annals of translational medicine, 6(11), 216. Zhou, L., Yang, X., Wang, P., & Liu, J. (2021). Biochar amendments for long-term carbon storage in agricultural soils. Biochar, 3, 127-142. Zhou, M., Wang, X., Li, P., Zhang, Y., & Chen, L. (2021). Biochar effects on carbon sequestration and greenhouse gas emissions in agricultural soils: A global meta-analysis. Journal of Cleaner Production, 288, 125593. Additional Declarations No competing interests reported. Supplementary Files Suplimentaryfile.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7307700","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":518056335,"identity":"16435e0a-6973-4f38-a2c9-c334b5a69227","order_by":0,"name":"Nallagatla Vinod Kumar","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABC0lEQVRIiWNgGAWjYDACZiDmYWCQ4QPzChjkQNSBB0Ro4WEDKzVgMAbTCYRsQtaS2ABi4NNicJz34IO3O+x42NiPX/z8wcAufX7Y4YdAW+zkdBtwaDnMl2w490wyDxtPTrHEAYPk3I230wyAWpKNzQ5g1yLZzGMmzdvGDHRYTgJQC3PuxtkJIC0HErfh11LPw8b/JvnHAYP6dMPZ6R/wauFnBms5zMMmkX4MaMvhBHnpHPy2ALUYG85tOw7U8obN4ozBccMN0jkFBxIMcPuFjf+M4YO3bdVy/Pzpj29UVFTLy89O3/zhQ4WdHC4tSIDHAEwZgFUaEFQOAuwPwJR8A1GqR8EoGAWjYAQBAHK/WaC9n4e1AAAAAElFTkSuQmCC","orcid":"","institution":"Telangana State Agricultural University PJTSAU","correspondingAuthor":true,"prefix":"","firstName":"Nallagatla","middleName":"Vinod","lastName":"Kumar","suffix":""},{"id":518056336,"identity":"7411f5cd-7c56-4e7b-8928-2ad3de266c72","order_by":1,"name":"Gajanan L. Sawargaonkar","email":"","orcid":"","institution":"International Crops Research Institute for the Semi-Arid Tropics","correspondingAuthor":false,"prefix":"","firstName":"Gajanan","middleName":"L.","lastName":"Sawargaonkar","suffix":""},{"id":518056338,"identity":"56ad937d-8679-4913-8326-95a46afe0b80","order_by":2,"name":"C. Sudha Rani","email":"","orcid":"","institution":"Agricultural Research Station","correspondingAuthor":false,"prefix":"","firstName":"C.","middleName":"Sudha","lastName":"Rani","suffix":""},{"id":518056339,"identity":"ed71669a-0635-4b36-8846-960eb3d3c9ec","order_by":3,"name":"T. Ram Prakash","email":"","orcid":"","institution":"AICRP on Weed Control, Professor Jayashankar Telangana State Agricultural University PJTSAU","correspondingAuthor":false,"prefix":"","firstName":"T.","middleName":"Ram","lastName":"Prakash","suffix":""},{"id":518056340,"identity":"3364fdd2-9aca-4445-9c62-9c506154e4d9","order_by":4,"name":"S. Triveni","email":"","orcid":"","institution":"Telangana State Agricultural University PJTSAU","correspondingAuthor":false,"prefix":"","firstName":"S.","middleName":"","lastName":"Triveni","suffix":""},{"id":518056342,"identity":"823b343d-036a-46e7-938d-21b22434198d","order_by":5,"name":"Ch. Sarada","email":"","orcid":"","institution":"ICAR- Indian Institute of Oil seed Research","correspondingAuthor":false,"prefix":"","firstName":"Ch.","middleName":"","lastName":"Sarada","suffix":""},{"id":518056343,"identity":"fa31a9d5-6f1b-4cfd-b580-8ecdf4c52d14","order_by":6,"name":"Ajith . S","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Ajith","middleName":".","lastName":"S","suffix":""},{"id":518056345,"identity":"b6251b8b-f996-49df-b822-8bfa0c220f63","order_by":7,"name":"Peace Raising","email":"","orcid":"","institution":"GD Goenka University","correspondingAuthor":false,"prefix":"","firstName":"Peace","middleName":"","lastName":"Raising","suffix":""},{"id":518056347,"identity":"44b47bcc-1e26-45fb-9f3d-2e5653062e04","order_by":8,"name":"M. Prabhakar","email":"","orcid":"","institution":"ICAR-Central Research Institute for Dry land Agriculture","correspondingAuthor":false,"prefix":"","firstName":"M.","middleName":"","lastName":"Prabhakar","suffix":""}],"badges":[],"createdAt":"2025-08-06 08:53:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7307700/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7307700/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91810239,"identity":"fb0025cf-9088-40cf-ba33-0b9e7c29a011","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7216270,"visible":true,"origin":"","legend":"","description":"","filename":"MLonly692025.docx","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/71acc91658b00f54dc1a833b.docx"},{"id":91810225,"identity":"ea8a83fa-f5cf-40fb-bb71-baa38b3ab889","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1391748,"visible":true,"origin":"","legend":"","description":"","filename":"RevisedFinalfigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/2cd04dfd0caa72e8affb013f.docx"},{"id":91810237,"identity":"a3e9cfd3-5aa1-4dcc-a178-75926f77f48a","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"json","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9741,"visible":true,"origin":"","legend":"","description":"","filename":"eb75ab95b1c14174974d9b2c77f07831.json","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/e8be43ffbab65131fb2cd469.json"},{"id":91810528,"identity":"99ef7bed-b449-4307-b398-2fa300f310af","added_by":"auto","created_at":"2025-09-22 04:23:17","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":40214,"visible":true,"origin":"","legend":"","description":"","filename":"Suplimentaryfile.docx","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/3a7781a178f51fa40d428201.docx"},{"id":91810229,"identity":"8f288c44-e436-4497-9990-117fdb221a70","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"xml","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":93546,"visible":true,"origin":"","legend":"","description":"","filename":"eb75ab95b1c14174974d9b2c77f078311enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/96832a8344e81c4e900b00cb.xml"},{"id":91810210,"identity":"4c12a4a7-d1a6-4e82-aeea-9b37bc2dbea8","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"eps","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":772,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage10.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/b2f10b28531ade4beac3e02b.eps"},{"id":91810244,"identity":"37c58b0e-c7da-4f65-9fea-a1e531905c52","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"eps","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":68429,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage11.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/2b79fcc911ff03e5c6feb30d.eps"},{"id":91810216,"identity":"4ca76107-2d48-4406-a0cb-d5c111f1a7c8","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"eps","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":781,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage12.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/c48e928e37adc1422d0949e6.eps"},{"id":91810543,"identity":"0ce332b2-0481-4d84-9864-9ff68681cb1f","added_by":"auto","created_at":"2025-09-22 04:23:18","extension":"eps","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":53315,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage13.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/a06222cb27fe4d40696a5182.eps"},{"id":91810529,"identity":"07e908ae-36b3-4869-b23c-2b6f64b530ef","added_by":"auto","created_at":"2025-09-22 04:23:17","extension":"eps","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":781,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage12.