In this study, we used parasitological, serological, and molecular tools, combined to geographical methods, to quantify the distribution and abundance of human and animal hosts, the vector, and the trypanosomes, in order to provide a comprehensive understanding of the epidemiological situation in the Guinean HAT focus of Boffa. The results show that the human population is young, very mobile, and that although T. b. gambiense is present and actively transmitted to humans in Boffa, it has not been found in tsetse, but has been suspected in two domestic animals, a pig and a goat. Remarkably, no pathogenic trypanosomes to animals have been observed in tsetse, nor in the 300 domestic animals sampled. Put together, and given the impossibility of eliminating T. b. gambiense by medical activities only, the results suggest that an integrated control strategy based on medical surveillance plus vector control will be required to protect people from HAT.
Human settlement, landscape and HAT
The settlement and the landscape of Boffa focus are very similar to those which can be found on other parts of the Guinean littoral. From the Mellacoree river mouth in the south, at the border with Sierra Leone, to the Rio Compony river mouth in the north, near the border with Guinea Bissau, the Guinean littoral is characterized by a mangrove ecosystem, except for the headlands of Conakry and Cape Verga 
and the important HAT foci in Guinea (Boffa, Dubréka and Forécariah) are all located in or around this mangrove ecosystem 
. Although the population of the littoral is multiethnic (Nalou, Baga, Temne ethnic groups), the Soussou group is clearly the most important. These populations conduct several activities in the mangrove that expose them to the bite of tsetse flies and so to the risk of getting HAT. In a recent study of Forecariah focus, in the southernmost part of the Guinean mangrove, the identified risk activities were rice cultivation, water supply at backwater and use of pirogue jetties 
. In Boffa, high human mobility between the mainland and the islands is certainly an important factor that increases the risk of transmission from infected tsetse to humans, and that favors the spread of the disease. First, because this mobility is probably responsible for high human/tsetse contact at pirogue jetties and in narrow mangrove channels (highly human- frequented places with high tsetse densities), and second, because this mobility also contributes to a low attendance of population to medical surveys, in particular the active part of the population, which is also the most at risk. In many instances, when the medical survey team arrived in a village, where important sensitization always came along census campaign that preceded medical survey, many of the people were not present, simply because their activities in the mangrove were more important for them. It has to be noted, as part of the explanation, that these mangrove activities, have more constraints than classical agricultural ones on the mainland. For instance, people have to take into account tide hours as a first priority for their movements. These results can certainly be generalized to other HAT foci of the African coast with mangrove habitat, e.g. Equatorial Guinea, Gabon, etc.
Tsetse distribution and population genetics
Considering tsetse distribution, two different types of habitats can be distinguished in the area: The northern and the southern ones. The northern one, on the mainland, looks like the classical distribution of the riverine species G. palpalis gambiensis
in humid savannahs such as northern Ivory Coast, Burkina Faso, Mali, with tsetse being strongly associated to the forest galleries bordering water courses, and being almost absent from the savannah itself (e.g. 
). Here the highest densities are indeed found along the rivers, and near sources which offer conserved vegetation and suitable host abundance. On the opposite, in the southern part, when approaching the coast and mangrove channels, tsetse become progressively almost evenly distributed, with highest densities in narrow mangrove channels with Rhizophora
vegetation, which represent hunting areas for tsetse, and also near natural watering points and pirogue jetties frequented by humans.
Given the absence of genetic differentiation observed between the tsetse sampled using 9 independent microsatellite loci, it is likely that Boffa provides a habitat for a single big panmictic tsetse population, which was a priori
questionable given the wideness of Rio Pongo River (sometimes more than 1 km) which could have acted as a barrier between the two banks of the river. This confirms earlier observation on another sleeping sickness focus of the Guinean littoral, Dubreka, where mainland mangrove was also found to host a single large and panmictic tsetse fly population 
. With such results, tsetse eradication at the scale of the focus of Boffa is clearly not the way to go because it is impossible to target the whole tsetse population, hence emphasis will be put to reduce (i.e. not eradicate) tsetse densities in order to reduce tsetse/human contact and stop T. b. gambiense
Epidemiology of HAT in the focus of Boffa
The medical survey results confirmed the endemic situation in the Boffa focus with an overall prevalence of 0.3% in humans. Transmission of T. b. gambiense probably occurs mainly at the mangrove/savannah interface which is believed to be the main sites of human contact with tsetse flies. The active population (20–45 years old) is by far the most affected. This part of the population comprises 70% of the cases detected, although only 40% attended the medical survey. Transmission is probably linked to high human tsetse contacts occurring during human activities conducted in the mangrove, mainly around sites such as pirogue jetties, water supply points, and mangrove channels where many tsetse were captured. These sites will constitute priority targets for any vector control operation aiming at reducing tsetse/human contact.
