The contradiction between humans and land is prominent in the karst area of Southwest China. Human destruction of vegetation eventually leads to serious water loss and soil erosion, resulting in a large number of rocky desertification. In order to control the expansion of rocky desertification, the Chinese government has carried out a series of projects, such as returning farmland to forest, natural forest protection and Yangtze River shelterbelt projects, which have caused corresponding changes in land use and vegetation cover. As one of the most important types of rocky desertification control in China, karst graben basins are usually characterized by drastic changes in basin-mountain topography, great climate contrast, severe soil flow/loss, poor vegetation site conditions, and the slow recovery of community ecosystem. Therefore, the study on the comprehensive control of rocky desertification at the karst graben basin became urgent. The ecological stoichiometry is a state-of-the-art science that analyzes the regional circulation patterns of biogeochemical elements from the perspective of element metrology, which provides a new insight to solve the problem of nutrient supply/demand and circulation in ecosystems. It has important application value in the study of soil nutrient circulation and balance mechanism, and is one of the hot spots in current ecological research. However, domestic studies on ecological stoichiometry mainly select desert steppe, Loess Plateau and Minjiang River estuary as research areas in recent years. There is a paucity of systematic studies on soil metrology characteristics in different stages of vegetation succession at karst graben basins, and the response of soil nutrients and ecological stoichiometry to different types of vegetation restoration is still unclear. In order to elucidate the content and stoichiometric ratios of C, N and P elements in the soils in different stages of vegetation succession at karst graben basins, Santang, located in Xiaojiang karst basin, Luxi county, Yunnan Province, was selected as a study area. This area is dominated by a subtropical monsoon climate, with the average annual temperature of 15.2 °C, the average annual sunshine of 2,122 h, and the average annual precipitation of 1,000 mm. The main tree species in the study area include Platycladus orientalis (L.) Franco, sticky alder, Yunnan pine, and so on. In January 2018, by the "space for time" approach, the soils from 5 different stages of vegetation succession (corn land, grassland, shrub land, planted forest and primary forest) were collected from different soil depths (0--10 cm and 10--20 cm) according to the sampling method of profile excavation. The soil organic carbon (SOC), total nitrogen (TN) and total phosphorus (TP) contents were determined by potassium dichromate oxidation with an external heating method, Kjeldahl method, and molybdenum antimony anti-colorimetric method. Excel 2019 software was used for data statistics. One-Way ANOVA, Duncan's new multiple range test method (LSD), Pearson correlation and Two-way ANOVA in SPSS 25.0 software were used to study the variance and correlations of soil ecological stoichiometric characteristics. The study results are expected to further reveal nutrient restriction and nutrient change patterns in the vegetation succession process, meanwhile, providing the scientific basis for vegetation restoration management and rational use of land resources in karst graben basins. There were significant differences in soil nutrient contents in different succession stages, mainly manifested in the 0--10 cm soil layer. SOC and TN contents showed an overall increasing trend with the increase of succession years, while TP content fluctuated and showed no obvious change pattern. Soil C and N contents differed significantly among soil layers. Specifically, SOC and TN contents in 0--10 cm soil layer were higher than those in 10--20 cm soil layer at each succession stage. Compared with C and N contents, soil P content did not differ significantly among soil layers at each succession stage, except for the content in the primary forest. In 0--10 cm soil layer, C/N, C/P and N/P showed significantly positive correlations with SOC content; C/P and N/P indicated significantly positive correlations with TN content; soil TP content was not significantly correlated with C/N, C/P and N/P. There was no significant correlation between soil physicochemical properties and soil stoichiometric ratio in the 10--20 cm soil layer. In the present study, the N/P ratios of 0--10 cm soil in different succession stages were between 2.71 and 5.93, and those of 10--20 cm soil were between 3.11 and 4.92. Essentially, all of the N/P ratios of soil at different depths were lower than 10, indicating that vegetation growth in the study area was more likely to be limited by N. However, given the fact that P content in karst soil is relatively low, vegetation growth should be first limited by P rather than N. Moreover, vegetation types and soil depths significantly affected SOC, TN and TP contents and their stoichiometric ratios in the study area. In summary, soil nutrient and stoichiometric ratios at karst graben basins change significantly with vegetation restoration chronosequence, and soil physicochemical properties have an important impact on soil stoichiometric ratios. These results are of great significance for further research on soil nutrient balance and limiting factors in karst graben basins. [ABSTRACT FROM AUTHOR]