Prevalence of Trypanosoma and Sodalis in Wild Populations of Tsetse Flies: Impact on Sit Programmes for Tsetse Eradication

Tsetse ies, the vectors of African Trypanosoma, have a highly regulated and dened microbial fauna composed of three bacterial symbionts that may have a role to play in the establishment of Trypanosoma infections in the ies and hence, may inuence the vectorial competence of the released sterile males. Sodalis bacteria seem to interact with Trypanosoma infection in tsetse ies. Field-caught tsetse ies of ten different taxa and from 15 countries were screened using PCR to detect the presence of Sodalis and Trypanosoma species and their interaction. The results indicate that the prevalence of Sodalis and Trypanosoma varied with country and tsetse species. Trypanosome prevalence was higher in east, central and southern African countries than in west African countries. Tsetse y infection rates with Trypanosoma vivax and Trypanozoon spp were higher in west African countries, whereas tsetse infection with Trypanosoma congolense and T. simiae, T. simiae (tsavo) and T. godfreyi infection prevalence were higher in east, central and south African countries. Sodalis prevalence was high in Glossina morsitans morsitans and G. pallidipes but absent in Glossina tachinoides. Double and triple infections with Trypanosoma taxa and coinfection of Sodalis and Trypanosoma were rarely observed but it occurs in some taxa and locations. A signicant Chi square value (< 0.05) seems to suggest that Sodalis and Trypanosoma infection correlate in Glossina palpalis gambiensis, Glossina pallidipes and Glossina medicorum. Trypanosoma infection signicantly increased the density of Sodalis in wild G. m. morsitans and G. pallidipes ies however no signicant impact of Sodalis infection on trypanosome density. The dynamics of Sodalis infections in tsetse colonies has been well studied and previous studies indicate that Sodalis is more frequently present in colonized tsetse ies than in wild tsetse populations 36,52 . The prevalence of Sodalis was 80 and 100% in colonized G. m. morsitans and G. p. gambiensis, respectively 52,53 , which is higher than the prevalence of Sodalis reported in wild populations of these tsetse species. This seems to indicate that the colonizing and rearing process of tsetse ies favours the transmission and spread of Sodalis infections. Recently, colonies of G. pallidipes, G. p. gambiensis, G. f. fuscipes, G. m. morsitans, G. m. centralis and G. m. submorsitans maintained at the FAO/IAEA Insect Pest Control Laboratory were screened for Sodalis infections and showed a prevalence of 100%, only G. brevipalpis had a lower prevalence of 95% (data not shown). Similarly, the prevalence of other symbionts was 100% for colonized species such G. m. G. m. centralis, and G. Our study showed that the prevalence of different Trypanosoma species and or subspecies can be different in different tsetse taxa. In G. tachinoides in Ghana, the Trypanosoma vivax infection was high (>10%) as well as the infections of the Trypanozoon (Tz) and Tsg group and the mixed infections of Tv-Tsg. However, the prevalence of T. congolense was very low. These results are in agreement with the prevalence of T. brucei s.l (11%) and T. congolense forest type (2.6%) reported in the same tsetse species in Cameroon. However, the same study reported a prevalence of 13.7% of T. congolense savannah type 35 , which was not observed in our study. Our results agree with the results of Lefrançois et al., 59 , i.e. in Côte d’Ivoire, G. tachinoides showed a high rate of Trypanosoma infection (61%) with Tv (27.2%), Tc (31.8%.), and Tz (2.3%) 60 . The high overall high prevalence of Trypanosoma and the high infection rate with Tv and Tz group agree with our results, but the high Tc prevalence conicts with our result. In our study, the Tc infection rate was high in G. m. morsitans and G. pallidipes, which does not agree with the reported Trypanosoma infection in G. m. morsitans from Malawi where T. brucei prevalence was 64.4% but all other Trypanosoma infections were < 10% 61 . The mixed infection of Trypanosoma species/subspecies is in agreement with previous reports 35,55,60,62 multiple) and of different as countries and tsetse taxa, were assessed using the matrix display and metric multidimensional scaling (mMDS) plot with bootstrap averages in PRIMER version 7+. The bootstrap averages plots were displayed with a Bray and Curtis matrix based on the square-root transformation of the Sodalis and Trypanosoma (single and multiple) infection abundance data 91 . The tests were based on the multivariate null hypothesis via the use of the non-parametric statistical method PERMANOVA 92 . The Permanova test was conducted on the average of the abundance data based on the country-species after excluding the data of Eswatini (low number of tested samples).


