The Epidemiology and Geographic Distribution of Relapsing Fever Borreliosis in West and North Africa, with a Review of the Ornithodoros erraticus Complex (Acari: Ixodida)

Background Relapsing fever is the most frequent bacterial disease in Africa. Four main vector / pathogen complexes are classically recognized, with the louse Pediculus humanus acting as vector for B. recurrentis and the soft ticks Ornithodoros sonrai, O. erraticus and O. moubata acting as vectors for Borrelia crocidurae, B. hispanica and B. duttonii, respectively. Our aim was to investigate the epidemiology of the disease in West, North and Central Africa. Methods And Findings From 2002 to 2012, we conducted field surveys in 17 African countries and in Spain. We investigated the occurrence of Ornithodoros ticks in rodent burrows in 282 study sites. We collected 1,629 small mammals that may act as reservoir for Borrelia infections. Using molecular methods we studied genetic diversity among Ornithodoros ticks and Borrelia infections in ticks and small mammals. Of 9,870 burrows investigated, 1,196 (12.1%) were inhabited by Ornithodoros ticks. In West Africa, the southern and eastern limits of the vectors and Borrelia infections in ticks and small mammals were 13°N and 01°E, respectively. Molecular studies revealed the occurrence of nine different Ornithodoros species, including five species new for science, with six of them harboring Borrelia infections. Only B. crocidurae was found in West Africa and three Borrelia species were identified in North Africa: B. crocidurae, B. hispanica, and B. merionesi. Conclusions Borrelia Spirochetes responsible for relapsing fever in humans are highly prevalent both in Ornithodoros ticks and small mammals in North and West Africa but Ornithodoros ticks seem absent south of 13°N and small mammals are not infected in these regions. The number of Ornithodoros species acting as vector of relapsing fever is much higher than previously known.


Introduction
Relapsing fever has long been recognized as major cause of disease and death in several regions of Africa [1][2][3][4][5]. Initially discovered in India in 1907 [6], the responsibility of Pediculus humanus in the transmission of Borrelia recurrentis, the Obermeier's spirochete, was confirmed the following year in Algeria [7]. Dramatic epidemics of louse-borne relapsing fever (LBRF) responsible for several millions of cases and a high fatality rate occurred throughout Africa after World Wars I and II when French and British colonial soldiers infected in Europe or North Africa returned to their countries [8][9][10][11][12]. More recently, several epidemics occurred in Sudan and LBRF still persists in the mountains of Ethiopia where it is endemic and can account for up to 27% of hospital admissions [5,[13][14][15][16]. Tick-borne relapsing fever (TBRF) was first recognized in East Africa in 1904 [17,18], in North Africa in 1928 [19], and in West Africa in 1932 [20]. Studies conducted between 1905 and 1960 progressively established the classical picture of TBRF in Africa, with three different vector/pathogen complexes involving soft ticks (Argasidae) of the genus Ornithodoros. In the savanna areas of eastern and southern Africa, from 16°N (Erythrea) to 34°S (Cape province, South Africa), TBRF is caused by Borrelia duttonii with O. moubata s.l. as vector [21]. In the wild, the two vector species of the O. moubata complex (O. moubata s.s. and O. porcinus) lives in large animal burrows of antbears, warthogs and porcupines [22]. They have adapted secondarily to human dwellings and domestic animal shelters, where they live in the cracks of walls and floors. There is no known mammal reservoir of the disease and O. moubata s.l. act both as the vector and the only reservoir of B. duttonii [23]. The annual incidence of the disease may reach up to 384 per thousand among children <1 year of age and 163 per thousand in children <5 years of age in Tanzania [24], and a lethality rate of 16% among pregnant women suffering from the disease has been reported from Rwanda [25]. In North Africa, from Morocco to Algeria and Tunisia, TBRF is classically caused by Borrelia hispanica with O. erraticus as vector and small mammals as reservoir host [2,[26][27][28][29][30][31]. O. erraticus is found both in large and small burrows, under stones, and has adapted to domestic animal shelters. Most human infections occur during summer among people sleeping in the fields or in farm buildings [32]. In West Africa, TBRF is caused by Borrelia crocidurae, with O. sonrai as vector and rodents and insectivores as reservoir host [33][34][35][36][37]. Most reports of the disease are from Senegal, but the few available data on the occurrence of the vector also include Mali, Mauritania, southern Morocco, Niger, Chad, Egypt and Kenya [23]. O. sonrai inhabits rodent burrows and like the other Ornithodoros species, it is a rapid feeder, blood meals lasting only a few minutes. People are generally infected during their sleep, when burrows open into their bedrooms [36].
The high incidence of TBRF in places where it was specifically investigated contrasts with the rarity of reports from other areas, suggesting than most cases remain undiagnosed because usually confused with malaria whose diagnosis is usually based only on clinical symptoms [38]. In Senegal, as a result of the ongoing drought since the 1970s, the tick has colonized the Sudan savanna, with the 750 mm isohyet as southern limit [36]. Rodent burrows colonized by O. sonrai occur in 87% of villages north of 13°30'N. In these areas, we have shown that the average incidence of TBRF at the community level is the highest described for any bacterial disease, reaching 11 per 100 person-years [39]. In out-patients clinics, TBRF is the second most common cause of fever after malaria [38]. These observations in Senegal have prompted us to investigate the distribution and epidemiology of TBRF in West, North, and Central Africa. Here we present the result of studies conducted between 2002 and 2012 in 17 African countries.

