Effects of herbaceous root growth and decay on gas conductivity of unsaturated soil: Experimental investigation and modelling

Gas conductivity of soil is vital for evaluating soil aeration and regulating oxygen delivery to plants, while plant roots affect the gas conductivity of surrounding soil. Hence, root decay during long-term root development is expected to affect gas conductivity accordingly. However, relevant experimental data are limited and models are lacking. Since herbaceous plants are more susceptible to root decay, this study focuses on measuring gas conductivity in unsaturated soil containing decayed roots of vetiver grass, alongside optical microscopic observation on root structural changes, and developing a model for gas conductivity incorporating decay effects. Microscopic observations reveal that most decayed secondary lateral roots were destroyed, forming hollow pores as preferential pathways. In contrast, decayed adventitious roots and primary lateral roots retained their epidermis but became contracted and irregular, creating pores at the root-soil interface and also facilitating gas transport along with hollow interiors. Therefore, root decay significantly increased the gas conductivity of soil across varying degrees of saturation. Specifically, the soil containing decayed roots (with a root decay ratio of about 26%) exhibited a gas conductivity of 6.7 × 10⁻⁷ m/s at the degree of saturation of 0.34, which was approximately 52% higher than that of bare soil (4.4 × 10⁻⁷ m/s). Gas conductivity ranked highest for soil containing decayed roots, intermediate without roots, whereas lowest with fresh roots. This resulted from roots obstructing large pores during plant growth, which reduced gas conductivity despite also enhancing pore connectivity via highly connected root-supported pores, leaving pore network with high connectivity after root decay. As regards modelling, the maximum gas conductivity considering root decay was derived by introducing the effect of root decay on soil pores into a pore-based model. Good agreements were observed between the measurements and predictions, under the root mean square error (RMSE) of 0.22 and the scatter index (SI) of 0.03.