We found a high seroprevalence of SFV in a semi-free-ranging colony of mandrills originating from and living in Gabon, central Africa. The habitat of mandrills is restricted to western central Africa, which is highly endemic for other retroviruses, such as SIV and STLV [42–47]. A seroprevalence of 89.5% was found in a small free-ranging macaque population (mostly adults) living in a temple in Bali, Indonesia, with a higher prevalence in adults than in juveniles [18, 31, 39]. A larger study provided evidence that Macaca tonkeana acquire SFV mainly through severe bites, mainly when young adults aged 5-8 years compete for sex partners . In a study of free-ranging colonies of chimpanzees, Liu et al. found a significant increase in SFV infection with age, with no evidence of vertical transmission to the young . In our study, there was a clear increase in SFV infection at 4-5 years of age. Altogether, these findings indicate horizontal rather than vertical (perinatal) transmission as the predominant route of SFV infection in these nonhuman primate communities. Nevertheless, some species or colony specificity may be found in natura among troops of nonhuman primates, which might change the relative importance of different modes and thus the timing of SFV transmission.
It is known that a similar virus can be transmitted quite differently in different nonhuman primate species: STLV-1 appears to be acquired mainly in breast milk in M. tonkeana  but is acquired mainly in adulthood in chimpanzees [18, 33, 49]; in mandrills, it is probably acquired through bites [42, 46–48, 50] and to a lesser extent by sexual contact, and a predator-prey system may sometimes be also involved . In our mandrill colony, about 50 animals were SFV-positive at the age of 1 year, perhaps due to exchange of saliva with their mother during feeding. It was reported recently that mandrills have a prominent muzzle-muzzle behaviour, usually between young naive and older individuals [51, 34–44]. It has also been reported that salivary glands are the major reservoir of SFV replication in monkeys [23, 26, 29]. We did not observe any difference in seroprevalence according to the sex of the animals. SFV seroprevalence increased significantly with age. These findings are similar to those on the seroprevalence of STLV-1 in this colony, which was evaluated at 13.4% .
Our study indicates that all except one integrase sequence of the SFV strains circulating in the colony are closely related, and some are identical. The probable explanation is related to the history of the colony, which was founded in 1983 with only a few animals, some of which probably harboured a virus originating from northern Gabon. The virus was therefore transmitted and spread in the colony during the past 25 years by the founders from the northern part of the country. Ten different strains are circulating in the northern group, with 96-99% sequence similarity. Similar observations have been made with regard to the circulation of several strains in other nonhuman primates, including monkeys and apes [2, 17, 27, 53].
The animal that harboured the eleventh strain circulating in the colony, which is quite different from the other strains, was a wild-born mandrill brought to the Primate Centre in 2003 from the southern part of the country at the age of 2 years. It was kept in quarantine for 6 months and then introduced into the mandrill colony. Dissemination of the virus could occur in several ways, as indicated above, but also because one of the infected mandrills is a dominant male in the colony. This hypothesis cannot, however, be confirmed, since no sample was available from the first mandrills introduced into the colony.
Our finding that two different strains exist in the colony suggests that mandrills living currently in northern and southern Gabon are infected by two different SFV strains. Similar situations have been reported for two other retroviruses that infect these monkey species, SIV  and STLV-1 . As seen in Figure 3 the cytochrome b study showed that most of the mandrills are from the south but are infected with a SFV strain from the north. This suggests that they were infected in the breeding colony by a SFV virus from a mandrill originating from the north (Figure 2), except for mandrill 31 (see above). In contrast, the origin of each wild mandrill (Figure 3) was concordant with the virus they harboured (Figure 5), confirming infection in their natural area. Furthermore, studies of cytochrome b polymorphism suggest that the Ogooué River separates mandrill populations into two different phylogenetic groups: one in the north (northern Gabon and Cameroon) and the other south of the River (southern Gabon and Congo) .