Ticks sampling
To investigate the geographic distribution of Ornithodoros ticks, we conducted four series of studies i) One transect study along the 14th parallel in Senegal, Mali, Burkina Faso, Niger and Chad, from 16°W (near the Atlantic coast of Senegal) to 22°E (near the border of Sudan in Chad). Sampling was conducted at each two degrees of longitude (i.e. at 16°W/14°N, 14°W/14°N, 12°W/14°N, and up to 22°E/14°N), either at the exact meridian/parallel junction point as determined with a Global Positioning System (GPS) receiver, or depending on the accessibility of this point and local environment, within a 10' radius around it (i.e. maximum distance: 20 km). The 14th parallel was selected since it was entirely located within the limits of the supposed range of O. sonrai (presumed southern limit: 750 mm isohyet, period 1970-2002) [36].
ii) Three North/South transect studies along three meridians: 12°W (in Morocco, Mauritania, Senegal, and Guinea), 2°E (in Mali, Niger, Burkina Faso and Benin), and 14°E (in Chad and Cameroon). Sampling was undertaken at each degree of latitude, from 10°N to 28°N, 6°N to 19°N, and 8°N to 14°N, respectively. In Morocco, the transect was further continued SW/NE up to the Mediterranean coast.
iii) Based on the results of the transect studies, additional surveys were conducted in Senegal, The Gambia, Mauritania, Mali and Niger in order to determine either on a 10 km square scale (Western Senegal) or on a half or quarter square degree scale the southern and eastern limits of the geographic distribution of Ornithodoros species. iv) We also undertook additional surveys in selected areas of Morocco, Algeria, Tunisia, Niger, Guinea, Guinea Bissau, Liberia, Ivory Coast, Burkina Faso and Togo. All surveys were conducted in predefined sites based on their geographic position. One site south of Spain, where the first studies on TBRF in Europe were conducted in the 1920s by Sadi de Buen [32], was also sampled for comparison of local tick population with African Ornithodoros species.
For each sampling site, as a general rule 30 to 60 burrows were investigated (60 burrows during transect studies when all burrows were negative for the presence of Ornithodoros ticks, 30 burrows when one or more burrows were positive; only 30 burrows even when all were negative during additional surveys), except in sites rapidly positive during additional Tick-Borne Borreliosis in West and North Africa PLOS ONE | www.plosone.org surveys in North Africa where only 5 to 15 burrows were investigated. Ticks were collected by introducing a flexible tube inside burrows and aspirating their contents using a portable petrol-powered aspirator. After collection, they were immediately stored in absolute ethanol (one tube per positive burrow) until morphological determination and DNA extraction. As a general rule, two or three sampling stations were selected in each site, including one village (except in most Saharan and North African sites) where several houses were surveyed for the occurrence of ticks both inside burrows and in cracks in the floor or walls. Outside villages, sampling stations were selected according to local environment either in natural or humanimpacted ecosystems and croplands or both of them.

