In the present study, phylogenetic and demographic profiling analyses were carried out using 76 Indian birds that belong to seven populations. We employed 11 microsatellite markers and also sequenced 650 bp of hyper variable region (D-loop) of mitochondrial genomes of two species of fowls namely G. sonneratii (n = 4) and G. gallus, which includes two subspecies G. g. murghi, (n = 56) and G. g. domesticus (n = 16).

Totally there were 197 alleles from 11 microsatellite loci, of which, 106 were from G. g. murghi, and 59 were from G. g. domesticus chicken. The locus MCW5 gave the maximum number of alleles in G. g. murghi and in G. sonneratii, while VITIIG2 locus revealed the maximum number of alleles in G. g. domesticus (Table 1). The total number of private alleles was maximum in G. g. murghi followed by G. g. domesticus and G. sonneratii, respectively. Population-wise mean number of alleles per locus ranged from 2.91 (G. sonneratii) to 6.09 (Birshi Kargah population of G. g. murghi, B-RJF), with a mean heterozygosity ranging from 0.481 (K-RJF) to 0.600 (B-RJF).

When the 76 birds were subdivided into seven groups (see methods section for the grouping of birds) and analyzed for Hardy-Weinberg equilibrium (HWE), of the 77 combinations (7 groups × 11 loci) 32 deviated from HWE, with 22 deviations occurring in G. g. murghi alone and 8 in G. g. domesticus group (See Additional file 1). We speculate that G. g. murghi is probably not in HWE because of inbreeding within the fragmented populations. The low F_ST values among G. g. murghi as compared to G. g. domesticus birds also suggested possible inbreeding within G. g. murghi populations. The average F_ST was 0.126 within G. g. murghi, while it was 0.154 within G. g. domesticus (Table 2). The AMOVA estimation based on 99 permutations using GenAlEx showed a significant (P = 0.01) within population variation (66% in 7G and 69% in 2G population). Upon grouping the 72 Indian birds as domestic and RJF, the microsatellite markers showed a significant variation 'among the groups' (i.e. domestic-RJF, 27%, P = 0.01) than 'within the group' (17%, P = 0.01).

Population-wise Nei's genetic distance calculated using GenAlEx program showed a higher average distance between G. sonneratii and G. g. domesticus (2.099) than between G. sonneratii – G. g. murghi (0.758) and G. g. murghi – G. g. domesticus (1.695) combinations (Additional file 1). Intra-population average distances were lower for both domestic and RJF groups than across the population distances, which is consistent with the observation that among population variation was more than within population variation. The results suggest very rare genetic exchange between the RJF and domestic chicken populations, at least in recent history.

A Maximum Likelihood (ML) tree obtained from the microsatellite data showed a clear separation of G. g. domesticus from the G. g. murghi, with G. sonneratii as an outgroup (Fig. 1a) suggesting the genetic distinctness of G. g. murghi. However, rare instances of hybridization between Indian RJF and domestic birds cannot be excluded in nature as seen in one case (K10-C1, Fig. 1a).

We also constructed a genetic distance based neighbor-joining (NJ) tree to obtain the genetic relationship among Indian birds. The result clearly points to the fact that hybridization between Indian chicken and Indian RJF G. g. murghi in the wild is extremely rare (Fig. 1b).

To understand the clustering pattern of birds, we also carried out genetic distance based principal component analysis (PCA) (Fig. 2) of domestic and RJF birds, using GenAlEx. The results showed clear segregation of all domestic birds into a single quadrate of the PCA (left upper quadrate in Fig. 2), which did not include any RJF. However, the population of RJFs did not segregate according to the location/population, suggesting the possibility of inbreeding within the RJF populations. This result further supports the absence of hybridization between RJF and domestic birds in India, at least in recent times.

We also sequenced and analyzed 650 bp of the D-loop region of 76 Indian birds to derive the matrilineal population history by using coalescent-based models.

The pairwise F_ST values calculated using Arlequin were very low within G. g. murghi when compared to G. g. domesticus (Table 2). The average F_ST values were more for G. sonneratii – G. g. domesticus combination than for G. sonneratii-G. g. murghi or G. g. murghi – G. g. domesticus combinations. Contrary to microsatellite based analysis, mitochondrial analysis showed a lower mean pairwise F_ST value for G. sonneratii – G. g. domesticus combination. These results again point out that there is hardly any genetic exchange across the three categories of birds, namely GJF, RJF and domestic birds.

