Population structure and origin of Brazilian hair sheep breeds

Transcription

1 Population structure and origin of Brazilian hair sheep breeds Tiago do Prado Paim¹, ²*, Samuel Rezende Paiva³ & Concepta McManus 4 ¹ Faculdade de Agronomia e Medicina Veterinária, Universidade de Brasília, Campus Darcy Ribeiro, Asa Norte, , Brasília/DF, Brazil. pradopaim@hotmail.com (Corresponding Author) ² Instituto Federal de Educação, Ciência e Tecnologia Goiano, Avenida Oeste nº 350 Parque União, , Iporá/GO, Brazil. ³ Embrapa Recursos Genéticos e Biotecnologia, Asa Norte, , Brasília/DF, Brazil. 4 Instituto de Biologia, Universidade de Brasília, Campus Darcy Ribeiro, Asa Norte, , Brasília/DF, Brazil. Summary Brazilian hair sheep possess a substantial genetic diversity among sheep breeds worldwide. These locally adapted sheep for the harsh environment of the Brazilian Northwest (semi-arid) maintain traits that are important for this region, such as parasite resistance, heat tolerance and high pelt quality. Genotypes (50K SNP chip) from seven Brazilian sheep breeds (5 hair and 2 coarse wool types) and 87 worldwide breeds were used to verify population structure, admixture and genetic diversity, using PCA and ADMIXTURE analyses. We constructed a phylogenetic tree and evaluated migration events between genetic groups using TREEMIX software. Somalis, a fat-tailed breed, was the unique breed with high relationship with East African breeds and formed a distinct cluster from other Brazilian breeds. Barriga Negra had a clear relationship with Barbados Blackbelly, which appeared as another group. Other Brazilian breeds seem to form a further genetic group with some recent admixtures. Morada Nova remained as a separate group, not showing a strong relationship with European or African breeds. Rabo Largo and Morada Nova share a common ancestor. Somalis and Rabo Largo are the most endangered sheep genetic resources in Brazil. Bergamasca and Santa Inês shared an important genetic background. Santa Inês animals came from an admixed population and recently received introgression from Suffolk animals. The genetic conservation efforts for these breeds is essential due to the specific traits and genes that these animals may carry, which would be useful to face further environmental constraints to commercial production. Keywords: genetic resources, migration, conservation, Santa Inês, Ovine Introduction There are no specific registers of sheep being taken to Brazil from Europe (McManus et al., 2010). It is probable that colonizers (mainly Portuguese and Spanish) brought animals with lower quality wool to the Latin America (New World). Other animals may have come from Africa (mainly Angola and Nigeria) with the slave trade. Sheep in Latin America underwent natural selection for almost five centuries, acquiring specific adaptive traits (such as parasite resistance, rusticity and prolificacy) (Mariante & Egito, 2002). Brazilian hair breeds (Santa Inês, Morada Nova, Somalis, 1

2 Barriga Negra and Rabo Largo) initially were reared in the Northeast region of Brazil, in a hot and dry climate, and some are considered parasite resistant. These local genetic resources have been jeopardized by the introduction of exotic breeds, often considered more productive and profitable. Therefore, the characterization of the genetic structure of these breeds is imperative for the preservation of these valuables genetic resources and developing viable conservation strategies. Our study attempts to establish the main origin of Brazilian breeds, evaluating their European and/or African ancestry. We analyzed a data set composed of animal genotypes of Brazilian hair sheep breeds and worldwide breeds using the 50K SNP chip. We aimed to verify the population structure, admixture and genetic diversity of Brazilian hair breeds and their relationships with breeds worldwide. Materials and methods Our initial data had genotypes of 4,014 animals in total, of which 1,149 animals were sampled in Brazil, 46 animals from Algeria presented by Gaouar et al. (2017) and 2819 animals from the Sheep HapMap project - ISGC (International Sheep Genomics Consortium) presented by Kijas et al. (2012). All genotypes were obtained with the OvineSNP50 BeadChip (Illumina, San Diego, CA). Initially, data was filtered using the following parameters in order: call rate per marker (<95%), minor allele frequency (MAF) <0.01 and sample call rate < 90%. Quality controls were performed using Golden Helix SNP & Variation Suite software v (Golden Helix Inc., Bozeman, Montana, USA). We limited the number of animals per breed to a minimum of 5 and maximum of 50. Sample location diversity was maintained with the Brazilian breeds. The final data set had 2,549 animals (with 47,480 SNPs) from 94 breeds. Linkage disequilibrium (LD) pruning was used with the following parameters: window size of 50 SNPs with window increment equal to 5 SNPs. The LD statistic and threshold used was r² > 0.5, using Composite Haplotype Method (CHM). A Principal Component Analyses (PCA) were conducted to visualize the breeds with higher relationship with Brazilian breeds. The parameters were set to find the first 10 components, normalizing each marker data by their actual standard deviation, utilizing an additive model. After two PCA, 66 breeds were removed from analysis, with 28 breeds with 490 animals remaining (Figure 1). Genetic relationships among breeds were evaluated through a model-based clustering algorithm implemented in the software ADMIXTURE v (Alexander et al., 2009). The cross-validation procedure (10-fold) estimate prediction errors for each K value (2 to 30). Coefficients of membership to each K cluster were visualized using the on-line CLUMPAK server with DISTRUCT for many K s (Kopelman et al., 2015). The tree-based approach was used to reconstruct historical relationships between the analyzed populations and to test for the presence of gene flow using TREEMIX software (Pickrell & Pritchard, 2012). The TREEMIX program was run on the dataset with animals classified in 29 breeds or populations. The previous 28 breeds used were joined with the Soay breed, used as a root. A variable number of migration events (M) (0, 1, 2, 3, 4, 5, 10, 20, 30, 40 and 50) were tested. Results and Discussion The final PCA with 28 breeds (Figure 1) demonstrated a relationship between 2

