Genetic diversity and population structure of indigenous chicken of Bangladesh using microsatellite markers

Objective The objectives of this study were to investigate the genetic diversity, population structure and relatedness among the five chicken populations of Bangladesh using microsatellite markers. Methods A total of 161 individuals representing 5 chicken populations (non-descript Deshi [ND], naked neck [NN], hilly [HI], Aseel [AS], and red jungle fowl [JF]) were included in this study to investigate genetic diversity measures, population structure, genetic distance and phylogenetic relationships. Genotyping was performed using 16 selected polymorphic microsatellite markers distributed across 10 chromosomes. Results The average observed and expected heterozygosity, mean number of alleles and polymorphic information content were found to be 0.67±0.01, 0.70±0.01, 10.7 and 0.748, respectively in the studied populations. The estimated overall fixation index across the loci (F), heterozygote deficiency within (FIS) and among (FIT) chicken populations were 0.04±0.02, 0.05 and 0.16, respectively. Analysis of molecular variance analysis revealed 88.07% of the total genetic diversity was accounted for within population variation and the rest 11.93% was incurred with population differentiation (FST). The highest pairwise genetic distance (0.154) was found between ND and AS while the lowest distance was between JF and AS (0.084). Structure analysis depicted that the studied samples can be categorized into four distinct types or varieties (ΔK = 3.74) such as ND, NN, and HI where AS and JF clustered together as an admixed population. The Neighbor-Joining phylogenetic tree and discriminant analysis of principal component also showed close relatedness among three chicken varieties namely AS, HI, and JF. Conclusion The results reflected that indigenous chicken of Bangladesh still possess rich genetic diversity but weak differentiation among the studied populations. This finding provides some important insight on genetic diversity measures that could support the designing and implementing of future breeding plans for indigenous chickens of Bangladesh.


INTRODUCTION
Poultry takes a vital place in global food chain. In fact, poultry is an integral part of rural households in the least developed countries around the world. There are about 255.31 million chickens in commercial and subsistence production systems in Bangladesh [1]. Among poultry, the chicken population is almost 90%, followed by ducks (8%) and rest 2% occupied by quail, pigeon, geese, and others [2]. The national share of commercial strains of chicken and indigenous family poultry in terms of egg production is almost equal (50:50) and that of meat production is 60:40 [3].About 89% of rural households keep indigenous chicken with an average flock size of 5.33 per holding under a backyard scavenging system that reflects the significance of indigenous chicken in Bangladesh perspective [4,5]. All elites, most city dwellers and some village level customers appreciate egg and meat produced from indigenous (deshi) chicken reared in scavenging system because their products are pro duced organically.
Indigenous chickens are lower in productivity but are well adapted to adverse tropical climate and fluctuating nutritional conditions compared to exotic chicken. However, the prefer ences for freerange indigenous chicken might be attributed to typical pigmentation, taste, leanness, firmness, high protein content and suitability for special dishes [3,6]. The identifi cation, collection, evaluation, and conservation of different genotypes are an insurance against future need for breeding [7]. In this regard, unselected random mated indigenous chickens are considered as a huge treasure of many known and unknown genotypes, which could be beneficial to provide valuable additional attributes to the future poultry production. According to Bhuiyan et al [5], nondescript Deshi (ND), naked neck (NN), hilly (HI), and Aseel (AS) chickens are noteworthy among the indigenous chicken genetic resources of Bangladesh. It is notable to mention that Bangladesh Live stock Research Institute (BLRI) has undertaken a planned and systematic native chicken breeding program since 2000 and has improved the productivity of three different native chicken varieties (ND, HI, and NN) [8]. The productivity of the said genotypes has increased remarkably over the gener ations through selective breeding onstation in comparison to existing indigenous chicken varieties. Based on the mito chondrial DNA sequence polymorphisms, Bhuiyan et al [5] suggested that Bangladeshi indigenous chickens still possessed abundant genetic diversity and have originated from multiple maternal lineages. Apart from this, genetic characterization of these varieties has not yet been sufficiently explored which is a prerequisite to knowing within and between popula tion differentiation in order to establish them as breeds or varieties.
Microsatellites are widely used DNA markers for exploring genetic variation and phylogeny among populations of same species [9,10]. The usefulness of microsatellite (MS) markers in estimating genetic relatedness and diversity in chickens have been demonstrated in several indigenous breeds, inbred strains and in commercial chicken lines [1114]. Although molecular genetic characterization of indigenous chicken of Bangladesh has been performed based on mitochondrial DNA, characterization using nuclear genomic DNA is absent until to date. Hence, it would be worthwhile to know the genetic diversity and population structures of available chicken varieties of Bangladesh for development of a proper con servation and breeding strategy. Here, we investigate three indigenous chicken varieties that have been maintained at BLRI along with red jungle fowl (JF) and AS chickens for assessing their genetic diversity, genetic distance and popu lation structure using 16 highly polymorphic MS markers.

