Genome-wide analysis across Indian camel populations reveals genetic distinctiveness of the Kharai camel breed

Abstract

Background

In India, indigenous camel populations have evolved under diverse ecological conditions, leading to the emergence of several distinct breeds with specialised adaptive traits. Understanding the genomic architecture and population structure of such camel populations is essential for their conservation and sustainable utilisation. In the present study, whole-genome sequencing data were generated and analysed for Indian dromedary camels representing nine recognised breeds, along with Indian Bactrian (double-humped) and Arabian Peninsula dromedaries with the primary objective to investigate population structure and to assess the genetic distinctiveness of the geographically isolated Kharai camel, a unique breed adapted to coastal and saline environments.

Results

The sequencing depth ranged from 11.3× to 25.4× for Indian dromedaries, with high mapping efficiency to the reference genome (mCamDro1). Population structure analyses consistently revealed limited genetic differentiation among most Indian dromedary populations, reflecting shared ancestry and historical gene flow. However, the Kharai camel exhibited clear genetic distinctiveness across multiple analytical approaches. Principal component analysis separated Kharai from other dromedaries, and admixture analysis at K = 2 demonstrated near-complete clustering of Kharai individuals to a distinct ancestral component. Pairwise genetic distances (F_ST) further supported moderate differentiation in Kharai (0.07–0.10), compared with the low differentiation observed among other populations. Neighbour-Net and phylogenetic tree analyses corroborated these findings, showing breed-specific clustering with clear separation of Kharai. Genome-wide diversity analysis revealed a predominance of low-frequency alleles (57.7% of SNPs with MAF 0.05–0.10), indicating population subdivision and localised demographic histories. Moderate genetic diversity was observed (Hₒ = 0.303 ± 0.001; Hₑ = 0.319 ± 0.0001), with evidence of a mild heterozygote deficit likely attributable to the Wahlund effect. Linkage disequilibrium (LD) analysis showed strong short-range LD (r² = 0.783 at 0–10 kb) with gradual decay up to 10 Mb, suggesting a moderate historical effective population size. LD-based reconstruction indicated a decline in effective population size from > 2,500 ancestral individuals to approximately 167 individuals in recent generations, highlighting demographic contraction.

Conclusions

These findings demonstrate substantial genetic homogeneity among Indian dromedary populations while highlighting the distinct genomic identity of the Kharai camel breed. The results underscore the evolutionary uniqueness of this breed and emphasise the need for targeted conservation and management strategies to preserve its genetic resources.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12864-026-13134-7.

Background

Animal husbandry is a cornerstone of India’s rural economy, providing food security, income, draught power, manure, and socio-cultural value [1]. Within this framework, pastoralism sustains livestock in marginal landscapes through mobility, grazing-based management, and adaptive breeding, making it one of the oldest and most climate-resilient livelihood strategies in arid and semi-arid regions [2]. Pastoral communities such as the Raika and Rabari have historically shaped camel husbandry in western India, maintaining populations adapted to extreme and heterogeneous environments [3].

The dromedary camel (Camelus dromedarius) remains integral to livelihoods and cultural traditions in India’s arid ecological terrain. India recognises nine indigenous dromedary breeds, including Bikaneri, Jaisalmeri, Kutchi, Jalori, Kharai, Marwari, Mewati, Mewari, and Malvi [4], valued for their endurance, heat tolerance, and multipurpose utility, including milk, transport, and draught power [5]. These breeds thrive under extreme conditions, tolerating temperatures above 40 °C and surviving nearly a month without water while losing up to 25% of body mass [6]. Among Indian dromedary camels, the Kharai camel represents a genetically unique and geographically isolated population, as this particular breed is restricted to the coastal and mangrove ecosystems of the Gulf of Kachchh, Gujarat [7, 8]. Kharai camels inhabit an environment fundamentally distinct from inland desert breeds. They are maintained primarily by Rabari, Fakirani Jat, and Maldhari pastoralists. Kharai camels exhibit remarkable ecological adaptations, including daily swimming to offshore mangrove islands and reliance on saline forage and brackish water [9, 10]. This semi-aquatic lifestyle sharply distinguishes them from other terrestrial Indian camel breeds, particularly in the case of extreme ecological specialization.

Despite the ecological and socio-economic importance, Indian camel populations have been observed to decline sharply, from about 0.40 million in 2012 to less than 0.25 million in 2019, a reduction of 37% following an earlier 22.5% decline between 2007 and 2012 [11]. Globally, by contrast, camel populations have nearly tripled over the past five decades, with dromedaries accounting for 95% of the 35 million animals [12]. Various factors contribute to the disappearance of the Indian camel, including the advent of mechanization in farming and rural lifestyles, shrinking grazing lands and migratory routes, climate change, and declining economic returns, which have disproportionately affected indigenous breeds. Geographically isolated populations, such as the Kharai, face a heightened risk of genetic erosion [13].