Abstract
Tsetse ies, the vectors of African Trypanosoma, have a highly regulated and de ned microbial fauna composed of three bacterial symbionts that may have a role to play in the establishment of Trypanosoma infections in the ies and hence, may in uence the vectorial competence of the released sterile males. Sodalis bacteria seem to interact with Trypanosoma infection in tsetse ies. Field-caught tsetse ies of ten different taxa and from 15 countries were screened using PCR to detect the presence of Sodalis and Trypanosoma species and their interaction. The results indicate that the prevalence of Sodalis and Trypanosoma varied with country and tsetse species. Trypanosome prevalence was higher in east, central and southern African countries than in west African countries. Tsetse y infection rates with Trypanosoma vivax and Trypanozoon spp were higher in west African countries, whereas tsetse infection with Trypanosoma congolense and T. simiae, T. simiae (tsavo) and T. godfreyi infection prevalence were higher in east, central and south African countries. Sodalis prevalence was high in Glossina morsitans morsitans and G. pallidipes but absent in Glossina tachinoides. Double and triple infections with Trypanosoma taxa and coinfection of Sodalis and Trypanosoma were rarely observed but it occurs in some taxa and locations. A signi cant Chi square value (< 0.05) seems to suggest that Sodalis and Trypanosoma infection correlate in Glossina palpalis gambiensis, Glossina pallidipes and Glossina medicorum. Trypanosoma infection signi cantly increased the density of Sodalis in wild G. m. morsitans and G. pallidipes ies however no signi cant impact of Sodalis infection on trypanosome density.

Background
Tsetse ies (Diptera: Glossinidae) are distributed in sub-Saharan Africa between 15° north and 26° south latitude 1 . Glossina spp. are the cyclic vectors 2 of unicellular protozoa of the genus Trypanosoma that cause African animal trypanosomosis (AAT) or nagana and human African trypanosomosis (HAT) or sleeping sickness 3,4 . Nagana in cattle is mainly caused by Trypanosoma congolense, Trypanosoma vivax and Trypanosoma brucei brucei 5 and causes annual losses to agriculture estimated at $4.75 billion 6 . In addition, around 35 million doses of trypanocidal drugs are administered to livestock per year for managing AAT 7 . Human African trypanosomosis is fatal without treatment 8 and is caused by two Trypanosoma subspecies, i.e. Trypanosoma brucei rhodesiense responsible for the acute form of HAT in East Africa and Trypanosoma brucei gambiense for the chronic form of HAT in western and central Africa 9 . The lack of effective vaccines and the development of resistance to the available trypanocidal drugs makes the control of AAT in the vertebrate host unsustainable 10,11 . Consequently, an effective tool to reduce Trypanosoma transmission would be the control of the tsetse vector. One effective method to manage populations of tsetse ies is the sterile insect technique (SIT) when used as part of an area-wide integrated pest management (AW-IPM) approach 12,13 . The SIT method relies on the mass-production and sterilization of male ies by ionizing radiation. The sterile males are released in the target area for mating with wild females and the absence of offspring will gradually reduce the density of the targeted tsetse populations 14 .
The biological transmission of the Trypanosoma species requires the parasite to undergo a series of proliferation and differentiation steps in the tsetse alimentary tract and nally mature into an infective form in the mouthparts (T. congolense) or salivary glands (T. brucei spp.) 15 . However, tsetse ies are refractory to Trypanosoma infection meaning that the probability that Trypanosoma ingested during a blood meal complete their developmental cycle in the y to result in a mature infection is rather low [16][17][18] . The endogenous bacterial microbiome seems important in providing tsetse ies the natural ability to mitigate Trypanosoma infections 19 . Three major endosymbiotic bacteria have been identi ed in tsetse ies, i.e. Wigglesworthia glossinidia, Sodalis glossinidius (hereafter mentioned as Sodalis) and Wolbachia pipientis 20 . Some studies suggested that the obligate mutualist Wigglesworthia must be present in the larval stage during the development of a mature tsetse y to properly develop a well-functioning immune system contributing to a refractory phenotype against Trypanosoma 5,19 .
Sodalis, the second mutualistic symbiont, can be found in the midgut, hemolymph, muscles, fat body, milk glands, and salivary glands of certain tsetse species and is inherited by the progeny through transovarial transmission 21 . The biological role/importance of Sodalis for tsetse remain unclear and needs to be clari ed 22 . This symbiont might provide some bene ts to the host as ies without Sodalis have a signi cantly shorter lifespan compared to ies with it 23 , however the establishment of Sodalis free colony was feasible 24 . Sodalis also presents many ideal characteristics to be used for expressing molecular effectors in paratransgenic tsetse 25  Zambia, and Zimbabwe and analyse these data in the context of a possible association between the occurrence of Sodalis and a Trypanosoma infection in tsetse. Such information might guide the decision maker for SIT programmes to take the appropriate action, if necessary, to minimize any potential risk of an increased transmission.
Regardless of tsetse taxon, in west African countries the average Trypanosoma prevalence was 20% (n = 3733), with the highest prevalence recorded in Ghana (61%) and the lowest recorded in Guinea (2.2%). The prevalence in Burkina Faso, Mali and Senegal was 21.9, 6.9 and 14.2% respectively ( Figure 1, and Table 2). In east, central and southern African countries, the Trypanosoma infection prevalence was a bit higher than in west African countries with an averaged infection of 31.5% (n = 3127), with the highest prevalence (53.6%) in Zimbabwe and lowest prevalence (2.9%) in DRC. No Trypanosoma infection was detected in Eswatini ( Figure 1 and Table 2). Regardless of the country, Trypanosoma prevalence varied from one taxon to another, and G. m. morsitans showed the highest Trypanosoma prevalence (41%) followed by G. pallidipes (38.5%) and the lowest prevalence was detected in G. brevipalpis (9.71%) in east, central and southern Africa. In west Africa, G. medicorum showed the highest Trypanosoma prevalence (39.5%) and the lowest prevalence was detected in G. p. palpalis (2.8%) ( Table 3).
Some tsetse taxa were collected from several countries as presented in Figure 2 and Table 4. The highest Trypanosoma prevalence was recorded in G. tachinoides in Ghana (61%). This was followed by high prevalence in G. m. morsitans collected from Zimbabwe (53.9%), Tanzania (53%) and Zambia (48,4%). G. pallidipes from Zimbabwe, Kenya, Zambia and Tanzania also showed high Trypanosoma prevalence of 52.7%, 50.9%, 45.2% and 37.3%, respectively. The lowest Trypanosoma prevalence was found in G. p. gambiensis from Guinea (2.2%). Based on the Trypanosoma prevalence presented in Figure 2 and Mixed infections of Trypanosoma groups (double or triple) are rare events with an average prevalence between 0.09 and 1.71% regardless of country or tsetse species. However, double infections seem to be more frequent in some countries than others (X 2 = 35.01, df = 14, P = 0.00) for Tv-Tz and in some tsetse species than others (X 2 = 21.20, df = 9, P = 0.012) for Tv-Tz (Supplementary le 1). The highest prevalence of the mixed infections Tv-Tz and Tc-Tz were observed in Ghana with 12.39% and 10.68%, respectively, regardless of tsetse species. Although the average Tc-Tsg prevalence was higher than that of Tv-Tz and Tc-Tz, the highest mixed infection with it was found in Zambia with 9.05%. Regardless of the country, the highest mixed infection of Tc-Tsg detected per tsetse species was ~5% in G. m. morsitans and G. pallidipes. The mixed infection of Tsg with either Tv or Tz or both was lower than 2% regardless of the country or tsetse species. Taking into account both the country and tsetse species, the highest mixed infection of Tc-Tsg (12.5%) was detected in G. m. morsitans in Zambia. However, the highest prevalence of Tc-Tz (10.68%) and Tv-Tz (12.39%) was detected in G. tachinoides from Ghana. Although the average prevalence of Tv-Tsg was low (0.54%), a relative high infection rate of 6.17% was found in G. m. morsitans from Tanzania.