Monkeys have a long co-existence with their SFV [2, 24, 28, 32, 33, 53, 55, 56], which would have started when mandrills in both the north and the south had a common ancestor and has persisted since their separation, about 800 000 years ago . These results for SFV infection in mandrills are supported by the fact that the same mandrills are infected with SIV  and STLV . Our analysis of the results for 15 wild mandrills caught in the northern and southern parts of Gabon clearly indicates the existence of two different variant strains of SFV. The discrepancy in our study between serological data and the absence of the SFV sequence in mandrill PBMCs may be due to a low viral load in blood samples. In some juveniles, it could be the result of high levels of maternal antibodies against SFV .
We also found that two of 20 people working at the Primate Centre were infected with SFVs: one with a mandrill strain and the second with a macaque virus. Only about 50 people worldwide have been shown to be SFV-infected (both serologically and molecularly) [13, 14], including people occupationally exposed to nonhuman primates [12, 25] and people at risk in natural settings, such as hunters in central Africa [15, 16]. Furthermore, only three other human infections with mandrill SFV have been reported. In the first case, a hunter living in Cameroon was found to be infected by a mandrill strain, but the route of infection was not documented . The second case was in a blood donor in Cameroon, also with no information on the route of infection . In the third case, a man aged 26 years had been bitten by a small monkey while hunting 1 year before the presence of mandrill SFV was found . We demonstrated the identity of the viral foamy strain in the donor (Mnd2ACDP) by molecular sequencing at the time of the bite that probably transmitted the virus, and in the human recipient 10 years later, with 99% similarity between the two sequences. This person had been bitten only once by mandrill Mnd2ACDP and not by other mandrills. The presence of a sequence from the clones of H1CIRMF (CIRMF1C9) among clone sequences from Mnd2ACDP, particularly Mnd2AC10Y0 (Additional file 1), sustains the hypothesis of the origin of H1CIRMF virus from Mnd2ACDP. No close sequence similarity was found between the H1CIRMF sequence and the three other sequences previously found in humans infected by a mandrill SFV [15, 16, 34] (Figure 6).
Only one molecular demonstration of SFV interspecies transmission has previously been reported, due to a bite by a chimpanzee to a zoo worker . Although the person infected by the mandrill virus in our study had also been bitten during his professional activity by a chimpanzee, we were unable to detect any chimpanzee SFV sequence in his PBMCs. 'Dual' risks with only one virus detectable by PCR have also been reported in hunters in south Cameroon . Co-infection with two different simian viruses was demonstrated recently in chimpanzees infected not only with their own chimpanzee SFV, but also with a Colobus strain . The second human was infected with a strain related to a macaque SFV. Despite the use of thousands of macaques in biomedical research, primate facilities and institutions for decades (in both Europe and North America), only one case of human infection with a macaque foamy virus has been reported (in a worker in Canada after a severe bite) . In contrast, recent studies in Asia showed transmission of macaque SFV to nine people, including zoo workers, owners of nonhuman primate pets, 'bush meat' hunters and temple workers [17, 18]. Mathematical modeling shows that, in Bali, about six of every 1000 visitors to monkey temples will be infected with SFV .
In our work, we also observed high stability of the integrase sequence of SFV over time (10 years in an infected mandrill as well as in an infected human), with neither genetic drift over time nor the presence of quasi-species. Foamy viruses are genetically very stable  and, with the exception of cross-species transmissions, have co-evolved with their hosts . Their high genome conservation often allows attribution to a particular monkey or ape subspecies through analysis of the appropriate foamy virus sequence [27, 32, 33]. Furthermore, in cross-species transmission to humans or apes, the transmitted virus can be easily traced back to the transmitting monkey species and appears to be genetically stable in the new host for decades [53, 58, 59].
In conclusion, we have shown that SFV is highly endemic in mandrills in Gabon, and this virus can be transmitted to humans. Further studies are being conducted to evaluate the prevalence of this virus in larger samples from various monkey species in central Africa. We are also studying the natural transmission of these viruses to human populations living in this geographical area, where consumption of 'bush meat' and hunting are common.