Exploring influential indicators for cultivating selenium-rich lily using a random forest model.
Hao Gong1, Liangliang Dai1, Jie Luo2
1China Geological Survey, Changsha General Survey of Natural Resources Center, Beijing, China.
Environmental Geochemistry and Health
|September 16, 2025
Summary
Selenium (Se) is essential for health, and its levels in lilies are influenced by soil factors beyond just Se. A random forest model accurately predicts Se content in lilies, aiding Se-rich crop planning.
Area of Science:
- Agricultural Science
- Environmental Science
- Geochemistry
Background:
- Selenium (Se) is a vital nutrient for human health.
- Crop Se levels depend on soil properties and plant-soil element interactions.
- Karst areas present unique soil conditions influencing Se bioaccumulation.
Purpose of the Study:
- To investigate factors affecting Selenium bioaccumulation in lilies.
- To develop a predictive model for Se content in lilies using soil geochemical data.
- To support the planning of Se-rich agricultural products in karst regions.
Main Methods:
- Utilized 1:50,000 land-quality geochemical survey data.
- Employed a random forest model to identify key soil indicators.
- Compared random forest model performance against multiple linear regression.
Main Results:
- Identified soil Selenium and nitrogen as key predictors of Se content in lilies.
- The random forest model demonstrated higher accuracy and precision in predicting Se content compared to multiple linear regression.
- Established a scientific approach for assessing Se bioaccumulation in crops.
Conclusions:
- The random forest model offers a robust method for predicting Se content in lilies.
- Findings support strategic agricultural planning for Se-rich specialty crops.
- Promotes sustainable development of regional agriculture through optimized Se-rich production.