The proportion of patients being in the 1st stage of the disease (39%) is greater than in other foci in littoral Guinea, since this proportion was only 2% in Dubreka 
, and 16% in Forecariah 
. This suggests that in Boffa, HAT transmission is still active, and this underlines the pressing need for active intervention in this focus. Despite important efforts to sensitize the communities, only half of the whole population attended the medical survey and was therefore medically screened. This attendance even decreases to 40.29% when considering only the active population, which represents the majority of the cases (69%). Given that the overall prevalence was 0.3%, and assuming the prevalence is the same in the non screened population (which is very conservative), we therefore expect around 60 more HAT cases to be still living in the area. Moreover, application of TL proved that an additional 14 SERO-TL+ individuals (at least) indeed had a contact with T.b. gambiense
and should be considered as potential additional carriers 
. They confirm once more (see also 
) the potential impact SERO subjects can have in the maintenance of transmission in HAT foci, especially with HAT control strategies targeting HAT patients only. In the absence of any prophylactic treatment against sleeping sickness, this number of T. b. gambiense
carriers is still growing since transmission is active. This emphasizes the need for intervention(s) that would overcome this constraint.
The presence of two 10 months old patients, the mother of whom was HAT diagnosed the year before (M.C., unpublished data), represents an additional indirect evidence for the anciently suspected existence of vertical T. b. gambiense
transmission (reviewed in 
Characterization of trypanosomes in the different hosts
The absence of pathogenic trypanosomes in domestic animals was not expected, but is very likely given both their absence using parasitology (BCT method) on more than 300 animals sampled, and also given the high mean PCV values found. These results suggest that animal trypanosomiasis is not a major veterinary problem in this area. The same result had been found in Loos islands, near Conakry, where no domestic animal had been found infected with trypanosomes out of 104 sampled 
The fact that none of the infected tsetse had trypanosomes identified using PCR suggests that the trypanosomes circulating may come from reptiles for instance, such as T. grayi for which we did not do PCR. Reptiles (crocodiles, monitor lizards) are numerous in this mangrove area. But it may also suggest that unrecognised, possibly pathogenic, trypanosome species exist that were not identified. This can be particularly evoked for the two midgut+proboscis infections which would have been interpreted as a Nannomonas infection if only parasitological methods had been used.
Hence the only pathogenic trypanosome identified in Boffa focus is T. b. gambiense
, its presence having been confirmed in humans, and in two domestic animals thanks to the trypanolysis test which is specific for T. b. gambiense
. T. b. gambiense
has not been found in tsetse, confirming the usual (but poorly understood) very low (<0.1%) mature infection rates of T. b. gambiense
in tsetse, even in active sleeping sickness foci (
. The same has been reported for T. b. rhodesiense
To summarize, our results suggest the absence of pathogenic trypanosomes in domestic animals in the focus of Boffa. We show the presence of T. b. gambiense
in humans, and a contact between T. b. gambiense
and some domestic animals (one pig and one goat in our study). The vectorial capacity of G. p. gambiensis
in this focus seems very low, confirming what was found on another area of the Guinean littoral, the Loos islands, where Kagbadouno et al.
also reported an absence of trypanosome in the G. p. gambiensis
dissected. This is in contrast with other areas where G. p. gambiensis
occurs, such as the savannah areas of Burkina Faso or Mali for instance, where G. p. gambiensis
can usually be found infected with animal trypanosomes, with infection rates generally ranging from 2 to 10% 
. The G. p. gambiensis
from littoral Guinea may also be genetically different from the one from the West African savannah, this being currently investigated (P.S., unpublished data). Nonetheless, even a very small number of tsetse having a T. b. gambiense
mature infection are able to spread these trypanosomes, since the infection by T. b. gambiense
alters tsetse saliva and modifies the behaviour of the tsetse, favoring the risk of human infection 
. The “good news” here is that, given the very low proportion of tsetse infected with T. b. gambiense
, a human needs to be bitten a great number of times before being bitten by an infective tsetse. Hence, any intervention that will reduce tsetse numbers and tsetse/human contact will also reduce the number of tsetse infected, and will protect people from an infective bite.
Interventions directed against animal trypanosomiasis have been advocated as a useful entry point for controlling HAT 
. However, in the littoral HAT foci of West Africa, and in many Central African foci, livestock are scant and/or trypanosomiasis is not an important constraint for livestock production. In these mangrove foci, interventions should be applied at a relatively small scale and have to be aimed specifically on eliminating HAT.
In conclusion, the prevalence of pathogenic trypanosomes in the human population, combined with low attendance at medical surveys and to an additional population of human carriers of T. b. gambiense
living in the community, highlights the importance of implementing new strategies. We suggest that in order to stop T. b. gambiense
transmission in Boffa and similar foci in West and Central Africa, vector control should be added to the current strategy of case detection and treatment. Such an integrated strategy of control will combine medical surveillance and vector control activities, the former finding and treating cases while the latter will protect people from the infective bites of tsetse. The recent development of insecticide-treated targets that are effective for G. p. gambiensis
offers the prospect of a method that can be applied in the mangrove systems. In addition there is an absolute need to follow up seropositive people, and to target more efficiently the population at risk if HAT is to be eliminated from the focus.