A total of 96 mixed indigenous Tswana chickens were used in the study. The indigenous chickens from the Southern part of Botswana were represented by normal strain (n = 39), naked neck strain (n = 32), dwarf strain (n = 13) of Tswana chickens kept under traditional free-running management system and commercial broiler (n = 12). The chickens used were approximately six months of age or older as per the information provided by the owners. Information on sampling locations and the number of samples per sampling location is available from Table 1.

Blood samples were collected from the medial metatarsal vein located on the leg of a chicken better suited for puncture using a 23-gauge, 1-in needle. The alternative site for blood collection was the brachial vein on the wings and for puncture, feathers in this area were plucked for smooth insertion of needle on the

veins of interest. All blood in vacutainer tubes containing EDTA and kept under cool environment of ice packs until they reach the laboratory where blood samples were kept at −20˚C until DNA extraction.

24 µl of NucleoMag® B-Beads and 360 µl MB2 Buffer were then added to the square-well Block and mixed by pipetting up and down, shaking for 5 minutes at room temperature. Magnetic beads were then separated against the wells by placing the square-well block on the NucleoMag SEP magnetic separator for at least 2 minutes. The supernatant was then removed from the wells and discarded by pipetting. The square-well block was then removed from the NucleoMag SEP magnetic separator and 600 µl of MB3 buffer was added to each of the wells, accompanied by shaking to completely resuspend the beads. Magnetic beads were again separated against the wells by placing the square-well block on the NucleoMag SEP magnetic separator for at least 2 minutes. The supernatant was again removed and discarded by pipetting.

The square-well block was removed again from NucleoMag SEP magnetic separator. 600 µl of MB4 buffer was then added to each of the wells and the beads were resuspended by shaking for 5 minutes. Magnetic beads were again separated by placing the square-well block on the NucleoMag SEP magnetic separator for at least 2 minutes and supernatant was removed and discarded by pipetting. 900 µl of MB5 buffer was then added to each of the wells while the beads were still attracted to magnets. After an incubation period of 50 seconds, the supernatant was aspirated and discarded. The square-well block was then removed from the NucleoMag SEP magnetic separator. 50 µl of DNA elution buffer was then added to each of the wells and shaking for 10 minutes at 56˚C to resuspend the beads. Magnetic beads were again separated by placing the square-well block on the NucleoMag SEP magnetic separator for at least 2 minutes. The supernatant containing purified genomic DNA was then transferred to the elution plate for SNP genotyping.

SNP genotyping was carried out at Agricultural Research Council-Biotechnology Platform in Pretoria according to the protocols described by [16] Khanyile et al. Briefly, SNP genotyping was carried out using the Illumina chicken iSelect SNP 60Bead chip using the Infinium assay compatible with the Illumina HiScan SQ genotyping platform. This Infinium assay is designed to analyze a large number of SNPs at many loci concurrently through multiplexing [17] http://www.illumina.com. SNP calling was done using Illumina Genome Studio v2.0. The genotype input file was converted into a PLINK (v1.07) [18] Purcell et al. input file using a plug-in compatible with Genome Studio program. SNP quality control was done in a series of steps depending on population genetic parameters estimated.

An original data set consisting of all four populations of Tswana chickens (naked neck, Normal and Dwarf populations) was filtered for SNPs that had minor allele frequency (MAF) ≤ 0.02 and this resulted in a total sample of 82 chickens across the three populations. There were 54,293 SNPs available to estimate observed and expected heterozygosity indices (Ho and He) and the inbreeding coefficient of each population. PLINK (v1.07) software [17] was used to estimate observed and expected heterozygosity, inbreeding coefficient and minor allele frequency distribution per population using the comprehensive data set before pruning for MAF. Bins were set for minor allele frequencies of 0 - 0.05, 0.05 - 0.1, 0.1 - 0.2, 0.2 - 0.3, 0.3 - 0.4 and 0.4 - 0.5 and the proportion of SNPs per bin was calculated by dividing the number of markers per bin by the total number of markers included in the MAF estimation according to [16] Khanyile et al.

A complete SNP data set with all four populations was filtered to remove SNPs that were on sex chromosomes or had their positions unmapped. Markers with missing data > 5%; that had a MAF ≤ 2% or were monomorphic were removed from the complete data set. SNPs that were in high linkage disequilibrium at a threshold of LD ≥ 0.2 were also filtered out of the complete data set. Individuals with missing genotypes of more than 5% and those that were closely related, as inferred by a kinship estimator ≥ 0.45 were also excluded from the analysis.