Although phenotypic, microsatellite, partial and complete mitochondrial DNA studies have provided initial insights into camel population structure [14–20], comprehensive genome-wide analyses remain limited. High-throughput whole-genome sequencing now enables detailed evaluation of genetic diversity, population structure, differentiation, and linkage disequilibrium [21], facilitating the study of breed-specific evolutionary adaptations. This study aimed to understand the genetic differentiation of Kharai camels from other camel populations, to decipher their population structure, and to determine the position of Kharai camels amongst other well-established, geographically closer Indian camel populations using whole-genome sequencing data.

Materials and methods

Animal resources and dataset compilation

A total of 32 Indian dromedary camels (Camelus dromedarius) and 3 double-humped camels (Camelus bactrianus), representing major breeds (or) populations across diverse geographical regions of India (Fig. 1), were included in this study. Samples belonged to several populations. Bikaneri, Kutchi, Mewari and Jaisalmeri camels were reared at ICAR-National Research Centre on Camels (NRCC), Bikaner, where the project is being executed. Other samples from populations like Mewati, Jalori, Malvi, Marwari, Sindhi and Double Humped camels were obtained from field herds (Suppl. Table 1). Informed consent from the owners to use the animals in this study was obtained. None of the animals used in this study were euthanised. The whole-genome sequencing (WGS) data for all Indian samples were generated and provided by the ICAR-National Research Centre on Camel (NRCC), Bikaner, India. Detailed information for these samples, including breed (or) population classifications and sampling locations, is presented in Supplementary file S1. Animals within each population were randomly sampled from the available herds to capture the existing genetic variation. Efforts were made to avoid deliberately sampling closely related individuals by using pedigree information, herd records, and information provided by herd managers/owners, whenever available. The sampled animals represented different geographical locations and management systems within their respective populations. In addition to the Indian camel populations, whole-genome sequence data from 5 Arabian Peninsula dromedary camels were incorporated for comparative analyses, retrieved from the European Nucleotide Archive (ENA) under the Bioproject accession number PRJEB47650 [22].

Fig. 1.

Geographical distribution of Indian camel populations

Processing of sequenced data

The raw whole-genome sequencing (WGS) reads (FASTQ files) were first evaluated for quality parameters, including per-base sequence quality, per-sequence quality scores, sequence composition, Phred quality distribution, and adapter content using FastQC [23]. Following quality assessment, the clean, high-quality adapter-free reads were aligned to the Arabian dromedary camel reference genome assembly mCamDro1 (GCF_036321535.1) [24] using BWA-MEM2 (version 2.3) [25], an optimised implementation of Burrows-Wheeler Alignment that provides comparable alignment accuracy with significantly increased computational speed. Aligned reads were converted to BAM format and sorted using SAMtools [26]. Duplicate reads were found and removed using Picard Tools [27]. Summary statistics for the mapped reads were calculated, including the proportion of the reference genome covered, the average sequencing depth, and the percentage of reads successfully mapped, using the coverage and flagstat utilities of SAMTools. Detection of genomic variants comprising Single Nucleotide Polymorphisms (SNPs) was performed with the GATK HaplotypeCaller in GVCF mode (version 4.6.2.0) [28], which were later combined and genotyped using the GATK CombineGVCFs and GenotypeGVCFs tools, resulting in a total of 29,011,075 variants. SNPs were identified from the combined variant calls by applying the -select-type SNP option in GATK HaplotypeCaller, resulting in a total of 25,288,792 SNPs. After extracting autosomal SNPs from the combined variant dataset, variants were subjected to hard filtering using the GATK VariantFiltration tool to remove low-quality calls. Filtering criteria included: (1) variants exhibiting significant strand bias between reference and alternate alleles (FS > 60); (2) variants with low quality relative to sequencing depth (QD < 2); (3) variants with poor root mean square mapping quality (MQ < 40); (4) variants with low Phred-scaled variant quality (QUAL < 30); (5) variants showing strand bias in mapping quality between reads supporting reference or alternate alleles (MQRankSum < − 12.5); and (6) variants where the alternate allele tends to occur near the ends of reads (ReadPosRankSum < − 8). These filtering steps ensured that only high-confidence 19,008,273 SNPs were retained for downstream analyses. Variants on the scaffold were also removed, resulting in a total of 17,340,233 SNPs.

