Genomic diversity and admixture patterns among six Chinese indigenous cattle breeds in Yunnan

Objective Yunnan is not only a frontier zone that connects China with South and Southeast Asia, but also represents an admixture zone between taurine (Bos taurus) and zebu (Bos indicus) cattle. The purpose of this study is to understand the level of genomic diversity and the extent of admixture in each Yunnan native cattle breed. Methods All 120 individuals were genotyped using Illumina BovineHD BeadChip (777,962 single nucleotide polymorphisms [SNPs]). Quality control and genomic diversity indexes were calculated using PLINK software. The principal component analysis (PCA) was assessed using SMARTPCA program implemented in EIGENSOFT software. The ADMIXTURE software was used to reveal admixture patterns among breeds. Results A total of 604,630 SNPs was obtained after quality control procedures. Among six breeds, the highest level of mean heterozygosity was found in Zhaotong cattle from Northeastern Yunnan, whereas the lowest level of heterozygosity was detected in Dehong humped cattle from Western Yunnan. The PCA based on a pruned dataset of 233,788 SNPs clearly separated Dehong humped cattle (supposed to be a pure zebu breed) from other five breeds. The admixture analysis further revealed two clusters (K = 2 with the lowest cross validation error), corresponding to taurine and zebu cattle lineages. All six breeds except for Dehong humped cattle showed different degrees of admixture between taurine and zebu cattle. As expected, Dehong humped cattle showed no signature of taurine cattle influence. Conclusion Overall, considerable genomic diversity was found in six Yunnan native cattle breeds except for Dehong humped cattle from Western Yunnan. Dehong humped cattle is a pure zebu breed, while other five breeds had admixed origins with different extents of admixture between taurine and zebu cattle. Such admixture by crossbreeding between zebu and taurine cattle facilitated the spread of zebu cattle from tropical and subtropical regions to other highland regions in Yunnan.


INTRODUCTION
Cattle are one of the most important livestock species in many aspects.Archaeological and genetic evidence suggested that modern domestic cattle descended from different aurochs populations [13], resulting in two main types -humpless/taurine cattle (Bos taurus) and humped/zebu cattle (Bos indicus) [4].Since domestication, taurine and zebu cattle each migrated out from its own domestication center, subsequently met with each other and hybridized at some contact regions such as Arabian Peninsula [5] and Southeast Asia [6].Yunnan is not only a frontier zone that connects China with South and Southeast Asia, but also represents an admixture zone between taurine and zebu cattle.Obviously, some indi viduals of Yunnan local cattle breeds seem to be offspring of crosses between taurine and zebu cattle.
According to China National Commission of Animal Ge netic Resources, there are six indigenous cattle breeds (Diqing, Zhaotong, Dengchuan, Dianzhong, Wenshan, and Dehong humped cattle) in Yunnan, Southwestern China.These breeds provide not only useful products such as cowhide, meat and milk to local farmers and communities, but also are used as the main means of transportation in remote villages.It should be emphasized that Dengchuan cattle is the only one local dairy cattle breed in China with high rate of milk production.In addition, these indigenous breeds exhibit excellent charac teristics such as tolerance to rough feeding, good adaptability under adverse climatic conditions, and resistance to some animal diseases and parasites [7,8].In recent years, the im plementation of artificial insemination, together with global climate change and frequent outbreaks of emerging animal diseases, has increased the risk of loss of Yunnan native cattle breeds.The number of some pure native breeds has been de creased from ten thousand head in the 20th century to several hundred head in present day [9].For instance, Dengchuan cattle are already on the brink of extinction, due to crossbreed ing with exotic commercial breeds such as Holstein, Simmental and Shorthorn cattle [9].As a result, it is urgent to under stand their genomic background before loss or extinction.Previous studies on Yunnan native cattle breeds focused mainly on mitochondrial diversity, origins, introgression and phy logeographic structure [7,10,11].However, information on genomic diversity and admixture patterns among six Chinese indigenous cattle breeds in Yunnan is still very limited.
Recently, highdensity single nucleotide polymorphism (SNP) genotyping arrays have become available for bovine genomic analysis, such as Illumina BovineHD BeadChip, which features 777,962 SNPs [12].The SNP chips have been used to investigate genomic diversity, population structure and admixture patterns of various native cattle breeds from different countries or geographic regions, such as Spain [13], Sudan [14], and Ireland [15].Herein, we first report a genome wide survey on six Chinese indigenous cattle breeds in Yunnan using BovineHD Genotyping BeadChip.The aims of this study were to evaluate the level of genomic diversity and quantify the extent of admixture in each breed.

