The procedures employed here were approved by the Ethics Committee for the usage of Animals in experiments of the Federal University of Paraiba (CEUA-UFPB) nº 135/2015.

Hair samples from 113 goats were collected, co­rresponding to four Brazilian local and two exotic breeds. Data were obtained from Canindé, n = 20, raised in Rio Grande do Norte state (CAN-RN); Moxotó, n = 19, raised in the Ceará state (MOX-CE); Marota, n = 24, raised in Piaui state (MAR-PI); Azul breed, n = 20, raised in Pernambuco state (AZU-PE) (Table 1). Two cosmopolitan exotic breeds were used in this study: Saanen, n = 20, raised in Pernambuco State (SAA-PE), and British Alpine, n = 10, raised in Bahia state (ALP-BA). Care was taken to distinguish pure native breeds from crossed or commercial breeds ( e.g ., cosmopolitan breeds) and non-descript breeds through accessing in­digenous knowledge shared by breeders.

The DNA of the 113 individuals was isolated from hair samples via alkaline extraction, according to Coelho et al . (2004) , and quantified using a spectrophotometer (NanoDrop®, Thermo Scientific).

The PCRs were performed according to Pereira et al . (2004) , with some modifications, using the forward pri­mer 5'-CGCTCGCCTACACACAAATA-3' and the rever­se primer 5'-GAAGAGTGGGCGATTTTAGG-3'. The PCR program consisted of an initial denaturation at 94°C for 3 min, followed by 38 cycles of 94°C for 30 s, 60°C for 45 s and 72°C for 1 min, with a final extension step at 72°C for 10 min. The PCR products were purified using the QIAquick PCR Purification Kit (Qiagen Inc., Valen­cia, CA, USA) according to the manufacturer's re­commendations. The amplified frag­ments were sequenced using the same primers used for the PCRs and an automatic ABI Prism 3730XL DNA analyzer (Applied Biosystems), according to the manufacturer's recommendations.

The sequences were edited, aligned, and compared using the Clustal W algorithm and MEGA® v. 7 soft­ware ( Tamura et al ., 2013 ). For the Brazilian local and exotic goat analyses, a fragment between the nu­cleotide positions 15,707 and 16,187 was taken into consideration, with 481 bp of the control region referring to 113 sequences. The concatenated sequences were aligned against the reference sequence AF533441 ( Parma et al ., 2003 ) and further trimmed using Clustal W implemented in the MEGA7 software.

To provide more complete information on the origin of Brazilian goats, new mtDNA sequences of both wild and domestic goats were included: haplotypes A (accessions AJ317736, AJ317661 and AJ317778, Luikart et al ., 2001 ); B1 (AJ317826, Luikart et al ., 2001 ; and EF618355 and EF617850.1, Naderi et al ., 2007 ); B2 (AJ317833, Luikart et al ., 2001 ); C (AJ317835, Luikart et al ., 2001 ; and AB110559, Sultana et al ., 2003 ); D (AB110587, Sultana et al ., 2003 ; and EF617701, Naderi et al ., 2007 ), F (DQ241349, Sardina et al ., 2006 ) and G (EF617728, Naderi et al , 2007 ), plus those of the Crespa breed (CRE) submitted to Genbank by Lopes et al . (2016) , (Crespa, n = 10, KM260525.1-KM260536.1; Anglo-Nubian, n = 6, KM260504.1-KM260509.1; and Boer, n = 6, KM260519.1- KM260524.1).

The comparison was performed according to Pereira et al . (2004) , using Capra pyrenaica as an outgroup (FJ207528; Hassanin et al ., 2009 ). As many published goat mtDNA sequences are substantially shorter or longer than our 481 bp sequence, fragments of 464 bp instead of 481 bp were considered to allow a comparison of sequences of the same size. These sequences were included to facilitate the recognition of haplogroups and to determine phylogenetic relationships. All fragments from the three datasets were aligned and trimmed using Clustal W in the MEGA7 software.

Polymorphic sites, number of haplotypes, haplotype diversity (h), and nucleotide diversity (p) were cal­culated using DnaSP 5.0 ( Librado & Rozas, 2009 ). The ha­plotypes found in each breed were deposited in GenBank under the accession numbers KM500901- KM500921. The molecular variance was analyzed (AMOVA), along with diversity measures, F _ST (es­timates of genetic diffe­rentiation among populations) distances and Fu’s Fs (test for an excess of rare alleles) values, using the Arlequin v. 3.5 software ( Excoffier & Lischer, 2010 ). We constructed a median-joining network for all sequences using the Network v. 4.6.1 software ( http://www.fluxus-engineering.com/sharenet.htm ). The de­mographic expansion was estimated using Fu’s F _S neutrality test.

The MEGA v.7 software ( Tamura et al ., 2013 ) was also used to generate a Maximum Likelihood tree to identify relationships between the sampled populations, using the Kimura 2-parameter model with 1,000 boot­strapping replicates. All positions containing gaps and missing data were deleted.

