Conducting mountain ecosystem zoning through ecosystem service bundles (ESBs) plays a crucial role in achieving coordinated management of multiple ecosystem services (ESs). However, research on the interrelationships and underlying mechanisms among ESs across different partitioned zones remains inadequate. In this study, we quantified the spatial and temporal changes of six ESs in the Taihang Mountains, including water yield, soil conservation, carbon storage, food production, net primary productivity (NPP) and biodiversity maintenance, and revealed their trade-offs/synergies based on the ESBs delineated by the Self-Organizing Map (SOM) method. We subsequently deployed the Mantel Test to pinpoint the key drivers of ESs for multi ESBs. The results showed that: (1) four of the six ESs demonstrated significant growth, with NPP and biodiversity maintenance increasing substantially across > 90% of the study area. Conversely, soil conservation fluctuated considerably overall, decreasing by 8.89% between 2000 and 2020. (2) Based on the characteristics features of six ESs, the Taihang Mountains were divided into NPP-biodiversity mutual enhancement (B1), major grain producing (B2), ecological core (B3), ecological fragile (B4), and water resources supply bundles (B5). The trade-off and synergies of ESs were not immutable, and varied with ESBs. A strong trade-off between water yield and food production was observed in B2 and B5, while a synergistic relationship was found in B4. The highest synergy effect appeared in the B1 pair. (3) This study proposed a differentiated zoning governance framework: whereas B3 requires strict ecological protection, bundles B1 and B4 should prioritize grassland restoration and conservation, while B5 necessitated curbing uncontrolled urban expansion to safeguard water yield service. Furthermore, grain production in B2 continued to increase in the context of regional warming. This study establishes the ecosystem management framework for typical arid/semiarid mountains, providing a scientific basis for territorial spatial planning and ES optimization.