Ethics statement
Ear skin tissues were randomly collected by veterinary prac titioners according to relevant international as well as national guidelines and under permission from the Guide for the Care and Use of Laboratory Animals of Yunnan University.All ex perimental procedures used in this study were approved by the Institutional Animal Care and Use Committee (IACUC) of the Yunnan University (ynucae 20140011).1).The 120 samples contained 50 males and 70 females, sampled from remote villages, where artificial insemination was not prac ticed.Care was taken to avoid sampling related individuals, based on information provided by local farmers.A tiny piece of ear skin tissue for each individual was cut by trained vet erinarians with the permission and in presence of the owners and deposited in tube with pure ethanol.Genomic DNA was extracted from ear skin tissues using the DNeasy Blood & Tis sue kit (QiaGen, Hilden, Germany).The DNA concentration and quality were checked with NanoDrop 2000 spectropho tometer (Thermo Scientific, Wilmington, DE, USA).

Data analysis
To estimate the level of genomic diversity in each breed, ob served heterozygosity (Ho), expected heterozygosity (He), and inbreeding coefficient (f) were calculated using PLINK v1.07 [19].Heterozygosity estimates are sensitive to various ascer tainment biases when SNPs discovered in one breed are used to genotype other breeds [18,20].Analyses of diversity indexes and genetic relationships based on ascertained SNP data would produce false positive results [21].Previous studies dem onstrated that pruning of SNPs in high LD would reduce the influence of ascertainment biases [22,23].Therefore, we used SNP datasets after quality control and LDpruning (by re moving closely associated SNPs) to evaluate the influence of ascertainment biases on diversity indexes.Furthermore, to discern genetic relationships among breeds, principal component analysis (PCA) was assessed using SMART PCA program implemented in EIGENSOFT v6.1.4[24].The software ADMIXTURE v1.30 [25] was used to reveal ad mixture patterns among breeds.The corresponding cross validation error value of cluster (K = 1 to 6) was calculated in ADMIXTURE v1.30.The optimal value of K would ex hibit a low crossvalidation error compared with those of other K values.The graphical representation of PCA and admixture patterns were depicted using ggplot2 function in R software package [26].

Level of polymorphisms within breeds
After quality control, 130,663 SNPs were excluded, in which 11,331 showed SNP call rate of less than 0.90, 109,315 had MAF less than 0.05 and 10,017 significantly deviated from HardyWeinberg equilibrium (p<0.001),we finally obtained 604,630 SNPs for subsequent analyses (Table 1).The exclud ed highest number of SNPs (251,637) that showed a MAF of less than 0.05 was found in Dehong humped cattle, while the excluded lowest number of SNPs (82,819) was observed in Zhaotong cattle.Additionally, all samples passed the pruned SNPs for LD, and obtained a dataset of 233,788 SNPs for ge nomic diversity, PCA and ADMIXTURE analyses.

Breed genomic diversity and ascertainment bias
Here we used two different datasets to assess the effect of as certainment biases on diversity indexes of each breed.Using a dataset of 604,630 SNPs, the average Ho ranged from 0.257 in Dehong humped cattle (Western Yunnan) to 0.366 in Zhao tong cattle (Northeastern Yunnan).Similarly, using a dataset of 233,788 SNPs, the lowest Ho (0.295) was found in Dehong humped cattle, while the highest Ho (0.360) was detected in Zhaotong cattle.Interestingly, there existed a declining trend of genomic diversity for these native breeds from Northeastern Yunnan to Western Yunnan (Table 2, Figure 2).In addition, when using LDbased pruning dataset (233,788 SNPs), we    2).

Principal component and Admixture analysis
The genetic relationships among six Yunnan native cattle breeds revealed by PCA are shown in Figure 3.The first (PC1) and second (PC2) principal components accounted for 6.9% and 2.5% of the total variation, respectively.The PC1 clearly sep arated Dehong humped cattle from other five breeds.The PC2 did not group the other five breeds into clearcut clusters, im plying that genetic admixture probably existed between other five breeds from different geographic regions in Yunnan.
To further examine the extent of admixture in each breed, we executed admixture analysis among breeds and obtained patterns of admixture for each breed (Figure 4).The lowest crossvalidation error (0.594) was detected at K = 2 (Supple mentary Figure S1).When K = 2, all six breeds were clustered into taurine and zebu lineages.When looking into details, we found that Dehong humped cattle showed 100% zebu ances try, while other five breeds showed 49% taurine ancestry and 51% zebu ancestry on average.The overall admixture patterns were in good agreement with those revealed by PCA.Notably, we identified three individuals in Diqing cattle and one indi vidual in Zhaotong cattle appeared to be 100% taurine ancestry.When K = 3, all 120 individuals were grouped into three clus ters, each for pure taurine, admixed and pure zebu cattle, with 100% zebu ancestry for Dehong humped cattle remained.In addition, when K = 4, it showed that Dehong humped cattle (i.e., zebu) component appeared in other five Yunnan native cattle breeds with different percentages.As expected, Dehong humped cattle remained to be pure zebu ancestry (around 100%), being consistent with those found in K = 2 and K = 3 (Figure 4).

