The thresholds of forest-to-grassland ratios can be critical for harmonizing ecosystem service relationships spatiotemporally in dryland regions

With extensive ecological restoration combating land degradation, known negative effects have caused these measures to shift from widespread vegetation planting to detailed structural reengineering. The specific type and quantity of ecological land have thus become the core issue of future ecological restoration to provide more ecological functions and resist land degradation. This requires adequate landscape structure analysis to identify appropriate patterns for harmonizing multiple ecosystem services (ESs). Identifying how the forest-to-grassland ratio (FGrat) affects ES correlation (EScor) can thus assist in such detailed ecological restoration in dryland regions. During landscape planning, the scale effect is also a problem that must be considered. In this paper, the changes in the trade-offs or synergies of three critical ESs (i.e., soil retention (SR), water yield (WY), and carbon sequestration (CS)) were compared and analyzed at three scales (grid, township and county) during 2000–2020 on the Loess Plateau (LP), which is a typical dryland region. Then, the responses of the EScors to time and the FGrat were analyzed via threshold identification in the different eco-regions. The results show that the direction and change trend of most EScors in different eco-regions were similar at each spatial scale over the past 20 years. Decreased synergy and increased trade-offs can be seen in the EScors of SR-WY and SR-CS in most regions. With the growth of vegetation and natural recovery, the EScors in plain area (PA) and earth-rocky mountainous area (EM) changed from having different trends to being similar to those in other eco-regions, which indicates that natural restoration gradually harmonizes vegetation coverage and naturally regulates the EScors. This highlights the importance of ecological restoration measures that should be based on natural situations. Moreover, the turning points of EScors between SR and WY mostly occurred around 2005 and were affected by artificial ecological restoration. The turning points of the EScors related to CS mainly occurred after 2010 and was also affected by vegetation growth and natural restoration. Spatially, most EScors were synergistic and strengthened as the scale increases, and most of the thresholds response to FGrat identified revealed no significant scale effects, which is significant and conducive to cross-scale research on the planning and management of ecological restoration in drylands. We also identified different FGrat thresholds in six eco-regions with different landscape structures and local environments. When the FGrat thresholds do not change significantly with scale, such as in the response of SR-CS or WY-CS to the FGrats, ecological restoration should mainly consider regional differences rather than scale. This study improves our understanding of how interactions between ESs depend on spatial scales and can inform decision-makers about how to control the FGrat to improve the reliability and sustainability of regional ecological restoration.