Ó The American Genetic Association. 2009. All rights reserved. For permissions, please email: [email protected].

Journal of Heredity 2010:101(2):201–210 doi:10.1093/jhered/esp104 Advance Access publication December 4, 2009

Analysis of STR Markers Reveals High Genetic Structure in Portuguese Native Cattle C ATARINA GINJA, LUI´S TELO DA GAMA,

AND

MARIA CECILIA T. PENEDO

From the Instituto Superior de Agronomia, Tapada da Ajuda, 1349-017 Lisboa, Portugal (Ginja); the Instituto Nacional dos Recursos Biolo´gicos, Fonte Boa, 2005-048 Vale de Santare´m, Portugal (Ginja and Telo da Gama); and the Veterinary Genetics Laboratory, University of California, One Shields Avenue, Davis, CA 95616 (Ginja and Penedo). Address correspondence to C. Ginja, Veterinary Genetics Laboratory, University of California, One Shields Avenue, Davis, CA 95616, or e-mail: [email protected].

Abstract Genetic structure and diversity of 13 Portuguese native and 3 imported cattle breeds were assessed with 39 microsatellites. Allelic richness per locus was high, with an overall average of 8.3 ± 2.5. The mean observed and expected heterozygosities were 0.673 ± 0.043 and 0.691 ± 0.034, respectively. The mean number of alleles per breed ranged between 5.36 ± 1.27 and 7.87 ± 2.66. Brava de Lide and Mirandesa breeds had the lowest genetic diversity, whereas Minhota, Arouquesa, and Mertolenga had the highest. Significant (P , 0.05) heterozygote deficit was detected in all breeds except Garvonesa, Marinhoa, Minhota, and Limousin. Hardy–Weinberg deviations are most probably due to inbreeding, particularly in Alentejana, Brava de Lide, Mertolenga, and Ramo Grande (Fis . 0, P , 0.0001). Based on the principal component and the Neighbor-Net analyses, Mirandesa was the most genetically distinct breed. Even though admixture was detected across all breeds (6.7%, q , 0.800), the molecular structure was consistent with original breed designations, with the exception of Cachena that had a clear influence of Barrosa˜ (K 5 15). Mertolenga showed substructure with independent clustering of red speckled animals. The percentage animals correctly assigned was 90 in all breeds except Cachena, Garvonesa, and Preta (q  0.800). The results obtained here confirmed that high levels of genetic diversity exist within Portuguese native cattle and that the breeds are highly structured. Conservation measures should be implemented for all native breeds to minimize inbreeding. Key words: native cattle, microsatellites, genetic structure, admixture, breed assignment

In Portugal, there are 13 breeds of native cattle that are raised in specific geographic regions and that have distinct characteristics described in the breed standards adopted in the herdbooks that maintain animal records for each breed. The introduction of imported breeds such as Friesian, Charolais, and Limousin has threatened native cattle populations in the mid 20th century (Ralo and Guerreiro 1981). For most native breeds, the number of breeding animals has declined, and these genetic resources are at some level of risk of extinction (Gama et al. 2004). Even though most of the Portuguese native cattle have not been under intensive selection, recent bottlenecks have been detected through pedigree analysis. For example, the Alentejana breed shows signs of genetic erosion, and based on demographic parameters, one male accounts for 60% of the Y chromosomes currently represented (Carolino and Gama 2008). The marketing of certified meat products for the last decade has helped counteract the decline tendency

with an increase in meat production from Portuguese native breeds of about 40% (Gama et al. 2004). Nonetheless, conservation measures are still needed to assure that these breeds are maintained. The genetic characterization of local genetic resources is not only important for the implementation of breed-specific conservation programs but also to generate information for comparisons of genetic diversity at wider geographic scales (Bruford et al. 2003). The origins of Portuguese cattle have been investigated through the use of mitochondrial DNA (mtDNA) (Cymbron et al. 1999) and Y chromosome sequence variation (Ginja et al. 2009). Influences from Central European and African cattle were detected, which is consistent with historical records (Zilha˜o 2001; Anderung et al. 2005; Beja-Pereira et al. 2006). Microsatellite (STR) markers are very informative and useful to characterize the genetic diversity and structure of domestic breeds (Bruford 2004). The genetic relationships among Iberian cattle and

201

Journal of Heredity 2010:101(2)

European breeds were investigated with STRs (Can˜on et al. 2001; Beja-Pereira et al. 2003; Cymbron et al. 2005), but no more than five Portuguese native breeds were analyzed. Only two studies (Mateus et al. 2004a; Mateus et al. 2004b) focused on the characterization of Portuguese cattle with STRs, but these did not include all native breeds. Information is lacking on the genetic structure of Portuguese native cattle and the degree of admixture between breeds. Fully Bayesian Monte Carlo Markov chain (MCMC) clustering methods are useful to estimate individual genotype probabilities, with or without prior information on source populations, and to detect admixture (Pritchard et al. 2000; Beaumont and Rannala 2004; Chikhi and Bruford 2005; Manel et al. 2005). These methods have been used to assess the genetic structure and assign individuals to breeds of several domestic species, such as cattle (Moioli et al. 2004; Negrini et al. 2007), sheep (Lawson Handley et al. 2007), goats (Can˜on et al. 2006), pigs (Vicente et al. 2008), chicken (Rosenberg et al. 2001), cats (Beaumont et al. 2001; Randi et al. 2001; Driscoll et al. 2007; Lipinski et al. 2008), and dogs (Koskinen 2003; Leroy et al. 2009; Pires et al. 2009). The performance of assignment methods depends on several factors, such as number and polymorphism of the markers, number of populations and their degree of differentiation, and number of individuals sampled; hence, it is necessary to assess their efficiency in specific contexts (Cornuet et al. 1999; Maudet 2001; Manel et al. 2002). The aim of this study is to contribute to the molecular characterization of Portuguese native cattle and to investigate their genetic structure with STRs. Fully Bayesian MCMC methods were used to investigate whether molecular structure mirrors breed designations and to detect within-breed substructure. The degree of admixture between breeds was assessed, as well as possible crossbreeding of native cattle with imported commercial breeds. A principal component analysis (PCA) from pairwise Fst values and a Neighbor-Net graph of breed clusters obtained from genetic distances were used to illustrate breed relationships.

