Horse DNA sequences used for comparison Sequence data from the ancient bones were compared with 363 sequences of the mtDNA D-loop region available from publications and/or from GenBank (http://www.ncbi.nlm. Ó 2005 International Society for Animal Genetics, Animal Genetics, 36, 203–209 Mitochondrial DNA analysis of horses nih.gov/GenBank).
We selected domestic horse sequences representing a variety of breeds from all over the world.
Central and East Asia: Tuva (AF481324–34); Mongolian (AF056071, AF014413–15, AF072996); Kazakh (AF072985); Korean Cheju (AF014405–8, AF014410–12, AY049718–19, AY246201–8, AF354425–41); Tsushima (AF169009–10); Chinese Yunnan (AF014416–17); Chinese Guanzhong (AY136785–86); Caspian (AY246195– AY246200).
Near East: Anatolian (AF481232–46); Cukorova (AF481259–70); Egyptian (AF481271–78); Fulani (AF481279–90).
Europe: Connemara (AF481246– 58); Lippizan (AY057408–AY057434); Belgian (AF246186–94, AF064631–32); Lusitano (AY293975– 89, AY246242–47); Cleveland Bay (AF246209–13); Clydesdale (AY246214–18); Garrano (AY246231–35); Haﬂinger (AY246236–41); Noriker (AY246248–52); Friesian (AY246225–27); Asturcon (AF46006); Andalusian (AF46007–8); Losino (AF46009–10); Mallorquina (AF466013–16).
South America: Argentinean Creole (AF465984–90); Peruvian Paso (AF465991–95).
We also included Ônative poniesÕ: Shetland (AF481291–AF481304, AY246253–AY246258 and AF072977–79); Sorraia (AY246259–66); Potoka (AF466011–12); Icelandic (AF072988); Fell (AF072981) and Exmoor (AY246219– 24, AF072992–93). ÔHot blooded breedsÕ: Thoroughbreds (AF481305–23, AY246266–71) and Arabian (AY246180–85; AF168689–98; AF064627; AF465996– 6005, AF72989), Ôcold blooded horsesÕ: Shire (AF072975– 76) and Suffolk Punch (AF072986), supposedly Ôrecent breedsÕ: Danish (AF072987); Welsh Cob (AF072983); Standardbred (AF072984) and Quarter Horse (AF072980), and Ôancient breedsÕ: Akhal-Teke (AY246174–79).
Finally we added: wild horse Przewalski’s sequences (AF014409, AF072994–95); ancient horse sequences from northern Europe (AF326676–79) and east Asia (AY049720), and late Pleistocene sequences from Alaska (AF326668–75). 205 Results The control region of the mtDNA was successfully ampliﬁed from the 13 ÔScythianÕ horse bones tested (all replicated sequences were identical), a result which can undoubtedly be attributed to the exceptional preservation of the archaeological remains recovered from the Berel site.
Examination of the 348 bp D-loop region showed that some of the ancient specimens shared the same mtDNA sequence.
Eight different haplotypes were determined, each differing from the reference sequence (GenBank X79547) by ﬁve to 11 sites.
Altogether, 24 polymorphic sites (representing 6.9 % of the total DNA sequence analysed) were identiﬁed; all of these variable positions were single nucleotide substitutions, none of them were transversions.
Moreover, no heteroplasmy was detected (Table 1).
Using the equation h¼ ð1 À Rx2 Þn ; nÀ1 Statistical analysis Sequences of the D-loop were handled using the BioEdit sequence alignment editor Version 5 (Hall 1999) and then incorporated in the NSA software (freely available at http://perso.wanadoo.fr/daniel.montagnon/NSASoft.htm) in which identical sequences were joined into the same haplotype (sequences with gap or missing data were excluded).
Phylogenetic analysis of the mtDNA haplotypes (133 altogether) was performed using MEGA Version 2.1 (Kumar et al. 2001).
Phylogenetic trees were constructed using the maximum parsimony and neighbour-joining method, the statistical conﬁdence of each node being estimated by 1000 random bootstrap runs.
Genetic distances were estimated using the Tamura–Nei distance (Tamura & Nei 1993) calculated on the basis of an equal substitution rate per site.
The donkey (Equus asinus) sequence was used as outgroup (Xu et al. 1993). where n is sample size and x is the frequency of each haplotype (Tajima 1989), the genetic diversity of the eight haplotypes was calculated to be 0.923, indicating substantial level of genetic variability between the ancient individuals.
Horses that shared an identical mtDNA sequence (and which were possibly from the same maternal line) were the following: BER01 and BER03 (which both belong to cluster A6), BER05 and BER09 (which differ from the previous ones by one transition at position 15597), BER04 and BER08 (both belonging to haplogroup D), and ﬁnally BER10, BER11 and BER13 [all three assigned to cluster C1 according to Jansen et al. (2002)].
