Genetic Variability in Local and Imported Germplasm Chicken Populations as Revealed by Analyzing Runs of Homozygosity

Simple Summary To maintain the uniqueness of conserved chicken populations of local and imported breeds is of great importance. In this study, we genotyped small populations belonging to 14 breeds and 7 crossbreds using an Illumina Chicken 60K SNP (Single Nucleotide Polymorphisms) BeadChip and looked for appropriate methods to characterize their purity/variability. It was not straightforward to identify crossbred individuals, and the best approach was based on calculating the length and number of homozygous regions, or runs of homozygosity (ROH), in the populations studied. The latter enabled most accurate identification of crossbreds and can be served as an effective tool in testing genome-wide purity of chicken breeds. Abstract Preserving breed uniqueness and purity is vitally important in developing conservation/breeding programs for a germplasm collection of rare and endangered chicken breeds. The present study was aimed at analyzing SNP genetic variability of 21 small local and imported purebred and F1 crossbred populations and identifying crossbreeding events via whole-genome evaluation of runs of homozygosity (ROH). The admixture models more efficiently reflected population structure, pinpointing crossbreeding events in the presence of ancestral populations but not in their absence. Multidimensional scaling and FST-based analyses did not discriminate properly between purebred populations and F1 crossbreds, especially when comparing related breeds. When applying the ROH-based approach, more and longer ROHs were revealed in purebred individuals/populations, suggesting this as an effective implement in genome-wide analysis of germplasm breed purity.


Introduction
Assessment of the genome-wide diversity plays a significant role in conserving local and imported genetic resources and maintaining an effective program for breeding commercial populations (e.g., [1][2][3]). Previously, in the framework of a poultry germplasm conservation and utilization research project, we assessed a broad spectrum of the world's chicken germplasm populations for genetic variation at economically important loci [4,5] and multiple single nucleotide polymorphisms (SNP) loci [6][7][8][9].
A comparative genetic evaluation of livestock populations of different origin, population history, and size is an important source of information for genetic changes in their genome, including degree of homozygosity across genome-wide regions [2,3,10]. This is especially desirable when breeding small conserved groups and evaluating crossbreeding/inbreeding effects [11,12]. Lengths of runs of homozygosity (ROH) in the genome of a particular animal eventually depend on selection, gene drift, and herd size in original population [12], and are useful for assessing information about the degree of inbreeding. Long ROHs are typical for inbred individuals, as haplotypes inherited from a common ancestor do not shorten during recombination. In contrast, short ROHs can inform studies of less pronounced inbreeding intrinsic to heterogeneous animals [13][14][15][16].
Since relevant ROH studies in small closed populations kept in a germplasm collection of local, rare, and endangered chicken breeds are missing, the goal of our study was a genome-wide evaluation of SNP genetic diversity in these populations, and a search for criteria to identify occasional hybridization events in population history in order to control breed purity. Herewith we investigated ROHs and linkage disequilibrium (LD) and hypothesized that, when monitoring intra-breed genetic structure, small purebred populations would have more and longer ROHs compared to F 1 crossbreds, which could be served as a marker to detect potential occasional hybridization events.

Materials and Methods
For this study, SNP genotypes of 673 birds were obtained for the following 21 small populations of local and imported breeds that represent a broad sample of the world's chicken germplasm and are maintained at the RRIFAGB Collective Use Centre 'Genetic Collection of Rare and Endangered Chicken Breeds'. These included 14 purebred populations: Amroks Cuckoo (A), Brahma Buff (BB), Brahma Light (BL), Bantam Mille Fleur (or Russian Korolyok) (BMF), White Cornish (C; two-way hybrid C1 bred inter se), Frizzle (F), Hamburg Silver Spangled Dwarf (HSSD), Leghorn Light Brown (or Italian Partridge) (LLB), Pushkin (P), Russian White (two populations, RW1 and RW2), Sussex Light (SL), Tsarskoye Selo (or Tsarskoselkaya) (TC), Uzbek Game (UG); and 7 groups of F 1 crossbred progeny: , and Tsarskoye Selo × (Sussex Light × Amroks Cuckoo) (TC × SL × A) as shown in Table 1.
In each population, both hens and cocks were genotyped using Illumina Chicken 60K SNP BeadChip (Illumina, USA) as performed with the assistance of the Geneseek/Neogen Corporation. SNPs were filtered and removed from the further analysis if they met the following criteria: minor allele frequency ≤0.05, Hardy-Weinberg equilibrium probability ≤1 × 10 −4 , and call rate ≤95%.
After completing the quality control procedures, 44,230 SNP markers from 28 autosomes were available for further analysis.
LD (or r 2 ) coefficients were calculated using the following formula in PLINK 1.9 software [17]: Search for homozygous regions was carried out using the PLINK 1.9 program according to the following algorithm: ROHs containing at least 100 SNPs and a total length of ≥1000 Kb were taken into account, each ROH having had at least one SNP per 50 Kb. The scan window contained 50 SNPs and no more than 1 heterozygous genotype.
Values of ROH-based inbreeding coefficients (F ROH ) were found using the following formula [18]: where k is the number of ROHs, length (ROH k ) is an individual ROH region length, and L is autosomal genome size covered by SNPs. Population variability of SNP markers in purebred and F 1 crossbred populations was estimated in the admixture 1.3.0 program [19]. To compute the maximum likelihood estimates, the most probable number of ancestral populations (K) was selected using genotype data from 44,230 SNPs. For this purpose, a cross-validation procedure was performed. The lowest cross-validation error was found at K = 10, and charts were plotted using the function barplot() in R [20].
Multidimensional scaling (MDS) was performed in Plink 1.9 software [21] using a matrix of identity by state (IBS) distances between samples. To eliminate the effect of individual's sex on IBS distances, SNP markers located on the sex chromosomes were excluded from the MDS analysis and, therefore, genotype data from 44,230 autosomal SNPs were employed.

