Genomic insights into the conservation and population genetics of two Chinese native goat breeds

Abstract

Chinese goats are an important group of goats worldwide. However, there are few studies on the conservation priority, genetic relationship, and potential gene flow between Chinese and global goat breeds. Here, we genotyped 239 goats from conservation populations of the Chinese Guangfeng and Ganxi breeds using the GoatSNP50 BeadChip. The conservation priority, population structure, selection signatures and introgression of these goats were analyzed in the context of 36 global goat breeds. First, we showed that Guangfeng and Ganxi goats had the largest effective population sizes across the global breeds 13 generations ago. Nevertheless, Ganxi goats have recently experienced a high degree of inbreeding, resulting in their conservation priority based on total gene and allelic diversities being lower than that of most other Chinese breeds (including Guangfeng goats). Population structure and admixture analyses showed that an average of 18% of Guangfeng genomic components were introgressed from Boer goats approximately 18-yr ago. Next, we reconstructed the subfamily structure of the core populations of Guangfeng and Ganxi goats, and proposed reasonable conservation strategies for inbreeding management. Moreover, a list of candidate genes under selection for fertility, immunity, growth, and meat quality were detected in Guangfeng and Ganxi goats. Finally, we identified some genes related to body development and reproduction, which were introgressed from Boer goats and may be beneficial for improving performance and productivity of Guangfeng goats. In conclusion, this study not only provides new insights into the conservation and utilization of Guangfeng and Ganxi goats but also enriches our understanding of artificial introgression from exotic goats into Chinese local goats.

Conservation and population genetic analyses uncover the conservation status, genetic architecture, selection signatures, and artificial introgression of two Chinese indigenous goat breeds in the context of 36 global breeds.

Introduction

The goat is the earliest domesticated ruminant and the most widely distributed domestic animal in the world (Zeder, 2008). China has a wealth of local goat breeds, and the number of goats ranked second in the world in 2019 (http://www.fao.org/). A total of 60 local goat breeds were recorded on the National Breed List of Livestock and Poultry Genetic Resources (2021 edition). These goats are well adapted to various agroclimatic environments and have excellent disease resistance and desirable meat quality (Du et al., 2011), providing a variety of genetic materials for scientific research and modern goat breeding. In recent decades, international commercial breeds (such as Angora, ANG; Boer, BOE; Nubian, NBN; and Saanen dairy, SAA) have experienced heavy selection for fast growth rates and high meat, milk, and fiber productivity. These competitive breeds have been widely introduced into China, posing a potential threat to the Chinese local goat industry. Increasing studies demonstrated that since the international commercial pig and chicken breeds were widely imported into China, the effective population sizes of Chinese local pigs (Ai et al., 2013; Yang et al., 2017; Xu et al., 2019; Liu et al., 2020) and chickens (Zhang et al., 2018a, 2020; Chen et al., 2019b) have decreased and many native breeds have been mixed with these exotic breeds. However, this has rarely been reported in Chinese goats.

Currently, the effective population sizes of some Chinese autochthonous goats are decreasing due to destruction of native habitats, inadequate management, and potential indiscriminate crossbreeding between commercial and indigenous breeds (Du et al., 2011). To better preserve these goat resources, many conservation farms have been established in China. Some conservation farms implement in situ preservation programs, restricting goats of this breed to at least one enclosed breeding farm to avoid gene flow from other goat breeds outside. However, the small and closed population inevitably results in a lower level of genetic variability and higher inbreeding coefficients than expected (Frankham, 2010; Ivy and Lacy, 2012), challenging long-term conservation. At present, conservation and population genetics based on molecular markers provide new opportunities for the preservation of goats. We can more precisely evaluate genetic parameters and then design effective conservation schemes (Benestan et al., 2016). Earlier conservation and population genetic analyses of goats used mitochondrial DNA (Chen et al., 2005; Liu et al., 2009) and low-density microsatellite markers (Fan et al., 2010; Xu et al., 2010; Wei et al., 2014), which have limited power to provide high-resolution results. Later, the extensive application of high-density single-nucleotide polymorphism (SNP) chips (Tosser-Klopp et al., 2014) provided a better approach for research on the genetic diversity, population structure, evolutionary history, and genetic bases of germplasm characteristics of goat resources. Some representative studies are related to the AdaptMap project (Colli et al., 2018; Stella et al., 2018; Talenti et al., 2018), which genotyped 4,653 individuals representing 169 global goat populations from 35 countries (no Chinese samples) on six continents using a 50K SNP chip (Tosser-Klopp et al., 2014), providing data for the global assessment of goat genetic diversity and population structure. In China, a growing number of studies have examined the population genetics of native goats using SNP chip data. For example, Islam et al. (2019) and Berihulay et al. ( 2019a, 2019b) used the 50K SNP chip data of six Chinese native goat breeds (including 24 Guangfeng [GF1] goats sampled in 2011) to analyze genetic diversity, population structure, and selection signatures for reproduction. Islam et al. (2020) reported the effective population size, inbreeding status, and population structure of three Chinese goat breeds using 50K SNP chip data. These studies facilitate improved conservation and genetic research on the Chinese local goat breeds, but the genetic relationships and potential gene flow among Chinese and global goat breeds have not been reported. It is worth noting that the recent rapid development of sequencing technology provides a great opportunity to investigate goat genetics using whole-genome sequencing data (Bickhart et al., 2017; Cai et al., 2020; Zheng et al., 2020). Although the cost of whole genome sequencing has reduced dramatically, it is still expensive and time consuming, which limits its widespread application.

Guangfeng (GF) and Ganxi (GX) goats are the only two indigenous goat breeds in Jiangxi Province, China. They are renowned for their outstanding prolificacy and environmental tolerance (Du et al., 2011). Recently, exotic goat breeds (such as Boer goats) were introduced to cross with Jiangxi goat breeds to improve their commercial traits. As a result, many native goats may have genetic components introgressed from exotic breeds. Moreover, the conservation populations of GF and GX goats are restricted to one and two government-supported conservation farms, which were established in 2014 to 2015. Previous studies researched the genetic diversity of GF and GX goats using low-density microsatellite markers or earlier samples (Fan et al., 2010; Wei et al., 2014; Berihulay et al., 2019a). The results revealed that GF and GX goats had a certain degree of inbreeding (Fan et al., 2010; Wei et al., 2014), and GF goats had a low level of effective population size (Ne; Berihulay et al., 2019a). However, the conservation status of current GF and GX goats, including genetic variability, inbreeding, admixture, and family structure, is largely unclear. In this study, we sampled 239 individuals from the GF and GX conservation populations. We then explored Illumina GoatSNP50 BeadChip data to uncover the conservation priority, genetic architecture, selection signatures and artificial introgression of GF and GX goats in the context of 36 global breeds.

Materials and Methods

All animal experiments presented in this study were authorized by the Animal Care and Use Committee of South China Agricultural University, Guangzhou, China (SCAU#2013-10).