To understand the demographic history of the two populations, we obtained pairwise mismatch distribution estimates. In an expanding population, mismatch distribution is expected to be a bell shaped smooth curve, whereas population at constant growth shows raggedness. In the present study, observed mismatch distribution was bell-shaped for 56 Indian birds of G. g. murghi (Fig. 3). However, appearance of multiple peaks suggested a population subdivision, which was more prominent in G. g. domesticus than in G. g. murghi. This also suggests homogenous population of G. g. murghi, again supporting our notion that inbreeding is more common in RJF than in domestic birds. Harpending's raggedness index was lower in G. g. murghi (0.001515) than in G. g. domesticus (0.0149), but was significant (P < 0.05) in both the populations reiterating population expansion.

In the present study we also compared the D-loop sequence of 76 birds with the corresponding sequences (approximately 400–440 bp) of 779 birds from NCBI GenBank to get insight into phylodemographic status of G. g. murghi and G. g. domesticus (India). Of the 855 sequences, 117 were from RJFs, 714 were from domestic chicken and the rest belonged to other species of Gallus. These 855 birds were classified into 13 groups based on geographical location and species/sub-species status as shown in Table 3. We included Japanese quail, Coturnix japonica as an outgroup whenever necessary.

All the 855 sequences resulted in 117 segregating sites and 146 haplotypes with a haplotype diversity of 0.940 and an average nucleotide diversity of 0.020. The NJ tree generated from the haplotype data revealed two distinct G. gallus clusters, one consisting of the G. g. bankiva group, and the remaining cluster containing all the other groups namely, G. g. spadiceus, G. g. gallus, G. g. murghi, G. g. jabouillei and G. g. domesticus. Irrespective of the geographic distribution and sub-species status, more than 95% of the birds clustered within this single largest group (Fig. 4 – see discussion). This result is consistent with the neighbor-joining tree of all 855 birds (Additional file 2), where we observed a single clade for all domestic and RJF birds. However, in contrast to Fumihito et al.'s work 5, we observed a clear segregation of a clade containing G. varius, G. lafayettei and G. sonneratii from all RJF/domestic bird combination, which reassures us about the absence of the hybridization. Such a topology was also evident from the haplotype-based tree (Fig. 4)

To obtain further insight into the haplotype relationships, we constructed a median-joining network using 146 haplotypes. In the haplotype network, most of the G. g. gallus formed individual nodes (e.g. H_1, H_2, H_78, H_79) and clustered independently, while many G. g. murghi haplotypes (H_86 to H_99 and H_102 to H_123) grouped around a major cluster (H_112) having 16 birds, giving rise to a star like phylogeny (Fig. 5). The largest haplotype H_3 (with 127 birds) confined mostly to Chinese G. g. domesticus birds without any representation of jungle fowls. Thirty nine birds of G. g. spadiceus belonging to 9 haplotypes did not have any representation from G. g. murghi birds. The second largest haplotype (H_61 with 102 birds) shared only a single individual of G. g. gallus and many G. g. domesticus from China, India, Japan and Indonesia. Forty-three G. g. domesticus of India (19 haplotypes) were distributed throughout the network. These haplotypes were connected with G. g. gallus, G. g. murghi and G. g. spadiceus. The haplotype H_63 contained G. g. domesticus (India) birds reported in the present study as well as from an earlier study 8. Interestingly, majority of the Indian G. g. domesticus haplotypes shared the birds from the three major G. gallus groups belonging to G. g. spadiceus (H_58, H_61), G. g. gallus (H_59, H_61), and also G. g. murghi (H_6). It is very likely that G. g. spadiceus has also contributed to the domestic chicken. Apart from Indian RJF, G. g. murghi haplotype H_6 also contained 5 Chinese, 1 Iranian and 1 Japanese domestic chicken. The distribution of Indian domestic chicken into different haplotypes could imply multiple origins of Indian domestic chicken breeds.

