2.5 Individual and population genetic diversity
Genome-wide genetic diversity was measured at both individual and population level. For the individual level, genome-wide observed heterozygosity (H O) was calculated by averaging the values of both variant and non-variant sites. The genotype data that include both variant and non-variant sites were re-analysed, and mono/bi-allelic SNPs with read depth ranging from 4 to 24 were retained for estimation of genetic diversity. H O at each locus was calculated as the generalized gametic heterozygosity which estimates the probability of randomly sampling two different alleles from an individual (Meirmans et al., 2018; Moody et al., 1993). Hence, a bi-allelic SNP had a H O value of 1 for diploids, and 2/3 for triploids because of three homologous chromosomes. For population level, genome-wide nucleotide diversity (π) was calculated using the R package PopGenome (Pfeifer et al., 2014) with a 10 kb window size.
2.6 Detection ofruns of homozygosity (ROH)
To assess whether inbreeding occurred in invasive populations, the genome-wide inbreeding level of each individual was estimated by detecting the total length and number of ROHs using the R packagedetectRUNs (Biscarini et al., 2018). As ROHs were detected only based on the genotype state (homozygous/heterozygous), the triploid VCF file was converted into diploid format which could be analyzed withdetectRUNs . The following constraints were applied to the sliding-window-based run detection: (1) The minimum number of SNPs in a ROH was 138 for diploids and 116 for triploids which was determined by the method described in Purfield et al. (2012) (Purfield et al., 2012). (2) One heterozygote and two missing genotypes were allowed in a window. (3) The maximum gap between consecutive SNPs was 300 kb. (4) The minimum length of a run was 500 kb. (5) The minimum SNP density was set to one SNP per 100 kb. The genomic inbreeding coefficient based on ROHs (F ROH) was calculated following the method described in McQuillan et al. (2008) (McQuillan et al., 2008):F ROH,i =L ROH/L AUTO, whereL ROH is the total length of ROHs in individuali , and L AUTO is the length of autosomal genome, here assumed to be 1.54 Gb. The ROH segment length (E (LIBD-H |gcA )) represents the number of generations from the common ancestor (gcA ), estimated asE (LIBD-H |gcA )= 41/(2gcA ) (Curik et al., 2014) assuming that 1 cM is roughly equivalent to 0.41 Mb (Liu et al., 2017a).