Climate variability and agricultural expansion are fundamentally reshaping ecosystem functions (EFs) in drylands. Xinjiang Region, a vast and typical dryland area in China, also faces dual pressures of natural and anthropogenic disturbances. Numerous local studies in Xinjiang have quantified individual EFs using biophysical models. However, research remains limited on the region-wide spatial distribution of long-term EF transitions, and on how nature-human interactions shape EF transitions across all of Xinjiang. In this paper, these gaps are addressed by using an EF classification framework to assess dominant EF patterns and reveal the ecosystem responses across Xinjiang, China from 2000 to 2024. A suite of multi-year EF maps with high accuracies (0.864-0.882) were produced using a two-level classification scheme with a random forest model, and Landsat imagery. EF maps showed that the water retention (WR) regions are concentrated in mountainous areas, the sandstorm prevention (SP) regions primarily occur in oasis-desert transition zones, and the agricultural production (AP) regions are dispersed in the desert (DS) periphery. Three major transition trajectories were identified: two climate-driven pathways, namely, greening ("DS-SP-WR") and browning ("WR-SP-DS") induced mainly by the water changes (R = 0.594 and -0.553 with annual precipitation, respectively), and a policy-driven cropland expansion pathway ("DS-AP") resulting from aggressive agricultural expansion, particularly cotton cultivation (R = 0.831). This "DS-AP" functional transition in Xinjiang may potentially introduce multiple ecological challenges, including biodiversity loss and increased water demand. These findings reveal how climate variability and agricultural intensification reshape dryland landscapes, with implications for sustainability across water-limited ecosystems globally. Our results underscore the urgent need for adaptive management strategies that balance agricultural development with ecosystem resilience in Earth's expanding drylands.