Populations and data

A total of 1,087 goats from 36 global goat breeds were analyzed in this study (Figure 1 and Table 1). Among them, 239 individuals were sampled from conservation populations of GF and GX goats in 2019: GF2 (N = 80) and GX (N = 159) goats. GX goats were collected from the Luxi (LX, N = 86) and Shangli (SL, N = 73) conservation farms, while GF2 goats were sampled from GF conservation farm. We preferentially selected all unrelated rams and ewes without consanguninty relationship within three generations and included some senior ewes of unknown origin. These 239 goats were selected, as much as possible, to cover all existent consanguinities of the three conservation populations, and the depth of pedigrees were up to three generations. We isolated genomic DNA from ear tissue using a phenol/chloroform method. SNP genotyping was accomplished using the Illumina GoatSNP50 BeadChip (containing 53,347 SNPs; Tosser-Klopp et al., 2014). Moreover, we downloaded the 50-K SNP data of 848 goats from six Chinese, six African, five American, two Oceanian, eight European, three Central Asian, four West Asian, and one Asian wild goat breed that were reported in previous studies (Colli et al., 2018; Stella et al., 2018; Berihulay et al., 2019a). According to the results (phylogenetic tree and admixture analyses) of previous study (Colli et al., 2018), global goat breeds with fewer admixtures and representative of their geographical classification were selected for this study. The number of breeds on each continent was as equal as possible, and these breeds were widely scattered on the map. The 848 goats included 24 GF1 goats, which were collected from many different places to represent the genetic diversity of GF breed (Berihulay et al., 2019a). We merged and filtered these SNP datasets using PLINK v1.9 software (Purcell et al., 2007; Chang et al., 2015) under the following criteria: (1) Individual and SNP call rates greater than 0.9, and (2) Minor allele frequency greater than 0.01. All SNPs were remapped to the ARS1 reference (Bickhart et al., 2017), and then the unmapped SNPs and those on sex chromosomes were eliminated. Finally, a set of 45,452 informative SNPs for all 1,087 goats was utilized for the subsequent analyses.

Figure 1.

Geographical distribution of the goat breeds tested in this study. Breed abbreviations are shown in Table 1.

Table 1.

Genetic diversity of 38 global goat populations

Group	Breed	Origin	Symbol	No	Fis	FROH	Ho	He	π (×10-6)	Ne
Africa	Boer	Australia	BOE	25	0.137	0.150	0.379	0.363	6.89	62
	Galla	Kenya	GAL	23	0.143	0.003	0.376	0.370	7.02	99
	Mubende	Uganda	MUB	23	0.178	0.032	0.361	0.358	6.71	90
	Nubian	Egypt	NBN	25	0.174	0.095	0.363	0.368	6.88	85
	Saidi	Egypt	SID	25	0.135	0.060	0.380	0.398	7.48	106
	Sofia	Madagascar	SOF	24	0.437	0.164	0.248	0.260	4.94	100
America	Caninde	Brazil	CAN	25	0.246	0.103	0.334	0.304	5.67	54
	Creole	Argentina	CRE	23	0.183	0.071	0.359	0.376	7.06	102
	Kiko	USA	KIK	11	0.026	0.043	0.429	0.405	7.88	52
	Moxoto	Brazil	MOX	25	0.211	0.065	0.349	0.333	6.20	72
	Spanish	USA	SPA	19	0.045	0.048	0.420	0.415	7.88	74
Central Asia	Bugituri	Pakistan	BUT	25	0.263	0.146	0.323	0.334	6.31	66
	Kamori	Pakistan	KAM	25	0.318	0.204	0.299	0.289	5.43	53
	Teddi	Pakistan	TED	23	0.206	0.116	0.348	0.33	6.24	58
China	Arbas Cashmere	China	AC	59	0.169	0.093	0.365	0.360	6.75	87
	Guangfeng1	China	GF1	24	0.265	0.130	0.323	0.343	6.49	71
	Guangfeng2	China	GF2	80	0.152	0.042	0.372	0.375	6.99	108
	Ganxi (Luxi)	China	LX	86	0.266	0.074	0.214	0.224	6.26	131
	Ganxi (Shangli)3	China	SL	62	0.229	0.025	0.226	0.228	6.39	130
	Jining Grey	China	JN	37	0.077	0.019	0.406	0.401	7.51	98
	Luoping	China	LP	24	0.297	0.040	0.308	0.312	5.92	103
	Nanjiang	China	NJ	23	0.139	0.028	0.378	0.370	7.02	90
	Qingeda	China	QG	24	0.094	0.014	0.397	0.394	7.48	98
Europe	Alpen	Italy	ALP	25	0.074	0.055	0.407	0.404	7.58	83
	Corse	France	CRS	25	0.089	0.026	0.400	0.401	7.58	99
	Girgentana	Italy	GGT	25	0.186	0.116	0.358	0.358	6.76	75
	Landrace	Finland	LNR	20	0.180	0.055	0.361	0.359	6.83	86
	Old_Irish	Ireland	OIG	19	0.221	0.136	0.343	0.323	6.12	52
	Poitevine	France	PTV	24	0.147	0.093	0.375	0.371	7.00	78
	Saanen	Switzerland	SAA	23	0.127	0.072	0.384	0.370	6.98	77
	Valdostana	Italy	VAL	24	0.160	0.096	0.369	0.376	7.14	84
Oceania	Cashmere	Australia	CAS	25	0.100	0.076	0.395	0.376	7.07	63
	Rangeland	Australia	RAN	24	0.032	0.027	0.425	0.416	7.81	84
West Asia	Angora	Turkey	ANG	25	0.107	0.064	0.392	0.372	7.03	68
	Ankara	Turkey	ANK	20	0.080	0.036	0.404	0.397	7.52	79
	Kil	Turkey	KIL	25	0.066	0.026	0.410	0.404	7.65	85
	Kilis	Turkey	KLS	25	0.079	0.029	0.405	0.400	7.58	87
Wild	Bezoar	Iran	BEZ	7	0.353	0.144	0.282	0.337	6.73	51

Previous population; ²Current population; ³Eleven admixed individuals were excluded in this table; No, sample number; Fis, SNP-based inbreeding coefficient; F_ROH, ROH-based inbreeding coefficient; Ho, observed heterozygosity; He, expected heterozygosity; π, nucleotide diversity; Ne, effective population size.

Inference of genetic distance and population structure

All 1,087 goats and 45,452 autosomal SNPs were used to construct two matrices containing genetic distance and genetic differentiation (F_ST) values (Weir and Cockerham, 1984) among individuals and populations using PLINK v1.9 software (Purcell et al., 2007; Chang et al., 2015). The genetic distance was calculated based on identical-by-state (IBS), and the PLINK option was “--distance-matrix”. A neighbor-joining (NJ) tree of individuals was generated via the neighbor program in PHYLIP v3.69 (Felsenstein, 1989) and was visualized by Figtree v1.4.2 (http://tree.bio.ed.ac.uk/software/figtree/). The F_ST matrix was used to display the population network with the NeighborNet option in SplitsTree v4.15.1 (Huson and Bryant, 2006). Principal component analyses (PCAs) were conducted using GCTA software (Yang et al., 2011) to illustrate the relationship among 1,087 individuals and 430 Chinese goats. The first two principal components were plotted. Eleven SL goats that clustered with Luoping (LP) goats in the NJ tree (Figure 3A) were removed in subsequent analyses. ADMIXTURE v1.30 (Alexander et al., 2009) was employed to estimate the most likely number of ancestral populations (K) for all tested populations with 100 bootstraps. We used 44,494 SNPs with pairwise linkage disequilibrium (LD) values of less than 0.5 to reduce the effect of ascertainment bias. According to previous descriptions (Wang et al., 2018; Chen et al., 2019a), we randomly retrieved the data from a subset of 10 goats from each population (retaining seven Asian wild goats) for ADMIXTURE analyses to avoid sampling bias. The cross-validation error revealed that the optimal number of hypothetical ancestries was 17. The results for K = 2 to 5 and 17 were visualized via R scripts, and the circular plot of all Ks was constructed using the R package BITE (Milanesi et al., 2017). Finally, three IBS-based NJ trees uncovered the family substructure of the GF2, LX, and SL populations, and then a conservation strategy was suggested for these populations according to the results mentioned above.

