A nature‐inclusive future with healthy soils? Mapping soil organic matter in 2050 in the Netherlands

• Published in: European journal of soil science Vol. 75; no. 4
• Main Authors: Helfenstein, Anatol, Mulder, Vera L., Hack‐ten Broeke, Mirjam J. D., Breman, Bas C.
• Format: Journal Article
• Language: English
• Published: Oxford, UK Blackwell Publishing Ltd 01.07.2024; Wiley Subscription Services, Inc
• Subjects: Agricultural land; Agricultural practices; agroecological farming; Depth; Digital mapping; digital soil mapping; Ecological farming; Grasslands; Land reclamation; Land use; Land use management; Learning algorithms; Machine learning; Mapping; nature‐inclusive; Organic matter; Organic soils; Peat; Peatlands; Reclaimed land; rewetting peatlands; scenario modelling; Soil; soil health; Soil mapping; Soil organic matter; Soils; space–time modelling; spatial planning; Spatial variations; Sustainability management; Netherlands
• ISSN: 1351-0754; 1365-2389
• DOI: 10.1111/ejss.13529

Nature‐inclusive scenarios of the future can help address numerous societal challenges related to soil health. As nature‐inclusive scenarios imply sustainable management of natural systems and resources, land use and soil health are assumed to be mutually beneficial in such scenarios. However, the interplay between nature‐inclusive land use scenarios and soil health has never been modelled using digital soil mapping. We predicted soil organic matter (SOM), an important indicator of soil health, in 2050, based on a recently developed nature‐inclusive scenario and machine learning in 3D space and time in the Netherlands. By deriving dynamic covariates related to land use and the occurrence of peat for 2050, we predicted SOM and its uncertainty in 2050 and assessed SOM changes between 2022 and 2050 from 0 to 2 m depth at 25 m resolution. We found little changes in the majority of mineral soils. However, SOM decreases of up to 5% were predicted in grasslands used for animal‐based production systems in 2022, which transitioned into croplands for plant‐based production systems by 2050. Although increases up to 25% SOM were predicted between 0 and 40 cm depth in rewetted peatlands, even larger decreases, on reclaimed land even surpassing 25% SOM, were predicted on non‐rewetted land in peat layers below 40 cm depth. There were several limitations to our approach, mostly due to predicting future trends based on historic data. Furthermore, nuanced nature‐inclusive practices, such as the adoption of agroecological farming methods, were too complex to incorporate in the model and would likely affect SOM spatial variability. Nonetheless, 3D‐mapping of SOM in 2050 created new insights and raised important questions related to soil health behind nature‐inclusive scenarios. Using machine learning explicit in 3D space and time to predict the impact of future scenarios on soil health is a useful tool for facilitating societal discussion, aiding policy making and promoting transformative change.