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/b17fe8a0677816f568f454c9.eps"},{"id":91810215,"identity":"b6fd471d-e191-411f-823d-ce9b8226d4e2","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"eps","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":59791,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage15.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/0fdf8fdc8a6314dcab73bb84.eps"},{"id":91810548,"identity":"6e6057c5-781c-4761-803e-df79ddb13286","added_by":"auto","created_at":"2025-09-22 04:23:19","extension":"eps","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":913,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage16.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/49f9ae61343825e27dbb6783.eps"},{"id":91810854,"identity":"32d42d92-c8c8-4de5-8b92-df419555e388","added_by":"auto","created_at":"2025-09-22 04:31:19","extension":"eps","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":77526,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage17.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/1778f4da52832b9470c6725c.eps"},{"id":91810213,"identity":"6ac3b01b-63ca-47a6-a826-ac25beae4268","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"eps","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":78820,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage2.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/4769797ba67cf8390f50f526.eps"},{"id":91810222,"identity":"4fcfd3fa-9b15-4f87-96d3-11d623ae9b34","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"eps","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":68922,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage3.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/599f050501201c083700cdce.eps"},{"id":91810249,"identity":"077b854b-2d7e-4ecf-a9bb-0f017205e4cd","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"eps","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":57565,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage4.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/533468dd7cbf663e96918e80.eps"},{"id":91810540,"identity":"dfd629fa-a9e2-4f5f-9a5f-812d4dab8b31","added_by":"auto","created_at":"2025-09-22 04:23:18","extension":"eps","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":60557,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage5.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/e3efd3cf6f23a4470ce754bf.eps"},{"id":91810220,"identity":"a18b848a-3063-4380-9c00-2d853145ee43","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"eps","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":60246,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage6.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/4173acef44891d9474f7f2a9.eps"},{"id":91811100,"identity":"e2475659-0e7e-45bb-9ac9-95f0c9ca6ae1","added_by":"auto","created_at":"2025-09-22 04:39:18","extension":"eps","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":59992,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage7.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/0fc90b4cf1c6a47c7a37815c.eps"},{"id":91810211,"identity":"32f36d26-0a4d-4955-8663-6771663e8b4f","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"eps","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":64529,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage8.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/b900bd46828bd0738bd1584d.eps"},{"id":91810221,"identity":"00ec7b3d-06c4-42e8-a6d0-3d4751238c87","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"eps","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":772,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage10.eps","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/a1c3dfc8543875ea505c188a.eps"},{"id":91810227,"identity":"4979eb40-9c7b-4f43-8130-14dc2cf21259","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":693511,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/96a3abdf9d7bd3a04cb5144f.png"},{"id":91810538,"identity":"f2133e7e-c533-4b35-8ea7-318a116d4264","added_by":"auto","created_at":"2025-09-22 04:23:18","extension":"jpeg","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":387270,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage10.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/eb37b52127e744d6bb9e32c1.jpeg"},{"id":91810217,"identity":"ad0444f9-3935-4dcd-b08e-fef10d8b41ba","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"jpeg","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":352113,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage11.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/475daf215bd6e4636a80f830.jpeg"},{"id":91810248,"identity":"16d5fdd7-ee80-4494-bec0-9400e1bc241d","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"jpeg","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":401009,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage12.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/3314ecfcf59e10ff46f94ad0.jpeg"},{"id":91810233,"identity":"82c1ca22-6a9c-48e4-8c17-229fdd99589b","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":457908,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/f2cfee69faa4ebff9fbd6877.png"},{"id":91810251,"identity":"06426f85-689d-402f-9984-b01b5031ad85","added_by":"auto","created_at":"2025-09-22 04:15:19","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1986636,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/13da52b4f911f0f03295a4d4.png"},{"id":91810247,"identity":"0a854717-9905-4738-a839-ec299d2dd299","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"jpeg","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2890,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/3192bc0faa598d1cc39d305f.jpeg"},{"id":91810533,"identity":"5861d078-b3ce-41bb-9762-a8a183e3f0d6","added_by":"auto","created_at":"2025-09-22 04:23:18","extension":"png","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":353204,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/77a0330c5247cb4dbb8cb718.png"},{"id":91810235,"identity":"019eb054-9dc6-4c35-a478-67b22e30e032","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"jpeg","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":832495,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/66e751dd61a12049b2fc543e.jpeg"},{"id":91810254,"identity":"dab35ab6-fbd7-4901-846e-c6b69463c6d3","added_by":"auto","created_at":"2025-09-22 04:15:19","extension":"jpeg","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":706051,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/8b97678fd7f2a83d30ee0b94.jpeg"},{"id":91810261,"identity":"201ffc8f-48da-4ac9-9708-d57cd33e8974","added_by":"auto","created_at":"2025-09-22 04:15:19","extension":"jpeg","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":560703,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/9543574389678c34859365af.jpeg"},{"id":91810550,"identity":"e50c74c0-739b-410c-9e17-fbc4e540ff2d","added_by":"auto","created_at":"2025-09-22 04:23:19","extension":"jpeg","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":524079,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/0e548e57080b9c808737714f.jpeg"},{"id":91810242,"identity":"f478249d-e5e8-4522-9bcc-12b232af12a7","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"jpeg","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":113403,"visible":true,"origin":"","legend":"","description":"","filename":"groupimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/aa380a8f9602c7cabe956af6.jpeg"},{"id":91810223,"identity":"520e5d27-7aae-4986-9ab3-3e81bb0428ca","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"png","order_by":35,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":143335,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/70abdc3b94774b1686c76b24.png"},{"id":91810852,"identity":"a6eb95b7-0a6d-43ba-9d24-1fb6859730bd","added_by":"auto","created_at":"2025-09-22 04:31:18","extension":"png","order_by":36,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":61893,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/ba971d4adf9218495b51688e.png"},{"id":91810544,"identity":"53c8a3e5-f9b3-4e8b-b95a-7f576b2d3f5e","added_by":"auto","created_at":"2025-09-22 04:23:19","extension":"png","order_by":37,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":54267,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage11.