Small mammals sampling
To investigate the reservoir of Borrelia, we captured rodents and insectivores in Senegal, Mali, Benin, Niger, Chad, Cameroon, Mauritania and Morocco. These collections were made along 12°W, 2°E and 14°E meridians and in further sites in Senegal, Mali and Morocco. Animals were trapped alive with lattice-work traps baited with peanut butter or onions. We adopted the method of trap lines for captures, with a 10-meter space between traps. In order to collect both diurnal and nocturnal species traps were placed in the afternoon and withdrawn late in the morning. Approximately 200 traps/days of captures were made in each site. In addition, hand captures of rodents were performed by night in most sites and during travel between sites. We necropsied trapped animals in the field by means of cervical dislocation and drew 1mL of blood from each by cardiac puncture. We immediately prepared a thick blood film for detection of Borrelia. Samples of brain were stored in nitrogen then at -80°C for further intraperitoneal inoculation to white mouse as previously described [40] and/or Borrelia molecular studies as described below.
The nomenclature here adopted follows Wilson & Reeder [41] unless otherwise mentioned [42]. Classical body measurements were taken and some individual rodents were kept alive for further chromosomal analyses, as described in Granjon & Dobigny [43]. Most of the rodents of the Sahelo-Sudanian region could be diagnosed to the species level based on our morphological, ecological and biogeographical knowledge on rodents from this geographic area [42]. However, molecular or chromosomal data were necessary to confirm unambiguously the specific determination of some specimens of the genera Arvicanthis, Gerbillus, Gerbilliscus, Mastomys, and Taterillus especially. This was done according to methods previously described [43][44][45][46][47]. Shrews were determined on the basis of external and skull criteria [Granjon et al, unpublished]. A selection of voucher specimens was kept in formalin or ethanol and is housed at IRD (UMR 22, Dakar and Montpellier). Organ samples as well as DNA and bone marrow cells extracts that have been used for molecular/ cytogenetic studies are also kept in the IRD tissue collection.

DNA isolation and PCR amplification in ticks
Ticks were individually washed in three sterile water baths, air dried and collected in sterile microtubes. DNA was individually crushed by shaking with a bead beater (mixer mill MM301, Qiagen, Hilden, Germany), and then DNA was isolated and purified using the DNeasy Blood and Tissue extraction Kit (Qiagen).
For each tick, 16S rRNA was amplified by PCR with Tm16S +1 (5'-CTGCTCAATGATTTTTTAAATTGC-3') and Tm16S-1 (5'-CCGGTCTGAACTCAGATCATGTA-3') primers designed by Fukunaga et al. [48]. The sequenced product size was around 450 bp long and PCRs were performed with the Multiplex PCR Kit (Qiagen) in a 25 µl volume containing 12.5 µl of 2x Multiplex PCR Master Mix (Qiagen), 2 µM of each Primer, 2.5 µl of Qsolution and 4 µl (40-100 ng/µL) of DNA template. The amplification cycle involved a denaturation step at 95°C for 15 min followed by 10 cycles of 1 min at 92°C, 1.5 min at 48°C, and 1.5 min at 72°C and 32 cycles of 1 min at 92°C, 1.5 min of 54°C, and 1.5 min of 72°C. A final extension step was carried out for 10 min at 72°C. The amplified products were detected by electrophoresis on 1.5% agarose gel in TAE 0.5X buffer and staining with Envision (Amaresco). Remaining reaction mixtures were stored at -20°C for direct sequencing.