SNPs quality control pruning

The filtered autosomal SNPs were subjected to further quality control (QC) using Plink v1.9 [29]. For diversity analyses, SNPs were removed if they exhibited a call rate of less than 95% across samples, deviated from Hardy–Weinberg equilibrium (p < 1 × 10⁻⁶), or had a minor allele frequency (MAF) ≤ 0.05. Table 1 summarises the number of SNPs retained in each dataset. After QC, 10,081,177 high-quality SNPs were retained for further analyses.

Table 1.

Quality control criteria and the number of excluded and remaining SNPs for the genetic diversity analyses

Dataset	Number of SNPs
Raw Autosomal SNPs	17,340,233

Quality Control Criteria	Number of SNPs removed
Call Rate < 95%	2,424,942
MAF < 0.05	4,713,596
HWE p-value < 1 X 10-6	120,518
SNPs remained after QC	11,1770,08

Population structure analyses

To determine whether the population of Indian camels under study forms genetically distinct groups, population structure analysis was performed. The estimates were obtained for each population separately. The population structure analysis was conducted using LD-pruned SNPs to reduce redundancy and create an independent set of SNPs. SNPs with LD (r² > 0.2) were pruned using the Plink v1.9 --indep-pairwise 50 10 0.2 command, resulting in a final panel of 468,436 independent SNPs as mentioned in Table 2.

Table 2.

LD pruning criteria and the number of excluded and remaining SNPs for the population structure analyses

Quality Controlled SNPs	10,081,177
SNPs removed after applying LD pruning criteria (r2>0.2)	9,612,741
SNPs remaining after LD pruning	468,436

Principal component analysis (PCA) was conducted using Plink v1.9 with the -pca 10 flag [30], and the resulting components were visualised using the ggplot2 package in R [31]. Genetic relatedness among livestock breeds can be assessed using the F_ST statistic, which quantifies the level of genetic differentiation between populations [32]. Pairwise Fixation index (F_ST) between different camel populations was estimated using the Weir and Cockerham [33] method in VCFtools v0.1.13 [34] with a window size of 10,000 kb.

where S² is the variance of allele frequency among populations and is the mean allele frequency across breeds (or) populations.

Patterns of population structure were assessed using unsupervised clustering in Admixture v1.3 [35], enabling inference of breed admixture and estimation of ancestral population proportion.

L (Q, F) =

where j are SNPs from i individuals, k is hypothesised ancestral populations, q_ik is the fraction of population k in individual i’s genome, and f_kj is the frequency of allele 1 of j SNP in population k. Genetic ancestry was assessed by testing K values from 1 to 12, and the optimal number of clusters was identified using cross-validation (CV). The K with the lowest CV error was considered to approximate the number of ancestral populations best. Individuals are considered independent; the log-likelihood (L) of the entire sample is given by Q = {q_ik} and F = {f_kj}, which have dimensions I × K and K × J, respectively.

To achieve a comprehensive visualisation of genetic relationships while accounting for potential phylogenetic signals, the Neighbour-Net approach [36] was followed. The analysis was performed using SplitsTree6 [37], applying the Neighbour-Net algorithm based on pairwise F_ST distances estimated using the method of Weir and Cockerham [33]. In addition, a dendrogram was constructed to further depict hierarchical relationships among populations based on genetic distances.

Diversity analyses

Due to the small sample size per population, genetic diversity estimates were obtained from pooled data across populations for minor allele frequency (MAF), observed heterozygosity (H_O), expected heterozygosity (H_E), linkage disequilibrium (r²), LD decay patterns, and effective population size (N_e). The minor allele frequency was estimated using the --freq flag in Plink v1.9. The H_O and H_E values were calculated in Plink v1.9 using the --het command, following standard procedures.

Where N_AA, N_AB, and N_BB are the number of individuals carrying AA, AB and BB alleles for each SNP, respectively.

where .

h_k is the population heterozygosity at the k^th locus, r is the total number of loci studied, n is the number of individuals per locus, and x_i is the frequency of the i^th allele at locus x in a sample population n [38].

Linkage disequilibrium estimates were obtained across chromosomes at distances up to 10 Mb using Plink v1.9. Pairwise r² was calculated for SNPs between 0 Kb and 10 Mb apart [39].

Where A and B are SNP pairs with two alleles; freq A, freq a, freq B, and freq b denote respective frequencies, and freq AB is the haplotype frequency.

The effective population size was estimated using SNeP v1.1 software [40], following a methodology that infers historical effective population size from linkage disequilibrium patterns derived from marker data [41].