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/fe12d1d05cc4b35828d2a7f6.png"},{"id":91810245,"identity":"641f36bf-56ae-4347-b2ff-a3c77f8476fc","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"png","order_by":38,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":65559,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage12.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/259cd2b6c4dbb30e5bb598e1.png"},{"id":91810258,"identity":"a50eeb1c-5562-4205-a20e-024f7b0fd90b","added_by":"auto","created_at":"2025-09-22 04:15:19","extension":"png","order_by":39,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":96328,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/51de3ebe0f9ec125aab8a9c2.png"},{"id":91810231,"identity":"d6568763-7e79-41fb-8018-d9c912751480","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"png","order_by":40,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":394117,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/47419e2848c9212e74c3371f.png"},{"id":91810240,"identity":"682279df-762f-46e0-83a7-02fa74d09990","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"png","order_by":41,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1056,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/f430c868be06a056c499551e.png"},{"id":91810535,"identity":"873e0dbb-8889-4ae1-b6fc-fbf4a04299c5","added_by":"auto","created_at":"2025-09-22 04:23:18","extension":"png","order_by":42,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":94550,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/3a2da898832a992784c0a7f0.png"},{"id":91810255,"identity":"5a7fe34b-824d-4175-b43f-7e6fe284cb24","added_by":"auto","created_at":"2025-09-22 04:15:19","extension":"png","order_by":43,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":135956,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/91e8d9295cfa517790051244.png"},{"id":91810537,"identity":"e608baab-0763-4380-b846-c582c938267a","added_by":"auto","created_at":"2025-09-22 04:23:18","extension":"png","order_by":44,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":150657,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/2c20c554f3fddf9ff967906d.png"},{"id":91810234,"identity":"ddbda99c-74d4-443c-b965-b79658c0c06d","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"png","order_by":45,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":119786,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/3dd0b90daf0033809aa871ac.png"},{"id":91810541,"identity":"21e661e9-66a0-424b-a314-7f8be6050163","added_by":"auto","created_at":"2025-09-22 04:23:18","extension":"png","order_by":46,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":107603,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/e96dc1cce3137493ed69f5e2.png"},{"id":91810253,"identity":"bbeb4d66-1b7a-4831-8cd5-672b11429efe","added_by":"auto","created_at":"2025-09-22 04:15:19","extension":"png","order_by":47,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":82082,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinegroupimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/aa08ebee31273562871a3452.png"},{"id":91810552,"identity":"57430667-126f-40cd-8d1a-4f73e7ea5ab6","added_by":"auto","created_at":"2025-09-22 04:23:19","extension":"xml","order_by":48,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":89147,"visible":true,"origin":"","legend":"","description":"","filename":"eb75ab95b1c14174974d9b2c77f078311structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/4b05c71631b6e9ca4e880b37.xml"},{"id":91810856,"identity":"cf42c617-c3ec-4055-9679-13a7e77265e1","added_by":"auto","created_at":"2025-09-22 04:31:19","extension":"html","order_by":49,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":107941,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/28aaf770702d5ae9638c4a6d.html"},{"id":91810218,"identity":"4ed5f4e4-c557-464b-affa-c3bb9b8e2d2f","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1735474,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ea. On-farm Production technology of biochar\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb. On-farm maize residue preparation\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/8a2723b0c89e8b39b56bf52b.png"},{"id":91810208,"identity":"448bcfd9-e6c5-4174-bd5e-bed184154740","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":490925,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFig.1\u003c/strong\u003e: Correlation heatmap among system productivity and soil parameters.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/f4977fe12d158ced1fb126d4.png"},{"id":91810209,"identity":"34333246-208a-4e6f-9cdb-1326d366096a","added_by":"auto","created_at":"2025-09-22 04:15:17","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":275913,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFig. 2:\u003c/strong\u003e Soil organic carbon influenced by various biological properties under different tillage practices (minimum and conventional tillage) and row ratios under varying residue management practices in pigeonpea + maize intercropping system\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/39d45fc6ede1e5d5e3f21a88.png"},{"id":91810238,"identity":"0ee37d21-17d5-47eb-a744-c04bf88cc5a4","added_by":"auto","created_at":"2025-09-22 04:15:18","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":548301,"visible":true,"origin":"","legend":"\u003cp\u003eFig. 3a. Partial Dependence Plot depicting the effect of soil physical, chemical and biological properties on the pigeonpea + maize intercropping system productivity (Mg ha⁻¹ yr⁻¹) under different tillage and residue management practices.\u003c/p\u003e\n\u003cp\u003eFig. 3b. Partial Dependence Plot depicting the effect of soil physical, chemical and biological properties on the pigeonpea + maize intercropping Carbon Sequestration (Mg ha⁻¹ yr⁻¹.) under different tillage and residue management practices.\u003c/p\u003e\n\u003cp\u003eFig.3c. Partial Dependence Plot depicting the effect of soil physical, chemical and biological properties on pigeonpea + maize intercropping Carbon Footprint (Mg CO₂-C ha⁻¹) under different tillage and residue management practices.