Material and Methods Sampling and DNA Extraction Blood samples were collected from a total of 675 animals of 13 native Portuguese cattle breeds: Alentejana (38), Arouquesa (70), Barrosa˜ (69), Brava de Lide (43), Cachena (51), Garvonesa (39), Marinhoa (46), Maronesa (47), Mertolenga (64), Minhota (50), Mirandesa (54), Preta (60), and Ramo Grande (44) from the Azores Archipelago. Samples were also collected from animals of the three major exotic breeds raised in Portugal: Charolais (58), Friesian (51), and Limousin (47). To minimize the degree of relationship among individuals, pedigree records were used for selection of nonrelated animals back to the second generation whenever possible. DNA was extracted using the Gentra kit (PuregeneÒ, Gentra Systems, Inc.) according to the manufacturer recommendations.

202

STR Genotyping We analyzed a total of 40 STR loci (BM0203; BM0888; BM1314; BM1818; BM1824; BM8125; BM2113; BM4208; BM5004; BRRIBO; CSSM036; CSSM042; CSRM060; CSSM066; CYP21; ETH003; ETH010; ETH152; ETH185; ETH225; HAUT027; HEL001; HEL009; ILSTS006; ILSTS011; INRA023; INRA032; INRA035; INRA037; INRA063; MGTG4B; MM12E6; RM006; RM067; SPS115; TGLA053; TGLA094; TGLA122; TGLA126; TGLA227) distributed across 21 cattle chromosomes (BTA). Primer sequences, chromosome location, repeat motif, reference, and dye label are shown in supplementary Table S1 (see Supplementary Material online). STR markers were amplified in multiplex polymerase chain reactions (PCRs) using fluorescence-labeled primers and the QiagenÒ multiplex PCR kit (Qiagen Inc., USA). PCRs were prepared in a total reaction volume of 12 ll containing about 60 ng of genomic DNA, primers at concentrations that varied between 0.05– 0.08 lM and 6 ll of the QiagenÒ PCR master mix. The PCR program included an activation step at 95 °C for 15 min; 5 cycles at 94 °C for 30 s, 60 °C for 90 s, and 72 °C for 60 s; 28 cycles at 94 °C for 30 s, 57 °C for 90 s, and 72 °C for 60 s followed by a final extension at 72 °C for 30 min. Negative and a positive DNA controls were included in all assays. PCR fragments were separated by capillary electrophoresis on ABI 3730 instruments (Applied Biosystems, Foster City, CA) according to the manufacturer recommendations. Allele sizes were determined with STRand software (Hughes 2000) using the internal size standard GeneScanä—500 LIZä (Applied Biosystems, Warrington, UK). Statistical Analysis Allele frequencies for all loci and breeds as well as the number of population-specific alleles (private alleles, PAs) found in each breed were determined with GenAlex v. 6 (Peakall and Smouse 2006). The frequency of null alleles (r) per locus within each breed was estimated with the expectation maximization algorithm of Dempster et al. (1977) in FreeNA software (available at http://www1.montpellier.inra.fr/URLB/; Chapuis and Estoup 2007). The presence of null alleles overestimates the number of homozygotes, decreases within population genetic diversity, and inflates estimates of population differentiation (Dakin and Avise 2004; Lawson Handley et al. 2007). The software Genetix v. 4.03 (Belkhir et al. 1996-2004) was used to estimate observed (Ho) and unbiased within-breed expected (He) heterozygosities, mean number of alleles per breed (MNA), and to calculate the Fis values per breed by Weir and Cockerham (1984) with P values obtained based on 1000 permutations. Fstat v. 2.9.3 (Goudet 2001) was used to estimate the F statistics per locus by Weir and Cockerham (1984), and P values were obtained based on 1000 randomizations. Allelic richness (Rt) over all breeds for each locus was also calculated with this software. Deviations from Hardy–Weinberg equilibrium (HWE) were assed with Genepop v. 3.4 software (Raymond and Rousset 2003).

Ginja et al.  Genetic Structure of Portuguese Native Cattle

Both global tests across populations and loci and tests per locus per population were done using the method of Guo and Thompson (1992), and the P values were obtained using a Markov chain of 10 000 dememorization steps, 500 batches, and 5000 iterations. Genotypic linkage disequilibrium (LD) was also calculated with this software and the same Markov chain settings. Genetic structure and the degree of admixture of Portuguese native breeds were investigated using the Bayesian clustering procedure of Structure (Pritchard et al. 2000). The most probable number of populations (K) given the observed genotypic data was estimated by performing 10 independent runs for each K (1 K  19) with burn-in length and MCMC iterations of 50 000 and 100 000, respectively. The parameter alpha (degree of admixture) was inferred from the data using the default settings and an admixture model with correlated allele frequencies (Falush et al. 2003). The method of Evanno et al. (2005) was used to identify the most probable K by determining the modal distribution of DK. The analysis with Structure was repeated for subsets of the data to assess within-breed substructure. The most informative loci were identified with Whichloci (Banks et al. 2003) using the allele frequency method after resampling 500 individuals from the original dataset in each population, and setting a minimum of 95% of correctly assigned individuals with a stringency of 2 (P value 5 0.01). Assignment tests were performed with Structure using prior information of source breeds and the settings described above for the MCMC. The proportion of each individual’s genotype in each cluster or breed (q) and the probability of ancestry in other breeds were estimated. The percentage of individuals correctly assigned to source breed was calculated for distinct threshold q values. A PCA to represent breed relationships based on pairwise Fst values was carried out in PCAGEN v. 1.2 (available at http://www2.unil.ch/popgen/softwares/pcagen.htm; Goudet 1999) and with P values for the axes obtained from 1000 randomizations of genotypes. Population pairwise DA distances were calculated in Populations (Langella 1999) and used to obtain a network of breed clusters following the Neighbor-Net method in SplitsTree4 v. 4.10 (Huson and Bryant 2006).