The sequenced regions of D-loop were well conserved in most of the birds. The 62 bp insertion element found in G. sonneratii was the only indel of considerable size present in the whole sequence. The salient mutation observed in G. g. murghi was the presence of 'T' in most of the birds (54 out of 56 i.e. 96.4%) (Fig. 4) that was present in only 33.3% of G. g. gallus, 24.4% of G. g. spadiceus and 13.1% of all the G. g. domesticus birds. Only three Indian G. g. domesticus (6.9%) had 'T' at this position. This position corresponds to the nucleotide number 360 (where there is 'C') in the complete mitochondrial genome sequence of G. gallus (Acc. No. NC_001323) available in GenBank. Across the portion of the sequenced mitochondrial genome we observed that transition to transversion ratio was lower in case of G. g. murghi (45/47 = 0.99) than in G. g. gallus (31/2 = 15.5) or G. g. spadiceus (23/4 = 5.75 – Table 3).

Mismatch distribution analyses carried out under sudden expansion model showed that mean pairwise differences were highest in G. g. murghi among G. gallus sub-species. However, another species of Gallus, G. sonneratii had the highest mean pairwise differences (Table 3). Both Tajima's D and Fu's Fs were significantly (P < 0.05) negative in G. g. murghi, suggesting the departure from neutrality (Table 3). Tajima's D was significant only in G. g. spadiceus and G. g. murghi. Fu's F_S, a better indicator for estimating the departures from neutral theory, showed a significant negative value (P of simF_S < = obsF_S is < 0.01) for G. g. murghi and G. g. domesticus (Indonesia). Nucleotide diversity was also high (0.030) in case of G. g. murghi (except for G. sonneratii) amongst all the groups studied (Table 3).

The F_ST values showed a very high differentiation between the out group C. japonica and the other 13 populations, with a value above 0.9. Most of the pairwise F_ST values were significant with a P value < 0.05. F_ST value 20 based NJ tree showed the divergence of G. g. murghi from G. g. gallus, G. g. spadiceus and G. g. domesticus (Additional file 3). AMOVA calculations showed that the majority of variation found within G. gallus subspecies was between the domestic and jungle fowl populations (76%), while 'among the group' variation was 7% with an overall F_ST value of 0.234 (P = 0).

Blood samples were collected from 3 geographically isolated populations of G. g. murghi i.e. Morni Hills ('M-RJF', n = 25), Kalesar Forest ('K-RJF', n = 10) and Birshi Kargah Forest ('B-RJF', n = 21) of Haryana State of India (Fig. 7), 3 populations of G. g. domesticus birds i.e. Chicken ('C', n = 5), Jodhpur ('J', n = 5) and Mirpur Bakshiwala ('M', n = 6) villages of Haryana. Grey Jungle fowl (G. sonneratii) blood samples (n = 4) were collected from different Zoological Parks of South India and were considered as a single population. Except for the G. g. domesticus birds, which were collected from specific villages, all the other birds were captive bred from the founder-lines that were obtained from the specified village. The 76 birds were considered as belonging to 7 populations (3 RJF, 3 domestic and a single population of GJF).

About 0.5 ml of blood was collected from the wing vein of live birds into vials containing 5 mM EDTA and genomic DNA was isolated as per standard protocols 25.

The microsatellite loci used in the study, the primer details and PCR conditions are available in Table 1. The PCR products were separated on 3.5% Metaphor agarose gel along with pUC/MspI digest (MBI Fermentas) ladder. For GeneScan analysis, the PCR products generated using fluorescent dUTPs were dissolved in 2 μl of formamide gel loading buffer with 0.3 μl of ROX-500™ GeneScan ruler (Perkin Elmer) and separated on a 5% polyacrylamide-7 M Urea gel.

The D-loop hypervariable region was PCR amplified using primers described elsewhere 7. The sequencing was carried out on both the strands. We also used an internal primer (5'GTGGAATATAGGTTAATGCC 3') to obtain the sequence information from 5' region without any ambiguity. 50 ng PCR product was used in a sequencing reaction that contained 8 μl of Ready reaction mix (BDT v 3.0, Applied Biosystems, Foster City, CA) and 5 picomoles of primer. The sequencing was carried out in ABI Prism 3100 Genetic Analyzer (Applied Biosystems).