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/0eda0ca3ad5d9332a8ab7d8d.png"},{"id":91810530,"identity":"dda6ae79-fde6-4367-8779-8abd864a0db5","added_by":"auto","created_at":"2025-09-22 04:23:17","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":330329,"visible":true,"origin":"","legend":"\u003cp\u003eFig. 4a. \u0026nbsp;Sensitiveness of pigeonpea + maize intercropping system productivity (Mg ha⁻¹ yr⁻¹) for changes in soil physical, chemical and biological properties under different tillage and residue management practices by using sensitive analysis.\u003c/p\u003e\n\u003cp\u003eFig. 4b. Sensitiveness of Carbon Sequestration (Mg ha⁻¹ yr⁻¹) to the changes in soil physical, chemical and biological properties in pigeonpea + maize intercropping under different tillage and residue management practices by using sensitive analysis\u003c/p\u003e\n\u003cp\u003eFig. 4c. Sensitiveness of Carbon Footprint (Mg CO₂-C ha⁻¹) to the changes in soil physical, chemical and biological properties in the pigeonpea + maize intercropping under different tillage and residue management practices\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/c55decfb951fad6da137572d.png"},{"id":94068791,"identity":"d904a113-972c-4edc-b478-e2433ffb4bbd","added_by":"auto","created_at":"2025-10-22 08:11:20","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3903460,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/f86b6636-d8dd-40e4-84dc-bd161aa3d203.pdf"},{"id":91810206,"identity":"31792de0-9f4d-4e47-847d-295b0f4fdc51","added_by":"auto","created_at":"2025-09-22 04:15:16","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":40214,"visible":true,"origin":"","legend":"","description":"","filename":"Suplimentaryfile.docx","url":"https://assets-eu.researchsquare.com/files/rs-7307700/v1/f65b15581fca9ab0a0755c5c.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine Learning Driven Optimization of Carbon Sequestration in Intercropping Systems Using XGBoost Modeling and Partial Dependence Analysis","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eClimate change mitigation through agricultural carbon sequestration has emerged as a critical strategy for achieving global sustainability goals while maintaining food security (Lal, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Smith et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Intercropping systems, characterized by the simultaneous cultivation of multiple crop species, offer substantial potential for enhancing soil organic carbon (SOC) storage compared to monoculture practices (Li et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zhang et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The integration of leguminous crops like pigeonpea with cereals such as maize creates synergistic effects that can significantly improve carbon sequestration rates through enhanced root biomass production, nitrogen fixation, and soil microbial activity (Kumar et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Sharma et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eTraditional agricultural management practices, including tillage operations and residue management, profoundly influence carbon dynamics in intercropping systems (Chen et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Minimum tillage practices have been shown to preserve soil structure and enhance carbon retention, while biochar application can provide long-term carbon storage benefits (Lehmann \u0026amp; Joseph, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Zhou et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). However, the complex interactions between these management factors and their collective impact on carbon sequestration remain poorly understood, necessitating advanced analytical approaches.\u003c/p\u003e\u003cp\u003eMachine learning techniques have revolutionized agricultural research by enabling the analysis of complex, non-linear relationships in ecological systems (Rashid et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Extreme Gradient Boosting (XGBoost) has demonstrated superior performance in predicting agricultural outcomes due to its ability to handle feature interactions and non-linear patterns (Liu et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Partial Dependence Plots (PDPs) complement machine learning models by providing interpretable insights into individual feature effects, making them particularly valuable for agricultural decision-making (Molnar, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Thompson et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eDespite growing interest in machine learning applications for sustainable agriculture, limited research has integrated XGBoost modeling with PDP analysis to optimize carbon sequestration in intercropping systems. The complex interactions between tillage practices, cropping patterns, and residue management require sophisticated analytical frameworks to identify optimal management strategies. Furthermore, understanding the sensitivity of carbon sequestration to key soil parameters is essential for developing targeted interventions that maximize climate benefits while maintaining productivity.\u003c/p\u003e\u003cp\u003eThis study addresses these knowledge gaps by employing XGBoost modeling coupled with Partial Dependence Analysis to optimize carbon sequestration in intercropping systems. The research objectives were to: (1) evaluate the performance of machine learning models in predicting carbon sequestration, system productivity, and carbon footprint; (2) identify key drivers of carbon sequestration through sensitivity analysis; and (3) quantify the individual effects of soil parameters on system outcomes using Partial Dependence Plots. The findings contribute to evidence-based agricultural management strategies that enhance both productivity and climate mitigation potential.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"2. Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003e2.7 Machine Learning processes\u003c/h2\u003e\n \u003cp\u003eMachine learning (ML) models were adopted to study the dynamics of systems productivity, CSQ and CF based on soil parameters under pigeonpea\u0026thinsp;+\u0026thinsp;maize inter cropping system. All the data related to soil was analyzed as per mentioned protocols in (Supplementary file, ST-1) used in its initial form to build ML models in this study. Eight ML models were separately trained and evaluated to predict each CSQ, CF, and SP using soil parameters as input. These models included linear approaches viz., Ridge Regression, LASSO Regression, and Elastic Net, alongside nonlinear techniques viz., Random Forest, Support Vector Regression, Extreme Gradient Boosting, Artificial Neural Network, and Decision Tree. In order to avoid the overfitting issues, the soil parameters that have a considerable predictive effect on the respective response variable were selected using partial least square regression (PLSR) analysis. The soil parameters with a Variable Importance in Projection (VIP) score greater than 0.9 for a given response were selected to model the respective response (Supplementary file, ST-3 \u0026amp; SF- 1). To ensure robustness, 80% of the dataset was used for training, while the remaining 20% was reserved for validation.