A principal component analysis (PCA) was then performed to establish relationships among different strains of Tswana chickens and the commercial broiler line using the Golden helix SNP variation suit (SVS) version 8.1 [19]. Furthermore, the Admixture 1.23 software [20] was used to estimate the most probable number of ancestral populations based on the SNP genotype data as described by [16] Khanyile et al. Admixture was run from K = 2 to K = 4 and the optimal number of clusters (K-value) was determined as that which had the lowest cross-validation error (CV error).

Pairwise identity by state (IBS) distances between all four chicken populations (naked neck, normal and dwarf strains of Tswana chicken and the commercial broiler) were calculated using PLINK v1.9. Genetic distances between the four populations were evaluated based on Nie’s (1987) unbiased. Genetic distance uses the R-package [21]. To evaluate pair-wise genetic differentiation, the fixation index Fst [22] 4 was calculated for all pairs of chicken populations.

Complete SNP data for the individual populations were filtered to remove SNPs on sex chromosomes or those were not mapped, those with MAF ≤ 5%, those that deviated from Hardy-Weinberg equilibrium (HWE) (P ≤ 0.001) and individual chickens with missing genotypes (>5%) and those with very close kinship (IBD ≥ 0.45) using PLINK (v1.07) [18]. After all quality control measures 46,084 out of 48,054 in the normal strain, 45,080 out of 46,909 in the naked neck strain, 38,781 out of 42,804 in the dwarf strain of Tswana chickens, and 36,401 out of 39,739 in the commercial broiler were available for the estimation of linkage disequilibrium. The resultant individual population data sets after quality control measures were used in the estimation of linkage disequilibrium and associated measures.

Pairwise r² estimation was used to measure LD between pairs of SNPs within a chromosome and population using PLINK (v1.07) program [18] 2007 for SNPs on chromosomes 1 - 28 that had passed quality control tests detailed above. According to [23] Lu et al., the r² measure, defined as the squared correlation coefficient of alleles at two loci was chosen because it is independent of allele frequency. Briefly, its calculation, considers two loci, A and B, each locus having two alleles (denoted A1, A2; B1, B2, respectively) [24]. The frequencies of the haplotypes will then be denoted as F11, F12, F21, and F22 for haplotypes A1B1, A1B2 and A2B2, respectively and as FA1, FA2, FB1 and FB2 for A1, A2, B1 and B2 alleles, respectively. From this, r² according to [16] Khanyile et al. were then calculated as shown in Formula (1) below:

r 2 = ( f 11 f 22 − f 12 f 21 ) 2 / F A 1 F A 2 F B 1 F B 2 (1)

PLINK by default only reports r²-values above 0.2 and to allow reporting of all r²-values observed in the populations, the −r²-window-ld0 option was used. An additional option, −r²-window-snp 5000 kb 10,000 described by [16] Khanyile et al., allowed for estimation of r² for SNP marker pairs separated by at most 5000 SNPs and within a 10 MB SNP interval.

The effective population size trends were estimated using the procedure described by [16] Khanyile et al. Briefly, the relationship between N_e, recombination frequency, and expected LD (r²) was determined using the equation from [25] Corbin et al. shown in Formula (2):

E [ r a d j 2 ] = ( α + 4 N e c ) − 1 (2)

where α = 1 when assuming no mutations and 2 if mutation was considered, r a d j 2 = r 2 − 1 / 2 n , c was the recombination rate, and n was the chromosomal sample size. The effective population size N_e, as 1/2c generations, was estimated from the adjusted r a d j 2 values related to a given genetic distance d in Morgans, assuming c = d [24]. For each pair of SNPs on each chromosome, recombination rate was estimated by converting physical marker interval length x_i (MB) to the corresponding genetic length c_i using the formula: c_i = ṓ_ix_i, where ṓ_i is the average ratio of Morgans per kilo base pair on chromosome I, which was taken from physical lengths of the chicken genome v74 [26]. The genetic length of chromosomes was adopted from [27]. The r² – values range from 0 and 1, whereby a zero value indicates uncorrelated SNPs while a value of one reflects SNPs that are perfectly correlated [24]. The trends in effective population sizes for each of the defined subpopulations were then estimated by setting bins at 10, 20, 40, 60, 100, 200, 500, 1000, 2000 and 5000 kb. The bins were designed to cover the genome in tens, hundreds, thousands and hundred thousand base pairs.