This approach estimates the effective population size (Ne) that existed t generations ago, where t = [42]. The recombination rate is calculated based on the physical distance between genetic markers. Specifically, the physical distance (δ) between two loci is converted into linkage distance (d), assuming that megabases (Mb) approximately correspond to centimorgans (cM). The LD measure (r²adj) is corrected for sample size, while the parameter α (set to 1, 2, or 2.2) accounts for the effects of mutation [43]. For the analysis, the minimum distance between SNPs (-mindist) was set to 100,000 bp and the maximum distance (-maxdist) to 100,000,000 bp. The number of bins (-numBINS) was fixed at 500, and the recombination modifier (-svedf) was applied to estimate recombination as c = d(1 – d/2), following Sved and Feldman [44].

Results and discussion

Summary Statistics of mapped sequence reads

The sequencing depth across the mapped reads of the Indian dromedary camel samples varied from 11.3x to 25.4x, with an average coverage of 18.5x, while the Bactrian camels had a mean of 21.6x (Supplementary file S2). The sequencing depth across the Arabian Peninsula Dromedary (APD) samples ranged from 28.1x to 34.5x, with a mean of 30.4x. Approximately 99.74% of the Dromedary sequence reads, and 99.8% of Bactrian sequence reads successfully aligned to the dromedary reference genome (mCamDro1), of which 98.12% and 97.43% in Dromedaries and Bactrians, respectively, were properly paired to the reference genome (Supplementary file S2).

Population structure study revealed a distinct grouping of Kharai camels from other Indian camel breeds

Principal component analysis (PCA) differentiated the double-humped camels from the dromedary populations along the first principal component (PC1), which explained 89.53% of the total variance, and PC2, which accounted for 1.44% of the total variance captured minor genetic structuring among the dromedary populations, with Kharai individuals forming a distinct cluster along the positive PC2 axis. (Fig. 2). To obtain a clearer picture, another analysis of only Indian dromedary camels was carried out, which revealed moderate genetic structuring, with the first two principal components explaining 13.33% (PC1) and 11.03% (PC2) of the total genetic variation, respectively (Fig. 3). PCA could significantly distinguish the Kharai camel from other dromedary populations. A clear separation of Kharai was observed along PC1, where individuals formed a distinct cluster on the positive axis, indicating marked genetic differentiation from the other populations. In contrast, the remaining breeds, including Bikaneri, Jaisalmeri, Jalori, Kutchi, Malvi, Marwari, Mewari, Mewati, and Sindhi, largely overlapped, suggesting substantial genetic admixture or shared ancestry among them. Some sub-structuring was evident along PC2, with partial dispersion among north-western breeds; however, the degree of separation was limited compared to that of Kharai. These findings are further supported by results reported in the study by Vijh et al. [45], in which Discriminant Analysis of Principal Components (DAPC) also revealed clear genetic differentiation of the Kharai camel from other Indian dromedary breeds. The consistent separation of Kharai across different multivariate analyses highlights its unique genetic background, likely shaped by its specialised adaptation to the coastal and mangrove ecosystems of Gujarat and the relatively restricted breeding tract of this population.

Fig. 2.

PCA plot of Indian dromedary camels in comparison with double-humped and Arabian Peninsula camels (Note:- APD – Arabian Peninsula Dromedary; DH – Double Humped)

Fig. 3.

PCA plot of Indian dromedary camels

Wright’s pairwise fixation index was estimated for all twelve populations (supplementary file S4). The heatmap (Fig. 4) indicated low to moderate genetic differentiation among most Indian dromedary breeds, with pairwise values largely below 0.05, reflecting substantial shared ancestry and gene flow as reported by Vijh et al. [45]. The pairwise genetic distances revealed a pronounced separation of DH from all other populations, with consistently high genetic distances (0.505–0.617), indicating substantial genetic divergence, also reported by Ming et al. [46] in their study. In contrast, the indigenous breeds formed a closely related cluster with generally low inter-breed distances (0.003–0.101). Notably, Kharai showed moderate differentiation from the remaining indigenous groups (0.076–0.101), positioning it as relatively distinct within the indigenous cluster while still far closer to them than to DH. Extremely low distances were observed among several breed pairs, particularly Sindhi–Jaisalmeri (0.003), Mewati–Jaisalmeri (0.015), and Malvi–Marwari (0.013), reflecting strong genetic proximity. This pattern may be attributed to the traditional extensive production system practiced in Rajasthan, where most Indian camels (except Kharai) are reared. Frequent herd migration, shared grazing grounds, exchange of breeding males, and limited reproductive isolation among populations promote gene flow and consequently reduce genetic differentiation among breeds. Arabian Peninsula Dromedaries (APD) showed moderate genetic distances from the Indian camel populations (0.031–0.099), with the closest relationships observed with Sindhi and Jaisalmeri, suggesting shared ancestry and possible historical gene flow between Arabian and Indian dromedary populations. The dendrogram topology further supported these relationships, grouping the indigenous breeds into a compact cluster, with Kharai occupying a slightly separated branch within this group. Overall, the results demonstrate high genetic homogeneity among the indigenous breeds, moderate but noticeable differentiation of Kharai within this cluster, and pronounced divergence of DH from all other populations.