Results and Discussion
Using whole-genome SNP genotyping, we found that the examined populations of local and imported chicken breeds significantly differed from each other in genetic architecture and total length of ROHs (Figure 1). Most conserved purebred populations had greater numbers of ROHs and longer ROHs as well as higher ROH-based inbreeding coefficients (F ROH ) than F 1 inter-breed crossbreds produced by crossing two or three breeds (Table 1). For example, HSSD, an old Dutch breed that underwent a strict intra-breed type of breeding, was characterized by a greater number of ROHs (63.9 ± 1.5 vs. 32.0 ± 5.2 averaged across other 13 breed populations) and a greater length of ROH (5735.0 ± 164.2 Kb vs. an average of 4193.8 ± 182.7 Kb), with the inbreeding coefficient F ROH being the highest one (0.324 ± 0.011 vs. an average of 0.141 ± 0.023; Table 1). All this suggests that this breed and its particular small population were affected by inbreeding both in earlier times and more recently. The presence of SNPs with higher rates of LD (0.438 ± 0.001; Table 1) also testifies to inbreeding effects in this population.
There were other small closed purebred populations with a higher content of ROHs in their genome, e.g., BB (F ROH = 0.167 ± 0.015), BMF (F ROH = 0.291 ± 0.015), etc. (Table 1). Among these, we found an exception with F ROH = 0.023 ± 0.005 in the decorative F breed known in Europe since 1676 and characterized by a specific structure of feathers due to the F (frizzling) gene. A lower content of ROHs in this breed population can be associated with a targeted introgression of genes of the other breeds to reduce the effect of feather fragility as reflected by the admixture analysis ( Figure 2).
The original Russian White population (RW1) existed from 1953-2003, had one founder and underwent intensive selection for chick tolerance to cold [7,26]. In 2005, in order to maintain genetic diversity in a small population of this breed, a single introductory crossing with White Leghorns resulted in producing the present population RW2. Accordingly, we observed a higher F ROH in RW1 (0.307 ± 0.014) and a lower one in RW2 (0.083 ± 0.006; p < 0.0001; Table 1). Also, these two populations differed in LD, with a lower value being in RW2 as compared to RW1 (p < 0.0001; Table 1). Despite a relatively small size (~200 animals), RW2 is not expected to raise homozygosity because of the previous crossing with White Leghorns that was likely to significantly increase its genetic diversity.
In two-breed F 1 crossbreds, we revealed a decrease in ROH metrics, while three-breed F 1 crossbreds appeared to show a further lowering in their genomic content of ROH (Table 1).  The original Russian White population (RW1) existed from 1953-2003, had one founder and underwent intensive selection for chick tolerance to cold [7,26]. In 2005, in order to maintain genetic diversity in a small population of this breed, a single introductory crossing with White Leghorns resulted in producing the present population RW2. Accordingly, we observed a higher FROH in RW1 (0.307 ± 0.014) and a lower one in RW2 (0.083 ± 0.006; p < 0.0001; Table 1). Also, these two populations differed in LD, with a lower value being in RW2 as compared to RW1 (p < 0.0001; Table 1). Despite a relatively small size (~200 animals), RW2 is not expected to raise homozygosity because of the previous crossing with White Leghorns that was likely to significantly increase its genetic diversity.
In two-breed F1 crossbreds, we revealed a decrease in ROH metrics, while three-breed F1 crossbreds appeared to show a further lowering in their genomic content of ROH (Table 1). The original Russian White population (RW1) existed from 1953-2003, had one founder and underwent intensive selection for chick tolerance to cold [7,26]. In 2005, in order to maintain genetic diversity in a small population of this breed, a single introductory crossing with White Leghorns resulted in producing the present population RW2. Accordingly, we observed a higher FROH in RW1 (0.307 ± 0.014) and a lower one in RW2 (0.083 ± 0.006; p < 0.0001; Table 1). Also, these two populations differed in LD, with a lower value being in RW2 as compared to RW1 (p < 0.0001; Table 1). Despite a relatively small size (~200 animals), RW2 is not expected to raise homozygosity because of the previous crossing with White Leghorns that was likely to significantly increase its genetic diversity.
In two-breed F1 crossbreds, we revealed a decrease in ROH metrics, while three-breed F1 crossbreds appeared to show a further lowering in their genomic content of ROH ( Table 1).
Number of ROHs could reflect the type of breeding program. Small local and imported conserved populations bred inter se had a greater (>20) number of ROHs (e.g., BMF, BB, BL, RW1 and C), meaning a reduced genetic diversity similar to what was shown in small livestock populations kept by individual breeders or in local breeds [2,[27][28][29][30][31]. In contrast, the genomes of two-and threebreed F1 crossbred progenies contained a lesser (<6) number of ROHs (Table 1; Figure 1), which was consistent with lower ROH metrics observed in other studies on crossbred animals (e.g., [3,32]).