Detection of Borrelia infections in ticks and small mammals
DNA isolation was conducted as above. Borrelia detection was based on nested PCR amplification of a 350 bp fragment of the flagella gene (FLA). This gene encodes the periplasmic protein peculiar to Borrelia. The amplification was performed using primers (Bfpad and Bfpdu for the first PCR, Bfpbu and Bfpcr for the second PCR) designed for B. duttonii [48]. Each PCR was performed in a 25 µl volume containing 5µl 5X buffer (Promega), 2 mM MgCl 2 , 200 µM of each dNTP, 0.2 µM of each primer and 2.5 unit of GoTaq DNA polymerase (Promega). 3µl of DNA template was added in the first reaction and 1 µl of the first amplified product was added in the second reaction. Amplification cycles consisted of an initial DNA denaturation step at 94°C for 3 min followed by 30 cycles of 40 sec at 94°C, 40 sec at 55°C for the first PCR and 51°C for the second PCR, and 40 sec at 72°C. A final extension step was carried out for 10 min at 72°C.
For the 16S-23S ribosomal RNA intergenic spacer (IGS), rrs rrlA intergenic spacer IGS/F and IGS/R for the first PCR and rrs rrlA IGS/Fn and IGS/Rn for the second PCR were amplified as previously described [49]. Each PCR was performed in a 25 µl volume containing 5µl 5X buffer (Promega), 2 mM MgCl 2 , 200 µM of each dNTP, 5 picomoles of each primer and 2.5 unit of GoTaq DNA polymerase (Promega). 2µl of DNA template was added in the first reaction and 1 µl of the first amplified product was added in the second reaction. Amplification cycles consisted of an initial DNA denaturation step at 94°C for 3 min followed by 35 cycles of 30 sec at 94°C, 30 sec at 56°C, 30 sec at 72°C and a final extension 5 min at 72°C for the first PCR. For the second PCR, amplification cycles consisted of an initial DNA denaturation step at 94°C for 3 min followed by 40 cycles of 30 sec at 94°C, 30 sec at 60°C, 30 sec at 72°C and a final extension 5 min at 72°C for the first PCR. A final extension step was carried out for 5 min at 72°C. The amplified products were detected by electrophoresis as above. Negative control was included in each nested PCR analysis to monitor contamination and false-positive amplification. To rule out amplicon carry-over, nucleotide-free water negative control was used throughout the steps of the protocol. The sequence-derived data reported herein were authentified as negative controls introduced in every PCR experiment remained negative, excluding the possibility of cross-contamination during the experiments. The amplified products for 16S rDNA, IGS and FLA were directly sequenced by Eurofins (Ebersberg, Germany).
Molecular distances with standard error estimate were conducted using the Kimura 2-parameter model among Ornithodoros 16S sequences and among Borrelia concatened IGS-FLA sequences as used in other phylogenetical pathogens/vectors studies [54,55]. In the final dataset, the analysis involved 165 different sequences grouping 821 16s RNA tick sequences and 105 different sequences grouping 216 IGS-FLA Borrelia sequences. Evolutionary analyses were conducted in MEGA 5 [56] and all positions containing gaps and missing data were eliminated.
In the goal of molecular diagnostic of Ornithodoros species of North and West African countries, we determined Single Nucleotide Polymorphism (SNP) based on 16S rRNA sequences using parsimony-informative site.

Analysis of environmental factors
We investigated relationship between environmental factors (latitude, longitude, mean annual temperature, mean annual rainfall [57], elevation and distance to the seashore) and tick species distribution using a linear discriminant analysis. All calculations were performed with R software [58].

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. The published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is: urn:lsid:zoobank.org:pub:583BB2C2-B859-4EB5-85D7-70945C923642. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.
The holotypes and paratypes of the new species are deposited at the Institut Royal des Sciences Naturelles de Belgique, Brussels, Belgium (IRSNB). Sequences obtained from a segment of leg of one or several specimens of the type series are deposited in GenBank.