As shown in Table 1, these individuals were not found to wear ornamentation of the same cultural inﬂuence.
The most striking example concerns BER04 and BER08.
Whereas BER04 was adorned with grifﬁn and horned lion, which seemed typical of the Achaemenid Persian art, BER08 showed ﬁnery characteristic from the steppe’s world and Altai.
From this it appeared that horses presumably belonging to the same maternal lineage were not similarly ornamented.
Likewise, some horses presenting decorative elements from the same cultural inﬂuence were found to have different mtDNA haplotypes (eg BER07, BER08, BER09 and BER11 were decorated with ornaments attributed to the Altaic art, although none of them share the same sequence).
Two samples from Yakut horses were also included in our study.
Both gave positive ampliﬁcation results.
The nucleotide sequences obtained differed between each other and with those of the Berel specimens (Table 1).
In order to determine the phylogenetic position of the ancient specimens studied, the ÔScythianÕ and the Yakut DNA sequences were compared with equine control region sequences retrieved from GenBank (see Materials and methods).
Alignment of the overall sequences allowed the deﬁnition of 203 haplotypes, more than half unique.
To optimize and simplify the comparative analysis of horsesÕ Ó 2005 International Society for Animal Genetics, Animal Genetics, 36, 203–209 206 Keyser-Tracqui et al.
Haplotypes, sequences with ambiguities, some of the unique sequences as well as three haplotypes were retrieved from the comparison data.
We ﬁnally conserved 133 haplotypes representing 300 sequences.
This haplotypes analysis revealed that BER10–11–13 perfectly matched 13 sequences present in the computed database: 11 from European or Northern European horses (among them a 1000–2000year-old specimen), one from Eastern Asian horse and one from a Near Eastern horse.
BER02 perfectly matched 12 other sequences present in the database: seven from European horses, one from a Southern American horse, three from Near Eastern horses and ﬁnally one from another Swedish 1000–2000-year old ancient horse.
BER01–03 had the same sequence that three European horses and one Egyptian horse.
BER05–09 shared their sequence with that of a European horse.
BER07 shared its sequence with three horses: a European one, an Anatolian one and a Tuva breed.
BER04–08 matched with 14 sequences: 11 from European horses, two from Near Eastern horses, and one from a South American horse.
BER06 had the same sequence that a Korean horse, a European horse and a Tuva breed.
Finally, BER12 is the only Berel sequence that matched none of the sequences retained in the database and which, like the two Yakut sequences, determined a unique haplotype.
Maximum parsimony and neighbour-joining trees were constructed to address the relationships between the ancient specimens studied and the breeds used for comparison.
As both trees produced similar patterns, only the neighbour-joining tree is represented (Fig. 1).
It shows that the 15 ancient ÔAsianÕ horse’s sequences obtained did not form a separate cluster.
Nine of them were distributed into ` et al. (2001).
BER02, ﬁve of the six clades described by Vila BER04 and BER08 samples clustered in haplogroup D with horses from different breeds (Caspian, Lusitano, Haﬂinger, Shetland, Thoroughbred, Tuva, Egyptian, Fulani, Anatolian, Argentinean-Creole, Peruvian Paso, Asturcon, Arabian, Lipizzan, Shire and Chinese Yunnan).
BER07 clustered in haplogroup B with Cheju, Mongolian, Arabian, Thoroughbred, Connemara, Mallorquina, Caspian and Sorraia breeds.
BER10, BER11 and BER13 were grouped in haplogroup C with horses from ÔmodernÕ breeds (Lusitano, Cleveland Bay, Andalusian, Losino, Argentinean-Creole, Peruvian Paso, Shetland, Connemara, Lipizzan and Exmoor) as well as with ancient ones (Akhal-Teke, Swedish horse from the Late Viking Age and Pleistocene specimens).
BER12, although devoid of mutations C–T at positions 15542 and 15635, was considered to belong to haplogroup A (deﬁned by a transition G–A at position 15666) and clustered with a group composed of Cheju, Anatolian, Lipizzan, Lusitano, Peruvian Paso, Garrano and Sorraia breeds.
YAK01 was found in haplogroup F in the subgroup formed by Korean Cheju, Mongolian, Chinese Guanzhong, Egyptian, Arabian and Lipizzan horses.
BER01 and BER03 as well as BER05 and BER09 had sequences that seemed to 15494 15495 15496 15526 15534 15538 15540 15585 15597 15602 15603 15604 15617 15632 15649 15650 15659 15666 15667 15703 15709 15718 15720 15771 15775 15777 15806 15809 15826 15827 Ó 2005 International Society for Animal Genetics, Animal Genetics, 36, 203–209
Read more about Single : 9 of the total DNA sequence analysed were identiﬁed all….: