**2. Materials and Methods**

#### *2.1. Fish Sampling*

We assessed five collections of yellowtail kingfish from China (wild and farmed), Japan (wild), and Australia (wild and farmed) (Figure 1): wild yellowtail kingfish (n = 33) from the Yellow Sea, off the coast of Dalian, China (39◦57 N, 123◦04 E); F1 generation yellowtail kingfish (n = 30) were provided by the Dalian Fugu Food Co., Ltd. (Zhuanghe, Dalian, China); wild yellowtail kingfish (n = 20) were collected off the Goto Islands, Japan ¯ (32◦45 N, 128◦27 E); wild yellowtail kingfish (n = 29) were collected from the Tasman sea off of New South Wales, Australia (33◦44 S, 151◦45 E); and F2 generation, farmed yellowtail kingfish (n = 31) were provided by the South Australian Department of Primary Industries and Regions. Pectoral fin clips were collected from the wild and farmed fish from China and Australia; the fin tissues were immersed in 95% ethanol and stored in a refrigerator at 4 ◦C before genomic DNA was extracted. Genomic DNA from the fin clips of wild fish from Japan was provided by a collaborator at the University of Tokyo.

**Figure 1.** Geographical distribution of the yellowtail kingfish collections sampled in this study.

#### *2.2. DNA Extraction, SNP Library Construction, and SNP Marker Typing*

DNA from fin clips was extracted using animal-tissue genomic DNA Extraction Kits (DP121221; Qiagen, Hilden, Germany). DNA concentration was measured using a NanoDrop 2000 (Thermo Fisher, Waltham, MA, USA), and DNA integrity was checked using 1% agarose gel electrophoresis. Extracted genomic DNA was stored at −20 ◦C until use. The 2b-RAD libraries were constructed by Qingdao OE Biotechnology Co., Ltd. (Qingdao, China), following the methods of Wang et al. [16]. In brief, 100–200 ng

of each DNA sample was digested with a Type IIB restriction enzyme (Bsa XI), and the digested products were ligated to restriction-site-specific adaptor sequences using T4 DNA Ligase (New England Biolabs Inc., Ipswich, MA, USA). Sample-specific barcodes were incorporated by PCR. PCR products were purified using the MinElute PCR Purification Kit and then pooled for paired-end sequencing using the Illumina Hiseq Xten sequencing platform (Illumina, CA, USA). Raw reads were cleaned by removing adaptor sequences, reads containing more than 8% unknown (N) bases, and low-quality reads (those where more than 15% of all bases had quality values lower than Q30). We used PEAR v0.9.6 (Heidelberg, Germany) [21] to stitch the pairs of clean reads together, and then reads corresponding to each sample were extracted based on sample location at the time of library building. Reads containing enzyme recognition sites were extracted. The selected samples were CW13, CW23, CW26, CW29, CW32, CW3, AF13, AF23, AF3, AW13, AW23, AW3, CF13, CF23, CF3. The ustacks module in the Stacks software package v1.34 (Eugene OR, USA) [22] was used to cluster the sequences and identify the reference sequences. The extracted reads containing enzyme recognition sites were aligned to the reference sequences using the SOAP software (Shenzhen, China) [23] with the following requirements: unique alignment, optimal alignment, and with a maximum of two mismatches allowed (−r 0 −M 4 −v 2). We performed 2b-RAD genotyping using the maximum likelihood (ML) method in RADtyping software [24]. To ensure the accuracy of subsequent analyses, we excluded the following SNP sites from all samples: those for which less than 80% of the associated individuals could be typed; those with MAF (minor allele frequency) values less than 0.01; those with one or four alleles; and those with more than one SNP within the tag.
