Objective This study aimed to investigate how topographic factors, stand structure, and soil nutrients drive understory vegetation diversity in alpine forest ecosystems, and to determine whether these driving mechanisms exhibit significant stratified differentiation between the shrub and herb layers.

Methods Based on the natural secondary forests in the eastern Qilian Mountains, a total of 64 plots were established across an elevational gradient of 2 700～3 100 m and across different topographic conditions. Stand structural characteristics, soil physicochemical properties (including pH, soil organic carbon, total nitrogen, alkali-hydrolyzable nitrogen, available potassium, and soil moisture content), and topographic variables (elevation and slope) were systematically measured. Four diversity indices, namely the Margalef richness index, Simpson diversity index, Shannon-Wiener diversity index, and Pielou evenness index, were calculated. Redundancy analysis (RDA), linear mixed-effects models (LMM), and variance partitioning analysis (VPA) were employed to comprehensively reveal the driving mechanisms underlying diversity patterns in the shrub and herb layers.

Results Understory vegetation diversity was generally higher in the herb layer than in the shrub layer, with pronounced differences among valleys. Shrub-layer diversity was highest in Qihan Gully, whereas herb-layer diversity reached its maximum in Tacang Gully. Results from the LMM and VPA indicated that elevation was the dominant factor affecting understory vegetation diversity, independently explaining 22.74% and 16.31% of the variation in shrub- and herb-layer diversity, respectively. Shrub-layer diversity was primarily driven by soil factors, while topographic and soil factors exerted relatively independent effects. In addition, the random effect of valley was significant (ICC = 24%). In contrast, herb-layer diversity was mainly influenced by the interaction between topography and stand structure, which explained 12.31% of the observed variation. Soil factors made a relatively small contribution, explaining only 4.63% of the variation, and showed no shared explanatory effect with other factor groups. Furthermore, herb-layer diversity exhibited a significant unimodal relationship with slope, reaching its maximum at approximately 25°, whereas no significant relationship was detected between shrub-layer diversity and slope.

Conclusion Understory vegetation diversity in alpine mountainous regions exhibits pronounced stratified differentiation. Herb-layer diversity is mainly regulated by cascading effects of topography through stand structure, and the associated understory light environment. In contrast, shrub-layer diversity is mainly controlled by the combined effects of soil nutrient availability and resource competition under varying topographic conditions. These findings provide a scientific basis for close-to-nature forest management and biodiversity conservation in alpine forest ecosystems.