Sample collection and DNA extraction
Institutional Animal Care and Use Committee (IACUC) re viewed the total experimental procedure and prior approval was obtained from IACUC for this experiment (Approval number: BLRIPCUC003). In total, 161 unrelated blood samples were collected from 5 different chicken populations of Bangladesh: ND (n = 41), NN (n = 44), HI (n = 39), AS (n = 18), and JF (n = 19). The Native Chicken Conservation and Improvement Project of BLRI maintains ND, NN, and HI flocks and these birds were used for sampling. Besides, blood samples of AS chicken were collected from Bangladesh Agri cultural University (BAU) research flock as well as from the farmers of Brahmanbaria district. JF sampling was performed from their breeding habitats, Chittagong Hill Tract regions (22.20 °N and 92.35 °E). According to Bhuiyan et al [5], two different subspecies of JF are available in Bangladesh; Gallus gallus gallus and Gallus gallus spadiceus. However, blood samples were not categorized as per subspecies level for this study. All possible precautions were taken during sampling from unrelated individuals to prevent cross contamination. The blood samples were collected from wing veins by 3.0 mL disposable syringe and were stored either in vacutainer containing ethylene diamine tetra acetic acid or in Flinders Technology Associates (FTA) classic card. The vacutainers and classic cards were then labeled with the name of chicken variety, age and sex of the birds and date of collection. The genomic DNA from whole blood samples and FTA card was extracted using both DNA isolation kit (Jena bioscience, Germany and GeNet Bio, Daejeon, Korea) and phenolchloro form method.

Polymerase chain reaction amplification and genotyping
Selection of primers and polymerase chain reaction amplification: Sixteen (16) highly polymorphic MS markers were selected from the list of markers previously reported in Cho et al [15], which were selected from the ark database (http:// www.thearkdb.org/usingarkdb/). Forward primers were labelled and modified with four types of fluorescence dye. The detailed information for primers used in this study is mentioned in Cho et al [15]. Multiplex polymerase chain reaction (PCR) for total of 161 DNA samples was performed in a 20 μL reaction volume. The PCR master mixture was consisting of 2.0 μL of 25 ng/μL gDNA, 10 μL of HS Prime multiplex PCR buffer (GeNet Bio, Korea) and 0.5 μL of 5 pmol forward and reverse primer and appropriate volume of triple distilled water to adjust the total volume. PCR was performed in an initial denaturation at 94°C for 10 min followed by 35 cycles of denaturation at 94°C for 30 s, an nealing temperature at 60°C for 30 s, extension temperature at 72°C for 30 s and final extension at 72°C for 20 min using the C1000 Thermal Cycler (BioRad, Irvine, CA, USA). PCR amplifications products were confirmed by 2.0% agarose gel electrophoresis.
Microsatellite genotyping: Fragment analysis was performed by capillary electrophoresis array using the Genetic analyser 3,130 xl (Applied Biosystems, Foster City, CA, USA). The genotyping reaction was performed in a 10 μL total volume including 1.0 μL of diluted PCR product, 10 μL of HiDi formamide (Applied Biosystems, USA) and 0.1 μL of the GeneScan500LIZ size standard (Applied Biosystems, USA). The genotyping results were obtained using GeneMapper software ver. 3.7 (Applied Biosystems, USA).