Results Molecular Markers Among the 40 markers, evidence of null alleles was detected only at INRA035, which had null alleles at high frequency (r . 0.2) across several breeds, and for this reason this marker was discarded. A total of 457 alleles were detected at 39 STR loci across all breeds. Measures of genetic variability as well as F statistics for each STR are shown in Table S1 (see Supplementary Material online). The total number of alleles (TNA) per locus ranged between 6 for ILSTS011 and BM1824 and 32 for CYP21. Allelic richness per locus was high, with an overall average of 8.3 ± 2.5. Several STRs had Rt . 10 (BM203, CSSM066, CYP21, MGTG4B, TGLA053, and TGLA122), whereas the lowest value was 4.6 (RM006).

The highest heterozygosity was found for TGLA053 (Ho 5 0.795 ± 0.074 and He 5 0.815 ± 0.050) and TGLA227 (Ho 5 0.781 ± 0.074 and He 5 0.820 ± 0.031) and the lowest for SPS115 (Ho 5 0.466 ± 0.211 and He 5 0.463 ± 0.190). The Fis value had an overall mean of 0.028 ± 0.005 (P , 0.01) and was significantly (P , 0.01) higher than zero for HAUT027 and INRA037. Fit and Fst per locus were mostly significant (P , 0.01), with overall means of 0.107 ± 0.006 and 0.081 ± 0.004, respectively. LD was significant (P , 0.0001) for 24 STR pairs; however, only the following corresponded to loci located in the same chromosome: BM0888/INRA037; CSRM060/INRA037; BM4208/ MM12E6; and HEL001/SPS115. Thus, the STRs INRA037, BM4208, and HEL001 were excluded from the analyses done with Structure. MM12E6 and SPS115 were selected for breed assignment analyses because, based on Whichloci results (see Genetic Structure and Admixture Analysis section), they showed better allele frequency distribution, and they were more informative than their counterparts BM4208 and HEL001. Within-Breed Genetic Diversity Estimates of within-breed genetic diversity are summarized in Table 1. The breeds Brava de Lide and Mirandesa had the lowest diversity (He 5 0.643 ± 0.150 and He 5 0.621 ± 0.147; Ho 5 0.577 ± 0.154 and Ho 5 0.607 ± 0.151; MNA 5 5.7 ± 1.8 and MNA 5 5.4 ± 1.3, respectively). Both Garvonesa and Alentejana had relatively low levels of diversity. The highest genetic diversity was found in Minhota (Ho 5 0.758±0.103 and He 5 0.739±0.088). Likewise, Arouquesa and Mertolenga had high genetic diversity and showed the highest MNA (7.9 ± 2.7 and 7.8 ± 2.4). PAs were detected in all breeds except Alentejana and Marinhoa; however, only 5 out of 54 PAs were detected in frequencies 0.05 (two in Friesian, one in Mertolenga, one in Minhota, and one in Ramo Grande). Tests for deviations from HWE showed significant (P , 0.05) heterozygote deficit in all breeds except Garvonesa, Marinhoa, Minhota, and Limousin. The number of loci that had significant heterozygote deficit (P , 0.05) varied between 1 (Marinhoa and Limousin) and 17 (Mertolenga). Out of the 39 STRs analyzed in each breed, only INRA037 showed significant (P , 0.05) HWE deviations across all breeds. Heterozygote excess was not significant for any of the locus/breed combinations tested. The HWE deviations detected are most probably due to inbreeding as the within-breed Fis estimates were significantly (P , 0.05) higher than zero in 10 populations. For Alentejana, Brava de Lide, Mertolenga, and Ramo Grande, the Fis . 0 was highly significant (P , 0.0001). Genetic Relationship Between Breeds Pairwise population Fst values are shown in Table S2 (see Supplementary Material online) and were all significant for P , 0.05 after a standard ‘‘Bonferroni’’ correction. The lowest Fst value was found among Cachena and Barrosa˜ (0.011), whereas the highest value was observed between Mirandesa and Friesian (0.177). Breed relationships based

203

Journal of Heredity 2010:101(2) Table 1. Observed (Ho) and expected (He) heterozygosities, MNAs, PAs, inbreeding coefficient (Fis), and number of loci showing heterozygote deficit (HWEd) estimated with 39 STRs in 13 Portuguese native and 3 imported cattle breeds Breed Portuguese native Northern Arouquesa (ARO) Barrosa˜ (BAR) Cachena (CAC) Marinhoa (MRH) Maronesa (MRO) Minhota (MIN) Mirandesa (MIR) Central and Southern Alentejana (ALE) Brava de Lide (BRV) Garvonesa (GAR) Mertolenga (MER) Azores Archipelago Ramo Grande (RGR) Imported Charolais (CHA) Friesian (FRI) Limousin (LIM) Overall

Estimated censusa

N

Ho±SD

He±SD

MNA±SD

PA

Fis

HWEd

4800 (2000) 7100 (3100) 850 (250) 2500 (1000) 5500 (1500) 6634 (2000) 5000 (1330)

70 69 51 46 47 50 54

0.708±0.109 0.673±0.133 0.709±0.134 0.672±0.136 0.654±0.142 0.758±0.103 0.607±0.151

0.734±0.090 0.691±0.115 0.718±0.113 0.685±0.118 0.671±0.124 0.739±0.088 0.621±0.147

7.87±2.66 6.64±1.81 7.36±1.97 6.26±2.07 6.38±1.79 7.69±2.10 5.36±1.27

5 3 3 0 3 7 3

0.035** 0.026* 0.014 0.019 0.025* 0.026 0.023*

9 3 4 1 7 2 3

7200 (120) 9000 (93) 145 (7) 14 983 (250)