The observed heterozygosity (H_o) values were 0.610 ± 0.012, 0.611 ± 0.014, 0.613 ± 0.0006 for normal, naked neck and dwarf strains of Tswana chickens respectively. There were no significant differences in H_o between the three strains of Tswana chickens. The H_o averaged 0.611 ± 0.016 across the three strains of Tswana chickens compared to H_o of 0.347 ± 0.023 in commercial broiler chicken. The expected heterozygosity (H_e) values were 0.613 ± 0.00012, 0.614 ± 0.00013, 0.608 ± 0.00021 for normal, naked neck and dwarf strains of Tswana chickens respectively. There were no significant differences in H_e between the three strains of Tswana chickens. The H_e averaged 0.612 ± 0.00015 across the three strains of Tswana chickens compared to H_e of 0.577 ± 0.00022 in commercial broiler chicken.

The H_o and H_e values for the three strains of indigenous Tswana chickens are similar with the H_o and H_e values found in indigenous chickens of Malawi, Zimbabwe and South Africa as reported by [16] Khanyile et al. Higher molecular diversity in indigenous Tswana chickens compared to commercial broiler chickens is consistent with [28] Al-Atiyat and Abudabos, who reported higher gene diversity in indigenous chickens of Jordan than in Ross broiler chickens (H_e of 0.54 vs 0.09). Higher genetic diversity in indigenous Tswana chickens than commercial broiler chickens might be due to inherent traditional breeding practices of natural and random mating of indigenous chickens. Indigenous Tswana chickens are also not subjected to intensive selection in various traits of economic importance which tends to promote diversity than uniformity. Lower genetic diversity in commercial broiler compared to indigenous Tswana chickens might be due to artificial selection for traits of economic importance such as meat production [28].

In the normal and naked neck strains of Tswana chickens, H_o was less than H_e while in the dwarf strain H_o was greater than H_e is indicating that the dwarf strain was significantly outbreed while the other two strains of Tswana chickens were inbred. In the commercial broiler chicken, H_o was also less than H_e also indicating that the broiler chicken was inbred. The inbreeding coefficients (Fis) were positive for the normal, naked neck and commercial broiler and negative in the dwarf strain of Tswana chicken. Of all the four strains only the dwarf strain was therefore outbred. The commercial broiler had the highest levels of inbreeding compared to indigenous Tswana chickens. The three strains of indigenous Tswana chickens generally had lower inbreeding coefficients than indigenous chickens of Malawi, Zimbabwe and South Africa reported by [16] Khanyile et al.. Higher levels of inbreeding in commercial broiler chickens than in indigenous Tswana chickens were expected as inbreeding is part of the breed development process. Intensive selection during development of commercial broiler chickens reduced diversity and increased uniformity partially as result of inbreeding.

The minor allele frequency (MAF) was also presented in Table 2 for each population from generated data set since it provides information to differentiate between common and rare variants in the population. The analysis of SNP markers indicated an average minor allele frequency (MAF) of 0.266, 0.266, 0.241 and 0.259 in the normal, naked neck, dwarf and broiler strains, respectively.

Pairwise population (F_ST) was calculated from filtered SNPs to investigate population differentiation among different strains of Tswana chickens. F_ST values are shown in Table 3 and varied from 0.013 to 0.084. According to [30] Hussein et al., it is generally recognized that F_ST values ranging from 0 to 0.05 indicate low genetic differentiation; those ranging between 0.05 and 0.15 medium differentiation; those between 0.15 and 0.25 indicate big differentiation and those values above 0.25 indicate very big genetic differentiation. There was low genetic differentiation between the normal strain of Tswana chicken and the commercial broiler chicken (0.031). There was, however, moderate genetic differentiation between the dwarf strain of Tswana chicken and commercial broiler chicken and between the naked neck strain of Tswana chicken and the commercial broiler chicken (0.084, 0.054 respectively).

Low genetic differentiation occurred between dwarf and naked neck (0.040), dwarf and normal (0.040), and between normal and naked neck (0.013) strains of Tswana chickens. Pairwise genetic distances between different strains of Tswana chickens and commercial broiler are highly correlated with genetic differentiation values. Generally, the higher the genetic differentiation between strains the higher the genetic distances. The low genetic distances occurred between normal and naked neck strains of Tswana chickens (0.013) and between normal strain of Tswana chicken and commercial broiler chicken (0.032). Moderate genetic distance occurred between naked neck and dwarf (0.042) and between normal and dwarf (0.043) strains of Tswana chickens. High genetic distances occurred between naked neck and commercial broiler (0.057) and between dwarf and commercial broiler (0.092). Generally, less genetic differentiation and fewer genetic distances occurred between the three strains of indigenous Tswana chickens as compared to between the commercial broiler and the three strains of Tswana chickens. The less genetic differentiation and genetic distances between the three strains of Tswana chickens support the hypothesis of closer evolutionary history of common origin of the three strains of Tswana chickens. Generally higher genetic differentiation and genetic distance between strains of

indigenous Tswana chickens and the commercial broiler chickens might be since indigenous Tswana chickens are not intensively selected while commercial broilers are heavily selected during their developmental process.