Fig. 4.

Pairwise F_ST genetic distance heatmap

Model-based population structure analysis was conducted using Admixture to infer shared ancestry and admixture proportions among the populations. The lowest cross-validation (CV) error was observed at K = 2 in unsupervised learning (Fig. 5). The admixture analysis at K = 2 revealed two major ancestral clusters across the studied camel populations (Fig. 6). All dromedary populations were predominantly assigned to a single ancestry component, V1, indicating a shared genetic background. In contrast, the double-humped camel (DH) formed a separate cluster (V2), with nearly complete assignment to the second component. This clear separation highlights strong genetic differentiation between dromedary and DH camel populations, reflecting their distinct evolutionary lineages and species-level divergence, consistent with previous whole-genome studies [22, 46]. While the analysis excluding DH and APD revealed most individuals from Bikaneri (BIK), Jaisalmeri (JAI), Jalori (JAL), Kutchi (KTC), Mewari (MWR), Malvi (MLV), Marwari (MARW), Mewati (MWT), and Sindhi (SIN) were predominantly assigned to a single ancestral cluster (V2), with high membership proportions and only minor introgression from the alternative component (V1) (Fig. 7). In contrast, Kharai (KHR) individuals showed a strikingly different pattern, with near-complete assignment to the ancestral cluster (V1), clearly distinguishing them from the other populations. Limited admixture was observed in a few individuals from Kutchi and neighbouring breeds, suggesting minor shared ancestry or gene flow; however, the overall pattern demonstrates strong genetic homogeneity among the non-Kharai populations and pronounced differentiation of Kharai at K = 2. These findings are consistent with the PCA and distance-based analyses, further supporting the distinct genetic identity of the Kharai population relative to the other camel populations.

Fig. 5.

Value of CV error values of K values ranging from 1 to 12

Fig. 6.

Admixture analysis of all populations under study indicated that the minimum cross-validation error occurred at 𝐾 = 2 (Note: BIK-Bikaneri; APD–Arabian Peninsula Dromedary; DH–Double Humped; JAI–Jaisalmeri; JAL-Jalori; KTC-Kutchi; KHR-Kharai; MWR-Mewari; MLV-Malvi; MARW-Marwari; MWT-Mewati; SIN-Sindhi)

Fig. 7.

Admixture analysis of Indian Dromedary populations at K = 2 (Note: BIK-Bikaneri; JAI–Jaisalmeri; JAL-Jalori; KTC-Kutchi; KHR-Kharai; MWR-Mewari; MLV-Malvi; MARW-Marwari; MWT-Mewati; SIN-Sindhi

Population structure analyses consistently revealed limited differentiation among most Indian dromedary breeds, reflecting shared ancestry and historical gene flow mediated by pastoral mobility. However, the Kharai camel consistently exhibited higher pairwise F_ST values (~ 0.07–0.10) and distinct clustering in PCA, highlighting localised adaptation and restricted gene flow. This localised adaptation may be attributed to the unique ecological conditions of the Kachchh region of Gujarat, where Kharai camels are reared. Unlike the other Indian dromedary breeds that predominantly inhabit the arid and semi-arid regions of Rajasthan, Kharai camels are adapted to a coastal ecosystem characterised by saline environments and mangrove vegetation. They are known for their ability to swim across seawater channels to access mangrove islands for grazing, a behaviour not observed in other camels. These distinct environmental conditions and management practices likely impose unique selective pressures, contributing to the genetic differentiation and distinct population structure observed in the Kharai breed. The admixture analyses revealed the distinct ancestry of the Kharai camels relative to other Indian dromedary camels. The local water adaptation of Kharai camels has kept them separate in their habitat, preventing mixing with other mainland camels in the region. Whereas other camels share the breeding tracts, and hence gene flow between known established breeds cannot be avoided. In contrast, whole-genome studies of Arabian Peninsula dromedaries have reported weak population structure and extensive shared ancestry among populations, largely attributable to high levels of admixture and historical gene flow [22, 47]. The comparatively stronger differentiation observed in Kharai camels therefore likely reflects the influence of local ecological adaptation and greater reproductive isolation within their restricted coastal habitat.