38 43 39 64

0.634±0.130 0.577±0.154 0.676±0.135 0.648±0.109

0.675±0.127 0.643±0.150 0.654±0.132 0.719±0.108

5.87±1.63 5.72±1.82 6.03±1.77 7.77±2.41

0 2 1 4

0.061*** 0.103*** 0.034 0.099***

7 15 2 17

597 (125)

44

0.688±0.091

0.719±0.087

7.46±2.20

6

0.044***

8

58 51 47 831

0.692±0.128 0.683±0.139 0.712±0.101 0.673±0.043

0.710±0.111 0.675±0.123 0.716±0.092 0.691±0.034

6.69±2.04 6.46±2.40 6.64±2.12 6.70±0.77

2 8 4 54

1235 51 900 3100

0.025* 0.012 0.006 0.027

4 3 1 89

Breeds are grouped according to geographic area. N: sample size. Significance levels are *(P  0.05), **(P  0.01), and ***(P  0.001). a

Number of breeding females registered in herdbooks and number of breeders as in 2004 census (Gama et al. 2004).

on pairwise Fst are depicted in the PCA graph of Figure 1. The northern breeds Mirandesa, Barrosa˜, and Cachena were the most distant. Marinhoa, Maronesa, Brava de Lide, and Ramo Grande had an intermediate position, whereas the remaining native breeds grouped at the center. Among the imported breeds, Friesian was the most distant and Limousin had a more central position closer to native breeds. Minhota and Ramo Grande showed a closer genetic relationship with imported cattle, which based on Axes 1 and 2 were split from most of the native breeds. Northern native cattle were split from southern breeds based on Axis 3, except for Marinhoa, Arouquesa, and Minhota, which occupied a more central position. The Neighbor-Net of DA genetic distances is shown in Figure 2, and pairwise population distance values are shown in Table S2 (see Supplementary Material online). The imported breeds appear separated from the Portuguese native breeds, except Minhota and Ramo Grande, with the latter clustering with Friesian. Barrosa˜ clustered with Cachena, and Mirandesa with Marinhoa, each pair forming a tight net. The southern breeds Alentejana, Garvonesa, and Mertolenga grouped together, but the first two breeds were genetically closer. Brava de Lide split from the center as an independent and more distant branch. The northern Maronesa also clustered separately, whereas Arouquesa was in the same branch as Mirandesa. Genetic Structure and Admixture Analysis For the analysis done without including prior information on sample origin and following Evanno et al. (2005), the highest DK was found for K 5 2 and K 5 15 (Figure S1, see

204

Supplementary Material online). The structure detected at K 5 2 is explained by the lowest likelihood variance, whereas the highest likelihood of the data was obtained for K 5 15. Based on individual q values, Barrosa˜ and Cachena animals formed one cluster whereas those from each of the remaining breeds formed independent groups (Figure 3a). Further analyses were done with subsets of the data to determine if individuals of these two breeds could be distinguished and to assess substructure within Mertolenga associated with different coat colors. The modal value of DK was obtained at K 5 3 for the analysis of Barrosa˜, Cachena, and Mirandesa. Mirandesa was included as a reference breed for this analysis because it is a well-differentiated population and thus useful to determine the most probable K value of this data subset. Clear evidence of admixture was detected between Cachena and Barrosa˜, although it was possible to identify a cluster of Cachena animals not related to Barrosa˜ (Figure 3b). For Mertolenga, substructure was confirmed through DK obtained at K 5 4 as shown in Figure 3c. Red speckled (Rosilho) animals formed an independent cluster, whereas red (Vermelho) and red- and white-spotted (Malhado) individuals clustered together. The results of the Bayesian cluster analysis done with Structure are summarized in Table 2. Average membership proportions in each cluster (Q) ranged between 0.692 ± 0.270 in Ramo Grande and 0.919 ± 0.073 in Mirandesa, with an overall average of 0.817 ± 0.080. Approximately 70% of the individuals were classified within their source cluster with q  0.800. The most heterogeneous breeds were Arouquesa, Cachena, Mertolenga, Minhota, Preta, and Ramo Grande, with ,55% of the individuals correctly assigned (q  0.800). Mirandesa had the highest

Ginja et al.  Genetic Structure of Portuguese Native Cattle

Garvonesa). Admixed animals were detected across all breeds, and among native breeds ranged between 2.6% (Alentejana) and 15.4% (Garvonesa).

Discussion

Figure 1. PCA based on pairwise Fst values. Breed codes are defined in Table 1.

percentage of animals assigned to the source cluster (93% with q  0.800). When prior information on source breeds was considered in the analysis, the overall Q was 0.944 ± 0.149 and varied between 0.865 ± 0.296 in Garvonesa and 0.981 ± 0.062 in Barrosa˜. Individual q values are depicted in Figure 3d. The overall percentage of individuals correctly assigned to the source breed was about 93% for q  0.800 and 82% for q  0.950. The percentage of correctly assigned animals with q  0.800 was 90% in all breeds except Cachena, Garvonesa, and Preta. Among native cattle, Barrosa˜ (94%), Mirandesa (89%), and Maronesa (92%) had the highest percentage of individuals correctly assigned with q  0.950. In fact, for Mirandesa and Barrosa˜, there were about 24% and 22% of animals with q  0.999, respectively. Overall, only 10% of the individuals could be assigned to the source breed with high stringency (q  0.999). When the assignment analysis was done using a subset of the 16 most informative markers (BM0888, BM2113, BM5004, BRRIBO, CSSM042, CSSM066, CY0P21, ETH003, ETH010, HEL009, INRA023, MGTG4B, MM12E6, TGLA053, TGLA122, and TGLA227), there was a slight increase in the overall percentages of correctly assigned animals at each threshold q value considered except at the highest stringency (q  0.999). Evidence of admixture was found for a total of 56 animals (6.7%) that had q , 0.800 in their breed of origin; however, only two of these animals were (mis)assigned to a breed other than the source (one Arouquesa and one

