A primary objective of this study was to explore whether non-invasive detection methods previously developed for SIV could be adapted to the identification of other infectious agents in endangered primates. We selected SFV infection of wild chimpanzees as a test case for several reasons: First, SFVs can infect humans who come in contact with primates and may thus represent suitable markers of human zoonotic exposure risks 
. Given that chimpanzees are naturally infected with several known human pathogens 
, determining the prevalence and genetic diversity of SFVcpz represented a first step toward examining the utility of this virus as a sentinel for human zoonoses. Second, although seemingly non-pathogenic in natural and non-natural hosts, SFVs could alter the course of SIV and HIV infections since dual SFV/HIV infections have been documented both in sex worker and blood donor cohorts in Africa 
. Thus, screening chimpanzees for both infections provided an opportunity to examine whether SIVcpz and SFVcpz are epidemiologically linked. Finally, foamy viruses are being explored as vaccine and gene therapy vectors for various human diseases 
. It thus seemed prudent to study at least one member of this virus group in its natural host. To this end, we developed new SFVcpz specific fecal detection methods and used these to conduct a large-scale molecular epidemiological survey of wild chimpanzees throughout equatorial Africa. Our results indicate that non-invasive screening strategies can be extended to other infectious agents and show more generally how endangered primates can be studied by non-invasive molecular approaches.
Although both SIVcpz and SFVcpz infected chimpanzees secrete antibodies and nucleic acids into their feces, we found marked differences in the detection sensitivities of these viral markers between the two infections. The most striking difference was the extreme variability with which fecal antibodies and/or viral RNA were detected in SFVcpz infected apes from different communities (). For example, at the TA and MH field sites nearly all SFVcpz infected chimpanzees were antibody positive (Western blot sensitivities of 100% and 92%, respectively), but only very few had detectable viral RNA in their feces (44% and 9%, respectively). In contrast, at the BB and LB field sites nearly all SFVcpz infected chimpanzees were vRNA positive (100% and 80%, respectively), but only very few had detectable antibodies in their feces (8% and 13%, respectively). A comparison of test sensitivities across all field sites indicated that these values were inversely correlated (). To determine whether this was due to a technical artifact, we re-analyzed nearly 300 antibody negative samples (including 74 specimens containing SFVcpz RNA) using newly produced Western blot strips and freshly prepared fecal extracts. Except for 17 weakly reactive samples, all others remained antibody negative. We also analyzed 34 IgG negative fecal samples for the presence of SFVcpz specific IgA. None of these were positive, consistent with the absence of SFVcpz specific IgA in other chimpanzee mucosal compartments 
. Finally, we repeated RT-PCR analysis on a select number of RNA negative samples, but failed to uncover new SFVcpz sequences. Thus, the observed differences in fecal antibody and vRNA detection sensitivities cannot be explained by uneven test performance. Instead, SFVcpz infected chimpanzees appear to shed virus specific antibodies and nucleic acids only intermittently. Whether these fluctuations reflect true temporal differences in fecal antibody secretion and virus replication, or are the consequence of generally lower production levels that sometimes fall below the limits of detection, will require further study. However, in light of the data in , it is tempting to speculate that the observed inverse correlations reflect, at least in part, different stages of recurring SFVcpz superinfection cycles where high titer viral replication at mucosal sites elicits an effective humoral (and possibly also cellular) immune response which reduces fecal viral load until the next infection cycle ensues. Regardless of the underlying mechanism(s), the observed fecal antibody and viral RNA fluctuations are in stark contrast to chronic SIVcpz infection where fecal antibodies are detected at all times with high sensitivity (92%), and where vRNA is amplified from virtually all antibody positive (non-degraded) fecal samples especially when different PCR primer sets are used 
. Thus, a screening algorithm consisting of an initial fecal antibody test followed by RT-PCR of only antibody positive samples (which is the standard approach for non-invasive SIVcpz surveys) is clearly not suitable for molecular epidemiological studies of SFVcpz. Instead, reliable non-invasive SFVcpz prevalence estimates require the use of both vRNA and antibody detection tests.