Data analysis
At first raw data were checked using MS tool kit (in excel) for detection of any genotyping error and null alleles. The evidence for significant deviation from population mutation drift equilibrium (bottleneck) at 16 neutral MS markers in all populations was checked after Bonferroni correction, with 2,000 replicates under the two phase mutation model with 95% stepwise mutations (Ps = 0.95) using BOTTLENECK program [16]. Measures of genetic diversity such as total number of alleles, allele frequencies, mean number of alleles (MNA), polymorphism information content (PIC), observed and expected heterozygosity (Ho, He) and fixation index (F) were computed using GenAlEx ver.6.5 [17]. Nei genetic dis tances were computed using the aforesaid software. Allele richness (AR) was calculated using FSTAT 2.9.3 [18]. Pairwise genetic differentiation (Wright's F statistics) and analysis of molecular variance (AMOVA) analysis were performed using Arlequin ver. 3.5 [19]. NeighborJoining phylogenetic tree was constructed based on Nei's standard genetic distance model with 1,000 bootstraps using the software POPTREE2 [20]. Genetic structure of the studied chicken populations was inferred by modelbased clustering using STRUCTURE ver. 2.3.4 [21] with a burn period of 20,000 generations and Markov Chain Monte Carlo simulations of 100,000 itera tions. The optimal K values were determined having the lowest crossvalidation (CV) errors (ΔK = 3.74). Discriminant analysis of principal component (DAPC) analysis in R, aims to provide an efficient description of genetic clusters using a few synthetic variables. These are constructed as linear combinations of the original variables (alleles) which have the largest betweengroup variance and the smallest within group variance [22]. Table 1 represents the morphological characteristics and dis tribution of the investigated chicken samples of Bangladesh. ND and NN are scattered throughout the country except in the hilly areas. Besides, HI chickens are predominantly avail able in Chittagong Hill tracts. The AS is mostly concentrated in Sarail Upazila of Brahmanbaria district as well as in the periurban and urban areas of Dhaka, Chittagong, and Sylhet division. All birds possess single comb type with few excep tions in AS chicken which have a strawberry comb. All the chicken varieties have whitish and/or yellowish shank color except red jungle fowl that possesses slate /blackish shank color. Bhuiyan et al [4], Faruque et al [23] and Sarker et al [24] previously reported the morphological features and geo graphic distribution of Bangladeshi indigenous chickens that support the present findings.    [29] reported higher values (0.69 to 0.86) in two chicken populations of Karnataka, India while lower estimates (0.44 to 0.58) were found in Thai and Bhutanese native chickens [7]. It is notable to mention that direct comparison of data from different studies is probably difficult due to the differ ent genetic backgrounds of the chicken populations studied and the different MS markers used [26]. The differences in MNA, Ho, and He between previous and present findings may be attributed with sample size, population structure, number of markers used, population specific alleles and/or allele scor ing bias (null allele or allele drop out) and sampling strategy. The higher MNA, and mean allele richness values estimated based on the minimum sample size 17 and weighted over 16 loci indicated relatively higher genetic diversity in all popu lations and therefore, Bangladeshi chicken populations are more diverse. Markers with PIC values >0. 5 and He values >0.6 provided high PIC for genetic diversity measures and were most reliable for population discrimination [25,30]. The present study confirmed that all markers had higher PIC and He values than the truncated level, except for ADL0304. Taken together, the observed high heterozygosity value, PIC, Shanon' s Information Index (Tables 2, 3) indicated that the selected 16 MS markers were reliable and informative for estimating the genetic diversity in indigenous chicken population of Bangladesh.

Genetic diversity measures within and among the
According to Weir and Cockerham [31] estimation, the overall within population inbreeding (F IS ) was found to be 0.046 considering all 16 loci and varied between -0.07 (LEI 0268 locus) and 0.14 (ADL0259 locus) ( Table 3). The overall heterozygote deficiency or total inbreeding (F IT ) and mean genetic distance (F ST ) were 0.16 and 0.12, respectively, over the 16 loci. Seo et al [25] discriminated five Korean native chicken lines using 15 highly polymorphic MS markers where the estimated mean F IS , F IT , and F ST values were 0.0093, 0.137, and 0.129, respectively and is comparable to the present study. Furthermore, Chen et al [27] reported the mean F IS and F IT values to be 0.002 and 0.18, respectively, in fifteen Chinese Indigenous chicken breeds and are comparable to the present study. The F IS represents a degree of nonrandom mating (deviation from HardyWeinberg equilibrium) where a posi tive value for F IS means selective breeding population. Only five (MCW0123, LEI0094, ADL0293, ADL0268, ROS0083) out of 16 markers showed a negative number. This result in dicated the absence of random mating in the investigated chicken populations, which was supported by Faruque et al [8] who reported that three indigenous chicken varieties (ND, NN, and HI) were under selective breeding for a de cade at the institutional flocks. Consequently, the rate of inbreeding has been increased over the generations. In ad dition, the higher F IT value of the present study was due to excessive homozygosity existing in six MS markers (MCW 0029, MCW0264, ROS0013, ADL0304, LEI0074, and LEI 0141).