The levels of within-breed diversity detected in Portuguese native cattle were high (average Ho 5 0.667 ± 0.046, He 5 0.689 ± 0.036 and MNA 5 6.7 ± 0.9) and comparable to those previously reported (Ho 5 0.645 ± 0.040, He 5 0.667 ± 0.027 and MNA 5 6.1 ± 0.4, Can˜on et al. [2001]; Ho 5 0.672 ± 0.056, He 5 0.697 ± 0.037 and MNA 5 7.0 ± 0.8, Mateus et al. [2004b]). Preta, Cachena, and Ramo Grande (Azores) were characterized here for the first time and showed levels of diversity comparable to other native breeds, even though the latter two have a census of ,900 cows (Gama et al. 2004). The genetic variability found in Portuguese cattle might reflect contributions from animals of distinct origins. Cymbron et al. (2005) found significantly higher diversity in Mediterranean cattle, which included three Portuguese breeds, when compared with Northern European cattle, as a result of Near Eastern and African influences. Analysis of mtDNA sequences and Y haplotypes have confirmed the heterogeneous genetic composition of Portuguese cattle through the detection both European and African lineages (Cymbron et al. 1999; Beja-Pereira et al. 2006; Ginja et al. 2009). The results obtained here also support African influence in Portuguese breeds through the detection of BM2113-123 allele (Alentejana, Garvonesa, and Mertolenga) that was previously associated with West African taurine cattle (MacHugh 1996). The presence of zebu BM2113-143 (Arouquesa, Garvonesa, Marinhoa, and Mirandesa) and ETH152-193 alleles (Alentejana, Mertolenga, and Preta), also found in African cattle (MacHugh 1996), suggests that zebu gene flow in Portugal might have an African origin. Despite the high genetic diversity detected within Portuguese native cattle, significant inbreeding was also observed in some breeds. Although the heterozygote deficit observed in Brava de Lide and Mertolenga could be the result of a Wahlund effect related to the presence of independent lineages in these breeds, there is evidence of genetic erosion caused by very low effective number of males in Alentejena (Carolino and Gama 2008) and very small population size in Ramo Grande (Gama et al. 2004). Thus, measures should be applied to minimize the negative effects of inbreeding and preserve these valuable genetic resources. The Bayesian clustering analysis confirmed that Portuguese breeds are highly structured, with slight admixture. Despite the small geographic area where these cattle are raised, there are low levels of gene flow among breeds. However, some level of admixture is expected because Portuguese cattle were threatened by crossbreeding with imported commercial breeds, and parentage DNA testing is not a common practice for registration of animals in herdbooks. The partition of the dataset was consistent with

205

Journal of Heredity 2010:101(2)

Figure 2.

Neighbor-Net graph of DA genetic distances among 13 Portuguese native and 3 imported cattle breeds.

the original breed designations. In the case of Barrosa˜ and Cachena breeds, although the most probable number of clusters was two, there was clear admixture between them. These two breeds were treated as one population for the structure analysis done with prior information on breeds. Historical evidence supports a close genetic relationship of these breeds. Cachena was considered a subtype of Barrosa˜ and only in 1998 was recognized as an independent breed (Leite and Dantas 2000). The substructure detected in Mertolenga reflects the tendency of breeders to separate herds according to distinct coat colors. Rosilho herds are distributed from the central to the southern region of Alentejo, Vermelho animals are raised throughout the central toward the north western area, whereas Malhado herds are fewer and dispersed throughout Alentejo (http://mertolenga. no.sapo.pt/raca_mapa.html). The structure analysis was useful to identify the more uniform breeds and confirm their genetic differentiation. Mirandesa was extremely homogenous, showing very high proportions of individual genotype memberships (Q 5 0.919 ± 0.073). The genetic distinctiveness of this breed is probably associated with the fact that it was the first among Portuguese native breeds to have a herdbook that was established in 1959 (Sousa 2000). Based on

206

a previous study, Mirandesa is genetically distinct from other European breeds and contributes significantly to the overall genetic diversity (Can˜on et al. 2001). Brava de Lide and Maronesa had a high percentage of individuals correctly assigned, even for the analysis done without prior information on sample origin, which indicates that these breeds are genetically distinct and is consistent with their placement depicted in the Neighbor-Net graph. Both breeds are considered to belong to a more ancestral group of the Iberian Peninsula as discussed by Mateus et al. (2004b). A recent study on mtDNA sequence variation in De Lidia cattle, the Spanish equivalent of Brava de Lide, concluded that these animals represent primitive Iberian cattle with multiple influences, such as Mediterranean and African (Cortes et al. 2008). The assignment tests done with prior population information resulted in high overall percentages of individuals correctly classified, except for the most stringent threshold q value of 0.999. The degree of genetic differentiation among Portuguese breeds and the results of the assignment tests indicate that it is possible to identify reference animals with high membership coefficients in their breed of origin. Even for closely related breeds, such as Mirandesa/Marinhoa and Alentejana/Garvonesa, it was

Ginja et al.  Genetic Structure of Portuguese Native Cattle

Figure 3. Results of the Bayesian cluster analysis with Structure. (a) Graphical representation of individual genotype membership coefficients (q) in each of 15 clusters (Barrosa˜ and Cachena are in the same group) without prior information on source breeds. (b) Detailed analysis of Barrosa˜ and Cachena using Mirandesa as reference. (c) Substructure within Mertolenga using the related Alentejana and Garvonesa breeds for comparison. (d) Assignment results when sample origin is included in the analysis.

