Soil health is a fundamental determinant of forest ecosystem stability and long-term sustainability. Understanding its spatiotemporal dynamics and underlying drivers across various forest types is essential for developing adaptive soil management strategies. In this study, we examined six representative forest types in the Huoditang area of the Qinling Mountains: Quercus aliena var. acuteserrata forest, Pinus tabuliformis forest, Betula albosinensis forest, Picea wilsonii forest, Pinus armandi-Quercus aliena var. acuteserrata mixed forest, and Tsuga chinensis-Pinus armandi mixed forest. Soil samples were collected from four depth intervals (0-60 cm), and long-term trends in soil physicochemical properties and understory plant diversity from 2010 to 2022 were analyzed. We calculated the Forest Soil Health Index (FSHI) for each forest type and applied sensitivity analysis combined with structural equation modeling (SEM) to quantify the responsiveness of key soil variables and to elucidate direct and indirect pathways affecting soil health. Results indicated pronounced temporal and vertical variation in soil properties. (1) Across all forest types, soil organic matter, total nitrogen, and total phosphorus showed a fluctuating decline between 2013 and 2019, followed by a marked increase in 2022. Alkali-hydrolyzable nitrogen exhibited divergent patterns: a significant decrease in the T. chinensis-P. armandi mixed forest in 2022 and a temporary decline in 2019 with subsequent recovery in the P. armandi-Q. aliena var. acuteserrata mixed forest. Both total and available nutrient contents consistently declined with increasing soil depth. (2) Plant diversity patterns revealed that, except in P. tabuliformis stands, the shrub layer exhibited a higher Margalef richness index than the herbaceous layer across most forest types. From 2013 to 2019, the Shannon-Wiener diversity index was higher in the shrub layer, whereas by 2022, herbaceous layer diversity surpassed that of shrubs, suggesting a structural shift in understory vegetation. (3) Among the six forest types, P. wilsonii and B. albosinensis forests demonstrated relatively higher and more stable FSHI values across years. Depth-wise, all forests exhibited a progressive decline in soil health with increasing depth. (4) SEM revealed that understory plant diversity influenced soil health through multiple pathways. In Q. aliena var. acuteserrata, B. albosinensis, and T. chinensis-P. armandi forests, diversity primarily enhanced soil health by directly improving soil physical attributes. Conversely, in P. wilsonii and P. armandi-Q. aliena var. acuteserrata mixed forests, diversity affected soil health mainly by altering chemical properties. Additionally, in T. chinensis-P. armandi forests, an indirect effect mediated via chemical properties was also detected.