Ethics Statement
The study protocol was approved by the Steering Committee of the IRD Special Programme Evolution Climatique et Santé (IRD, Montpellier, France), reference project ATI-ECS-07-H/ 2002. Animals were treated in a humane manner, and in accordance with authorizations and guidelines of the American Society of Mammalogists (Animal Care and Use Committee 1998). Captures of rodents and insectivores (Senegal, Mali, Benin, Niger, Cameroon, Chad, Mauritania, Morocco) and ticks (Senegal, Mali, Burkina Faso, Niger, Cameroon, Chad, Guinea Bissau, Guinea, Gambia, Liberia, Côte d'Ivoire, Togo, Benin, Mauritania, Morocco, Algeria, Tunisia, Spain) excluded national parks and protected areas, did not involve endangered or protected species (CITES, UICN, and national guidelines), and were conducted as part of research agreements between IRD, national institutions, national ministries of health and/or scientific research. All tick, rodent and insectivores captures inside houses and private lands were conducted after the owner of the house and land gave permission to conduct the study on this site.

Geographic distribution of Ornithodoros ticks
We investigated the occurrence of Ornithodoros ticks in 9,870 burrows from 484 sampling stations in 282 study sites in 12 West African countries (Mauritania, Senegal, The Gambia, Guinea Bissau, Guinea, Mali, Burkina Faso, Niger, Ivory Coast, Benin, Togo, Liberia), three North African countries (Morocco, Algeria, Tunisia), two Central African countries (Cameroon, Chad), and one European country (Spain) ( Table S1).  (Table 1). In West and Central Africa, all burrows located east of 01°E and south of 13°N were negative ( Figure 1). The southernmost limits were 13°32'N, 13°35'N and 13°54'N in Senegal, The Gambia and Mali, respectively, and the easternmost limit south of Sahara was 00°34'E in Mali. In all countries where we found Ornithodoros ticks, their distribution was generally contiguous, with rarely a negative site between two positive sites, and thus it has been possible to establish in the field precise distribution limits (i.e. at a few km scale in western Senegal and at one quarter degree scale in other areas of Senegal and in western Mali). In central and eastern Mali, we found Ornithodoros ticks only in burrows located close to the flood plain of the Niger River. In other areas, the occurrence of Ornithodoros ticks in burrows was independent of any hydrographic system either active or fossil. Based on classical morphological criteria, all Ornithodoros ticks collected in burrows were attributable either to O. sonrai (Senegal, The Gambia, Mali, Mauritania, and the most arid areas of Morocco, Algeria and Tunisia), O. erraticus (northern Morocco, northwestern Algeria and a few ticks from northern Tunisia) or O. normandi (northeastern Algeria and northern Tunisia). We never observed O. moubata.   As previously reported, brain samples of the 140 specimens collected in Morocco were also studied by PCR [40]. A Borrelia infection was demonstrated in 57 animals from Senegal, Mali, Mauritania or Morocco (4.1 % of the 1,386 specimens tested by at least one method) belonging to 13 rodent species and two insectivore species (Table 2). All infected rodents and insectivores were collected in sites or areas where we documented the presence of Ornithodoros ticks. In these areas, we tested 868 animals and the proportion of those found infected by thick blood film examination, brain inoculation and PCR was 3.3% (28/850), 4.2% (21/499), and 8.6% (12/140), respectively. Except for Senegal, where 22.8% (21/92) of brain inoculations to white mice were positive (all were performed within a few days after sampling), all other brain inoculations to white mice were negative, including those from specimens from Mali and Mauritania with positive thick blood films, this certainly in relation to badly preserved spirochetes in brain tissue after sampling.

Molecular analysis of Ornithodoros ticks and Borrelia infections
A total of 1,801 Ornithodoros ticks from 113 study sites distributed in 7 African countries (Mali, Senegal, The Gambia, Mauritania, Morocco, Algeria and Tunisia) and in Spain were tested for molecular species determination and associated Borrelia infection. We observed the presence of Borrelia infections in 295 ticks (16.4%) distributed in all countries and 37.2% of sites (42/113). We obtained 820 sequences of 16S rRNA of Ornithodoros ticks, 216 concatened partial sequences of FlaB gene, and partial sequences of Intergenic spacer for the Borrelia detected (Table 3).