Alignment of 481 bp of the hypervariable region, with 83 control region sequences (15,707-16,188) from local Brazilian goats, showed 33 polymorphic sites. The fragment showed a high-frequency polymorphism in the Brazilian goats, with a haplotype diversity (h) of 0.8240, revealing 16 different haplotypes. Of these, 88% were different from each other. The most frequent sequence appeared 27 times, while 11 sequences occurred only once, signifying that they were unique to a breed (Canindé, n = 6; Marota, n = 3; Azul, n = 1; and Moxotó, n = 1) (Fig. 1). The most frequent haplotype (H2) was shared among all local Brazilian breeds, while the second most sha­red haplotype (H5) was also common to all local breeds studied here. The haplotypic diversity index (Hd) for each population was high (above 0.6), and previous studies have verified that the index ranges from 0.6842 (Moxotó) to 0.8842 (Canindé) (Table 1). Nucleotide diversity ranged from 0.0164 (Marota) to 0.0188 (Canindé); the lowest nu­cleotide diversity was observed in the Marota goat breed.

The neutrality test was not statistically significant for the populations analyzed (Table 1). Table 1 shows the number of polymorphic sites and Hd, π, Tajima´s D and Fu’s F _S values for all breeding populations. We observed no tendency toward breed structuring.

The high degree of diversity found in Brazilian caprine breeds, with some haplotypes sharing lineage with foreign breeds, is consistent with our knowledge about the initial formation of the national herd. Spe­cifically, it seems to be the result of European/African animals that were intro­du­ced into the country. This argument is consistent with historical accounts of the introduction of domestic goats into Brazil by Portuguese settlers in 1515 ( Simonsen, 1937 ).

In addition to the colonial history of Brazil, Saa­nen, British Alpine, Anglo-Nubian, Boer, and other breeds were introduced over the decades to increase production. Many breeders used them in crosses to improve the production of local breeds, and this inclusion of exotic, non-targeted material culminates in genetic dilution.

The analysis revealed 61 polymorphic sites with 40 different haplotypes, with 75% of all haplotypes different from one another. The most common se­quence appeared 31 times, reaching a frequency of 22.79%, while 30 sequences occurred only once. Of these 40 identified haplotypes, 11 were unique to northeastern Brazilian breeds, including Canindé (H26, H27, H28, H29, H30, and H31), Marota (H32, H33, and H35), Azul (H13) and Moxotó (H36) (Figs. 1 and 2).

The most common haplotype (H12) was shared among 31 individuals of the Canindé, Azul, Moxotó, Marota, and Saanen breeds. The second most shared haplotype (H2) was absent only in the ANG breed (Fig. 2). There were no haplotypes common to all local and exotic breeds studied here.

The values of F _ST and numbers of haplotypes shared among all breed pairs were used to establish the genetic distances among all studied groups (Table 2). The most notable example of genetic differentiation was observed between ALP-BA and MOX-CE breeds (F _ST = 0.4022), while the most compelling example of significant weak differentiation was observed between MAR-PI (Population 4) and AZU-PE (Population 2) breeds.

A phylogenetic tree was constructed for the a­vailable mtDNA control region sequences to analyze the rela­tionship between the goat breeds raised in Brazil (local and exotics) and those from other world regions. Additionally, it was used to compare the local goat di­versity, allowing researchers to compare local Brazilian breeds with other domestic and wild goats.

We identified some haplotypes that are present in the Iberian goat breeds ( Parma et al ., 2003 ). The local Brazilian goat breeds presented specific haplotypes in Similar results have also been found in previous studies ( Pereira et al ., 2005 ; Amills et al ., 2009 ).

All Brazilian goat breeds were included in maternal lineage A and separated from the remaining groups, which clustered together (Fig. 3). The haplogroup A is the most frequent group and has been identified in all goat breeds reported so far ( Luikart et al ., 2001 ; Wang et al ., 2015 ; Deng et al ., 2018 ). The diversity among members of haplogroup A results from the beginning of domestication in the fertile crescent region and its global expansion ( Luikart et al ., 2001 ). The ma­jority of the goat breeds in the world belong to this haplogroup, which agrees with the theory that the place of domestication of goats is the fertile crescent region ( Othman & Mahfouz, 2016 ). After domestication, the goats that already carried differentiated gene pools spread to Europe, Africa and Asia ( Colli et al ., 2018 ).

The highest bootstrap value for external and internal branches was 98%, sequences referring to C. pyrenaica as well, and haplogroups (Fig. 3). We did not detect the mitochondrial lineages B, C, D, F, and G since they are rarely found in most of the European continent ( Naderi et al ., 2007 ; Amills et al ., 2009 ). More than 90% of the goats studied with the D-loop region worldwide are from haplogroup A. On the other hand, the B lineage has been found only in Asia and South Africa, while the C lineage is found in southern Europe, the D lineage in Asia, the F lineage in Italy and the G lineage in southwest Asia and North Africa ( Groeneveld et al ., 2010 ; Pakpahan et al ., 2016 ).

Recent studies using other molecular markers have found divergences regarding the ancestry of Brazilian goats. Colli et al . (2018) , using SNP markers, found that Canindé (CAN-RN) and Moxotó (MOX-PE) breeds strongly exhibit ancestry with West African animals, while Ginja et al . (2017) , with microsatellite markers, showed a European origin of these breeds. The present study revealed European goats as the dominant stock for Brazilian goats, but there are some haplotypes within haplogroup A, clustering with African and Asian goat breeds.

The large variety of local goats represent an im­portant genetic reservoir. Even in areas far from do­mestication centers, there is a significant genetic divers­ity that can benefit conservation programs. These results could be suitable for creating a strategic con­servation program, potentially benefitting future bree­ding programs.