\u003c/p\u003e\n \u003cdiv id=\"Sec4\" class=\"Section3\"\u003e\n \u003ch2\u003e2.7.1 Performance metrics used for the ML models\u003c/h2\u003e\n \u003cp\u003eCoefficient of determination (R\u003csup\u003e2\u003c/sup\u003e) or Explained Variance score and other metrics of Errors \u003cem\u003eviz\u003c/em\u003e., Root Mean Squared Error, Mean Absolute Error, Mean Absolute Percentage Error, and Mean Absolute Bias Error. The best-performing model was identified based on the highest R\u0026sup2; and the lowest error metrics for both training and validation datasets. The selection of the most suitable model also provided insights into the response behavior of SP, CSQ and CF for change in soil parameters. If a linear model such as Ridge, LASSO, or ENET demonstrated superior performance, it suggested a proportional effect of soil properties on the response variables across their entire range. Conversely, if a nonlinear model like RF, SVR, XGBoost, ANN, or DT exhibited the best predictive capability, it indicated complex interactions, where input variables influenced the response in a non-linear manner.\u003c/p\u003e\n \u003cp\u003eTo further interpret the model outputs, PDPs and SA were generated using the best-fitted model, providing a visual representation of how variations in specific soil parameters influenced CSQ, CF, and yield while keeping other variables constant. The PDP is generated as discussed below (Zhang et al., 2018),\u003c/p\u003e\n \u003col style=\"list-style-type: lower-roman;\"\u003e\n \u003cli\u003eLet there are \u0026lsquo;k\u0026rsquo; soil parameters and \u0026lsquo;n\u0026rsquo; observations of each such that j=1, 2\u0026hellip;n. Let the X\u003csub\u003ei\u003c/sub\u003e is the parameter of interest.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe values of other (k-1) parameters are replaced with their average values for all the n observations.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eThe response values are predicted through the best-fitted model using the new dataset.\u003c/li\u003e\n \u003cli\u003ePlot the predicted values against the levels of the parameter of interest X\u003csub\u003ei.\u003c/sub\u003e\u003c/li\u003e\n \u003c/ol\u003e\n \u003cp\u003eThese PDP facilitates a deeper understanding of the underlying relationships between soil health indicators and key agricultural sustainability metrics (Friedman, 2001).\u003c/p\u003e\n \u003cp\u003eThe sensitivity plots are generated by predicting CSQ, CF, and yield through the best-fitted model for each soil parameter while holding other parameters at the maximum and minimum values.\u003c/p\u003e\n \u003cp\u003eThe details of the metrics, source codes of which can be found in Scikit-learn are explained below:\u003c/p\u003e\n \u003cp\u003eRoot Mean Square Error (RMSE):\u003c/p\u003e\n \u003cdiv id=\"Equa\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e$$\\:\\text{R}\\text{M}\\text{S}\\text{E}=\\:\\sqrt{\\frac{{\\sum\\:}_{i=1}^{N}\\:{\\left({Predicted}_{i}-{Actual}_{i}\\right)}^{2}}{n}}$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003ewhere, N is the total number of observations.\u003c/p\u003e\n \u003cp\u003eMean Absolute Error (MAE):\u003c/p\u003e\n \u003cdiv id=\"Equb\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e$$\\:\\text{M}\\text{A}\\text{E}=\\left(\\frac{1}{n}\\right)\\text{*}\\sum\\:\\left|{Actual}_{i}-{\\:Predicted}_{i}\\right|\\:$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003ewhere, n is the total number of observation and \u0026lsquo;i\u0026rsquo; is the i\u003csup\u003eth\u003c/sup\u003e observation.\u003c/p\u003e\n \u003cp\u003eCoefficient of determination or R\u003csup\u003e2\u003c/sup\u003e:\u003c/p\u003e\n \u003cdiv id=\"Equc\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e$$\\:{\\text{R}}^{2}=\\:\\:\\:\\left(1-\\frac{SSR}{SST}\\right)$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003ewhere, SSR is the sum of squared of the residual errors and SST is the total sum of the errors.\u003c/p\u003e\n \u003cp\u003eMean Absolute Percentage Error (MAPE):\u003c/p\u003e\n \u003cdiv id=\"Equd\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e$$\\:\\:\\:\\:\\text{M}\\text{A}\\text{P}\\text{E}=\\:\\left(\\frac{1}{n}\\right)\\:*\\:\\sum\\:\\left(\\frac{\\left|Actual-Predicted\\right|}{\\left|Actual\\right|}\\right)\\:*100$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003ewhere, n is the total number of observations.\u003c/p\u003e\n \u003cp\u003eMean Absolute Bias Error (MABE) = \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\frac{1}{n}\\sum\\:\\left|Predicted-Actual\\right|\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003e2.8 Statistical analysis\u003c/h2\u003e\n \u003cp\u003eThe carbon footprint (CF) was calculated based on the average data of carbon sources and sinks, along with their respective carbon equivalents. The ML models, PDP and sensitivity plots are generated using the R software version 4.4.3 (R Core Team, 2024)\u003c/p\u003e\n\u003c/div\u003e"},{"header":"3. Results","content":"\u003cp\u003e\u003cstrong\u003e3.4. Machine Learning Model Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4.1. Partial Dependence Plot Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSystem Productivity\u003c/strong\u003e: Available nitrogen showed a strong positive relationship with productivity, which increased from 3.0 to 3.4 Mg ha⁻\u0026sup1; y⁻\u0026sup1; as nitrogen levels rose from 150 to 300 kg ha⁻\u0026sup1; (Fig. 3a). Available potassium exhibited a positive but moderate influence, with productivity rising from 3.32 to 3.38 Mg ha⁻\u0026sup1; y⁻\u0026sup1; as potassium increased from 150 to 400 kg ha⁻\u0026sup1;. In contrast, phosphorus displayed an inverse relationship with productivity. SOC showed a negative effect on productivity when isolated, with yields decreasing from 3.8 to 3.2 Mg ha⁻\u0026sup1; y⁻\u0026sup1; as SOC rose from 0.425% to 0.525%. Smaller water-stable aggregates (\u0026lt;0.25 mm) enhanced productivity from 3.0 to 3.6 Mg ha⁻\u0026sup1; y⁻\u0026sup1; as their percentage increased from 4% to 12%, whereas larger aggregates (0.25-2 mm) reduced productivity from 3.5 to 3.3 Mg ha⁻\u0026sup1; y⁻\u0026sup1;. Field capacity demonstrated a negative relationship with productivity, while bulk density (BD) showed a positive trend. Bacterial abundance was negatively correlated with productivity, while actinomycetes demonstrated a positive effect. Dehydrogenase activity exhibited a non-linear relationship with productivity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCarbon Sequestration\u003c/strong\u003e: SOC exhibited clear threshold increases in sequestration capacity, rising from 2.5 to 4.0 Mg C ha⁻\u0026sup1; as SOC increased from 0.425% to 0.525% (Fig. 3b). BD demonstrated a notable increase in sequestration from 3.2 to 3.9 Mg C ha⁻\u0026sup1; when density reached 0.125 g cm⁻\u0026sup3;. Bacterial abundance showed a positive effect on sequestration, which rose from 3.45 to 3.85 Mg C ha⁻\u0026sup1; as populations increased from 30 to 65 CFU g⁻\u0026sup1; soil. Actinomycetes exhibited a negative trend, with sequestration decreasing from 4.2 to 3.7 Mg C ha⁻\u0026sup1; as their counts increased. Fungi showed stepwise increases in sequestration from 3.7 to 4.0 Mg C ha⁻\u0026sup1; as their abundance rose. Enzyme activities showed varying relationships with sequestration. Dehydrogenase activity remained relatively stable until exceeding 15 \u0026mu;g TPF g⁻\u0026sup1; soil 24 hr⁻\u0026sup1;, beyond which sequestration sharply increased. Microbial biomass carbon exhibited a dramatic threshold response, with sequestration sharply rising from 3.7 to 4.0 Mg C ha⁻\u0026sup1; when biomass exceeded 500 \u0026mu;g C g⁻\u0026sup1; soil.