Phylogenetic analysis of Ornithodoros ticks
The phylogenetic analysis based on 820 16S rRNA sequences divided on 165 unique sequences which clustered in nine entities of Ornithodoros ticks (Figure 2). The genetic distance estimates (Kimura 2-parameters model) between entities varied from 0.062±0.013 (O. costalis -O. marocanus distance) to 0.251±0.030 (O. merionesi -O. normandi distance) ( Table 4). The nine entities were genetically differentiated and hybrids between populations of these areas were known to be infertile [29], indicating that they should be considered as nine different species, including five previously undescribed species (see below and Appendix for designation of types, description of species and criteria for diagnosis):    Figure 3 shows that high specific diversity was observed in North Africa where the nine species were found and presented either a wide distribution in the most arid areas of Morocco, Algeria and Tunisia (one species: O. sonrai), or various restricted distribution patterns in coastal and/or inland areas of North Africa and Spain (eight species).
In West Africa, only O. sonrai was observed, and analysis performed on the 16S sequences of 533 O. sonrai ticks from 72 study sites in Sudanian, Sahelian and Saharan areas showed a large distribution of O. sonrai in West and North Africa ( Figure 3). However, over the whole molecular range observed for this species, one group was genetically distinct (bootstrap >90). These ticks were sampled in an area extending from northwestern Senegal (Dagana, Richard-Toll) to northwestern Mauritania (Nouâdhibou) and the most southern part of Morocco (Adrar Souttouf) and we consider that they constitute a distinct sub-species of O. sonrai (Figure 3).

Phylogenetic analysis of Borrelia
The phylogenetic analysis of Borrelia was based on 105 different sequences among 216 IGS-FLA sequences obtained. Figure 4 shows that three Borrelia species were present in sampled ticks: B. crocidurae, B. hispanica, and a third species, B. merionesi, a rare spirochaete that was formerly described from Ornithodoros ticks collected in the same area of southern Morocco than our specimens [59,60] (Figure 5). The genetic distance estimates among Borrelia species varied from 0.003±0.002 (B. recurrentis -B. duttonii distance) to 0.076±0.010 (B. hispanica -B. recurrentis distance) (

Diagnostic SNPs of Ornithodoros ticks
To allow a rapid molecular identification of Ornithodoros species, we highlighted 25 single nucleotide polymorphisms among the 9 species in 16S rRNA sequence dataset (Table 6). We detected at least one diagnostic SNP for each species, i.e. at a given position of the 16S sequence a nucleotide type is present in all sequences of a given species and absent in other species. According to species, the total number of diagnostic SNPs that can be used for taxonomic purpose of soft tick of the genus Ornithodoros varied from 1 to 8 (Table 6).

Environmental factors and tick species distribution
To investigate the relationship between environmental factors and species distribution, we included in the analysis ticks from 110 study sites. Figure 6 shows the projection of the individual ticks on the first factorial plane. The first two axes