possible to correctly assign individuals with high threshold q values (e.g., q  0.950). The assignment analyses were useful to detect admixture and to identify the animals that conform to the genetic profile of the breed, which could be selected for breeding to help maintain the integrity of each breed. For the more heterogeneous breeds, the percentages of correctly assigned individuals were lower (,79%, q  0.950). Because local breeds might have multiple origins, some degree of heterogeneity is anticipated and should be considered as part of their history, rather than a depreciative feature (Econogene 2006). This is most probably the case of Mertolenga, which is heterogeneous at the molecular and phenotypic levels. However, low genotype memberships can also reflect dilution of a gene pool due to crossbreeding, which seems to be the case of Minhota and Ramo Grande. Even though the source of the admixture was often difficult to determine because of low q values in each breed, the PCA and the Neighbor-Net representations of breed clusters are consistent with this interpretation. Historical evidence supports crossbreeding in Minhota, mainly with German Yellow and Charolais animals (Machado 2000). Ramo Grande appears associated with Friesian, and crossbreeding among these two breeds is

supported by the large predominance of Friesian cattle in the Azores Archipelago. A recent survey of Y haplotypes in Portuguese cattle showed prevalence of Friesian patrilines in Ramo Grande (Ginja et al. 2009). Nonetheless, the genetic composition of Ramo Grande can also mirror the initial contribution of distinct continental breeds because some of the admixed animals had recent ancestry in other native breeds. The admixture detected in Preta appears to reflect crossbreeding with related southern breeds such as Alentejana and Garvonesa. The heterogeneity observed in Garvonesa is the result of genetic erosion of this native breed, which is highly threatened and only a few individuals persist (Gama et al. 2004). In conclusion, Portuguese native breeds retain high genetic diversity and are highly structured. Nonetheless, some breeds showed signs of genetic erosion due to inbreeding or crossbreeding mainly with imported cattle. These results can be used to assist authorities in the implementation of national conservation measures and breeding programs. The genotype dataset assembled here is also useful for the assignment of unknown samples to source breeds. Assignment methods can also be used for validation of certified meat products, but further investigation is still needed to validate the performance of such methods to assure high confidence in breed allocation.

207

Journal of Heredity 2010:101(2) Table 2.

Summary results of the Bayesian cluster analysis done with Structure Without prior information

Breed names Portuguese native Alentejana Arouquesa Barrosa˜ Brava de Lide Cachena Garvonesa Marinhoa Maronesa Mertolenga Minhota Mirandesa Preta Ramo Grande Imported Charolais Friesian Limousin Overall

With prior information on breeds

With prior information on breeds and 16 STRs

Q±SD

% Corr. assign. q 0.800

Q±SD

q 0.800

q 0.950

q 0.999

% Adm.

Q±SD

q 0.800

q 0.950

q 0.999

0.836±0.154 0.724±0.225 0.892±0.098 0.875±0.136 0.743±0.186 0.794±0.223 0.847±0.138 0.877±0.093 0.716±0.245 0.769±0.194 0.919±0.073 0.708±0.271 0.692±0.270

71.1 51.4 91.3 81.4 45.1 64.1 73.9 87.2 51.6 54.0 92.6 51.7 53.8

0.963±0.112 0.931±0.173 0.981±0.062 0.945±0.183 0.927±0.132 0.865±0.296 0.948±0.144 0.968±0.093 0.948±0.095 0.920±0.173 0.957±0.156 0.904±0.215 0.922±0.167

97.4 92.9 97.1 95.3 88.2 84.6 93.5 95.7 93.8 90.0 94.4 86.7 90.9

78.9 77.1 94.2 86.0 76.5 71.8 84.8 91.5 75.0 74.0 88.9 75.0 75.0

7.9 0.0 21.7 14.0 2.0 7.7 4.3 6.4 7.8 2.0 24.1 13.3 11.4

2.6 7.0 2.9 4.6 11.8 15.4 6.5 4.2 6.3 10.0 5.5 13.3 9.3

0.970±0.071 0.921±0.180 0.983±0.025 0.946±0.164 0.925±0.126 0.886±0.250 0.943±0.131 0.937±0.171 0.957±0.061 0.956±0.066 0.965±0.094 0.951±0.091 0.950±0.081

94.7 91.4 100 93.0 88.2 84.6 91.3 93.6 96.9 96.0 94.4 95.0 93.2

86.8 78.6 94.2 79.1 72.5 74.4 82.6 85.1 68.8 74.0 85.2 75.0 75.0

2.6 0.0 0.0 4.7 0.0 2.6 2.2 4.3 0.0 0.0 7.4 3.3 0.0

0.879±0.096 0.910±0.068 0.885±0.064 0.817±0.080

77.6 90.2 89.4 70.4

0.978±0.046 0.969±0.097 0.975±0.062 0.945±0.149

96.6 96.1 97.9 93.2

89.7 88.2 89.4 82.3

15.5 15.7 8.5 10.1

3.4 3.9 2.1 6.7

0.973±0.067 0.972±0.067 0.972±0.031 0.951±0.118

96.6 98.0 100 94.2

89.7 84.3 80.9 80.4

5.2 11.8 0.0 2.7

% Corr. assign.

% Corr. assign.

Average genotype membership proportions (Q), the percentage of correctly assigned (% corr. assign.) animals, and the percentage of admixed (% Adm.) individuals are shown for each breed and overall.

Supplementary Material Supplementary material can be found at http://www.jhered. oxfordjournals.org/.

Funding Grant from the Fundac xa˜o para a Cieˆncia e a Tecnologia (Ref. SFRH/BD/13502/2003 to C.G.). This work was partially funded by the Veterinary Genetics Laboratory, University of California, Davis. The sample collection and establishment of the DNA bank were funded by Direccxa˜o Geral de Veterina´ria and Sociedade Portuguesa de Recursos Gene´ticos Animais.

Acknowledgments We gratefully acknowledge the collaboration and assistance of the Portuguese breed associations in the sampling of animals.

References Anderung C, Bouwman A, Persson P, Carretero JM, Ortega AI, Elburg R, Smith C, Arsuaga JL, Ellegren H, Gotherstrom A. 2005. Prehistoric contacts over the Straits of Gibraltar indicated by genetic analysis of Iberian Bronze Age cattle. Proc Natl Acad Sci USA. 102:8431–8435. Banks MA, Eichert W, Olsen JB. 2003. Which genetic loci have greater population assignment power? Bioinformatics. 19:1436–1438.