Figure 3.

Population structure of Guangfeng and Ganxi goats. (a) The ancestral lineage compositions of Guangfeng and Ganxi goats in a worldwide panel of goats. The assumed number of ancestries from 2 to 5 and 17. Four Guangfeng and Ganxi populations, LX, SL, GF1, and GF2, are highlighted by red wireframes. (b) Cross-validation errors of the admixture analysis for different K values. K = 17 represents the optimal number of assumed ancestors. (c) Population admixture and splits of the nine Chinese goat populations, four international goat breeds and one Asian wild goat population. Breed abbreviations are shown in Table 1.

Evaluation of genetic diversity indices and conservation priorities

The expected heterozygosity (He), observed heterozygosity (Ho), runs of homozygosity (ROH), inbreeding coefficient (Fis), nucleotide diversity (π), and Ne were computed to estimate the genetic diversity of 1076 goats from 38 global populations. PLINK v1.9 (Purcell et al., 2007; Chang et al., 2015) was used to evaluate He and Ho with default settings. ROH were identified using a sliding window approach in PLINK v1.9 (Purcell et al., 2007; Chang et al., 2015) with the following parameter settings (Bertolini et al., 2018; Xu et al., 2019): 1) The number of SNPs in a sliding window was 50; 2) One heterozygote and five missing calls were allowed per window; 3) The minimum length of ROH was set 1 Mb, and each ROH contained at least 15 SNPs. The ROH-based inbreeding coefficient (F_ROH) for each goat was calculated using the following method (McQuillan et al., 2008): F_ROH = L_ROH/L_auto, where L_ROH is the total length of ROHs in the genome of each animal, and L_auto is the total length of 29 autosomes of goats covered by SNPs, which was 2.4 Gb (Islam et al., 2019). Three ROH categories were further divided (Pertoldi et al., 2014): 1–5 Mb (F_ROH1-5), 5–10 Mb (F_ROH5-10), and >10 Mb (F_ROH>10). We pruned the dataset with the command “indep-pairwise 50 10 0.5”, and Fis was calculated using 44,494 SNPs in PLINK v1.9 (Purcell et al., 2007; Chang et al., 2015) with the “--het” command. π was computed in a 500-kb sliding window with 250-kb steps via VCFtools software (Danecek et al., 2011). The SNeP v1.11 program (Barbato et al., 2015) was employed to estimate Ne using the following formula (Corbin et al., 2012):

$${\displaystyle \begin{array}{l}{\displaystyle N_{T(t)}=\frac{1}{4f(C_t)}(\frac{1}{E\left[r_{adj}^2|C_t\right]}-1)}\end{array}}$$

where $N_{T(t)}$ is the evaluated Ne t generations ago, $C_t$ is the recombination rate t generations ago inferred by the program, $E\left[r_{adj}^2|C_t\right]$ is the linkage disequilibrium (LD) expectancy adjusted for sampling bias, and $f(C_t)$ is a modified function of the recombination rate based on the genetic distances with default values of 1 Mb = 1 cM.

To estimate the contribution of each subpopulation to the total genomic diversity, we used _METAPOP2 v2.4 software (López-Cortegano et al., 2019a) to calculate the percentage of loss (+) or gain (–) of gene diversity and allelic diversity after removal of each subpopulation. The loss of diversity after removing a subpopulation indicates that the subpopulation contributes positively to diversity, while the gain of diversity means a negative contribution (Petit et al., 2010; López-Cortegano et al., 2019a). We first computed the gene diversity within (H_S) and between (Nei’s minimum genetic distance, D_G) subpopulations, and then total gene diversity was defined as H_T = H_S + D_G. Analogously, total allelic diversity was denoted as A_T = A_S + D_A. The intra-subpopulation component (A_S) is estimated from the average number of alleles segregating in the subpopulations minus one. The inter-subpopulation component (D_A) is calculated based on the average number of unique alleles present in the subpopulation compared with other subpopulations. In addition, we calculated the maximal expected heterozygosity (H) and total number of alleles (K) to estimate the percentage of individuals contributing from each subpopulation to a pool of synthetic population via _METAPOP2 v2.4 (López-Cortegano et al., 2019a). For a better understanding, we set the size of the synthetic pool equal to the number of test samples.

Detection of introgression

Given the complex population structure of goats worldwide, we retrieved data of 524 goats from seven Chinese breeds, four international commercial breeds, and one Asian wild goat population (BEZ) to infer introgression in GF and GX goats. The international commercial goat breeds included ANG, BOE, NBN, and SAA goats. As the main exotic breeds, they have been widely introduced into China for crossing with local goats (Du et al., 2011). TreeMix software (Pickrell and Pritchard, 2012) was first employed to estimate the genetic pattern of population splitting and migration events in Chinese goats using BEZ as the outgroup with 100 bootstraps. Maximum likelihood (ML) tree was plotted using the treemix.bootstrap function of BITE package (Milanesi et al., 2017). Then, the f3 statistic (David et al., 2009) was calculated to infer the significance of admixture in GF and GX goats via TreeMix software (Pickrell and Pritchard, 2012) with the default parameters. Z-scores of f3 less than –2 were considered significant. We set 3 yr as one generation, and ALDER v1.03 software (Loh et al., 2013) was used to evaluate the time of admixture between populations under the default setting. Patterson’s D test (Patterson et al., 2012) was applied to assess the introgression from four populations with the relationship (((P1, P2), P3), O). Where O refers to an outgroup, and P1 is the reference group with no gene flow with P3 and is closer to P2 than P3. A Positive D-value indicates that P2 shares more derived alleles with P3, while a negative D-value indicates that P1 and P3 have more common alleles. The tree topology (((GX, GF), BOE), BEZ) was used to test whether GX and GF were admixed with the candidate introgressor BOE goats. The GX (P1) goat group included 86 LX and 62 SL individuals. We excluded three goats with Asian genetic components less than the mean minus two standard deviations (SDs) and then set 101 GF (24 GF1 and 77 GF2) goats as P2. BOE goat was used as P3 because it was shown to have gene flow with GF according to the results of ADMIXTURE, TreeMix, and f3. BEZ was set as the outgroup (O) because the Bezoar populations were believed to be the wild progenitor of domestic goats (Zheng et al., 2020). To identify the introgressed regions, the f_dM index reported by Malinsky (Malinsky et al., 2015) was computed using a 500-kb sliding window with 250-kb stepping. Positive f_dM values indicated introgression between BOE and GF goats. The windows with the top 1% of f_dM values were defined as significantly introgressed regions, and the adjacent windows were merged into concatenated introgressed regions. The threshold P value of f_dM (0.850) estimated by Z-transformation was 0.013.

