Spatiotemporal heterogeneity and driving factors of ecosystem service relationships and bundles in a typical agropastoral ecotone

Complex geographical environments and intensified human activities have induced significant spatiotemporal heterogeneity in ecosystem services (ESs) and their relationships worldwide, especially in agropastoral ecotones. Identifying the ecological and socioeconomic driving factors of the spatiotemporal dynamics of the relationships and bundles among multiple ESs is essential for effective agropastoral ecotone management and restoration. Herein, five key ESs, including water yield (WY), carbon sequestration (CS), sand fixation (SF), soil conservation (SC), and food production (FP), in the northern piedmont of the Yinshan Mountains (NPYM), a typical agropastoral ecotone, were quantified from 1990 to 2018. Ecosystem service bundles (ESBs) were identified by the Kmeans algorithm, and the trade-offs/synergies among the different ESs in each ESB were investigated using Spearman's correlation analysis. Moreover, the main driving factors of the spatiotemporal dynamics of the ES relationships and ESBs were identified by the geographical detector model. Our results showed that the five ESs in the NPYM exhibited high spatiotemporal heterogeneity over the last three decades, with WY, SC and FP significantly improving over time but CS and SF slightly decreasing. Considering the ecological features and spatial distribution, the NPYM could be divided into four ESBs: the grassland protection bundle (GPB), desertification control bundle (DCB), agricultural production bundle (APB) and forest conservation bundle (FCB). Although trade-offs between WY and CS and synergy between SC and CS occurred in almost all ESBs, significant differences in the relationships between the other ESs were observed in the different ESBs. Generally, land use type was one of the dominant drivers of most ES relationships in all ESBs, especially in the APB. However, climate, vegetation and socioeconomic factors also played important roles in the spatiotemporal heterogeneity of various ES relationships in some bundles. Moreover, the substantial changes in the different ESBs over the last three decades were mainly due to climatic and socioeconomic factors. Overall, our results indicated that identifying the spatiotemporal heterogeneity and driving factors of ES relationships and ESBs could provide useful guidance for developing targeted zoning management strategies in agropastoral ecotones.