SFV infection is latent in most tissues, except for lung and tissues of the oral pharynx which express large quantities of viral RNA (up to 104
copies per cell) and thus represent primary sites of SFV replication 
. SFV replication has also been observed in the mesenteric lymph nodes and small intestine of SIVmac infected macaques 
; however, even in these severely immune compromised animals, there was no evidence of SFV replication in the large intestine 
. In light of these data, our finding of SFVcpz RNA in a large number of fecal samples comes as a surprise. Passage through the stomach would be expected to degrade both cell and virion associated SFVcpz RNA. It is thus highly unlikely that the fecal RNA that we observe is produced in the oral mucosa. Instead, it seems more likely that gut epithelial cells represent a primary site of SFV replication, at least at some stage during natural infection. Given the apparent fluctuations in fecal RNA shedding, it is easy to envision how this could have previously gone unrecognized 
. We did not determine the copy number of SFVcpz RNA in the feces and thus cannot estimate how many cell equivalents are required to account for the detected amounts. However, in addition to SFVcpz, we also amplified SFV RNA from a limited number of bonobo, gorilla and mandrill stool samples, all of which were collected in the wild (). It is thus clear that fecal RNA shedding is a common property of this entire group of viruses. It will be interesting to determine whether SFV RNA containing stool samples are infectious. This could explain why some zoo workers and animal handlers who never had direct physical contact with non-human primates were found to be SFV infected 
In addition to its production site, the source of the SFVcpz RNA in stool samples remains a mystery. Unlike in other retroviruses, reverse transcription of the SFV genome takes place during budding and virion assembly, resulting in the production of SFV particles that contain both viral DNA and RNA 
. The viral RNA that we detect may thus derive from cell free virions and/or from mRNA and genomic RNA present in productively infected cells that are sloughed off into the feces. However, since SFV particles often bud at intracellular membranes 
, we would expect to also detect viral DNA. Instead, we found SFVcpz DNA in only 2 of 40 fecal samples from captive chimpanzees, and in none of 173 samples (including 87 SFVcpz RNA positive specimen) from wild chimpanzees. Thus, it remains unknown whether the SFVcpz RNA present in fecal samples is cell-derived, particle-derived, or a combination of both. Given our findings, it may also be of interest to determine whether currently used in vitro
culture systems accurately reflect SFV replication in vivo
Our survey of 25 different chimpanzee communities revealed high prevalence rates of SFVcpz infection across equatorial Africa. This observation, together with the lack of geographic clustering of most SFVcpz strains, and the obvious propensity of SFVcpz to superinfection and recombination, indicates that SFVcpz is a highly transmissible virus. Previous studies have indicated horizontal routes as the primary mode of SFV transmission 
. Our findings in Gombe National Park are consistent with these observations. The fact that we detected SFVcpz in each of 13 adult chimpanzees, but in only 3 of 14 infants and juveniles indicates a clear increase of SFVcpz prevalence with age. In addition, we found no conclusive evidence for perinatal transmission. Two of the three infected offspring were SFVcpz negative at the time of first analysis, and the third one harbored a virus that was genetically indistinguishable in the pol
-IN region from viruses infecting unrelated chimpanzees. Thus, perinatal transmission of SFVcpz, if it occurs at all, appears to be uncommon in wild-living chimpanzees. Instead, chimpanzees appear to acquire SFVcpz by horizontal routes, most likely by exposure to saliva (or feces), as has been proposed for other primates 
. Indeed, young chimpanzees stay with their mothers until they are 8 or 9 years old and often share food. Thus, infants and juveniles are frequently exposed to their mother's saliva, which may constitute a common source of infection. In contrast, SIVcpz appears to be transmitted primarily by sexual (and sometimes perinatal) routes (
; Keele et al., unpublished). In light of these differences, the absence of an epidemiological link between SIVcpz and SFVcpz infections is perhaps not too surprising. Examining seven different communities, we found no indication that infection with one of these viruses increased or decreased the likelihood of infection by the other.
Simian foamy viruses are believed to have co-evolved with their respective primate hosts for millions of years 
, and our finding of subspecies-specific SFVcpz lineages is consistent with this hypothesis. Remarkably, all of the 120 newly characterized SFVcpz strains clustered according to their subspecies of origin. This included one strain from a site (WE) just north of the Sanaga River (i.e., within the range of P. t. vellerosus
) infecting an individual with P. t. troglodytes
mtDNA, indicating gene flow, but not viral flow, across a subspecies boundary. This monophyly of SFVcpz strains from each subspecies contrasts with the mtDNA phylogeny where P. t. schweinfurthii
sequences lie within the P. t. troglodytes
radiation. While the validity of classifying chimpanzees into subspecies has been questioned 
, the SFVcpz phylogeny corroborates the existence of four geographically isolated chimpanzee populations and the absence of SFVcpz transmission between subspecies argues that they are effectively separated, especially since such transmissions are frequently observed in captive settings (e.g., see DEB and MUS SFVs in ). The SFVcpz and mtDNA phylogenies (, , , S1
) differed with regard to the relationships among the four subspecies. However, these differences do not undermine the co-evolution hypothesis. When successive speciation events occur over a relatively short timescale, persistence of polymorphism from one event to the next means that any one genetic marker may not have the same phylogeny as the species 
; this phenomenon is even more likely with recent subspeciation events. Thus, even if there has been complete co-evolution of SFVcpz with chimpanzees, discordance between the SFVcpz and mtDNA phylogenies may appear because either, or both, differ from the true historical relationships among the subspecies. In fact, the apparently shorter coalescence time of SFVcpz indicated by the reciprocal monophyly of P. t. troglodytes
and P. t. schweinfurthii
viruses suggests that SFVcpz could be less susceptible to this problem than mtDNA. Thus, SFVcpz may emerge as a more sensitive marker of population structure that may be useful for chimpanzee systematics as well as conservation strategies.