208

Beaumont M, Barratt EM, Gottelli D, Kitchener AC, Daniels MJ, Pritchard JK, Bruford MW. 2001. Genetic diversity and introgression in the Scottish wildcat. Mol Ecol. 10:319–336. Beaumont MA, Rannala B. 2004. The Bayesian revolution in genetics. Nat Rev Genet. 5:251–261. Beja-Pereira A, Alexandrino P, Bessa I, Carretero Y, Dunner S, Ferrand N, Jordana J, Laloe D, Moazami-Goudarzi K, Sanchez A, et al. 2003. Genetic characterization of southwestern European bovine breeds: a historical and biogeographical reassessment with a set of 16 microsatellites. J Hered. 94:243–250. Beja-Pereira A, Caramelli D, Lalueza-Fox C, Vernesi C, Ferrand N, Casoli A, Goyache F, Royo LJ, Conti S, Lari M, et al. 2006. The origin of European cattle: evidence from modern and ancient DNA. Proc Natl Acad Sci USA. 103:8113–8118. Belkhir K, Borsa P, Chikhi L, Raufaste N, Bonhomme F. 1996–2004. GENETIX: logiciel sous Windows TM pour la ge´ne´tique des populations. Montpellier (France): Laboratoire Ge´nome, Populations, Interactions, CNRS, Universite´ de Montpellier II. Bruford MW. 2004. Conservation genetics of UK livestock: from molecules to management. In: Simm G, Villanueva B, Townsend S, editors. Farm animal genetic resources. Nottingham: Nottingham University Press. p. 151–169. Bruford MW, Bradley DG, Luikart G. 2003. DNA markers reveal the complexity of livestock domestication. Nat Rev Genet. 4:900–910. Can˜on J, Alexandrino P, Bessa I, Carleos C, Carretero Y, Dunner S, Ferrand N, Garcı´a D, Jordana J, Laloe D, et al. 2001. Genetic diversity measures of local European beef cattle breeds for conservation purposes. Genet Sel Evol. 33:311–332. Can˜on J, Garcı´a D, Garcı´a-Atance MA, Obexer-Ruff G, Lenstra JA, Ajmone-Marsan P, Dunner S. 2006. Geographical partitioning of goat diversity in Europe and the Middle East. Anim Genet. 37:327–334.

Ginja et al.  Genetic Structure of Portuguese Native Cattle Carolino N, Gama LT. 2008. Indicators of genetic erosion in an endangered population: the Alentejana cattle breed in Portugal. J Anim Sci. 86:47–56. Chapuis MP, Estoup A. 2007. Microsatellite null alleles and estimation of population differentiation. Mol Biol Evol. 24:621–631. Chikhi L, Bruford MW. 2005. Mammalian population genetics and genomics. In: Ruvinsky A, Marshall-Graves J, editors. Mammalian genomics. Oxford: CABI Publishing. p. 539–584. Cornuet JM, Piry S, Luikart G, Estoup A, Solignac M. 1999. New methods employing multilocus genotypes to select or exclude populations as origins of individuals. Genetics. 153:1989–2000. Cortes O, Tupac-Yupanqui I, Dunner S, Garcia-Atance MA, Garcia D, Fernandez J, Canon J. 2008. Ancestral matrilineages and mitochondrial DNA diversity of the Lidia cattle breed. Anim Genet. 39:649–654. Cymbron T, Freeman AR, Isabel Malheiro M, Vigne JD, Bradley DG. 2005. Microsatellite diversity suggests different histories for Mediterranean and Northern European cattle populations. Proc Biol Sci. 272:1837–1843. Cymbron T, Loftus RT, Malheiro MI, Bradley DG. 1999. Mitochondrial sequence variation suggests an African influence in Portuguese cattle. Proc Biol Sci. 266:597–603. Dakin EE, Avise JC. 2004. Microsatellite null alleles in parentage analysis. Heredity. 93:504–509. Dempster AP, Laird NM, Rubin DB. 1977. Maximum likelihood from incomplete data via the EM algorithm. J R Stat Soc B. 39:1–38. Driscoll CA, Menotti-Raymond M, Roca AL, Hupe K, Johnson WE, Geffen E, Harley EH, Delibes M, Pontier D, Kitchener AC, et al. 2007. The Near Eastern origin of cat domestication. Science. 317:519–523. Econogene. 2006. Marker-assisted conservation of European cattle breeds: an evaluation. Anim Genet. 37:475–481. Evanno G, Regnaut S, Goudet J. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Mol Ecol. 14:2611–2620. Falush D, Stephens M, Pritchard JK. 2003. Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics. 164:1567–1587. Gama LT, Carolino N, Costa MS, Matos CP. 2004. Recursos gene´ticos animais em Portugal. Relato´rio Nacional. Vale de Santare´m (Portugal): Food and Agriculture Organization of the United Nations. p. 68. Ginja C, Gama LT, Penedo MCT. 2009. Y chromosome haplotype analysis in Portuguese cattle breeds using SNPs and STRs. J Hered. 100:148–157. Goudet J. 1999. PCAGEN software. Available from: http://www2.unil.ch/ popgen/softwares/pcagen.htm Goudet J. 2001. FSTAT a program to estimate and test gene diversities and fixation indices. Updated from Goudet J. 1995. J Hered. 86:485–486. Guo SW, Thompson EA. 1992. Performing the exact test of HardyWeinberg proportion for multiple alleles. Biometrics. 48:361–372. Hughes SS. 2000. STRand nucleic acids analysis software. University of California, Davis, CA.

