The WA-PLS (weighted averaging partial least squares) method was applied to model the vegetation coverage (indicated by NDVI; Normalized Derivation Vegetation Index) based on the correspondence between modern vegetation coverage and pollen assemblage in topsoil (Cao et al., 2022; Cheng et al., 2021; Zheng et al., 2014). The modern NDVI data are derived from Global Inventory Modelling and Mapping Studies (GIMMS; https://ecocast.src.nasa.gov/data/pub/gimms/) with a spatial resolution of 8 km × 8 km and a temporal resolution of 15 days. The annual mean NDVI values in and around the Daling River Basin from 1982 to 2015 were obtained from the monthly NDVI data by the maximum value composite method. A modern pollen database of China, far from human-affected areas, is collected from references (Herzschuh et al., 2019; Yang et al., 2016).Before calculation via R software, a percentage of less than 2% in more than 3 samples, aquatic pollen, and Poaceae larger than 40 μm, were removed, which cannot indicate climate change for its low content and human impact as a planted crop (Wen et al., 2013). Seven calibration-set modern pollen assemblages around the study area (500 km/600 km/700 km/800 km/900 km/1000 km/all samples) were applied to determine the spatial extents (Table S10). Performances of the climate-vegetation transfer mode WA-PLS showed that small-scale calibration sets (<700 km and <1000 km) could be the most appropriate to accurately reconstruct regional NDVI by cores in the NQ-1 and Sihailongwan Lake, respectively (Table S3). The code has been depicted in the reference (Cao et al., 2022).