Signatures of selection

In this study, we selected 101 GF (24 GF1 and 77 GF2), 148 GX (86 LX and 62 SL), and 167 other Chinese goats mentioned above to detect signatures of selection. First, the integrated haplotype score (iHS; Voight et al., 2006) was calculated within the GF and GX breeds via selscan v1.2.0a (Szpiech and Hernandez, 2014) under default settings. Beagle v5.0 (Browning and Browning, 2007) was employed to impute missing and phase genotypes. Then, Weir and Cockerham’s F_ST (Weir and Cockerham, 1984) was calculated between 101 GF and 167 other Chinese goats and between 148 GX and 167 other Chinese goats using PLINK v1.9 (Purcell et al., 2007; Chang et al., 2015). Furthermore, we used the command “-xpehh” (Sabeti et al., 2007) under the default setting in selscan v1.2.0a software (Szpiech and Hernandez, 2014) to detect the signatures of selection in 101 GF and 148 GX goats compared with 167 other Chinese goats. norm software (Szpiech and Hernandez, 2014) was used to normalize the iHS and cross population extended haplotype homozygosity (XPEHH) scores with default parameters. The |iHS|, XPEHH, and F_ST values were visualized with circular Manhattan plots using the R package rMVP (Yin et al., 2021). For the three methods, SNPs at the empirical threshold of the top 1% distribution were identified as potentially selected regions (100-kb upstream and 100-kb downstream). We merged the continuous windows using BEDTools v2.26.0 (Quinlan and Hall, 2010). The significantly selected regions identified by two or three methods were treated as a shortlist.

Annotation of candidate genes

Ensembl BioMart MartView (ARS1; http://asia.ensembl.org/index.html) was used to annotate the genes within the significant candidate regions. Enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Gene Ontology (GO) terms were then detected using the Metascape database (Zhou et al., 2019), and the significance threshold was set as P < 0.01.

Results

Population structure and evolutionary history of Guangfeng and Ganxi goats

A genetic distance based (IBS) NJ tree was first constructed to examine the genetic relationships among 1,087 goats. In the NJ tree, all goats from the same population clustered together except for eleven SL goats that were grouped with LP goats (Figure 2A). Two GF (GF1 and GF2) populations clustered in one clade and showed intimate relationships with North Chinese goat breeds, including Jining Grey (JN), Arbas Cashmere (AC), Nanjiang (NJ), and Qingeda (QG) goats. GF goats exhibited a larger genetic distance than southern Chinese LP goats when compared to GX goats. The F_ST-based population dendrogram was consistent with the NJ tree. Goat breeds from each geographical group always had closer phylogenetic relationships (Figure 2B). The two pairs of GF (GF1 vs. GF2, F_ST = 0.028) and GX (LX vs. SL, F_ST = 0.015) populations grouped on one branch and displayed the lowest F_ST among the 38 tested goat populations.

Figure 2.

Phylogenetic relationships of global goat populations. (a) Neighbor-joining tree of 1,087 individuals from 38 goat populations. (b) F_ST-based NeighborNet tree of 38 goat populations. (c) Principal component analysis of 1,087 goats from 38 goat populations. The eight geographical groups are denoted by different colors. (d) Principal component analysis of 430 goats from nine Chinese goat populations. The seven Chinese goat breeds are indicated by different ellipses. Breed abbreviations are shown in Table 1.

The PCA results were in agreement with the NJ tree, and all individuals clustered in eight geographical groups (Figure 2C). PC1 explained 8.4% of the variance between Asia and other continents, and PC2 accounted for 3.2% of the variance between Europe and other continents (Figure 2C). Similar to a previous report (Colli et al., 2018), in the PCA plot, BEZ goats were at the center, and the groups from Europe, Asia, and Africa were located in the three corners. GF and GX goats overlapped with other Chinese goats in Figure 2C. In the PCA plot of Chinese goats (Figure 2D), the two GF populations (GF1 and GF2) overlapped with each other and showed a close relationship with JN goats, whereas the two GX populations (LX and SL) clustered and had a close relationship with LP goats.

The genetic compositions of 377 goats from 38 worldwide populations were plotted with Ks (Figures 3A and Supplementary Figure S1). With an increase from K = 2 to 5, the ancestral lineages of Asia (LP), Europe (OIG), Africa (SOF), America (CAN), and Central Asia (KAM) appeared one by one. Both GF1 and GF2 individuals displayed obvious signatures of admixture with exotic goat breeds (K = 2 to 5; Figure 3A). In contrast, LX and SL goats showed few signatures of introgression. When K ≥ 10, BOE goats represented an independent lineage, and the introgression signal of the BOE was consistently present in the GF1 and GF2 populations (Supplementary Figure S1). The cross-validation error test indicated that K = 17 was the optimal number of assumed ancestors (Figure 3B), and the GF and GX goats formed an independent ancestral lineage. LX and SL goats contained ancestry fractions of LP goats, and GF1 and GF2 goats had genetic components of AC, LP, and BOE origin. GF2 goats displayed a higher ancestral proportion of BOE lineages than GF1 goats.

To further investigate the population split and admixture of GF and GX goats, the TreeMix approach was used to infer a maximum-likelihood (ML) tree and potential migration events for nine Chinese goat populations, four commercial goat breeds, and one Asian wild goat population. We plotted the ML tree with the number of migration events set as 2, which explained a large fraction (99.9%) of the variance in relatedness between populations (Figure 3C). In the inferred ML tree, Chinese and commercial goat breeds were grouped into two separate clades. GF populations and GX populations were clustered on one branch, consistent with the F_ST-based phylogenetic results. Both migration events were from BOE to Chinese goats. The strongest introgression signal revealed a genetic contribution from BOE goats to Chinese JN goats, with a migration weight (w) of 29.9%. Another significant admixture event was gene flow from BOE goats into the two GF populations (w = 18.5%).

Genetic diversity and conservation priority of Guangfeng and Ganxi goats

To evaluate the genetic variability of GF and GX goats, we first calculated Ho, He, Fis, and π among 38 global goat populations (Table 1). Of note, LX (Ho = 0.214, He = 0.224) and SL (Ho = 0.226, He = 0.228) goats had the first and second lowest values of Ho and He, and their Fis (Fis_LX = 0.266, Fis_SL = 0.229) levels were higher than those of most other breeds. The previously sampled GF goats (GF1) had lower values of Ho (0.323) and He (0.343) that were even smaller than those of international commercial goat breeds (ANG, BOE, NBN, and SAA goats). The π values of LX (π = 6.26 × 10^–6), SL (π = 6.39 × 10^–6) and GF1 (π = 6.49 × 10^–6) were lower than those of most other goat breeds. These results showed that GX goats, especially the LX population, had lower levels of genetic diversity. In contrast, the current GF goats (GF2) displayed higher levels of Ho (0.372), He (0.375), and π (6.99 × 10^–6) and lower Fis (0.152) values than the GF1 population. Intriguingly, goat breeds from South China showed lower genetic diversity than those from North China, except for GF2 goats. The Ne ranged from 51 in BEZ goats to 131 in LX goats for the past 13 generations (Table 1). LX, SL (Ne = 130) and GF2 (Ne = 108) goats had the top three largest Ne values among the goat breeds worldwide. F_ROH was also estimated in this study, and the results showed that F_ROH had significant positive correlationships with Fis in GF2 (R = 0.83, P = 2.15 × 10^–13), LX (R = 0.93, P < 2.20 × 10^–16) and SL (R = 0.71, P < 2.20 × 10^–16) goats (Supplementary Figure S2), respectively. When ROH were divided into three classes, F_ROH with ROH > 10 Mb (F_ROH>10) showed the largest value of correlationships with F_ROH in GF2 (R = 0.97, P = 3.30 × 10^–16), LX (R = 0.99, P < 2.20 × 10^–16), and SL (R = 0.99, P = 1.32 × 10^–13) populations (Supplementary Figure S2), respectively.

