Influence of artificially-induced biocrust development on soil matrix infiltration: insights from a long-term field experiment and random forest modeling

Abstract

Soil water availability is crucial for ecosystem sustainability in arid and semi-arid regions, making an understanding of soil infiltration processes essential for effective water management. This study investigated the impact of artificially-induced biocrusts, an innovative desertification mitigation and ecosystem restoration strategy, on soil matrix infiltration in the Qubqi Desert. Specifically, we focused on the development of artificially-induced biocrusts and compared the matrix infiltration dynamics of cyanobacteria- vs. moss-dominated biocrust communities after 16 and 22 years of restoration. Field matrix infiltration tests (54 tests total) were randomly conducted using a Mini Disk Infiltrometer (MDI) at –5 cm tension, and the results showed distinct patterns in soil matrix infiltration across different stages of biocrust restoration. As biocrusts formed and developed, the initial infiltration rate (IIR), steady infiltration rate (SIR), and sorptivity (S) all exhibited a decreasing trend, with moss-dominated biocrusts showing a more pronounced reduction compared to cyanobacteria-dominated biocrusts. In the early restoration stages, state transitions in the biocrust community (e.g., initial biocrust formation and succession from cyanobacteria- to moss-dominated types) caused distinct changes in soil hydrological properties, whereas infiltration changes stabilized during later stages as biocrust communities reached a relatively steady state. Key factors influencing soil matrix infiltration characteristics were identified, including biocrust thickness, chlorophyll-a (Chl-a) content, and dissolved organic carbon (DOC). Random forest modeling further verified these variables as critical predictors of biocrust matrix infiltration, achieving high predictive accuracy (R² > 0.95). The findings underscore the potential of monitoring these factors to assess the impact of biocrust restoration (e.g., after cyanobacteria inoculation) on regional hydrological cycles. Moreover, our findings demonstrate that the formation and development of artificially-induced biocrusts significantly alter soil structure and infiltration behavior, potentially promoting preferential flow pathways and enhancing soil water retention, particularly during the early stages of restoration. Overall, these findings suggest artificially-induced biocrusts as a practical strategy for sustainable land management in regions that are at risk of desertification and provides a predictive framework for assessing their eco-hydrological impacts at large-scales.