It is significant to identify the response thresholds of ecosystem services and influencing factors for regional ecosystem restoration and optimization. However, current research often overlooks the scale effects on the response thresholds of ecosystem services and influences. In this study, the composite of ecosystem services (CES) in central Yunnan Province from 1990 to 2020 was first quantified using various methods, such as the integrated valuation of ecosystem services (InVSET), the revised universal soil loss equation (RUSLE), and the normalized summation method, and then its spatial and temporal changes were analyzed. The influencing factors of CES were identified using the random forest model. The nonlinear relationship between CES and different influencing factors was explored using the constraint line method. Finally, the scale effect of CES with the response thresholds of different influencing factors was explored in the scale range of 1km to14km. The main findings are as follows: (1) CES in central Yunnan Province fluctuated and then decreased from 1990 to 2020, exhibiting a spatial distribution pattern of ‘low value in the middle and high value on both sides’. (2) Fractional Vegetation Cover (FVC) and annual precipitation were identified as the most important factors affecting CES, with mean importance values of 0.5472 and 0.1893, respectively. (3) The constraint line between FVC and CES showed a positively convex pattern, while the constraint lines between CES and factors like annual precipitation, average annual temperature, standardized precipitation evapotranspiration index (SPEI), elevation, and slope exhibited a hump-shaped pattern. The constraint lines between GDP per capita, urbanization rate, and CES demonstrated a negatively convex pattern. (4) Within the scale range of 1 to 14 km, the shapes of most constraint lines remained consistent as spatial scale increased, but the R2 values fluctuated and decreased. The response thresholds of CES to annual precipitation, elevation, and slope showed fluctuations and decreasing trends, while the thresholds to annual mean temperature fluctuated and increased. There is no significant regularity in the response thresholds of CES to SPEI. This indicates that there is a scale effect between CES and the thresholds of its influencing factors. The findings provide new insights into how changes in CES and scale sensitivity influence the relationship between CES and its driving factors in central Yunnan Province. In summary, this study not only bridges the gap of scale effects in the study of response thresholds of ecosystem services and influencing factors, but also can provide scientific references for ecosystem service management and ecological restoration.