Espe´cies bovina e equina. Salamanca (Spain): Fundacxa˜o Rei Afonso Henriques. p. 77–90. Leroy G, Verrier E, Meriaux JC, Rognon X. 2009. Genetic diversity of dog breeds: between-breed diversity, breed assignation and conservation approaches. Anim Genet. 40:333–343. Lipinski MJ, Froenicke L, Baysac KC, Billings NC, Leutenegger CM, Levy AM, Longeri M, Niini T, Ozpinar H, Slater MR, et al. 2008. The ascent of cat breeds: genetic evaluations of breeds and worldwide random-bred populations. Genomics. 91:12–21. Machado MA. 2000. Galega/Minhota. In: Garcia EY, editor. Cata´logo de rac xas auto´ctones de Castela e Lea˜o (Espanha)—Regia˜o Norte de Portugal: Espe´cies bovina e equina. Salamanca (Spain): Fundacxa˜o Rei Afonso Henriques. p. 117–129. MacHugh DE. 1996. Molecular biogeography and genetic structure of domesticated cattle. Department of Genetics, Trinity College. Dublin: University of Dublin. Manel S, Berthier P, Luikart G. 2002. Detecting wildlife poaching: identifying the origin of individuals with Bayesian assignment tests and multilocus genotypes. Cons Biol. 16:650–659. Manel S, Gaggiotti OE, Waples RS. 2005. Assignment methods: matching biological questions with appropriate techniques. Trends Ecol Evol. 20:136–142. Mateus JC, Eding H, Penedo MC, Rangel-Figueiredo MT. 2004a. Contributions of Portuguese cattle breeds to genetic diversity using marker-estimated kinships. Anim Genet. 35:305–313. Mateus JC, Penedo MC, Alves VC, Ramos M, Rangel-Figueiredo T. 2004b. Genetic diversity and differentiation in Portuguese cattle breeds using microsatellites. Anim Genet. 35:106–113. Maudet C. 2001. Diversite´ et caracte´risation ge´ne´tique des races bovines et caprines originaires de la re´gion Rhoˆne-Alpes. Laboratoire de Biologie des Populations d’Altitude de Grenoble. Grenoble (France): Universite´ Joseph Fourier. p. 165. Moioli B, Napolitano F, Catillo G. 2004. Genetic diversity between Piedmontese, Maremmana, and Podolica cattle breeds. J Hered. 95:250–256. Negrini R, Milanesi E, Colli L, Pellecchia M, Nicoloso L, Crepaldi P, Lenstra JA, Ajmone-Marsan P. 2007. Breed assignment of Italian cattle using biallelic AFLP markers. Anim Genet. 38:147–153. Peakall R, Smouse PE. 2006. Genalex 6: genetic analysis in Excel. Population genetic software for teaching and research. Mol Ecol Notes. 6:288–295. Pires AE, Amorim IR, Ginja C, Gomes M, Godinho I, Simo˜es F, Oom M, Petrucci-Fonseca F, Matos J, Bruford MW. 2009. Molecular structure in peripheral dog breeds: Portuguese native breeds as a case study. Anim Genet. 40:383–392. Pritchard JK, Stephens M, Donnelly P. 2000. Inference of population structure using multilocus genotype data. Genetics. 155:945–959. Ralo A, Guerreiro A. 1981. Racas exoticas e cruzamentos. In: Rodrigos A, editor. Bovinos em Portugal. Lisboa: Direcc xa˜o Geral dos Servic xos Veterina´rios.

Huson DH, Bryant D. 2006. Application of phylogenetic networks in evolutionary studies. Mol Biol Evol. 23:254–267.

Randi E, Pierpaoli M, Beaumont M, Ragni B, Sforzi A. 2001. Genetic identification of wild and domestic cats (Felis silvestris) and their hybrids using Bayesian clustering methods. Mol Biol Evol. 18:1679–1693.

Koskinen MT. 2003. Individual assignment using microsatellite DNA reveals unambiguous breed identification in the domestic dog. Anim Genet. 34:297–301.

Raymond M, Rousset F. 2003. GENEPOP: population genetics software for exact tests and ecumenicism. Updated from Raymond M, Rousset F. 1995. J Hered. 86:248–249.

Langella O. 1999. POPULATIONS. France: CNRS UPR9034. Lawson Handley LJ, Byrne K, Santucci F, Townsend S, Taylor M, Bruford MW, Hewitt GM. 2007. Genetic structure of European sheep breeds. Heredity. 99:620–631.

Rosenberg NA, Burke T, Elo K, Feldman MW, Freidlin PJ, Groenen MA, Hillel J, Maki-Tanila A, Tixier-Boichard M, Vignal A, et al. 2001. Empirical evaluation of genetic clustering methods using multilocus genotypes from 20 chicken breeds. Genetics. 159:699–713.

Leite JV, Dantas RV. 2000. Cachena. In: Garcia EY, editor. Cata´logo de rac xas auto´ctones de Castela e Lea˜o (Espanha)—Regia˜o Norte de Portugal:

Sousa FJR. 2000. Mirandesa. In: Garcia JEY, editor. Cata´logo de rac xas auto´ctones de Castela e Lea˜o (Espanha)—Regia˜o Norte de Portugal: I.

209

Journal of Heredity 2010:101(2) Espe´cies bovina e equina. Salamanca (Spain): Fundac xa˜o Rei Afonso Henriques. p. 65–89.

Zilha˜o J. 2001. Radiocarbon evidence for maritime pioneer colonization at the origins of farming in west Mediterranean Europe. Proc Natl Acad Sci USA. 98:14180–14185.

Vicente AA, Carolino MI, Sousa MC, Ginja C, Silva FS, Martinez AM, Vega-Pla JL, Carolino N, Gama LT. 2008. Genetic diversity in native and commercial breeds of pigs in Portugal assessed by microsatellites. J Anim Sci. 86:2496–2507.

Received April 28, 2009; Revised August 17, 2009; Accepted October 23, 2009

Weir BS, Cockerham CC. 1984. Estimating F-statistics for the analysis of population structure. Evolution. 38:1358–1370.

Corresponding Editor: James E. Womack

210