Trait-based prediction of targeted species distributions facilitates environmentally adaptive vegetation restoration on the loess plateau

Widespread afforestation has been used to prevent soil erosion on the Loess Plateau. However, neglecting environmental heterogeneity in these efforts has led to mismatches between artificial forests and local conditions, consequently aggravating soil water shortages and compromising ecosystem sustainability. Therefore, selection of appropriate species that match environmental conditions is critical for vegetation restoration. We established artificial vegetation sites corresponding to natural vegetation sites along environmental gradients within steppe, forest-steppe, forest zones in the Yanhe River catchment. We assessed community-level traits and trait–environment relationships to clarify differences in adaptive strategies between natural and artificial vegetation. Thereafter, community assemblages suited to local environmental conditions were predicted to guide species selection for vegetation restoration. Specific leaf area (SLA), leaf nitrogen content (LN), and root nitrogen content (RN) of artificial vegetation (SLA: 162.24 cm 2 g −1 , LN: 26.67 g kg −1 , RN: 9.97 g kg −1 ) in the steppe zone were significantly higher than natural vegetation (SLA: 99.80 cm 2 g −1 , LN: 15.88 g kg −1 , RN: 8.44 g kg −1 ), a pattern not observed in the forest zone. Traits of natural vegetation exhibited significant correlations with mean annual temperature and precipitation. However, artificial vegetation in the steppe zone lacked trait values conferring adaptability to the local environment. Using Traitspace model, we successfully predicted the distributions of common species and potential community assemblages across the three vegetation zones (R 2 = 0.59). This study emphasized functional traits as key predictors for community assemblages and provide a reference standard for environmentally adaptive vegetation restoration and management in ecologically vulnerable areas. • Artificial vegetation in steppe zone lacked adaption compared to natural vegetation. • Artificial vegetation showed weak trait-environment relationships. • Targeting natural vegetation traits improved species selection for revegetation. • Suitable species were successfully predicted for revegetation on the Loess Plateau.