The Neighbour-Net network revealed the genetic relationships among the studied camel populations (Fig. 8), with most individuals clustering according to their respective breeds. This analysis revealed clear clustering patterns corresponding to breed identity, while also highlighting varying degrees of genetic relatedness among populations. Individuals are largely grouped by breed, forming distinct radial clusters. Notably, Kharai samples formed a distinct cluster separated from most other Indian dromedary populations, indicating moderate genetic differentiation and supporting its unique genetic background. Similarly, coherent clustering was observed for Kutchi, Mewati, Mewari, Malvi, Marwari, Bikaneri, Jaisalmeri, Jalori and Sindhi, though the spatial proximity among several of these groups suggests closer genetic affinities and possible shared ancestry. The relatively shorter branch lengths and reticulated central positioning of many non-Kharai breeds indicate lower levels of divergence among them, whereas the more distinct positioning of Kharai supports its unique genetic background. The Double-Humped camel formed a highly divergent branch separated from all dromedary populations, reflecting substantial genetic divergence between the two camel species. Overall, the network structure corroborates the patterns observed in PCA and admixture analyses, demonstrating strong breed-specific clustering, with clear differentiation between Kharai and comparatively greater genetic homogeneity among the remaining camel populations. The dendrogram (Fig. 9) further supported the major patterns observed in the PCA, F_ST, and Neighbour-Net analyses. Double-Humped camels formed a clearly distinct lineage separated from all dromedary populations. Among the dromedaries, individuals generally showed clustering tendencies according to population; however, several breeds exhibited close relationships and partial intermixing, reflecting shared ancestry and historical gene flow. Kharai individuals displayed a tendency to cluster together more frequently than many other breeds, supporting their relative genetic distinctiveness, although complete breed-specific separation was not observed.

Fig. 8.

Neighbour-Net tree of populations under study

Fig. 9.

Dendrogram of the populations under study

Genetic diversity across the camel population

Genetic diversity analysis was performed on pooled samples of Indian origin. The MAF estimates of the filtered SNP dataset were evaluated by classifying variants into five defined MAF intervals (Table 3). A large majority of variants were found in the lowest-frequency class (MAF 0.05–0.10), accounting for 57.7% of all SNPs, indicating a predominance of skewed allele frequencies (Fig. 10). Approximately 19.28% of SNPs were observed in the 0.10–0.20 interval, while progressively fewer markers were detected at higher frequency classes. Chromosome-wise MAF showed limited variation across the genome, with a mean value of 0.151 ± 0.00021 (Table 4). Chromosome-wise distribution of SNPs after quality control has been depicted in Fig. 11. The observed MAF distribution indicates that a substantial proportion of genetic variation in Indian dromedary camels is represented by low-frequency alleles, providing an overview of allele frequency patterns across the genome.

Table 3.

Distribution of SNPs according to MAF after quality control analysis

S.No	MAF class interval	No. of SNPs	% of SNPs
1	0.05-0.1	5528035	57.7
2	0.1-0.2	1847026	19.28
3	0.2-0.3	825272	8.61
4	0.3-0.4	698124	7.29
5	0.4-0.5	681891	7.12

Fig. 10.

Distribution of SNPs according to MAF (0.05 to 0.50) after QC

Table 4.

MAF chromosome-wise with their standard errors

CHR	MAF ± SE
1	0.147 ± 0.0001
2	0.144 ± 0.0001
3	0.149 ± 0.0001
4	0.143 ± 0.0001
5	0.147 ± 0.0001
6	0.146 ± 0.0001
7	0.146 ± 0.0001
8	0.157 ± 0.0001
9	0.147 ± 0.0001
10	0.16 ± 0.0001
11	0.145 ± 0.0002
12	0.146 ± 0.0001
13	0.151 ± 0.0001
14	0.146 ± 0.0002
15	0.145 ± 0.0002
16	0.154 ± 0.0002
17	0.145 ± 0.0002
18	0.143 ± 0.0002
19	0.158 ± 0.0002
20	0.148 ± 0.0002
21	0.149 ± 0.0002
22	0.154 ± 0.0003
23	0.149 ± 0.0003
24	0.15 ± 0.0002
25	0.154 ± 0.0003
26	0.142 ± 0.0003
27	0.142 ± 0.0003
28	0.165 ± 0.0003
29	0.146 ± 0.0002
30	0.143 ± 0.0003
31	0.152 ± 0.0003
32	0.149 ± 0.0003
33	0.16 ± 0.0003
34	0.153 ± 0.0003
35	0.177 ± 0.0005
36	0.161 ± 0.0004
µ	0.151 ± 0.00021

Fig. 11.