A summary of r² values for the 28 chicken autosomal chromosomes in the three strains of Tswana chickens and commercial broiler chicken are shown in Table 4. Overall population LD over all chromosomes ranged from 0.067 to 0.241 in indigenous Tswana chickens and from 0.342 to 0.407 in commercial broiler chicken and averaged 0.128 ± 0.056 and 0.375 ± 0.0152 in indigenous Tswana chickens and commercial broiler chicken, respectively. The highest LD values were found in chromosome 16 in the normal and naked neck strains of Tswana chickens and in chromosome 23 in the dwarf strain of Tswana chicken. Higher LD values on chromosome 16 for naked neck and normal strains of Tswana chickens are consistent with [16] Khanyile et al. who also reported higher LD values in chromosome 16 for both village chickens and conservation flocks. The lowest LD values were found in chromosome 23 in the naked neck and normal strains of Tswana chickens and in chromosome 25 in the dwarf strain of Tswana chicken. Low LD values on chromosome 25 in the dwarf strain of Tswana chicken are consistent with [16] Khanyile et al. who also reported low LD values on chromosome 25 for both village and conservation flocks. Chromosome 17 had high LD in the commercial broiler strain while chromosome 16 had low LD. Consistent with [16] Khanyile et al., the current study also indicates that evolutionary forces affecting LD act differently on different chromosomes and different strains. Commercial broiler chicken had higher LD compared to the three strains of indigenous Tswana chickens probably because of the effects of artificial selection for higher meat yield. On the other hand, natural selection could be a major evolutionary force in the three strains of Tswana raised under free-running management systems with minimal artificial selection [16]. There was no significant difference in LD between the normal and naked neck strains of Tswana chickens. However, the two strains of Tswana chickens had significantly lower LD than dwarf strain of Tswana chicken. Of the four chicken strains, the commercial broiler chicken had significantly higher LD compared to the three strains of Tswana chickens.

Higher LD in commercial broiler compared to the three strains of Tswana chickens is consistent with [16] Khanyile et al. who found significantly higher LD in conservation flocks compared village chicken populations kept by small holder farmers. Differences in LD between commercial broiler and the three strains of Tswana chickens could be due to their different evolutionary histories under the influence of random genetic drift, selection, and mutations [16]. The dwarf strain of Tswana had higher LD across the 28 autosomal chromosomes compared to normal and naked neck strains of Tswana chickens. Higher LD in the dwarf strain compared to the naked neck and normal strains of Tswana

chickens is consistent with its low diversity as measured by both observed and expected heterozygosity (H_o and H_e) and its low effective size (Figure 4). Lower effective population size in the dwarf compared to the normal and naked neck strains of Tswana chickens are consistent with the actual population of dwarf chicken in the general population of Tswana chickens. The general population of Tswana chickens comprises mostly the normal strain followed by naked neck strain and lastly the dwarf strain of Tswana chickens [13]. Across the four strains under investigation the normal strain had the lowest LD (0.087) and the commercial broiler had the highest LD (0.375).

Plots of estimated effective population size (N_e) at t-generations ago for various strains of Tswana chickens and commercial broiler chickens are shown in Figure 4. The adjusted Linkage Disequilibrium (LD) based estimates of N_e indicated low effective population sizes of 123, 738 and 748 for dwarf, naked neck, and normal strains of Tswana chickens, respectively, 19 generations ago. Higher Ne of 613, 2640, 2650 was found for the dwarf, naked neck, and normal strains of Tswana chickens respectively, 120 generations ago. The effective population size (N_e) was even higher with values of 3768, 7477 and 7891 for the three strains of Tswana chickens, respectively, 982 generations ago. The commercial broiler had N_e of 147, 859 and 5994 at 19, 120 and 982 generations ago, respectively. The graphs illustrate a steady decrease in N_e from over 980 to below 20 generations ago in both indigenous Tswana chickens and commercial broiler chickens. The naked neck and normal strains of indigenous Tswana chickens showed similar patterns of decreasing N_e values but had higher N_e values than the dwarf strain. In comparison with the three strains of indigenous Tswana chickens, the commercial broiler chicken had higher N_e values at all generations than the dwarf strain. The LD patterns are consistent with effective population size and diversity patterns in the commercial broiler and the three strains of Tswana chickens. Generally, higher LD patterns are associated with low effective population sizes and lower diversity in the populations.