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCarbon Footprint\u003c/strong\u003e: The isolated effect of pH was substantial, with CF becoming less favorable (from -1.9 to -1.0 Mg CO₂-Ce ha⁻\u0026sup1;) as pH increased from 7.5 to 9.0 (Fig. 3c). SOC showed a strong pattern, with carbon footprint improving (more negative) from -0.8 to -2.0 Mg CO₂-Ce ha⁻\u0026sup1; as SOC increased from 0.425% to 0.5%. BD independently improved the carbon footprint from 0 to -2.5 Mg CO₂-Ce ha⁻\u0026sup1; as density increased from 0.11 to 0.14 g cm⁻\u0026sup3;. Soil moisture showed a U-shaped curve with the lowest carbon footprint (-1.9 Mg CO₂-Ce ha⁻\u0026sup1;) at approximately 0.05% moisture content. Among microbial indicators, actinomycetes showed worsening carbon footprint as their abundance increased, while fungi displayed a U-shaped relationship. Urease activity was associated with an improved carbon footprint, decreasing from 0 to -2.0 Mg CO₂-Ce ha⁻\u0026sup1; as activity increased. Microbial biomass carbon showed a strong inverse correlation with carbon footprint.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4.2. Sensitivity Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSensitivity analysis revealed distinct patterns in how soil and microbial parameters influenced system productivity, carbon footprint, and carbon sequestration (Fig. 4). System productivity showed the greatest sensitivity to carbon footprint parameters, with maximum values reaching ~5.5 Mg ha⁻\u0026sup1; y⁻\u0026sup1; during carbon sequestration phases (Fig. 4a). Carbon sequestration was most significantly influenced by SOC content, with maximum values approaching 3.8 Mg C ha⁻\u0026sup1; (Fig. 4b). Bulk density was associated with the lowest sequestration values (1.5 Mg C ha⁻\u0026sup1;). Narrow gaps between maximum and minimum values for acid phosphatase indicated a consistently strong effect on sequestration regardless of external variability. Carbon footprint remained negative across most parameters, reflecting net sequestration potential in the system (Fig. 4c). Alkaline phosphatase had the most positive influence on carbon dynamics, while SOC and urease activity exhibited the most negative influence, with footprint values reaching as low as -2.5 Mg CO₂-Ce ha⁻\u0026sup1;.\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003e\u003cstrong\u003e4.4. Machine Learning Insights into Soil-Plant-Microbial Interactions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4.1. Partial Dependence Relationships\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe analysis of soil-property interactions using partial dependence plots provided valuable insights into the complex relationships between soil properties and agricultural sustainability metrics. Soil physical properties, particularly water-stable aggregates, showed significant influence on key outcomes. Improved soil structure resulted in yield increases, enhanced carbon sequestration, and reduced carbon footprint, aligning with Bronick and Lal (2005), who emphasized that soil structure governs key processes such as water infiltration, aeration, and microbial habitat provision. The relationship between bulk density and agricultural outcomes suggested that moderate compaction can enhance root-soil contact while preserving adequate pore space, consistent with Reynolds et al. (2009), who highlighted optimal physical quality ranges for soil productivity. Field capacity and soil moisture dynamics underscored the importance of water management in influencing both crop performance and carbon cycling, supporting findings by Rawls et al. (2003). Among soil chemical properties, available nitrogen demonstrated a strong positive correlation with productivity, affirming Robertson and Vitousek\u0026apos;s (2009) conclusion on its foundational role in crop systems. Soil pH emerged as a significant regulator of carbon footprint, aligning with Lal (2004), who noted the role of pH in moderating greenhouse gas emissions through microbial-mediated processes. SOC proved to be the most influential property, driving increases in yield, reductions in carbon footprint, and improvements in carbon sequestration. These results validate Lal\u0026apos;s (2016) assertion that managing SOC offers a dual solution for boosting productivity while mitigating climate change impacts. The study also highlighted the pivotal role of soil biological processes, with microbial populations and enzyme activities showing distinct relationships with agricultural outcomes. These findings agree with Bardgett and van der Putten (2014), who emphasized the essential contributions of soil biota to ecosystem functions and Burns et al. (2013), who documented threshold effects in enzyme activities.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4.2. Sensitivity Analysis Implications\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur sensitivity analysis revealed significant variations in agricultural ecosystem metrics in response to changes in soil and microbial parameters. System productivity demonstrated remarkable sensitivity to carbon dynamics, with up to 187% variation between minimum and maximum values when influenced by carbon footprint parameters. This significant range highlights the potential for climate-friendly management practices to deliver dual benefits: optimizing agronomic outputs while achieving environmental sustainability (Lal, 2020). Carbon sequestration potential showed a 98% increase in response to changes in SOC content, revealing powerful positive feedback mechanisms that can accelerate carbon capture in well-managed semi-arid soils. This self-reinforcing relationship provides tremendous opportunity for climate change mitigation through enhanced agricultural carbon storage (Paustian et al., 2016). The carbon footprint analysis revealed predominantly negative values, indicating the potential for agricultural systems to act as carbon sinks rather than sources. Even modest increases in SOC from 0.40% to 0.50% resulted in a 136% improvement in carbon footprint. This non-linear relationship emphasizes how minimal investments in enhancing soil carbon can yield disproportionately large environmental benefits, providing robust support for policies incentivizing soil carbon enhancement as a climate mitigation strategy (Smith et al., 2019). The identification of threshold effects, particularly step increases in carbon sequestration associated with specific parameter thresholds, suggests that targeted interventions could yield disproportionate environmental benefits. Increasing soil microbial biomass carbon beyond 500 \u0026mu;g C g⁻\u0026sup1; soil triggered an 8% jump in sequestration, highlighting the potential value of management practices that enhance soil microbial abundance and activity.