DISCUSSION
In this study, six Chinese indigenous cattle breeds from Yunnan show a high level of genomic diversity.The heterozygosities (Ho = 0.327,He = 0.352 for 604,630 SNPs) in Yunnan six native cattle breeds (Table 2) was slightly higher than those reported in Spanish beef cattle breeds (He = 0.30) [13] and lower than those in Irish cattle (Ho = 0.379) by using bovine highdensity SNP chips [15].Our results revealed that the Ho ranged from the lowest in Dehong humped cattle (Western Yunnan) to the highest in Zhaotong cattle (Northeastern Yunnan), showing a declining trend of diversity indexes for six native cattle breeds from Northeastern Yunnan to Western Yunnan (Figure 2), being consistent with that trend revealed by mitochondrial Dloop sequences [11].Interesting, we found the Ho values of Dengchuan (0.356) and Dehong humped cattle (0.257) were higher than those from mediumdensity bovine SNP chip assay (0.243 in Dengchuan cattle and 0.161 in Dehong humped cattle) [27].This difference may be mainly attributed to dif ferent numbers of SNPs genotyped.
Comparing to other five cattle breeds (Table 2), Dehong humped cattle showed the lower level of polymorphism, MAF and heterozygosities.This might be directly related to ascer tainment biases, caused by relatively less SNPs from zebutype breeds designed in BovineHD SNP chip.This observation and interpretation were in agreement with previous reports [18,28], despite that LDpruning could minimize the effect of ascer tainment biases [22,23].As expected, pruned SNP dataset slightly increased the level of diversity in Dehong humped cattle than previously unpruned dataset.
The results of the PCA and admixture analyses indicate that other five Yunnan native breeds except for Dehong humped cattle are weakly differentiated.This observation was consis tent with previous studies on phylogeographic structure of Yunnan native cattle breeds [11].The PCA and admixture analyses could separate the breeds into clusters.In this study, the PC1 and admixture analyses clearly separated Dehong humped cattle from other five breeds.This is explained by the fact that Dehong humped cattle is a pure zebu breed (100% zebu ancestry) as revealed by Li et al [11] and that this breed distributes in Dehong Prefecture (a region close to Myanmar) of Yunnan Province, belonging to tropical and subtropical climatic conditions.These facts indicated that Dehong humped cattle are descendants of zebutype ancestral cattle without influence of taurine cattle, because taurine cattle poorly adapt in such tropical and subtropical climatic conditions.In addi tion, when K = 2, the admixture analysis separates all six breeds into taurine and zebu lineages, corresponding to two independent ancestries of modern domestic cattle [1,29,30].When K = 3, all samples were grouped into three clusters, each for pure taurine, admixed and pure zebu cattle, which pro vided clear evidence for hybridization between taurine and zebu cattle [31].

CONCLUSION
In this study, we found considerable genomic diversity in six Yunnan native cattle breeds except for Dehong humped cattle from Western Yunnan.Our genomewide analysis further confirmed that Dehong humped cattle is a pure zebu breed with 100% zebu ancestry and no influence from taurine cattle.We also found that five breeds (Diqing, Zhaotong, Dengchuan, Dianzhong, and Wenshan cattle) had admixed origins with different degrees of admixture between taurine and zebu cattle.Such admixture by crossbreeding between zebu and taurine cattle facilitated the spread of zebu cattle from tropical and subtropical regions to other highland regions in Yunnan.Future studies are needed to detect chromosomal segments or genomic regions that have adaptive significance in admixed individuals.

Table 1 .
Breed, sampling locality, number of samples (n) and SNPs excluded and remaining after quality control based on 735,293 SNPs SNP, single nucleotide polymorphism; CR, call rate; MAF, minor allele frequency; HWE, Hardy-Weinberg equilibrium (p < 0.001).

Table 2 .
The genetic diversity indexes (Ho and He) and inbreeding coefficient in the studied indigenous cattle populations