Abstract:As a key component of the terrestrial phosphorus cycle, soil available phosphorus (AP) is crucial for ecosystem stability and security. However, a systematic analysis of its spatiotemporal patterns and its impact on ecosystem services remains limited. To address this, we constructed a long-term (2000-2020), spatially-continuous AP dataset (0-20 cm depth) for the karst region of China by integrating 33,114 site observations, coupling geospatial and remote sensing data with the Random Forest algorithm. Concurrently, we quantified key ecosystem services-carbon sequestration (CS), soil retention (SR), and biodiversity (BI)-using the InVEST model, and further assessed ecosystem multifunctionality (EMF). By employing correlation analysis and the elasticity coefficient method, we revealed the driving effects of soil AP dynamics on these services. The results indicated that: (1) From 2000 to 2020, CS, SR, and BI demonstrated an overall increasing trend, with annual mean increases of approximately 3.2g C m-2 a-1, 6.8 t hm-2 a-1, and a fluctuating rise for BI, respectively. The spatial mean of EMF was 0.416 but showed a slight decreasing trend. (2) Spatially, AP exhibited a distribution pattern characterized by "high in the east and west, low in the north and south" (mean: 13.28 mg/kg), while temporally, it displayed a fluctuating decreasing trend. (3) Overall positive correlations were observed between AP and the ecosystem services, suggesting that changes in AP have promoted the enhancement of ecosystem services in the karst region. Furthermore, the influence of AP on ecosystem services exhibited a double-threshold effect: a trough occurred at an AP level of approximately 11.6 mg/kg, and a peak was reached at about 20.3 mg/kg. This study clarifies the complex role of AP in influencing ecosystem services and highlights the existence of threshold effects in their relationships, thereby providing a scientific basis for formulating targeted phosphorus management strategies and promoting sustainable development in karst regions.