We estimated the loss or gain of gene diversity and allelic diversity by removing each subpopulation from the dataset to elucidate the contribution of each population (Petit et al., 2010; López-Cortegano et al., 2019a). Among the eight global groups (Figure 4A, Supplementary Table S1), the Chinese group exhibited the lowest H_S (–1.715%) and H_T (–1.171%) upon removal, even though it had the second largest D_G (0.544%). Consistent with the gene diversity results, the Chinese group contributed negatively to A_S (–0.030%) and little to A_T (0.049%). The second and third largest D_G and D_A (0.078%) were displayed between the Chinese and other groups. Then, we found that the Chinese group had the second highest individual contribution to global goats based on maximal gene diversity (H = 19.889%) and maximal number of alleles (K = 15.985%). Focusing on Chinese goat breeds (Figure 4B, Supplementary Table S1), LX (H_S = –1.431%, D_G = –0.203%, H_T = –1.634%), and SL (H_S = –0.682%, D_G = –0.335%, H_T = –1.017%) goats had the most negative contribution to gene diversity across all Chinese goat breeds. The contributions to allelic diversity of LX (A_S = –0.332% and A_T = –0.128%) and SL (A_S = –0.166% and A_T = –0.081%) goats were also negative. The second and third largest contributions to allelic diversity were shown between subpopulations in LX (D_A = 0.204%) and SL (D_A = 0.085%) goats. GF2 goats made positive contributions to gene and allelic diversities both within subpopulation (H_S = 1.045% and A_S = 0.269%) and total populations (H_T = 0.587% and A_T = 0.129%). However, for the components of gene diversity and allelic diversity between subpopulation, GF2 had negative values (D_G = –0.458%, D_A = –0.140%). Finally, South Chinese goat breeds had no or a small percentage of individuals contributing to the metapopulation (Figure 4B). For GX goats, LX hardly contributed any individuals to the synthetic population with H (0) and K (0.477%). SL goats contributed no individuals to the pool with H, but they made a 16.945% contribution of individuals based on K. This is consistent with the more negative contributions of LX goats than SL goats to H_T and A_T.

Figure 4.

Contributions to genomic diversity was tested in 1076 individuals. (a) Contribution of eight global groups to genomic diversity. (b) Contribution of nine Chinese goat populations to genomic diversity. The top and middle panels show the percentage of loss (+) or gain (-) of gene and allelic diversity after removing each subpopulation. H_S and A_S denote within-population gene and allelic diversities (indicated by bars), respectively. D_G and D_A denote between-subpopulation gene and allelic diversities (indicated by bars), respectively. H_T and A_T denote total gene and allelic diversities (indicated by points), respectively. The bottom panel shows the contribution of individuals (%) from each group to a synthetic pool of individuals with maximal gene diversity (H) or number of alleles (K). Breed abbreviations are shown in Table 1.

Subfamily structure and diversity contribution of Guangfeng and Ganxi goats

ADMIXTURE analysis was implemented to estimate the ancestral lineage compositions of 24 GF1 and 80 GF2 goats. Similar to the aforementioned method, we randomly retrieved the data from a subset of 10 goats from each population (GF1 was divided into three groups, and the third group had six identical animals to the second group; GF2 was equally divided into eight groups). When K = 2, Asian and European goats appeared as two differentiated lineages. GF1 goats contained a proportion of Asian genetic components ranging from 68.76% to 96.98%, with an average value of 85.54% and an SD of 10.13% (Supplementary Table S2). The Asian genetic compositions of GF2 goats ranged from 60.67% to 98.41%, with a mean of 82.07% and an SD of 8.78% (Supplementary Table S2). Three GF2 goats harboring a lineage purity value less than the average value minus two SDs (64.52%) should be removed from the conservation population. For GX goats, we removed the 11 SL individuals due to their clustering with LP goats in the NJ tree. The IBS-based NJ trees were further reconstructed for 77 GF2, 86 LX, and 62 SL goats (Figure 5). According to previous descriptions (Wang et al., 2018; Liu et al., 2020), there should be a certain genetic distance among subfamilies, and the numbers of individuals in each subfamily should be equal and include at least one male when possible. GF2, LX, and SL goats were clustered into seven, six, and seven clades in the NJ trees, thus representing seven, six, and seven subfamilies in these three conservation populations, respectively. All families of LX goats had at least one ram, but two families of GF2 and SL goats lacked rams.

Figure 5.

Family substructure of Guangfeng and Ganxi goats. (a) Seventy-seven Guangfeng goats are classified into seven subfamilies in the NJ tree. (b) Eighty-six Luxi goats are assigned to six subfamilies in the NJ tree. (c) Sixty-two Shangli individuals are clustered into seven subfamilies in the NJ tree. In these three NJ trees, subfamilies are shown in different colors, while rams are shown in black font. (d–f) The distribution of F_ROH, Fis, the loss (+) or gain (–) of genomic diversity of H_T and A_T, and the contribution of individuals with H or K in each subfamily of GF2, LX, and SL goats. The dotted and solid lines in the plots of F_ROH/Fis denote the average and average minus two SDs of F_ROH/Fis values, respectively. H_T, A_T, H, and K denote the total gene diversity, total allelic diversity, maximal gene diversity, and maximal number of alleles, respectively.

Finally, for conservation management, we calculated the Fis, F_ROH, H_T, A_T, and contribution of individuals with H and K for each subfamily of GF2, LX, and SL goats (Supplementary Table S1). Of note, six, two, and three individuals presented as outliers (> 2SDs) based on the average Fis values for GF2, LX, and SL goats (Figure 5D, E and F), respectively. In addition, seven, three, and five outliers were identified with the F_ROH patterns (mean plus 2SDs) of GF2, LX, and SL goats, respectively. We also found that two, three, and four subfamilies contributed negatively to global gene diversity and/or allelic diversity in GF2, LX, and SL goats (Figure 5D, E and F), respectively. These subfamilies basically had a higher degree of inbreeding and a lower contribution of individuals.