3 Somalis (BSO) and some African breeds. The Morada Nova breed (BMN) was seen isolated in both components. Santa Inês (BSI), Rabo Largo (BRL), Pantaneira, Bergamasca, Barriga Negra (BBN), Suffolk, Comisana, Barbados Blackbelly and St. Elizabeth were seen as relatively close. The results of ADMIXTURE showed, at K=3, African, European and Brazilian components in genetic structure of the populations (Figure 2). Kijas et al. (2012) also found that sheep from the Americas (Brazil and Caribbean) cluster separately from European, African, or Asian populations. At K=16 (the lowest prediction error), BSO, BRL, BBN and BMN differentiated from other groups and between themselves. The inferred tree with 29 breeds in TREEMIX software (Figure 3) showed BSO in an African branch, next to Red Maasai and Ethiopian Menz. BBN remained close to the Barbados BlackBelly as observed in other analyses. Other Brazilian breeds formed a separate branch and Algerian breeds grouped together on another branch. With four migration events, a contribution of Afrikaner breeds ancestor to BRL constitution is highlighted. Also, some migration events between Brazilian breeds can be seen, such as BSO to BMN and BRL ancestor, and from this ancestor to BBN. BSO and BRL are fat-tail hair breeds, and seem to have different origins. Brazilian Somalis was close to African breeds in all analyses, while Rabo Largo remained close to other Brazilian breeds. Both breeds represent unique genetic architecture in cluster analysis, forming an exclusive cluster from K=4 for BSO and K=14 for BRL. These results may be related to the small number of animals remaining in each breed (McManus et al., 2014) and, consequently, the strong genetic drift suffered in both cases. Rabo Largo shares a common ancestor with Morada Nova (Figure 3), which is expected as both are reared in similar regions (McManus et al., 2014). BRL received a migration from Afrikaner breeds, justifying the fat-tail phenotype. BRL animals are reared in the hottest climate with the lowest precipitation index and had the lowest number of flocks and the lowest distance from mid-point of breed occurrence (McManus et al., 2014). These previous findings with the unique genetic structure observed here highlights the need for genetic conservation efforts for this breed. In the cluster analysis (Figure 2), BMN breed appear in a specific cluster. Spangler et al. (2017) showed a relationship between Morada Nova and a Nigerian breed (Djallonke), disclosing the possible west African origin of Morada Nova. More studies with other African breeds should be conducted to uncover the Morada Nova origin. Santa Inês is the main commercial hair breed in Brazil and its origin is still unknown (McManus et al., 2014). Here, BSI showed some admixture with Bergamasca, Pantaneira, Rabo Largo, Comisana and Suffolk. The Bergamasca (Italian breed) is clearly the main composition of the BSI animals. There is no specific records about the importation of this breed to South America, so little is known about their arrival and rearing in Brazil. Bergamasca is known to have been used for crossbreeding in many European countries because of its large frame (Ciani et al., 2014). Probably, Bergamasca was used for crossbreeding with locally adapted sheep in Northwest, aiming to increase the size of the animals and them backcrossing to hair breeds was carried out avoiding animals bearing wool. The backcrossing of Bergamasca animals in Brazil is also a possibility, which would be produced animals better adapted to the hot climate. Carneiro et al. (2010) and Paiva et al. (2005) stated a possible recent crossbreeding of Santa Inês with meat-selected breeds, mainly Suffolk. This hypothesis can be confirmed here as the BSI genetic structure has a contribution from Suffolk. 3