Discussion
Genetic distances between the nine clades of Onithodoros ticks collected in this study ranged from 5.2 to 25.1%, a distance similar to those found between Ornithodoros species from other parts of the world. For comparison, the genetic distance between O. puertoricensis and O. rioplatensis, two species from South America, based on 16S rRNA sequencing, is approximately 12.7% [61]. Morphologically, it was easy to separate these nine species in two groups, the first one comprising two small species distributed in northeastern Algeria and northwestern Tunisia and characterized, among other diagnostic characters, by a transverse groove on coxae I, the second one comprising the seven other species and characterized by an oblique groove on coxae I, as previously illustrated elsewhere [31]. These two groups are confirmed by our molecular study (Figure 2). The two species of the first group includes respectively (a) all specimens from La Calle, (PhyML 100 bootstraps, available at http:// mobyle.pasteur.fr/cgi-bin/portal.py). The colored triangle estimated the Borrelia species diversity. We included in the B. merionesi clade (9 seq.) a Borrelia detected in a rodent captured in El Argoub (Morocco). The colored full circle correspond to the tick species determined by the 16S phylogenetic analysis of this study. The phylogram was constructed using a maximum-likelihood method from concatenated sequence data (220 sequences including GenBank reference sequences, 779 nucleotides). Bootstrap values  [62,63]. None of 75 specimens of these two species were found infected by B. hispanica, B. crocidurae or B. merionesi in this study nor in a previous study in Tunisia [31], suggesting that O. erraticus sensu strico and O. normandi may be poor vectors or reservoirs of TBRF. By contrast, six of the seven species of the second morphological group of Ornithodoros ticks were found infected by one of these Borrelia species, most often with high prevalence rates (minimum 4.6% for O. costalis, maximum 36.6% for O. sonrai).
Among the seven species of the second group, only two species were previously known: O. marocanus Velu, 1919, type locality "vieille casbah" near Casablanca, Morocco and O. sonrai Sautet and Witkovski, 1943, type locality Gao, Mali [64,65]. However, since the work of Colas-Belcour [66], O.    Table 6. Twenty-five Single Nucleotide Polymorphisms (SNPs) and their position which distinguish the nine Ornithodoros species, based on 820 16S rRNA aligned sequences (457 nucleotides). marocanus was considered as a junior synonym of O. erraticus [67,68]. Although most Algerian arthropods collected by Lucas [62] were deposited at the Museum National d'Histoire Naturelle in Paris, part of the collection was lost, including the three O. erraticus syntypes. The original description of these specimens was very short and O. erraticus was later redescribed by Neumann [69] on the basis of specimens not from La Calle but from Nemours (= Ghazaouet) and Marnia in   [29] in the early 1950s indicated that all hybrids of the first or second generation were either sterile or have very low fertility [29]. The results of these experiments combined with our data and the partial sympatry of several of these species clearly support the full specific rank of the seven species of the O. marocanus / O. sonrai group identified in our study ( Figure  7). Interestingly, our data confirm the tick-spirochete species specificity highlighted by previous authors [70]. B. crocidurae was only detected in O. sonrai, and both this vector and this spirochete were genetically very close North and South of the Sahara desert, certainly because they present a continuous distribution through Mauritania and Western Sahara (but probably not through central Sahara, a much dryer area [57] where we failed to find O. sonrai). The only apparent disjunction in the distribution of this vector was observed in central Mali, where we failed to find a connection between the populations of O. sonrai bordering the Niger River and those distributed from western Mali to North Africa. B. hispanica was detected in O. marocanus, O. occidentalis and O. kairouanensis, a complex of three large species (female adult length 5.5 -7.5 mm) distributed in typically Mediterranean areas of Morocco, Algeria, Tunisia and Spain and previously confused with O. erraticus [2,23,[28][29][30][31][32]60,[66][67][68][69]. Finally, the poorly known B. merionesi [59,60] appears restricted to Atlantic coastal areas of the Sahara desert where it is transmitted by two Ornithodoros species previously confused with O. sonrai: O. costalis and O. merionesi. Interestingly, no Ornithodoros species was found infected by more than one Borrelia species despite the high prevalence of infected ticks.