Signatures of selection in Guangfeng and Ganxi goats

We identified 393, 454, and 449 candidate regions according to the empirical threshold of the top 1% for the iHS, F_ST, and XPEHH methods, respectively. Then, 100 uniquely significant regions on 26 chromosomes were detected by at least two statistics (Supplementary Table S3). The GO terms of 92 annotated genes were mainly enriched in metabolism, such as “cellular biogenic amine metabolic process (GO: 0006576)” and “organic hydroxy compound metabolic process (GO: 1901615)”, and some were associated with response to external stimulus (“cellular response to abiotic stimulus (GO: 0071214)” and “cellular response to environmental stimulus (GO: 0104004)”) (Supplementary Table S4). Among these candidate genes, fourteen (SPATA16, AKAP9, GLRA1, G3BP1, FKTN, LBH, ASPH, PLAG1, MOS, SRD5A1, VRK1, CDKN1A, ARMC5, and SPATA19) were related to reproduction, four (ICA1, HAVCR2, ITCH, and ITGAD) were involved in the immune response, two (LAMA2 and BARX2) were linked to muscle development, and one (SRSF3) was related to lipid metabolism (Figure 6A and Supplementary Table S5). Notably, only two genes (LAMA2 and BARX2) related to muscle development were detected on the basis of all three statistics (Figure 6A and Table 2).

Figure 6.

Genomic signatures of selection in Guangfeng and Ganxi goats. (a) Circular Manhattan plot of genome scans for Guangfeng goats. (b) Circular Manhattan plot of genome scans for Ganxi goats. From inside to outside, the statistics are F_ST, iHS, and XPEHH. The red dotted lines represent the top 1% threshold. Two candidate genes identified by all three methods are shown in red in the Guangfeng and Ganxi goats. Two candidate genes simultaneously present in Guangfeng and Ganxi goat breeds are indicated in blue.

Table 2.

Candidate genes under selection in Guangfeng and Ganxi goats

Position, Mb	Gene1	Method	Trait	Breed	Reference
3:80.00–80.23	COL11A1	iHS, FST, XPEHH	Reproduction	GX	(Hafez et al., 2015)
8:95.17–95.21	FKTN	F ST, XPEHH	Reproduction	GF, GX	(Kurahashi et al., 2005)
9:54.46–54.98	LAMA2	iHS, FST, XPEHH	Muscle development	GF	(Miyagoe et al., 1997)
21:60.96–61.04	VRK1	iHS, FST	Reproduction	GF, GX	(Wiebe et al., 2010)
26:37.33–37.36	KIF11	iHS, FST, XPEHH	Reproduction	GX	(Castillo and Justice, 2007)
29:31.93–32.01	BARX2	iHS, FST, XPEHH	Muscle development	GF	(Meech et al., 2012)

This table shows only the genes detected with all three methods or in two Jiangxi goat breeds.

Similar to the results for GF goats, 122 significant regions on 26 chromosomes were identified in GX goats by at least two of the iHS, F_ST, and XPEHH methods (Supplementary Table S3). A total of 181 candidate genes were annotated. The GO and KEGG results showed that these genes were mainly enriched in protein production (“protein kinase binding (GO: 0019901)” and “ubiquitin-like protein conjugating enzyme binding (GO: 0044390)”), “developmental maturation (GO: 0021700)” and “glycoprotein metabolic process (GO: 0009100)” (Supplementary Table S4). We also identified a list of functional genes, including 12 fertility-related genes (COL11A1, CABS1, FKTN, TFAP2C, SUN5, SPAG4, AK7, VRK1, ARIH2, MDFI, KIF11, and LDHC); four growth-related genes (CHPF, IHH, GDF5, and CBFB); two immune-related genes (BACH2 and DNMT3B); and one gene (RAPGEF3) associated with fat deposition (Supplementary Table S5). Only two fertility-related genes (KIF11 and COL11A1) were found on the basis of all three statistics (Figure 6B and Table 2). Moreover, two fecundity-related genes (FKTN and VRK1) were shared by GF and GX goats (Table 2).

Introgression from boer to Guangfeng goats

We detected 3 out of 33 extreme Z scores for the f3 statistic of GF goats, including –2.50, –2.54 and –3.74 for the (GF2; GF1, BOE), (GF2; SAA, GF1), and (GF2; BOE, LX) groups, respectively. The ALDER results indicated that the time of admixture between GF and BOE goats was 5.97 ± 0.79 generations ago (17.91 ± 2.37 yr). The D-statistic results also showed clear evidence of gene flow from BOE to GF goats ((((GX, GF), BOE), BEZ), D = 0.116, P < 2.2 × 10^–16 (t test)). Then, the modified f-statistic (f_dM) revealed 19 significant introgressed regions on 14 chromosomes (Figure 7A and Supplementary Table S6). A total of 10 GO terms and 1 KEGG pathway were detected for 60 significant introgressed genes (Figure 7B and Supplementary Table S4). These functional categories were mainly associated with growth and development. Notably, 20 out of 60 genes were enriched in the top significant GO terms (–log₁₀(P) = 12.5) of embryonic skeletal system morphogenesis (GO: 0048704). Among the 20 genes, ten genes are members of the family of homeobox genes (HOXB1–HOXB9 and HOXB13). Interestingly, we also observed one significant selected region (29:32385176–32487669) according to the iHS and F_ST statistics in the significant introgressed region detected by f_dM (29:32250001–32750000). This 100-kb region harbored a spermatogenesis-related gene (SPATA19) (Figure 7A).

Figure 7.

Introgression from Boer to Guangfeng goats. (a) Introgressed regions identified in GF goats using the f_dM statistic. The red dotted line denotes the threshold (top 1%). (b) Top 10 significantly enriched GO terms and KEGG pathway of significant introgressed genes.

Discussion

Population structure and evolutionary history of Guangfeng and Ganxi goats

Goats were domesticated in the Fertile Crescent ~ 10,000 yr BP and dispersed throughout the world with human migrations (Zeder and Hesse, 2000). Subsequently, a large number of breeds with different characteristics have been formed through colonization, geographical and reproductive isolation, genetic drift, natural and artificial selection and migrations and/or importations (Colli et al., 2018). Ancient DNA analysis suggested that Chinese goats originated from the eastern Fertile Crescent between ~8,000 and 4,400 yr BP (Cai et al., 2020). Chinese modern goat breeds show little genetic turnover compared with that of their Fertile Crescent ancestors, especially South Chinese goat breeds (Cai et al., 2020). Gene flow from European into northern Chinese goats has been reported and may have occurred during the expansion of the Mongol Empire in Eurasia (May, 2012; Cai et al., 2020). However, our results confirmed that gene flow occurred from European lineage into Chinese GF goats, a breed raised in southern China. This is why GF goats showed a close relationship with North Chinese goats in the NJ tree and PCA. Although GF1 and GF2 goats were clustered together, indicating equivalent genetic backgrounds, the introgression proportion increased from 14.46% in GF1 goats to 17.93% in GF2 goats. The ADMIXTURE, TreeMix, and f3 results suggested that introgression was mainly from BOE goats, and the admixture degree was approximately 18%. The ALDER program estimated the time of admixture between GF and BOE goats to be approximately 5.97 generations (18-yr) ago. These estimates were consistent with historical records that BOE goats were widely imported for crossing with Chinese indigenous goats after 1995 (Du et al., 2011). GF goats likely indiscriminately crossbred with introduced BOE goats during this period. Individuals with a high level of admixture generally show better production performance. These goats may have been preferentially used in the mating programs and increased the introgression rate of the current GF population. In addition, some external GF goats have recently been introduced into the conservation farm in an attempt to improve the genetic diversity of the GF conservation population. These goats may contain some individuals with high BOE ancestry, leading to an increase in admixture fraction of the conservation population. Therefore, it is necessary to use genomic data to monitor and manage the genetic structure of GF goats. Eleven SL goats may be admixed with LP goats, and these 11 SL goats should be eliminated from the conservation population of SL goats. This result could be explained by the fact that SL goats were recently collected from different places in the GX area of Jiangxi Province. Some hybrids were retained because they were indistinguishable from pure GX goats in terms of appearance. Last, the LX and remaining SL goats displayed a similarly low level of admixture based on the population structure results. This result will benefit the conservation project for Chinese indigenous goats, as the government can support two conservation populations for GX goats.

