as a function of time
We inferred the demographic history of the B. schroederi by using the WGS data generated by DNBSEQ-T1 from one individual. We simultaneously perform the same analysis on a giant panda by using resequencing short reads of an individual download from SRA database (accession SRA053353). For this analysis, we used BWA (v0.7.13-r1126) (H. Li & Durbin, 2009) to map the clean reads to each genome with the default parameters. Next, the PSMC method (H. Li & Durbin, 2011) was used to evaluate the dynamic changes of the effective population size (Ne ) of B. schroederi and the giant panda. Following Li’s procedure (H. Li & Durbin, 2011), we applied a bootstrapping approach, repeated sampling 100 times to estimate the variance of simulated results for both B. schroederi and giant panda. We used 0.17 and 12 years per generation (g) and a mutation rate (μ ) of 9×10-9 and 1.29×10-8 for B. schroederi and giant panda, respectively (Cutter, 2008). Since fluctuations in the effective population size of giant pandas have been reported to closely reflect changes in climate and atmospheric dust (S. Zhao et al., 2013), we added the mass accumulation rate (MAR) of Chinese loess over the past 250,000 years for comparison. In addition, we implemented the MSMC2 (Schiffels & Durbin, 2014) which can infer the recent effective population size history. We phased all SNPs of each individual by using beagle (v5.0) (Browning & Browning, 2007) using the following parameters: -i 20 -t 6 -p ’10*1+15*2’. The mutation rate (μ ) of B. schroederi for MSMC2 was the same as used for PSMC.