The considerable abundance of Ornithodoros TBRF vectors in North and West Africa -from 39.5 % of rodent burrows colonized by vector ticks in Morocco to 13.7 % in West Africa (Senegal, The Gambia, Mali and Mauritania) -coupled with the high infection rate of these vectors by at least two Borrelia species pathogenic for humans suggest that TBRF is a common disease in the whole range of distribution of these vectors. In rural Senegal, we have previously shown that the incidence of B. crocidurae TBRF reached 11 per 100 personyears and was the highest described in the world literature for any bacterial disease [39]. In northwestern Morocco, we found that 20.5% of patients with an unexplained fever had B. hispanica TBRF [30]. Clearly, similar incidence rates are likely to occur in most areas where we demonstrated the occurrence of one or more of the seven species of the O. marocanus -O. sonrai complex. It is not known if B. merionesi is also pathogenic for humans. However, a few previous experiments with this spirochete failed to induce TBRF in humans [59].
We consistently failed to find any Ornithodoros tick inside burrows or houses in West Africa south of 13°N and east of 01°E. We also failed to find these ticks in Cameroon and Chad. In the literature, there were in these areas only two mentions of Ornithoros species known as potential vectors of TBRF, one doubtful report of two larvae of O. erraticus collected on Arvicanthis niloticus in Ndjamena before 1959 [71] and the mention of one nymph of O. sonrai collected in a burrow of Xerus erythropus in Niamey in 1956 [23]. The later report was consistent with the current known distribution of O. sonrai along the Niger River in Mali, but our extensive investigations (132 burrows examined) in Niamey and along the Niger River between Niamey and the Malian border were negative. In Mali, a recent comprehensive study of the small mammal reservoir of B. crocidurae in 20 villages from different areas of the country indicated that 11% of 663 rodents and 14% of 63 shrews had antibodies to spirochetes and 2.2% had confirmed active B. crocidurae infections [37]. Interestingly, distribution of seropositive mammals and O. sonrai ticks supported our own data, except for some villages of southern Mali where a few rodents were found seropositive although collected south of the known distribution limit of O. sonrai. Since low rates of serological cross reactivity and/or dispersal or accidental introduction of rodents south of places where they were infected cannot be excluded, further studies are needed to confirm local transmission of TBRF in these areas. In the Malian study, three rodent species (M. erythroleucus, M. natalensis, P. daltoni) and two insectivores species (C. olivieri, C. fulvastra) were found infected by thin blood film examination, a method where the amount of blood examined in each microscopic field is approximately 20-fold less than by thick blood film examination [72]. In Senegal, additional mammal species found infected during previous studies include A. albiventris, R. norvegicus, G. gambianus and G. gracilis [35,36]. With these species and C. fulvastra found infected in Mali [37], the known small mammal reservoir of TBRF in West and North Africa currently comprises 21 species.
A recent study in Togo indicated that 7.9% and 1.2% of 239 patients with fever from all regions of the country had B. crocidurae and B. duttonii infections, respectively, according to PCR blood testing [73]. However, in wet savanna and forest areas of West Africa, Ornithodoros ticks have never been reported [23,67,74,75], and this was confirmed by our surveys in Togo, Benin, Guinea, Guinea Bissau, Ivory Coast and Liberia where a total of 900 burrows and several hundred household examined were negative. If the occurrence of local transmission of B. crocidurae and B. duttonii in these areas is confirmed (since thick blood films failed to show the presence of spirochetes in PCR positive blood samples in the Togo study), this would suggest the existence of unknown vector / reservoir systems of Borrelia infections in wet savanna areas of West Africa.
Finally, our study as those previously conducted in northern, western, central, eastern and southern Africa [25,39,59,76,77] highlight the need of much more attention from health services, researchers and funding agencies to the continued burden of relapsing fevers due to Borrelia spp. in African populations.
Holotype: IRSNB/RBINS IG.32.280/004/1, a female of 5.3 mm long collected on 18 July 2010 inside a rodent burrow in a pine woodland located at 36°38'25"N / 04°30'41"W near Torremolinos, Spain by G. Diatta  Diagnosis: An Argasidae tick of the O. marocanus group of the genus Ornithodoros, characterized by an oblique groove on coxae I., and belonging to the subgroup of large sized species (average female length > 5.2 mm, average male length >3.0 mm) of this complex (subgroup O. marocanus complex). Morphologically indistinguishable from O. marocanus, but genetically differing from these species and from all other species of the O. marocanus group by the unique position A/C on 16S rRNA aligned sequences (position 244, see Table 6).
Geographic distribution: Coastal areas of Morocco, Algeria and Tunisia.
Ornithodoros erraticus. (Lucas, 1849). The three syntypes of Ornithodoros erraticus are not in Paris MNHN. According to M. Judson (curator for arthropods, personal communication), they were certainly destroyed during transport to Paris as many other specimens of Lucas's collection of Algerian arthropods.
The types of O. marocanus were sent by Velu [64] to the Pasteur Institute of Paris after he described this species. They could not be traced either in Paris or Casablanca and are now considered as lost (F. Rhodain, personal communication).