Genetic diversity of and conservation priorities for Guangfeng and Ganxi goats

Preservation of genetic diversity is an important component of biological conservation programs (López-Cortegano et al., 2019b). The previous literature suggests that conservation approaches should take population subdivision into account and focus on global management, including potential migrations among subpopulations (Frankham et al., 2010). Here, we estimated the genetic diversity of GF and GX goats in a global goat panel. LX and SL goats had the largest Ne numbers for the past 13 generations among goats worldwide. Strangely, the LX and SL goats had low levels of genetic diversity on the basis of Ho, He, Fis, and π, with values even lower than those observed for international commercial breeds (ANG, BOE, NBN, and SAA), which are undergoing intensive selection for economic traits. ROH are widely used to infer the inbreeding history according to the formula ROH_length = 100/(2g*cM), where ROH_length is the length of ROH segment, g is the generation ago, and 1 cM is approximately equal to 1 Mb (Howrigan et al., 2011). Short ROH fragments indicate ancestral inbreeding, while long ROH fragments reflect recent inbreeding (Keller et al., 2011). F_ROH>10 made a major contribution to F_ROH of GF2, LX, and SL populations, revealing the recent (last five generations) reduction in genetic diversity and increase in homozygosity in these three populations. GF2 goats had more abundant genetic variability and a higher Ne than GF1 goats. One possible explanation is the sample size bias. Another convincing explanation is that increasing admixture also increased the genetic variability in GF2 goats. The latter hypothesis is supported by the ADMIXTURE results mentioned above. We also investigated the contributions to the genomic diversity of each group (subpopulation) to discuss conservation priorities. Chinese goats showed a low contribution of genomic diversity to the global goat pool, which may be partially attributed to SNP ascertainment bias. The SNP panel used in this study was developed based on the sequence data of diverse goat breeds, including Alpine, BOE, Creole, Katjang, SAA, and Savanna goats (Tosser-Klopp et al., 2014), which reduced the ascertainment bias to a certain extent. Previous work (Colli et al., 2018) used the AdaptMap dataset of 4,653 goats from 169 populations in 35 countries across six continents to estimate genetic diversity, revealing that SNP ascertainment bias did not have a significant impact on the major results. We believe that ascertainment bias had little influence on our results. Moreover, we cannot rule out the possibility that our samples underrepresent the genetic diversity of Chinese goats. GF2 goats contributed more total gene diversity and allelic diversity than the two GX populations, SL and LX, indicating the importance of preserving GF2 goats. LX and SL goats made negative contributions to the total gene and allelic diversities, and they contributed no individuals based on H to the synthetic pool. Although conservation priorities could differ depending on diversity criteria (López-Cortegano et al., 2019a; Zhao et al., 2021), in general, GX goats had a lower conservation priority than other Chinese goats. The lower levels of genetic diversity and higher degrees of inbreeding may underestimate the genetic contribution of these goats to total diversity, which also occurs in LP goats. As GX goats are a local representative goat breed that has made an important contribution to local economic development, more attention should also be paid to their long-term conservation.

Sustainable conservation strategy for Guangfeng and Ganxi goats

An accurate understanding of the genetic diversity and genealogical data of the managed population can help establish an effective conservation program (Goyache et al., 2003). To facilitate the management of the population structure of GF and GX goats, we reconstructed seven, six and seven subfamilies for 77 GF2, 86 LX and 62 SL core individuals. In China, state goat conservation farms should keep at least six unrelated pedigrees (subfamilies) in the core conservation population. All the conservation populations of GF and GX goats had at least six subfamilies. However, the subfamilies of GF2, LX and SL goats had unevenly distributed ram numbers, inbreeding levels, contributions to genomic diversity and individuals. Therefore, more elaborate breeding programs should be implemented to solve these problems. Small, isolated, and fragmented populations tend to have increased inbreeding and reduced genetic variation (Liu et al., 2021). High levels of inbreeding can cause inbreeding depression, which reduces population viability (Pekkala et al., 2014). A low degree of genetic variation in populations can also decrease their local adaptation and elevate their risk of extinction (Hamilton and Miller, 2016; Robinson et al., 2019). The results from previous studies suggested that family rotational mating could effectively reduce inbreeding (Honda et al., 2004; Windig and Kaal, 2008) and sustain most (90%) of the genetic variability in a livestock population for more than 100 yr (Lu, 2013). To better maintain the genetic variability of GF and GX goats, similar to previous studies (Wang et al., 2018; Xu et al., 2019), we suggest ram-mediated rotational crossing among all subfamilies of these populations and retaining similar numbers of rams and ewes in each subfamily generation by generation. Due to the heterogeneous genetic background of GF goats, reducing the mating opportunities of individuals with a high proportion of admixture can guarantee the purity of the conservation population. To avoid inbreeding depression as much as possible, goats with high levels of Fis and F_ROH should be replaced by individuals with low inbreeding levels in mating programs. In the case of two GX conservation populations, LX and SL, individuals from each population can be appropriately introduced into the other population to enrich the genetic diversity of both populations.

Candidate genes putatively under selection in Guangfeng and Ganxi goats

Natural and artificial selection leave evolutionary footprints on livestock genomes. Detection of these selection footprints can provide insights into the underlying genetic mechanisms of typical germplasm characteristics to better guide animal conservation and breeding (Berihulay et al., 2019a). Here, we calculated iHS, F_ST, and XPEHH statistics to detect genomic signatures of selection in 101 GF and 148 GX goats. The iHS statistic is powerful for detecting recent positive selection, but the selected alleles have risen only to moderate frequencies (Voight et al., 2006), while F_ST (Weir and Cockerham, 1984) and XPEHH (Sabeti et al., 2007) can identify selected alleles that have risen to fixation or near fixation. Due to differences in the statistics of each method (Mastrangelo et al., 2020), we detected few identical candidate regions with the three methods. The low SNP density (45K SNPs used in the present study) could also miss some important candidate selected regions. Therefore, similar to the previous literature (Mastrangelo et al., 2020; Wang et al., 2021), we focused on significant genes identified by at least two methods. These candidate positively selected genes may be associated with the formation of phenotypic characteristics in GF and GX goats. Functional enrichment analyses revealed that the candidate genes were mainly enriched in GO terms and KEGG pathways related to metabolic, developmental, and reproductive processes. Although some candidate genes were not associated with significant functional processes, we still focused on genes that have been reported to be associated with phenotypic traits in other species.