Characterization of genetic differentiation in conserved populations of local and imported breeds and monitoring of their purity/variability are an important component of breeding/conservation programs [1,2]. In the presence of ancestral purebred populations, use of the ADMIXTURE program effectively detects F1 crossbred progenies (Figure 2). However, in the absence of Number of ROHs could reflect the type of breeding program. Small local and imported conserved populations bred inter se had a greater (>20) number of ROHs (e.g., BMF, BB, BL, RW1 and C), meaning a reduced genetic diversity similar to what was shown in small livestock populations kept by individual breeders or in local breeds [2,[27][28][29][30][31]. In contrast, the genomes of two-and three-breed F 1 crossbred progenies contained a lesser (<6) number of ROHs (Table 1; Figure 1), which was consistent with lower ROH metrics observed in other studies on crossbred animals (e.g., [3,32]).
Characterization of genetic differentiation in conserved populations of local and imported breeds and monitoring of their purity/variability are an important component of breeding/conservation programs [1,2]. In the presence of ancestral purebred populations, use of the admixture program effectively detects F 1 crossbred progenies ( Figure 2). However, in the absence of ancestral populations, it is difficult to determine whether it is an F 1 crossbred progeny or a purebred population (data not shown). If data on the origin are not taken into account, F 1 crossbred progenies are difficult to identify. In contrast, when conventional MDS and F ST -based analyses (Figure 3a,b) are performed, one cannot always describe properly the population structure, differentiation, and gene flow due to crossbreeding, especially when comparing related breeds.  SNPs and values of (a) identity by state (IBS) matrix (Multidimensional scaling (MDS) plot using the first two principal components, c1 and c2) and (b) FST matrix (Neighbor Joining tree) that did not clearly discriminate between purebred and F1 crossbred populations. Abbreviations of all populations are given in Materials and Methods. There are many chicken breeds/populations in the world that could be similar in phenotype, but different in origin. This may cause a problem of identification and discrimination between breeds/populations, which can be solved with a high accuracy by comparing them against the available global database of SNP genotypes. The differentiation of individuals and their affiliation to a certain population or populations can be assessed using the MDS method. The accuracy of this analysis is significantly affected by the genetic background of populations selected for comparison, the breeding history of populations, and their effective population size. Therefore, there may be a bias in the assessment or false conclusions. For SNPs and values of (a) identity by state (IBS) matrix (Multidimensional scaling (MDS) plot using the first two principal components, c1 and c2) and (b) FST matrix (Neighbor Joining tree) that did not clearly discriminate between purebred and F1 crossbred populations. Abbreviations of all populations are given in Materials and Methods. There are many chicken breeds/populations in the world that could be similar in phenotype, but different in origin. This may cause a problem of identification and discrimination between breeds/populations, which can be solved with a high accuracy by comparing them against the available global database of SNP genotypes. The differentiation of individuals and their affiliation to a certain population or populations can be assessed using the MDS method. The accuracy of this analysis is significantly affected by the genetic background of populations selected for comparison, the breeding history of populations, and their effective population size. Therefore, there may be a bias in the assessment or false conclusions. For There are many chicken breeds/populations in the world that could be similar in phenotype, but different in origin. This may cause a problem of identification and discrimination between breeds/populations, which can be solved with a high accuracy by comparing them against the available global database of SNP genotypes. The differentiation of individuals and their affiliation to a certain population or populations can be assessed using the MDS method. The accuracy of this analysis is significantly affected by the genetic background of populations selected for comparison, the breeding history of populations, and their effective population size. Therefore, there may be a bias in the assessment or false conclusions. For example, the distribution of F1 crossbreds in Figure 3a does not directly reflect their origin. The tree topology based on FST analysis (Figure 3b) also does not clearly suggest the original breeds used for producing F1 crossbreds.

Conclusions
Exploration of the length and number of ROHs can be helpful in discriminating between purebred and F 1 crossbred animals, and this approach can be used as a tool in selecting purebred individuals for conserved propagation of local and imported breeds. ROH-based characterization in individuals and in the whole population can be used in adjusting germplasm breeding/conservation programs and identifying events of inter-breed gene transfer.

Conflicts of Interest:
The authors declare no conflict of interest. The funder had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.