Phylogenetic analyses identified discordant branching orders for several SFVcpz strains, suggesting co-infection or recombination 
. To examine whether this was indeed the case, we selected a subset of samples for repeat RT-PCR analyses, including single genome amplification (SGA) of re-extracted fecal viral RNA. SGA amplifies single viral templates, is not subject to Taq
polymerase induced nucleotide substitutions and recombination, and thus provides an accurate representation of the viral population in the individual 
. Adapting this approach to fecal RNA provided new insights into SFVcpz biology. SGA analysis formally documented infection with more than one virus in two chimpanzees. One of these apes (MF1279) was infected with two distinct SFVcpz strains, while the other (DP157) harbored at least four genetically diverse viruses. In both cases, predominant viral forms were identified by bulk RT-PCR (red in ), but SGA was required to characterize the full extent of viral diversity in the sample, including the relative proportion of different variants. Repeat RT-PCR and SGA analyses also documented mosaic genome structures in several SFVcpz strains and demonstrated that these did not represent PCR artifacts. Although preliminary, these results suggest that superinfection and recombination occur rather frequently. As mentioned above, successive superinfection cycles may account at least for some of the observed fluctuations in fecal antibody and viral RNA detection in different chimpanzee communities. It will be interesting to test this hypothesis in chimpanzees from Gombe National Park where longitudinal samples from SFVcpz infected apes are available.
Because they are avid hunters, chimpanzees are also frequently exposed to SFV strains from other primate species. Testing 392 fecal samples for SFVcpz viral RNA, we found one male chimpanzee to harbor an SFV strain (LB309) that was closely related to viruses previously identified in captive DeBrazza's and mustached monkeys (). The finding of LB309 RNA indicated a productive viral infection in the chimpanzee host. Similar findings were recently reported for chimpanzees from the Taï Forest where 3 of 12 apes studied harbored SFV strains from sympatric western red colobus monkeys 
. Interestingly, these apes (all males) were also coinfected with SFVcpz; however, it was not determined whether the dual infections were productive since viral DNA (and not RNA) sequences were amplified from spleen necropsy specimens using strain specific PCR primers 
. Since we did not use strain specific primers, it is likely that our data grossly underestimate the frequency of SFV cross-species transmission in the wild. Moreover, the failure of these cross-species infections to initiate secondary spread suggests that their replication (and thus fecal detection) may be limited. However, the examples demonstrate that chimpanzees, like humans, are susceptible to SFVs from other primate species, and the fact that all cross-infected apes were males (who hunt more frequently and eat more meat than females) strongly suggest that these transmissions occur in the context of predation. These findings may be of use to primatologists interested in chimpanzee hunting behavior and prey preferences in the wild.
Finally, SFVs are of public health interest because people in sub-Saharan Africa are routinely exposed to these viruses in the context of primate bushmeat hunting 
. We show herein that SFVcpz infection is highly prevalent in wild chimpanzee populations throughout their natural range. Thus, monitoring humans for SFVcpz infection should be informative as to the locations where human/chimpanzee encounters are most frequent and where additional cross-species transmissions should be anticipated. One such area is southern Cameroon where chimpanzees are endemically infected with SIVcpz strains that have already crossed the species barrier to humans, in one case (HIV-1 group M) with devastating consequences 
. Screening humans for SFVcpz infection may also provide new insight into the environmental circumstances that underlie cross-species transmissions. For example, if the frequency of human SFVcpz infection were significantly lower in east compared to west central Africa, this would argue for lower exposure rates and, in turn, provide a reason why SIVcpz strains from P. t. schweinfurthii
apes have not emerged as human pathogens. Thus, human SFVcpz infection should be formally investigated a sentinel for ape-derived pathogens, including new SIVcpz/HIV-1 outbreaks.