The number of SNPs per chromosome after quality control

The observed heterozygosity (H_O) and expected heterozygosity (H_E) across the studied dromedary populations were 0.303 ± 0.001 and 0.319 ± 0.0001, respectively, further supporting moderate genome-wide diversity within Indian dromedaries (Table 5). The pooled-sample estimates were considered more reliable owing to the larger sample size, whereas breed-wise H_O estimates were less reliable due to the limited number of individuals per breed. Breed-wise H_E estimates (0.216–0.217), however, remained relatively consistent (Supplementary Table 3). Therefore, breed-specific diversity estimates should be interpreted with caution. Future studies incorporating larger numbers of individuals from each population will improve the precision of genetic diversity estimates and facilitate a more comprehensive assessment of within-breed variation. Compared to Arabian Peninsula dromedaries, which exhibit weak population structure due to extensive historical gene flow [22, 47], Indian dromedaries show moderate genetic differentiation, with observed heterozygosity (H_O = 0.303) slightly lower than expected (H_E = 0.319), suggesting a mild heterozygote deficit likely due to the Wahlund effect [48, 49], which may reflect localized breeding practices, ecological adaptation to the arid and semi-arid landscapes of the Indian subcontinent, and relatively restricted gene flow between breeding tracts. Due to the limited number of Bactrian camel samples included in this study, robust comparisons of genomic diversity between Bactrian and dromedary camels could not be made. Nevertheless, the moderate heterozygosity observed in the present study indicates that Indian dromedaries still retain a substantial reservoir of genetic diversity, likely maintained through traditional pastoral management, herd mobility, and periodic exchange of breeding animals among pastoral communities.

Table 5.

Summary of observed heterozygosity (HO) and expected heterozygosity (HE) for the pooled Indian camel population

Population	Mean Ho ± SD	Mean HE ± SD
Indian dromedary camels	0.303 ± 0.001	0.319 ± 0.0001

LD decay (r²) was plotted across inter-marker distance classes from 0 kb to 10 Mb (Fig. 12) for the pooled data. LD decay pattern in the indicine dromedary camels showed a clear decline in mean r² with increasing physical distance between SNPs (Table 6). LD was highest at 0–10 kb (r² = 0.783) and gradually decreased across 10–25 kb (r² = 0.746), 25–50 kb (r² = 0.731), 50–100 kb (r² = 0.724), and 100–500 kb (r² = 0.705). Beyond 0.5 Mb, LD declined further to 0.661 (0.5–1 Mb), 0.616 (1–5 Mb), and 0.514 (5–10 Mb). The decay curve showed a sharp initial decline, followed by a gradual reduction, indicating moderate LD persistence and relatively slow genome-wide LD decay in the Indian camel population. Comparable LD decay patterns have been reported in genomic studies of genome-wide analyses of Arabian Peninsula dromedaries, which showed a gradual decline in LD [46, 50]. Reviews of camel genomics highlight that pastoral mobility and gene flow among herds contribute to moderate LD and gradual LD decay, with similar patterns observed in other camel populations where high LD at short genomic distances progressively declines with increasing inter-marker distance due to recombination accumulated over generations [51–53]. Overall, the LD decay pattern observed in the present study closely resembles those reported in other dromedary populations, supporting moderate LD persistence shaped by historical gene flow and demographic processes.

Fig. 12.

LD decay across Indian dromedary camels

Table 6.

Linkage disequilibrium (r2) values observed up to 10 Mb base pair distance

Inter-Marker Distance	Mean r2
0-10kb	0.78328
10-25kb	0.74583
25-50kb	0.73101
50-100kb	0.724274
100-500kb	0.704986
0.5-1 Mb	0.66123
1-5 Mb	0.615655
5-10 Mb	0.513673

LD-based historical N_e was plotted across 1000 generations in the past (Fig. 13). The analysis indicated strong short-range LD with gradual decay, consistent with moderate historical effective population size and demographic constraints. LD-based reconstructions revealed a pronounced decline from > 2,500 ancestral individuals to ~ 167 individuals approximately 13 generations ago, suggesting a recent demographic contraction. Such declines in N_e have been reported across several livestock populations and are often attributed to intensified selection, population fragmentation, and reduced numbers of breeding individuals in modern breeding programs. However, LD-based estimates of N_e should be interpreted with caution because they can be influenced by sample size, marker density, and population structure. A further limitation of the present study is the relatively small number of individuals sampled per population. Although the sampling was broadly balanced across populations and sufficient for detecting major patterns of population structure observed in this study, larger sample sizes would improve the accuracy of breed-specific diversity parameters and demographic inferences. In particular, small breed-specific sample sizes can increase sampling variance in LD estimates, potentially leading to less precise estimates of N_e. This limitation has also been noted in genomic studies of livestock populations, including the analysis of Indian cattle breeds reported by Eva M. Strucken et al. [54], where limited sample representation for some breeds affected the robustness of LD-based demographic inferences. Similar methodological considerations regarding LD-based N_e estimation have also been discussed [55, 56], that highlighted finite sample sizes and marker ascertainment bias to inflate LD and N_e estimates.