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study successfully demonstrated the effectiveness of machine learning approaches, particularly XGBoost modeling coupled with Partial Dependence Analysis, for optimizing carbon sequestration in intercropping systems. The key findings provide valuable insights for sustainable agricultural management and climate change mitigation strategies. The XGBoost model achieved superior performance in predicting carbon sequestration (CSQ), while Artificial Neural Networks excelled in forecasting system productivity (SP) and carbon footprint (CF). These results confirm the presence of complex, non-linear relationships within intercropping systems that require advanced analytical techniques for accurate prediction. The high predictive accuracy of these models validates their potential for real-world agricultural decision support systems. Machine learning analyses revealed complex relationships between soil parameters and sustainability metrics, with SOC emerging as the key driver of both carbon sequestration and reduced carbon footprint. Sensitivity analysis identified threshold effects and quantified trade-offs, providing valuable guidance for precision agriculture approaches that optimize both productivity and environmental services.\u003c/p\u003e\u003cp\u003eFuture research should focus on validating these models across different geographical regions and extending the analysis to include economic considerations. Long-term studies examining the temporal dynamics of carbon sequestration under varying climate scenarios would further enhance the practical applicability of these findings. Additionally, integration of remote sensing data could expand the spatial scale of model applications. This research contributes significantly to the growing body of knowledge on sustainable agriculture and climate-smart farming practices. The demonstrated effectiveness of machine learning approaches for optimizing carbon sequestration provides a foundation for developing evidence-based policies and management recommendations that address both food security and climate change challenges.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;No funding is available.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data is available upon request\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCRediT authorship contribution statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eNallagatla Vinod Kumar\u003c/strong\u003e: Conceptualization, Formal analysis, Investigation, Methodology, Data curation, Writing -original draft, Writing - review \u0026amp; editing. \u003cstrong\u003eGajanan Sawargaonkar, C. Sudharani\u003c/strong\u003e: Conceptualization, Methodology, Resources, Supervision, Writing - review \u0026amp; editing. \u003cstrong\u003eT. Ram Prakash, S. Triveni, Ch. Sarada\u003c/strong\u003e: Methodology, Validation, Writing - review \u0026amp; editing.\u0026nbsp;\u003cstrong\u003eAjith S\u003c/strong\u003e\u003cstrong\u003e,\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ePeace Raising and M. Prabhakar\u003c/strong\u003e: Formal analysis, Methodology, Visualization, Writing -review \u0026amp; editing.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe acknowledge the ICRISAT and PJTSAU for providing experimental field and laboratory facilities during the experimentation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial registration:\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics and Consent to Participate declarations:\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author declares that there is no confict of interests regarding the publication of this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData will be made available on request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to Participate and Consent to Publish\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInformed consent to participate and consent to publish, including the use of personal information and photographs (Fig. 1), were obtained from all participants. No participants under the age of 18 years were involved in this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eBardgett, R. D., \u0026amp; van der Putten, W. H. (2014). Belowground biodiversity and ecosystem functioning. Nature, 515, 505-511.\u003c/li\u003e\n \u003cli\u003eBronick, C. J., \u0026amp; Lal, R. (2005). Soil structure and management: A review. Geoderma, 124, 3-22.\u003c/li\u003e\n \u003cli\u003eBurns, R. G., DeForest, J. L., Marxsen, J., Sinsabaugh, R. L., Stromberger, M. E., Wallenstein, M. D., ... \u0026amp; Zoppini, A. (2013). Soil enzymes in a changing environment: Current knowledge and future directions. Soil Biology and Biochemistry, 58, 216-234.\u003c/li\u003e\n \u003cli\u003eChen, X., Liu, Y., Wang, Z., \u0026amp; Zhang, H. (2022). Tillage effects on soil carbon dynamics in intercropping systems: A meta-analysis. Soil and Tillage Research, 218, 105-118.\u003c/li\u003e\n \u003cli\u003eChen, Y., Li, M., Zhang, H., Wang, L., \u0026amp; Liu, S. (2022). Effects of tillage and residue management on soil carbon dynamics in intercropping systems: A meta-analysis. Agriculture, Ecosystems \u0026amp; Environment, 327, 107825.\u003c/li\u003e\n \u003cli\u003eFriedman, J.H., 2001. Greedy function approximation: a gradient boosting machine.\u0026nbsp;Annals of statistics, pp.1189-1232.\u003c/li\u003e\n \u003cli\u003eKumar, A., Singh, P., Patel, R., \u0026amp; Gupta, S. (2021). Nitrogen fixation and carbon sequestration in legume-cereal intercropping systems. Agriculture, Ecosystems \u0026amp; Environment, 312, 107-122.\u003c/li\u003e\n \u003cli\u003eKumar, A., Singh, R., Patel, N., \u0026amp; Sharma, K. (2021). Nitrogen fixation and carbon sequestration in legume-cereal intercropping systems: Mechanisms and management strategies. Field Crops Research, 276, 108371.\u003c/li\u003e\n \u003cli\u003eLal, R. (2004). Soil carbon sequestration impacts on global climate change and food security. Science, 304, 1623-1627.\u003c/li\u003e\n \u003cli\u003eLal, R. (2016). Soil health and carbon management. Food and Energy Security, 5, 212-222.\u003c/li\u003e\n \u003cli\u003eLal, R. (2020). Soil carbon sequestration to mitigate climate change. Geoderma, 123, 1-22.\u003c/li\u003e\n \u003cli\u003eLal, R. (2020). Soil organic carbon sequestration impacts on global climate change and food security. Science, 304(5677), 1623-1627.\u003c/li\u003e\n \u003cli\u003eLehmann, J., \u0026amp; Joseph, S. (2015). Biochar for environmental management: Science, technology and implementation (2nd ed.). Routledge.\u003c/li\u003e\n \u003cli\u003eLehmann, J., \u0026amp; Joseph, S. (2015). Biochar for environmental management: Science, technology and implementation (2nd ed.). Routledge.\u003c/li\u003e\n \u003cli\u003eLi, M., Chen, Q., Zhang, L., \u0026amp; Wang, J. (2020). Intercropping enhances soil organic carbon storage through improved root systems. Field Crops Research, 245, 107-115.\u003c/li\u003e\n \u003cli\u003eLi, W., Zhang, P., Ye, J., Li, L., \u0026amp; Baker, P. A. (2020). Impact of two different types of rural land use change on soil organic carbon in a semi-arid region. Plant and Soil, 452(1), 149-166.\u003c/li\u003e\n \u003cli\u003eLiu, H., Wang, S., Chen, M., \u0026amp; Li, X. (2023). XGBoost applications in agricultural prediction: A comprehensive review. Computers and Electronics in Agriculture, 198, 106-118.\u003c/li\u003e\n \u003cli\u003eLiu, X., Wang, Y., Chen, M., Zhang, L., \u0026amp; Thompson, R. (2023). XGBoost applications in agricultural systems: A comprehensive review of predictive modeling approaches. Computers and Electronics in Agriculture, 205, 107634.