Abstract: Trade-offs and synergies among ecosystem services are often subject to the scale effects in modern agriculture. But only a few studies have explored from the perspective of the optimal spatial scale, thereby limiting their applicability in policy making. Taking Zhangbei County, a representative Grain for Green areas in China, as a case study, two ecosystem services were quantitatively evaluated from 2000 to 2020, namely windbreak and sand fixation, and food production, both of which were essential to maintain the ecological stability and rural livelihoods in the ecologically fragile agro-pastoral transition zones. Spatial continuous wavelet transform was applied to determine the optimal spatial scale, at which the trade-offs and synergies were characterized for the spatiotemporal features. Ecological zoning was finally proposed under the differentiated regulation. The results showed that: 1) The two ecosystem services exhibited the opposing trends from 2000 to 2020. Windbreak and sand fixation services increased overall, while the food production services slightly declined. Spatially, there was the decrease in the windbreak and sand fixation services from the central to peripheral areas, indicating the close association with the regional topography, climatic conditions, and land-use pattern, whereas the food production services displayed a fragmented pattern that aligned well with the spatial distribution of cropland. 2) 1.2 km was identified as the optimal scale for the trade-offs and synergies between the two ecosystem services in the study area. There was the phased pattern over the 20 years, with the initial trade-offs, followed by synergies, and then a return to trade-offs. Spatially, the trade-offs were widely distributed in a patchy pattern in the county, with a continuous contraction. Synergies were concentrated in the eastern mountainous and hilly areas, exhibiting a strengthening trend. Structurally, the weak trade-offs increased rapidly, whereas the moderate and strong trade-offs declined significantly. In contrast, there was increase in the weak, moderate, and strong synergies. 3) According to the decision trees of the trade-offs and synergies, an ecological zoning decision tree was constructed to take the grid cells as the basic analytical units. Time series were adopted as the main analytical thread. The trade-offs and synergies patterns were identified as the hierarchical classification nodes. Four ecological zones were then classified, including subzone I, subzone II, subzone III, and subzone IV. The targeted strategies were optimized for each subzone. The findings can provide a methodological pathway for the interactions among ecosystem services, thereby enhancing the scientific basis for ecological zoning under the Grain for Green Program. The specific locality can be reduced after optimization. The approach can be extended into the similar regions for the sustainable development goals, such as SDGs 2 and 15. Future work should focus on the trade-offs and synergies among a broader set of ecosystem services, such as the carbon sequestration, water conservation, and soil retention. The optimal pathways can be identified for the coordinated multi-service development at the regional scale. In addition, the ecological and practical scale can balance the ecological zones and regional decision-making at the optimal spatial scale.