The InVEST-PLUS model was coupled to calculate and analyze the spatiotemporal dynamics of four types of ecosystem services in Guizhou Province from 2000 to 2030 under different scenarios. Based on Spearman’s correlation analysis, the trade-offs/synergies between various service functions were examined, and self-organizing feature mapping networks were utilized to reveal the ecosystem service bundles. The results showed that from 2000 to 2020, the changes in land structure were mainly characterized by the continuous decrease of arable land and the fluctuating changes of forest and grassland, with a significant expansion of construction land. The natural development scenario showed the most significant increase in construction land among the three scenarios, with an increase of 25.39%, while in the ecological protection scenario, the main ecological land such as forest and grassland were protected. The water yield and soil conservation showed a fluctuating trend of first decreasing and then increasing from 2000 to 2020, while habitat quality and carbon storage showed a declining trend. The average value of carbon storage decreased from 234.05t/hm² to 231.70t/hm², and the average value of habitat quality decreased from 0.64 to 0.60. Relatively, in the ecological protection scenario, except for a slight decrease in water yield, other ecological function indicators improved, especially carbon storage and habitat quality, which reached the highest values. The trade-offs (3pairs) and synergies (3pairs) between ecosystem service functions in the study area showed a relatively balanced state. Under the ecological protection scenario, the correlation coefficients between different ecosystem services gradually increased, indicating an overall trend of environmental improvement in the region. Based on the clustering results, four types of ecosystem service bundles (ESB) were identified in Guizhou Province. From 2000 to 2030, the spatial pattern of various ecosystem service clusters in the study area remained basically stable, but there was spatial and temporal heterogeneity among different ESBs.