TL;DR
This study quantifies the global potential for soils to sequester additional carbon through the sorption of dissolved organic carbon (DOC) to mineral surfaces. By compiling laboratory experiments and using machine learning, the researchers estimate that soils could potentially store an additional 107 ± 13 Pg C, representing a 7% increase in current soil carbon stocks, with mid- and low-latitude soils showing higher potential.
What Problem Does This Paper Address?
Soil carbon storage is a critical component of the global carbon cycle and a key target for climate change mitigation strategies. While increasing soil carbon is a major goal, understanding the upper limit of this storage capacity is essential for setting realistic policy targets. One particularly stable form of soil carbon is that which sorbs to mineral surfaces. However, a global-scale estimate of this "additional sorption potential" – the capacity for soils to sorb more DOC beyond what is already present – has been lacking. This knowledge gap hinders our ability to fully assess the role of soils in carbon sequestration and to inform effective land management practices.
Methods
The researchers employed a two-pronged approach:
- Data Compilation and Langmuir Modeling: They compiled data from 402 laboratory sorption experiments that measured the adsorption of dissolved organic carbon (DOC) onto mineral soils. They then fitted these data to the Langmuir equation, a standard model for describing adsorption processes. Crucially, they adapted the Langmuir equation to specifically estimate the additional sorption potential (Qsp), accounting for the DOC already present in native soils.
- Machine Learning for Global Prediction: To scale up these laboratory findings to a global level, they trained Random Forest machine learning models.
- Exploration Phase: Initially, they used 17 predictors (climate and soil geochemical variables) from the laboratory experiments to identify the most influential factors affecting Qsp and the equilibrium constant (k).
- Training Phase: They then built a global model using only 5 predictors (percent clay, pH, soil order, mean annual temperature (MAT), and mean annual precipitation (MAP)) that are available at global scales. This model was trained on data from 14,631 soil profiles.
- Prediction Phase: This trained model was then used to predict Qsp and k for global soil profiles and gridded soil data.
- Uncertainty Estimation: To quantify uncertainty, they generated multiple predictions using alternative datasets for the predictors and multiple Random Forest models.
Datasets
- Laboratory Sorption Experiments: A compilation of 402 laboratory sorption experiments from five studies, representing 111 locations across North America and Europe, covering six soil orders (Alfisols, Entisols, Inceptisols, Mollisols, Spodosols, Ultisols). This dataset was used to fit the Langmuir equation parameters (Qsp and k).
- World Soil Information Service (WOSIS) Database: Provided data for 14,631 soil profiles globally, including variables like % Clay, pH, Soil Order, Bulk Density, and SOC stock. This was used for the global prediction of Qsp and k.
- SoilGrids Data Products: Provided gridded data (10 km resolution) for % Clay, pH, Soil Order, Bulk Density, and SOC stock, used for global mapping.
- WorldClim Version 2 and CHELSA: Provided gridded climate data (MAT and MAP) at 30-second resolution.
- Alternate Gridded Predictor Datasets: Used for uncertainty estimation (e.g., alternate MAT, MAP, %Clay, and pH datasets).
Key Insights and Ideas
- Sorption as a Major Carbon Sink: The study strongly reinforces the idea that mineral-associated carbon, formed through DOC sorption, is a significant and potentially stable pool of soil carbon.
- Subsoils Have Higher Potential: A key finding is that subsoils (30 cm - 1 m depth) generally have a greater capacity to sorb additional DOC (higher Qsp) than topsoils. This is attributed to higher mineral concentrations and available surface area in deeper soil layers, coupled with lower existing soil carbon concentrations.
- Climate and Soil Properties Matter: The research highlights that mid- and low-latitude soils, and those with higher clay content, exhibit a greater capacity for DOC sorption. This suggests that climate and soil mineralogy are critical controls on this carbon sequestration pathway.
- The "4 per 1000" Goal: The estimated additional sorption potential could theoretically help achieve the "4 per 1000" goal (increasing global soil carbon by 0.4% annually) for 12 years, underscoring the potential magnitude of this process.
- Model Limitations: The study acknowledges that laboratory batch experiments, while useful, may overestimate field-scale potential due to soil homogenization and the disruption of natural soil structure.
Key Results
- Global Additional Sorption Potential: The estimated global additional DOC sorption potential for six soil orders to 1 m depth is 107 ± 13 Pg C.
- Percentage Increase in SOC: This potential represents a 7% increase in the existing total soil organic carbon (SOC) stock of these soil orders (which is estimated at 1615 Pg).
- Influential Predictors (Lab Scale): For laboratory experiments, dithionite-extractable Fe, mean annual temperature (MAT), and clay percentage were the most influential variables for predicting Qsp.
- Influential Predictors (Global Scale): For global predictions, percent clay was the most important predictor of Qsp, followed by MAT, MAP, pH, and soil order.
- Soil Order Influence: Alfisols, Mollisols, and Ultisols showed higher average Qsp compared to other soil orders.
- pH Threshold: Qsp showed a threshold relationship with pH, with low Qsp in very acidic soils (pH < 4).
- Subsoil vs. Topsoil: Qsp was higher in subsoils (1.14 g C/kg) compared to topsoils (0.99 g C/kg), while the equilibrium constant (k) showed no significant difference.
Limitations and Future Work
- Field Realizability: The authors explicitly state that the estimated sorption potential is a maximum and may not be fully realizable in field conditions. Batch experiments homogenize soil, disrupting natural structures that limit mineral access to DOC.
- Excluded Soil Orders: Gelisols and Histosols, common in high latitudes, were excluded due to poor representation in the laboratory data. These soils might have different sorption characteristics.
- Data Gaps: The lack of widespread gridded data for reactive metals like Fe and Al limited their inclusion as predictors in the global models, despite their importance in laboratory experiments.
- Nutrient Limitations: The rate of C sequestration via sorption is ultimately limited by the availability of nitrogen and phosphorus, which could constrain the practical realization of this potential.
- Future Work: The authors suggest expanding measurements to include more soil orders (e.g., Andisols, Oxisols, Aridisols), investigating the role of different clay mineral types, and developing models that better integrate various soil C stabilization mechanisms beyond simple sorption. Further research is also needed to understand the importance of sorption relative to other mechanisms and how it varies with climate and management.
Summary updated