4 Conclusions Somalis breed represents an East African group within the Brazilian hair breeds, related to Red Maasai, Ethiopian Menz and Afrikaner breeds. Barriga Negra had a clear relationship with Barbados BlackBelly. Morada Nova has a unique genetic structure, which was shared mainly with Santa Inês and Rabo Largo. Bergamasca and Santa Inês shared an important genetic background. Santa Inês animals came from an admixed population and recent introgression with Suffolk animals. Genetic conservation efforts for these breeds are necessary due the specific traits and genes that these animals may carry, which would be useful to face further environmental constraints to commercial production. Acknowledgments To CAPES for providing the scholarship for the first author. To CNPq for providing financial support. To IF Goiano for funding the first author. To The International Sheep Genomics Consortium for providing some genotypes used in the analyses. List of References Alexander, D.H., J. Novembre & K. Lange, Fast model-based estimation of ancestry in unrelated individuals. Genome Res. 19(9): Carneiro, H., H. Louvandini, S.R. Paiva, F. Macedo, B. Mernies & C. McManus, Morphological characterization of sheep breeds in Brazil, Uruguay and Colombia. Small Rumin. Res. 94(1 3): Ciani, E., P. Crepaldi, L. Nicoloso, E. Lasagna, F.M. Sarti, B. Moioli, F. Napolitano, A. Carta, G. Usai & M. D'Andrea, Genome-wide analysis of Italian sheep diversity reveals a strong geographic pattern and cryptic relationships between breeds. Animal Genetics 45(2): Gaouar, S.B.S., M. Lafri, A. Djaout, R. El-Bouyahiaoui, A. Bouri, A. Bouchatal, A. Maftah, E. Ciani & A.B. Da Silva, Genome-wide analysis highlights genetic dilution in Algerian sheep. Heredity 118(3): Kijas, J.W., J.A. Lenstra, B. Hayes, S. Boitard, L.R. Porto Neto, M. San Cristobal, B. Servin, R. McCulloch, V. Whan & K. Gietzen, Genome-wide analysis of the world's sheep breeds reveals high levels of historic mixture and strong recent selection. PLoS Biol. 10(2): e Kopelman, N.M., J. Mayzel, M. Jakobsson, N.A. Rosenberg & I. Mayrose, Clumpak: a program for identifying clustering modes and packaging population structure inferences across K. Mol. Ecol. Resour. 15(5): Mariante, A.S. & A.A. Egito, Animal genetic resources in Brazil: result of five centuries of natural selection. Theriogenology 57(1): McManus, C., P. Hermuche, S.R. Paiva, J.C. Ferrugem Moraes, C.B. de Melo & C. Mendes, Geographical distribution of sheep breeds in Brazil and their relationship with climatic and environmental factors as risk classification for conservation. Braz. J. Sci. and Technol. 1(1): 3. McManus, C., S.R. Paiva & R.O. Araújo, Genetics and breeding of sheep in Brazil. Rev. Bras. Zootec. 39: Paiva, S.R., V.C. Silvério, A.A. Egito, C. McManus, D.A. Faria, A.S. Mariante, S.R. Castro, M.S.M. Albuquerque & J.A. Dergam, Genetic variability of the 4

5 Brazilian hair sheep breeds. Pesq. Agropec. Bras. 40: Pickrell, J.K. & J.K. Pritchard, Inference of population splits and mixtures from genome-wide allele frequency data. PLoS Genetics 8(11): e Spangler, G.L., B.D. Rosen, M.B. Ilori, O. Hanotte, E.S. Kim, T.S. Sonstegard, J.M. Burke, J.L.M. Morgan, D.R. Notter & C.P. Van Tassell, Whole genome structural analysis of Caribbean hair sheep reveals quantitative link to West African ancestry. PLoS ONE 12(6): e Figure 1. Principal component analysis (PCA) using genomic data of 28 breeds genetically close to Brazilian hair sheep breeds. The number between parentheses in each axes means the amount of the variance explained by each eigenvector (EV). 5

6 Figure 2. Clustering performed with ADMIXTURE software on genotypic data from 28 sheep breeds (K, number of clusters, from 2 to 16). Figure 3. Inferred trees from 29 sheep breeds using TREEMIX software simulating 4 migration events and using Soay breed as root. 6