\u003c/li\u003e\n \u003cli\u003eMolnar, C. (2022). Interpretable machine learning: A guide for making black box models explainable (2nd ed.). Leanpub.\u003c/li\u003e\n \u003cli\u003eMolnar, C. (2022). Interpretable machine learning: A guide for making black box models explainable (2nd ed.). Lean Publishing.\u003c/li\u003e\n \u003cli\u003ePaustian, K., Lehmann, J., Ogle, S., Reay, D., Robertson, G. P., \u0026amp; Smith, P. (2016). Climate-smart soils. Nature, 532, 49-57.\u003c/li\u003e\n \u003cli\u003eRashid, M., Ahmed, K., Hassan, N., \u0026amp; Ali, S. (2023). Machine learning revolution in sustainable agriculture: Current trends and future prospects. Precision Agriculture, 24, 892-915.\u003c/li\u003e\n \u003cli\u003eRashid, M., Ahmed, S., Khan, A., \u0026amp; Ali, M. (2023). Machine learning approaches for sustainable agriculture: Applications, challenges, and future perspectives. Agricultural Systems, 208, 103652.\u003c/li\u003e\n \u003cli\u003eRawls, W. J., Pachepsky, Y. A., Ritchie, J. C., Sobecki, T. M., \u0026amp; Bloodworth, H. (2003). Effect of soil organic carbon on soil water retention. Geoderma, 116, 61-76.\u003c/li\u003e\n \u003cli\u003eReynolds, W. D., Bowman, B. T., Drury, C. F., Tan, C. S., \u0026amp; Lu, X. (2002). Indicators of good soil physical quality: Density and storage parameters. Geoderma, 110, 131-146.\u003c/li\u003e\n \u003cli\u003eRobertson, G. P., \u0026amp; Vitousek, P. M. (2009). Nitrogen in agriculture: Balancing the cost of an essential resource. Annual Review of Environment and Resources, 34, 97-125.\u003c/li\u003e\n \u003cli\u003eSharma, P., Kumar, S., Singh, A., \u0026amp; Patel, R. (2023). Intercropping effects on soil microbial communities and carbon cycling: Insights from pigeonpea-maize systems. Soil Biology and Biochemistry, 179, 108986.\u003c/li\u003e\n \u003cli\u003eSharma, V., Patel, K., Reddy, M., \u0026amp; Singh, R. (2023). Synergistic effects of pigeonpea-maize intercropping on soil carbon dynamics. Agroforestry Systems, 97, 445-462.\u003c/li\u003e\n \u003cli\u003eSmith, P., Adams, J., Beerling, D. J., Beringer, T., Calvin, K. V., Fuss, S., ... \u0026amp; Arneth, A. (2019). Land-management options for greenhouse gas removal and their impacts on ecosystem services and the sustainable development goals. Annual Review of Environment and Resources, 44, 255-286.\u003c/li\u003e\n \u003cli\u003eThompson, K., Martinez, L., Brown, J., \u0026amp; Wilson, D. (2023). Partial dependence plots in agricultural machine learning: Interpretation and applications. Precision Agriculture, 24(2), 456-478.\u003c/li\u003e\n \u003cli\u003eThompson, R., Davis, L., \u0026amp; Miller, K. (2023). Interpretable machine learning for agricultural decision support systems. Agricultural Systems, 208, 103-117.\u003c/li\u003e\n \u003cli\u003eWang, H., Liu, Y., Chen, X., Zhang, S., \u0026amp; Lee, K. (2022). Machine learning in precision agriculture: Current applications and future prospects. Smart Agricultural Technology, 3, 100098.\u003c/li\u003e\n \u003cli\u003eWang, T., Liu, G., Zhang, F., \u0026amp; Zhou, Y. (2022). Deep learning applications in crop yield prediction and optimization. Journal of Agricultural Science, 160, 234-248.\u003c/li\u003e\n \u003cli\u003eZhang, Q., Liu, H., Wang, J., Chen, F., \u0026amp; Li, Y. (2022). Carbon sequestration potential of intercropping systems: A global synthesis. Global Change Biology, 28(12), 3652-3668.\u003c/li\u003e\n \u003cli\u003eZhang, W., Liu, K., Wang, H., \u0026amp; Chen, Y. (2022). Carbon sequestration potential of intercropping systems: A global synthesis. Global Change Biology, 28, 1245-1261.\u003c/li\u003e\n \u003cli\u003eZhang, Z., Beck, M. W., Winkler, D. A., Huang, B., Sibanda, W., Goyal, H., 2018. Opening the black box of neural networks: methods for interpreting neural network models in clinical applications.\u0026nbsp;Annals of translational medicine,\u0026nbsp;6(11), 216.\u003c/li\u003e\n \u003cli\u003eZhou, L., Yang, X., Wang, P., \u0026amp; Liu, J. (2021). Biochar amendments for long-term carbon storage in agricultural soils. Biochar, 3, 127-142.\u003c/li\u003e\n \u003cli\u003eZhou, M., Wang, X., Li, P., Zhang, Y., \u0026amp; Chen, L. (2021). Biochar effects on carbon sequestration and greenhouse gas emissions in agricultural soils: A global meta-analysis. Journal of Cleaner Production, 288, 125593.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"System productivity, Carbon sequestration, Carbon footprint, Partial Dependence Plot and Sensitivity","lastPublishedDoi":"10.21203/rs.3.rs-7307700/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7307700/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study integrates XGBoost modeling with Partial Dependence Plots to optimize carbon sequestration. The experiment was conducted using a split-split plot design replicated thrice with the main plot compressing tillage practices \u003cem\u003eviz.\u003c/em\u003e, minimum tillage (M\u003csub\u003e1\u003c/sub\u003e) and conventional tillage (M\u003csub\u003e2\u003c/sub\u003e), subplot consists of row ratios viz. pigeonpea\u0026thinsp;+\u0026thinsp;maize (1:2 ratio) (R\u003csub\u003e1\u003c/sub\u003e), pigeonpea\u0026thinsp;+\u0026thinsp;maize (1:3 ratio) (R\u003csub\u003e2\u003c/sub\u003e), and sole pigeonpea (R\u003csub\u003e3\u003c/sub\u003e) and sole maize (R\u003csub\u003e4\u003c/sub\u003e), sub-subplot consists residue management practices viz. on farm produced biochar application (S\u003csub\u003e1\u003c/sub\u003e), on farm produced residue application (S\u003csub\u003e2\u003c/sub\u003e), and control with no biochar or residue application (S\u003csub\u003e3\u003c/sub\u003e). Machine learning models of XGBoost accurately predicted CSQ while Artificial Neural Network performed best for SP and CF, confirming nonlinear relationships. PDPs showed available nitrogen increased productivity from 3.0 to 3.4 Mg ha⁻\u0026sup1; y⁻\u0026sup1; (150\u0026ndash;300 kg ha⁻\u0026sup1;), while SOC reduced it from 3.8 to 3.2 Mg ha⁻\u0026sup1; y⁻\u0026sup1; (0.425\u0026ndash;0.525%). Sensitivity analysis identified SOC, moisture, and enzyme activities as key drivers, with SOC enhancing sequestration by ~\u0026thinsp;1.5 Mg C ha⁻\u0026sup1; and improving footprint by ~\u0026thinsp;1.2 Mg CO₂-Ce ha⁻\u0026sup1;. These insights enable targeted, efficient soil management for productivity and climate benefits. Machine learning-based Partial Dependence Plot (PDP) and sensitivity analysis (SA) revealed SOC, BD, and microbial parameters as key drivers of these outcomes.\u003c/p\u003e","manuscriptTitle":"Machine Learning Driven Optimization of Carbon Sequestration in Intercropping Systems Using XGBoost Modeling and Partial Dependence Analysis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-22 04:15:11","doi":"10.21203/rs.3.rs-7307700/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e439c18e-723b-41a8-8b73-2384f0517b66","owner":[],"postedDate":"September 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-10-22T08:08:16+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-22 04:15:11","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7307700","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7307700","identity":"rs-7307700","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}