Fig. 13.

The effective population size of the Indian Camel population up to 1000 generations in past

Despite these limitations, the observed downward trend in N_e likely reflects genuine demographic pressures acting on the population, including reduced breeding populations, habitat constraints, and environmental changes. Declining N_e is an important indicator of potential loss of genetic diversity and increased risk of inbreeding, which can ultimately affect the long-term adaptive potential of livestock populations. Therefore, continued monitoring of genetic diversity and implementation of sustainable breeding and conservation strategies are essential to maintain the genetic health and evolutionary resilience of these populations.

Conclusion

The present genome-wide analysis provides valuable insights into the population structure, genetic diversity, and demographic history of Indian camel populations included in this study. While most Indian camel populations exhibit low to moderate genetic differentiation and substantial shared ancestry, likely shaped by historical pastoral mobility and overlapping breeding tracts, the Kharai camel breed consistently emerged as genetically distinct across PCA, Admixture, F_ST, neighbour-net, and phylogenetic analyses. This distinctiveness may reflect a combination of historical isolation, restricted gene flow with mainland populations, and ecological factors associated with its unique coastal habitat; however, the specific mechanisms underlying this differentiation require further investigation through dedicated selection signature and association studies. Despite moderate overall genetic diversity, the predominance of low-frequency variants and LD-based estimates of declining effective population size indicate demographic pressures and potential vulnerability of these populations. The observed recent contraction in effective population size underscores the importance of strategic conservation and sustainable breeding programs. Taken together, this study highlights both the shared genomic foundation of Indian dromedary camel populations and the Double-Humped population, as well as the distinct genetic position of the Kharai camel breed. These findings provide a valuable genomic framework that can support breed conservation and informed management strategies for Indian camel populations. However, given the sample sizes and population representation available in the present dataset, these findings should be interpreted with appropriate caution and further validated using larger and more geographically extensive datasets. They also provide a valuable foundation for future research aimed at exploring adaptive traits, genome-wide selection signatures, and genotype–phenotype associations that may contribute to the observed genetic distinctiveness of the Kharai breed.

Authors' contributions

All authors were involved in the conception, design and execution of this work. G. R. Gowane and Ved Prakash developed the research framework, provided methodological guidance, and critically reviewed the manuscript. Dr. Basanti Jyotsana was invloved in data generation and project implementation. Amisha carried out the data analyses and drafted the original manuscript. Ashish Yadav and Karan Mahar contributed to the analytical work. Shubham Kumar Singh heled for paper editing and read the manuscript. Rani Alex offered constructive comments and suggestions that strengthened the manuscript and also read the draft. All authors revised the manuscript critically for important intellectual content and approved the final version for publication.

Funding

The present work is an outcome of the research Project titled “Indian dromedary camel genome diversity analysis and development of customized low density SNP chip for camel’ (IAEA Research Contract No: 24845) under the Coordinated Research Project (CRP): “Improving efficiency of animal breeding programs using nuclear related genomic information –practical applications in developing countries (D31030)”. The financial assistance to carry out this work by International Atomic Energy Agency (IAEA), Vienna, Austria is duly acknowledged.

Data availability

All the data has been compiled and presented in the form of figures and tables inside the manuscript. The datasets used and analyzed during the were previously published and available in the European       Nucleotide Archive (ENA) (Project numbers: PRJEB47650) ( [https://www.ebi.ac.uk/ena/browser/text-  search?query=PRJEB47650](https:/www.ebi.ac.uk/ena/browser/text-%20%20search?query=PRJEB47650) ).

Declarations

Ethics approval and consent to participate

The informed consent from the owners of the camel farm was obtained for use of the samples in this study. The collection of blood samples required for whole genome sequencing to carry out this study was approved by the Institutional Animal Ethics Committee (IAEC) of ICAR-NRCC with proposal number NRCC/IAEC/2023/01/03 dated 03 January 2023.

Competing interests

The authors declare no competing interests.