A total of 469 chickens, belonging to 19 different chicken breeds from eight different countries and three imported breeds were used in this study. The 22 chicken breeds included six Korean native breeds (KNG, KNB, KNR, KNW, KNY, KNO), four Laotian (LYO, LCH, LBB, LOU), three Indonesian (INK, INS, ING), two Vietnamese (VTN, VNH), one Kyrgyzstani (KGPS), one Mongolian (MGN), one Nepalese (NPS), one Sri Lankan (SBC), and three imported (RIR, Rhode Island Red; WLG, White Leghorn; CON, Cornish). Detailed information on these breeds can be found in Table 1. Ulnar venous blood of the six Korean native breeds and the three imported breeds was collected from the Animal Genetic Resources Research Center at NIAS. Genomic DNA was extracted from the blood using the Wizard Genomic DNA purification Kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Genomic DNA of the other chicken breeds was obtained from each country, for the purposes of the AFACI AnGR project. The DNA concentrations were quantified by UV Spectrophotometer (Nanodrop ND-1000; Thermo Scientific, Waltham, MA, USA) and the samples were diluted to a final concentration of 10 ng/μL in distilled water. This experiment was conducted with the approval of the NIAS Committee on the Ethics of Animal Experiments (approval number: 2018-048).

Ten MS markers were selected from among the International Society for Animal Genetics / Food and Agriculture Organization of the United Nations (ISAG/FAO) recommended markers. Another Ten MS markers were selected based on their high heterozygosity in the Ark database website (Roslin Bioinformatics Group, Edinburgh, UK). The information for the twenty MS markers used in this study is available in Supplementary Table A. Extracted DNAs were amplified by the GeneAmp PCR 9700 system (Applied Biosystems, Foster, CA, USA) using AccuPower Negative dye PCR PreMix (Bioneer, Daejeon, Korea), including DNA polymerase, dNTP, Tris-HCl, KCL, and MgCl₂. The polymerase chain reaction (PCR) reactions were performed in a total reaction volume of 20 μL containing 2 μL of template DNA, and 0.4 to 2.6 μL (2 pmol/μL) of primer based on the multiplex combinations. The initial denaturation was performed at 95°C for 5 min, followed by 35 cycles of 60 s at 95°C, 45 s of annealing at 58°C to 62°C based on the multiplex combination, 60 s of extension at 72°C, a final extension at 72°C for 30 min, and then cooling to 4°C.

After PCR amplification, the genotyping reaction mixtures were made using 1 μL of the PCR products, 10 μL of Hi-Di Formamide (Applied Biosystems, USA), and the GeneScan 500 LIZ Size Standard (Applied Biosystems, USA) mixture. The genotyping reaction mixture was denatured for 10 min at 95°C and then immediately placed in ice. Electrophoresis was performed using capillary arrays in an ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems, USA). The allele sizes were determined using GeneMapper Software 5 (Applied Biosystems, USA) and was analyzed statistically.

The allele frequencies, the number of alleles, expected heterozygosity (H_exp), observed heterozygosity (H_obs), and polymorphism information content (PIC) values for each of the chicken breeds across the 20 loci were calculated using the MS Tool Kit [23]. Nei’s D_A genetic distances between breeds were calculated using the DISPAN software [24]. The output file for the neighbor-joining (NJ) phylogenetic tree was generated using the PHYLIP package [25] and visualized using TreeView 1.6 [26].

The genetic structures and the degree of admixture among the 22 chicken breeds were analyzed using the Bayesian clustering procedure of STRUCTURE ver 2.3.4 [27]. Twenty independent runs were performed for each K value from 2 to 22. For all runs, the admixture models had a burn-in period of 20,000 repeats, followed by 100,000 repeats of the Markov chain Monte Carlo algorithm. To identify the K value that best fits the data, STURCTURE HARVERSTER [28] was used, which implements the Evanno method [29]. The CLUMPP program ver 1.1.2 [30] was used to align the 20 repetitions of each K value. The CLUMPP output files were visualized using the DISTRUCT program ver 1.1 [31]. Principal coordinate analysis (PCoA) was conducted using the adegenet package [32] in R Studio [33].

To obtain insight into the genetic diversity and population structures, the H_exp, H_obs, and the PIC value for each locus were calculated using the MS Tool kit (Table 2). From the 20 MS markers, a total of 193 alleles were identified in the 22 chicken breeds. The number of alleles ranged from 3 (MCW0103) to 20 (LEI0192) (mean 9.65). The means of H_exp, H_obs, and PIC were 0.598±0.022, 0.591±0.020, and 0.523±0.023, respectively, for the 20 MS markers. The lowest values of H_exp (0.348±0.036), H_obs (0.382±0.046), and PIC (0.272±0.027), were all found at the MCW0103 locus, whereas the highest values of H_exp (0.714±0.021) and PIC (0.640±0.022) were found at the ADL0176 locus, and the highest value of H_obs (0.735±0.034) was found at the MCW0193 locus. Moreover, MCW0103 showed low polymorphism levels in previous studies [34–36]. Botstein et al [37] reported that MS markers with PIC≥0.5 and H_exp≥0.6 were highly informative for genetic analysis. Our study demonstrated that 13 of the 20 MS markers were highly informative for discrimination analysis, and would be appropriate for the analysis of the 22 chicken breeds.