GF and GX goats are renowned for their desirable fertility (Supplementary Table S7; the lambing rates are 172% to 300% and 151% to 285%, respectively; Du et al., 2011; Islam et al., 2019). As expected, we detected a list of functional genes related to reproduction in GF and GX goats. Of note, COL11A1 and KIF11 were detected with all three methods in GX goats, and the pregnancy-related genes FKTN and VRK1 were detected for both GF and GX goats. COL11A1 and KIF11 are associated with embryonic development and embryogenesis (Castillo and Justice, 2007; Hafez et al., 2015), respectively. FKTN-null mice exhibit embryonic lethality (Kurahashi et al., 2005). Mice are infertile due to deficiency in the serine/threonine protein kinase VRK1 (Wiebe et al., 2010). Islam et al. (2019) also reported several candidate genes associated with fertility in Chinese goat breeds using 50-K SNP data. None of these genes were identified in our study. This disparity may be due to differences in computational parameters, tested populations and sample sizes. Several genes related to the immune response were detected in GF and GX goats, including ICA1, HAVCR2, ITCH, and ITGAD in GF goats, and BACH2 and DNMT3B in GX goats. These immune-related genes may play important roles in the adaptation of GF and GX goats to local mountain environments. Additionally, genes related to the regulation of lipid metabolism (SRSF3; Sen et al., 2013) and fat deposition (RAPGEF3; Yan et al., 2013) were detected in GF and GX goats, respectively, which may relevant to their meat quality. Four bone development genes (CHPF, IHH, GDF5, and CBFB) were identified in GX goats, which may play significant roles in its body size traits. Two genes (LAMA2 and BARX2) detected by the three methods may be involved in muscle development and growth in GF goats. LAMA2 is associated with muscle fiber degeneration (Miyagoe et al., 1997), while BARX2 is required for normal muscle growth and regeneration (Meech et al., 2012). These findings provide some theoretical basis for further research on the molecular mechanism of the germplasm characteristics in GF and GX goats.

Introgressed BOE genes possibly enhance the growth and reproductive performances of Guangfeng goats

BOE goats are native to the Republic of South Africa and have spread all over the world as one of the best-known commercial goat breeds. BOE goats have a fast growth rate and excellent adaptability, disease-resistance, fertility, and meat production (Casey and Vanniekerk, 1988; Erasmus, 2000; Malan, 2000). Currently, BOE goats are widely used to improve native goats and boost their economic traits (Du et al., 2011; Basinger et al., 2019). The introduction of this highly productive breed has led to genomic introgression into some indigenous goat breeds, such as GF goats. Introgressed genomic regions play an important role in improving productivity and performance in recipient breeds (Zhang et al., 2018b; Wang et al., 2020). Using the f_dM statistic, we detected a list of introgressed genes related to the GO term for embryonic skeletal system morphogenesis. These genes include a family of homeobox genes (HOXB1–HOXB9 and HOXB13) that play crucial roles in vertebrate embryonic development (Medina-Martínez et al., 2000). The HOXB1 gene is related to teat number in pigs (Bovo et al., 2021) and spermatogenesis and/or male fertility in rats (Pandey et al., 2019). HOXB2 and HOXB4 genes are associated with sternum development, and nearly 75% of the Hoxb2 homozygous mutant mice die within 24 h of birth, owing to split sternum (Barrow and Capecchi, 1996). HOXB3 plays an important role in skeletal development of chickens (Li et al., 2021). HOXB5 plays an important role in specifying the position of limbs along the anteroposterior axis of the vertebrate body (Rancourt et al., 1995). The HOXB6 gene controls several independent aspects of skeletal patterns in mice (Kappen, 2016). HOXB6 and HOXB7 are related to human infertility (Li et al., 2022). HOXB7 and HOXB8 are implicated in the control of the abdominal and thoracic segments, which exhibit the most striking morphologic defects (Gibby et al., 2009). HOXB9 may be implicated in the control of oocyte maturation (Paul et al., 2011). The HOXB13 gene is associated with the development of limbs and bones in sheep (Ahbara et al., 2018). We identified a shared region of significant introgression and selection on chromosome 29 harboring a spermatogenesis-related gene (SPATA19). The spermatogenesis-associated (SPATA) family of genes plays significant roles in spermatogenesis, sperm maturation, or fertilization (Sujit et al., 2020). SPATA19 (spermatogenesis associated 19) is critical for sperm mitochondrial function and male fertility (Mi et al., 2015). The high productivity of admixed individuals was likely further retained and improved by heavy selection, and then the frequency of favorable alleles was increased or even fixed (Zhang et al., 2018b). Our results suggested that human-driven cross-breeding introduces fertility and body development-related genes from BOE goats into GF goats, which possibly enhances the performance and productivity of GF goats.

Conclusion

In summary, this study explored the conservation status, population structure, evolutionary history, signatures of selection and artificial introgression of GF and GX goats in a comprehensive manner. The results indicated that GF goats had more abundant genetic variability due to introgression from Boer goats, and the conservation priority based on total gene and allelic diversities was higher than that of GX goats. For conservation management, we inferred subfamily structure and proposed reliable conservation strategies for GF and GX goats. Then, analysis of selection signatures revealed numerous candidate genes under selection for fertility, growth, immunity, and meat quality in GF and GX goats. We further identified eleven genes related to fecundity and body development that were introgressed from BOE into GF goats. This work provides a starting point for conservation and germplasm characteristic research on Jiangxi indigenous goat breeds. Further in-depth studies, such as RNA sequencing, whole-genome resequencing and phenomics, will be beneficial to confirm and refine our results.

Supplementary Data

Supplementary data are available at Journal of Animal Science online.

Figure S1. Circular representation of Admixture software results for K = 2 to 38. Breed abbreviations are shown in Table 1.

Figure S2. Scatterplots (lower triangle) and correlations (upper triangle) of the genomic inbreeding coefficients between the Fis and F_ROH (F_ROH, F_ROH1-5, F_ROH5-10, F_ROH>10) in GF2 (a), LX (b) and SL (c) goats. *P < 0.05, **P < 0.01, ***P < 0.001.

Table S1. Contribution to diversity and individuals

Table S2. Inbreeding coefficient and Chinese lineage purity of Guangfeng and Ganxi goats

Table S3 Candidate selected regions for Guangfeng and Ganxi goats

Table S4. GO terms and KEGG pathways for the candidate genes based on genome scans and introgression

Table S5. Candidate genes associated with reproduction, immunity, growth, and meat quality in Guangfeng and Ganxi goats

Table S6. Significant regions putatively under introgression from Boer into Guangfeng goats detected by the f_dM statistic

Table S7. Some body size and reproduction traits of Chinese goat breeds

Glossary

Abbreviations

A_S

allelic diversity within subpopulations

A_T

total allelic diversity

D_A

allelic diversity between subpopulations

D_G

gene diversity between subpopulations

F_ST

genetic differentiation

Fis

inbreeding coefficient

GO

Gene Ontology

H_S

gene diversity within subpopulations

H

maximal expected heterozygosity

H_T

total gene diversity

He

expected heterozygosity

Ho

observed heterozygosity

IBS

identical-by-state

iHS

integrated haplotype score

K

maximal total number of alleles

KEGG

Kyoto Encyclopedia of Genes and Genomes

LD

linkage disequilibrium

ML

maximum-likelihood;

Ne

effective population size

NJ

neighbor-joining

PCA

principal component analysis

π

nucleotide diversity

SNP

single-nucleotide polymorphism

XPEHH

cross population extended haplotype homozygosity

Conflict of Interest Statement

The authors declare no real or perceived conflicts of interest.