In cold and arid regions, the mechanical properties and influencing mechanisms of the root–soil interface during the thawing stage remain poorly understood. This study focuses on Alhagi sparsifolia root–clayey sand composites to investigate the effects of temperature (−10 °C to 25 °C), initial soil water content (4–12%), and naturally varying root diameter (4.50–5.05 mm) on root pullout behaviour, and integrates endoscopic macro-observation, environmental scanning electron microscopy (ESEM), soil water migration tests, and nuclear magnetic resonance (NMR) techniques to reveal the dominant influencing mechanisms. Key findings reveal the following: (1) An increase in soil water content from 4% to 12%, and a temperature rise from −10 °C to 25 °C led to a maximum reduction in the average peak pullout force (F_T) of roots exceeding 95%. (2) There is a non-monotonic relationship between root diameter and pull-out force, which can be attributed to two distinct failure modes: a newly observed failure mode known as root bark peeling, occurring under high soil moisture conditions (≥8%), and a commonly observed failure mode referred to as partial soil detachment, occurring under low soil moisture conditions (≤6%). (3) The coupling effects of temperature and water content reveal that the increase in temperature predominantly contributes to strength loss (>63%) during the ice–water phase transition (−10 °C to 0 °C), while soil water content primarily influences root pullout behaviour in the liquid water stage (5 °C to 25 °C). (4) As the temperature rises, in soils with low water content (4–6%), the reinforcing effect of roots appears to stabilize at −1 °C, whereas in soils with high water content (8–12%), stabilization occurs only beyond 5 °C. These findings enhance the understanding of root–soil interactions in thawing environments and provide a theoretical basis for soil bioengineering techniques aimed at slope stabilization in cold and arid regions.

In response to the problems of loose soil structure and insufficient water and nutrient retention capacity of sandy bank slopes in arid regions, which constrain vegetation establishment and long-term slope stability, this study focuses on typical sandy soils in arid northwestern China. The desert plant Alhagi sparsifolia, characterized by clonal root sucker reproduction, was selected as the study species to construct and optimize a composite-amended sandy substrate suitable for ecological restoration of bank slopes. Based on an orthogonal experimental design, carboxymethyl cellulose sodium (CMC), straw fibers (SF), and fly ash (FA) were combined at different proportions to assess (i) the vertical distribution of soil water and nutrients in the A. sparsifolia growth habitat, (ii) aggregate structure, (iii) plant trait responses to environmental regulation, and (iv) the shear strength of root–soil composites. The results indicate that when the contents of CMC, SF, and FA were 0.5%, 1.0%, and 5.0%, respectively, the substrate environment promoted a vertically oriented root system with pronounced lateral root development in A. sparsifolia, and the plants adopted an adaptive strategy that balances resource acquisition efficiency and environmental constraints by regulating aboveground growth allocation. This growth pattern reduced the risk of disturbances to slope stability caused by excessive aboveground biomass while maintaining the sand-fixing function of root morphological traits. This study provides a plant functional trait-based regulation strategy for ecological restoration of typical sandy slopes in arid regions, and the proposed composite substrate optimization scheme offers a feasible reference for improving vegetation establishment and substrate performance in sandy habitats.

Vegetation restoration can reduce soil erosion and improve soil stability on erosion-prone slopes, but the contrasting roles of herbaceous and shrub species in surface erosion resistance and subsurface reinforcement remain unclear. This study compared two herbaceous species, Melilotus officinalis (L.) Pall (MOP) and Suaeda glauca (Bunge) Bunge (SGB), and two shrub species, Alhagi camelorum Fisch. (ACF) and Nitraria tangutorum Bobrov (NTB), with bare soil (CK) as the control. Soil detachment capacity (Dc), root–soil shear strength, root traits, and soil properties were measured using planting experiments, flume scouring tests, and direct shear tests. Compared with CK, the vegetation treatments showed lower bulk density and higher total porosity and water-stable aggregates content, while soil organic matter exhibited differences among treatments. Herbaceous species developed dense fine roots in surface soil, whereas shrubs had deeper, coarser root systems. Vegetation significantly increased root–soil shear strength, mainly by increasing cohesion; herbs provided stronger surface reinforcement, whereas shrubs maintained more stable reinforcement at depth. All vegetation treatments significantly reduced Dc, and MOP showed the best overall anti-erosion performance. Random forest analysis further showed that root traits dominated Dc variation, accounting for 84.9% of total variable importance, with root length density and root surface area density as the leading factors. A regression model based on shear stress, total porosity, bulk density, and root length density predicted Dc well (R² = 0.86, NSE = 0.65, RMSE = 0.083). These findings support site-specific vegetation selection: herbs for surface erosion control, shrubs for slope-stability concerns, and shrub–herb combinations where both risks coexist.