Genetic distances and clustering of the indigenous chicken varieties
Genetic distances: The populations pairwise distances (F ST ) among the studied populations were found significant (p< 0.01) and the values ranged between 0.084 and 0.154 ( Table  4). The highest pairwise genetic distance (0.154) was found between ND and AS while the lowest distance was observed between JF and AS (0.084). Nei's unbiased genetic distance results were also in agreement with pairwise genetic distances showing highest distance (0.578) between ND and AS, whereas the lowest genetic distance (0.285) was between AS and JF. The AMOVA results showed the genetic variation among the 5 populations was 11.93% while the remaining 88.07% of the total variation was accounted for within population diversity with an overall F ST value of 0.12 (p<0.001) ( Figure  1). Several previous studies reported higher F ST values (0.15  to 0.26) in Indian [14], Chinese [27] and African indigenous chicken populations [10] and depicted that comparatively lower genetic distances existed among the Bangladeshi chicken varieties. On the contrary, Halima et al [26] reported the ge netic distances in seven Northwest Ethiopian native chicken populations varied from 0.073 to 0.13. Seo et al [25] found pairwise genetic distance among five Korean native chicken lines to be 0.083 to 0.171. The above two findings are similar to the present study. In addition, the current results are sup ported by the study of Yamamoto et al [32] who reported the lower genetic differentiation among the different Ban gladeshi chicken populations. Bhuiyan et al [5] found low F ST value (0.1084) for Bangladeshi chicken varieties using mtDNA Dloop sequence analysis and strongly justified the results of this study as evidenced by population pairwise F ST and AMOVA analysis. Altogether, the low genetic dif ferentiation among the Bangladeshi chicken populations might be due to recent gene flow among themselves as well as common ancestor for constructing the populations. Phylogenetic analysis: The Nei's unbiased genetic distance matrices and dendrogram have shown that NN and ND were grouped into the same branch (Table 4, Figure 2). The three other chicken varieties AS, HI, and JF made a separate cluster which revealed the close relatedness among them selves. Notably, HI was separated from AS and JF with a node that signifies HI as a distinct chicken variety. Bhuiyan et al [4] and Faruque et al [8] reported ND, NN, and HI were distinct indigenous chicken varieties of Bangladesh where HI are geographically isolated from others and supports the present findings. This phenomenon could also be explained from the geographic history, as both JF and HI are the in habitants of Chittagong Hill Tract regions and might be separated from the common origin in the recent past. AS is an established fighting breed but has a close relationship with JF. This might be due to the fact of inter se mating be tween AS and JF, a long tradition to the farmers, for increasing vigor and aggressiveness in AS birds (personal communi cation). The phylogenetic tree also suggested the possible introgression of JF to AS breed and compatible with the previous mitochondrial DNA study [5]. Besides, ND and NN are evenly distributed throughout the country. Gene flow between these two populations is not unlikely and there fore, they belong to the same cluster due to their close genetic relationships.
Structure program was employed to investigate the genetic structure of the 5 chicken populations of Bangladesh ( Figure  3). The results showed better agreement in structure output at K values between 4 and 5. The output at K = 5 seems plau sible; it clearly distinguished each chicken population. This result is supported by the findings of Cho et al [15] and Seo et al [25] where they reported Korean native chicken breeds (five or six) had an equal number of underlying genetic clus ters (K = 5 or 6). The ΔK statistic was 3.74 at a maximum K value (K = 5), which suggest the most probable number of inferred clusters. Structure data also indicated a certain portion of genetic admixture in each population from other varieties. Moreover, structure analysis also depicted remark able introgression of JF into AS, which was supported by the findings of Bhuiyan et al [5] and Sarker et al [24]. Kmeans based genetic clustering in DAPC analysis was given at K mean = five (lowest BIC value) and group membership values (data not shown) were consistent with structure results. Effi cient description of genetic clusters among Bangladesh chicken illustrated in DAPC plot (Figure 4) also provides member ship probabilities of individuals for the different groups based on the retained discriminant functions. DAPC result was similar to structure results indicating that all groups were discriminated, however, a close proximity between AS, JF, and HI was observed. It is noteworthy to mention that the genetic basis of the founder population of these five chicken varieties was not well defined as well as a bit complicated. Eggs were screened from the farmers merely based on phe notypic information from the random mated population. They were being reared in scavenging condition and there fore, some gene flow in neighbouring regions or populations possibly exists. Taken together, this study provides important information on genetic background of indigenous chicken genetic resources of Bangladesh that could be utilized for conservation and subsequent improvement.

IMPLICATIONS
The molecular study showed that indigenous chicken of Bangla desh still possesses rich genetic diversity. Analysis of molecular variance analysis revealed only 11.93% of the total genetic diversity accounted for between population differentiations (F ST ) that indicates poor genetic variability among the inves tigated populations. The phylogenetic tree and discriminant