Furthermore, the genetic diversity parameters of the 22 chicken breeds were calculated using the MS Tool Kit (Table 3). The mean number of alleles ranged from 2.15±0.93 (MGN) to 6.50±3.02 (NPS). The imported breed WLG showed the lowest H_exp (0.372±0.055), H_obs (0.384±0.019) and MGN showed the lowest PIC (0.306±0.048), while NPS showed the highest H_exp (0.718±0.027) and PIC (0.663±0.030). Correspondingly, although the population size was small, LCH had the highest H_obs (0.690±0.039). Overall, the diversity of the imported breeds was lower than that of local chicken breeds in 8 countries, except for MGN. This is because the imported breeds have been strongly selected for their performance characteristics and breeding purpose (meat type, egg type, etc.) for many decades. In our study, the genetic diversity of the Korean native chicken (KNC) breeds was lower than that of the other Asian chicken breeds tested. The reason for this may be that five KNC breeds (KNY, KNR, KNW, KNG, and KNG) were restored in 2008 by NIAS, according to the genetic fixation of their different feather colors [38], and this fixation was absent in the other Asian chicken breeds tested. Thus, we further investigated whether the genetic distances could be discriminated amongst between the 22 chicken breeds.

To further investigate the genetic divergences among the breeds using the 20 MS marker allele frequencies, we estimated Nei’s D_A genetic distance between pairs of breeds, for all 22 chicken breeds, using the DISPAN program (Supplementary Table B); the shortest genetic distance was between VNH and VTN at 0.086, and the longest was between KNG and MGN at 0.503. Moreover, to understand the evolutionary relationships among the chicken breeds, an NJ phylogenetic tree was constructed using the PHYLIP program based on the D_A genetic distance (Figure 1). In our study, three main branches appear in the phylogenetic tree. The first main branch comprised the KNC breeds (except for KNO), and the RIR and SBC breeds. The WLG, CON, KNO, MGN, and KGPS breeds constituted the second major branch, and this branch was further subdivided into the WLG and CON groups. KNO was grouped with CON, whereas MGN and KGPS were grouped with WLG. The third main branch was comprised of the other Asian chicken breeds. This suggests that the KNC breeds are clearly genetically separated from the other Asian chicken breeds that we studied.

We conducted clustering analysis using Bayesian clustering, which provided more accurate estimates of relatedness of the breeds [39]. According to the STRUCTURE analysis the most probable number of inferred clusters and the K value (ΔK), was K = 20 (46.92). The genetic structures of each chicken breed (for K = 2, 4, 9, 14, and 20) were visualized using DISTRUCT (Figure 2). For K = 2, the KNC breeds (excepting KNO) and SBC were clustered with RIR, whereas the other Asian chicken breeds were clustered with WLG and CON. WLG and MGN were distinguished from the other breeds at K = 4. These two breeds were found to differ the most in terms of genetic composition, compared with the other breeds. Differentiation from the other breeds began for RIR and KGPS at K = 9, and for CON at K = 14. Based on the CLUMPP analysis at K = 20 (Supplementary Table C), 11 breeds (RIR, WLG, CON, KNG, KNB, KNR, KNW, KNY, KNO, KPGS, and LCH) were detected in independent cluster and each of these breeds occurred predominantly in one cluster (with more than 84% of its membership in one cluster). Moreover, MGN had 88.9% and WLG had 95.1% membership in cluster 18, and the genetic distances between them were short (0.193). It was difficult to distinguish between WLG and MGN, suggesting that MGN was derived from WLG. SBC was also detected independently in cluster 1, but with a relatively low proportion of membership (68.3%). The proportion of membership of the other breeds ranged from 25.3% to 79.5% in cluster 13 (LOU) and 17 (INS), respectively. Generally, the genetic uniformity of the imported breeds (RIR, WLG, CON) and KNC breeds was higher than that of the other Asian chicken breeds, except for MGN, KPGS, and LCH. This is probably because these Asian chicken breeds have not gone through genetic fixation processes via strong selection processes. It means they were crossbreed. If genetic uniformity is low, it is difficult to determine whether a breed is distinct from other breeds. Therefore, to increase genetic uniformity, as long as it does not reduce genetic diversity through planned breeding, these Asian chicken breeds should be selected according to their specific purposes.

To assess the relatedness of breeds, we carried out PCoA analysis using the allele frequencies of the 20 MS markers (Figure 3). These results were similar to those of our phylogenetic tree and structure analysis. The percentages in the label of each axis indicate the variance explained by the axis. Three axes explained 56.2% of the variance; the first two explained 19.17% and 18.92%, respectively, and the third axis explained 18.11%. Remarkably, on the first axis, the Laotian, Vietnamese, and Indonesian chicken breeds were clearly separated from the other breeds. In addition, WLG and MGN were completely isolated from the other breeds on the second axis, whereas the Asian chicken breeds were not clearly separated from each other on this axis. On the other hand, CON and KNO were separated from the other breeds on the third axis.

In conclusion, the genetic diversity of the KNC and imported breeds was lower than that of the other Asian chicken breeds, whereas their genetic uniformity was higher, except in the MGN, KGPS, and LCH. Therefore, selection for the specific characteristics of Asian chicken breeds used in this study is necessary to increase their genetic uniformity. Moreover, estimating the genetic diversity between the 22 Asian chicken breeds is the first step in a strategic plan for the genetic characterization and conservation of these breeds. Although the sample size of some of the breeds was small, our findings are meaningful in that the study was conducted using various breeds from different countries. Additionally, this study may be useful as an initial guide for defining conservation objectives, for designing future investigations of genetic variation, and for developing conservation strategies for 22 Asian chicken breeds. However, further research is required to elucidate the specific reasons for the genetic differences between the Asian breeds and the native Korean and imported breeds.