Population structure
Two approaches were used to examine the population structure of the black-faced spoonbill as sampled from Taiwan and Hong Kong. Firstly, we used ADMIXTURE v1.22(Alexander & Lange, 2011), which is based on a Bayesian clustering method, to infer the number of ancestors of the black-faced spoonbills sampled in current study. Only unlinked autosomal SNPs were used for the admixture analysis. We used PLINK (Purcell et al., 2007) to filter out linked autosomal SNPs. The window size to calculate pair-wise SNP-SNP r 2 was set to 10 Kb, and the step of the sliding window was set to 2 Kb. The threshold ofr 2 was set to 0.2. SNPs that significantly deviated from the Hardy-Weinberg expectation (p < 0.001) were also filtered out. In total 25,172 unlinked autosomal SNPs (r 2 < 0.2) with > 10× coverage were used in the admixture analysis.
Secondly, we used GCTA (Yang, Lee, Goddard, & Visscher, 2011) to carry out principal components analysis (PCA). The same set of 25,172 unlinked autosomal SNPs was analyzed. The score of the first component (PC1; 1.33% of variance) was plotted against the score of the second component (PC2; 1.28% of variance) of each individual to illustrate the genetic affinity between the black-faced spoonbill samples.