Prevalence of Sodalis infection
The prevalence of Sodalis infection based on the PCR results varied signi cantly with country (X 2 = 108.02, df = 1, 14, P << 0.001) and tsetse species (X 2 = 69.60, df = 9, P < 0.001). The best glm model (lowest AICc) selected for the overall Sodalis prevalence retained the countries, the species and their interaction (where possible) as variables that tted the data well (AICc = 1296.12). Similar to the prevalence of Trypanosoma, the average Sodalis prevalence in east, central and southern Africa (24.6%) was higher than in west Africa (2.70%). Regardless of tsetse species, the highest prevalence of Sodalis infection was found in Tanzania Figure 2 and Table 5. The samples showing high Sodalis prevalence includes G. m. morsitans from Kenya (63.5%) and Tanzania (76.5%) and G. pallidipes from Tanzania (74.6%) and Uganda (75%), however the samples with no Sodalis infection includes G. austeni from Eswatini, G. p. gambiensis from Mali and Senegal and G. tachinoides from Burkina Faso and Ghana indicating that there is 95% con dence that the infection rate is less than or equal to 10%, 0.82%, 0.55%, 1.28% and 0.36%, respectively.

Interactions between Sodalis and Trypanosoma infections
Prevalence of co-infections of Sodalis with Trypanosoma The screening results indicated that the single infection rate was 9.3% (n = 638) and 21.9% (n = 1503) for Sodalis and Trypanosoma, respectively, over all taxa and countries ( Figure 3A). No Sodalis infection was found in G. tachinoides, and therefore was excluded from the analysis. A Cochran-Mantel-Haenszel test for repeated tests of independence showed that infection with Sodalis and Trypanosoma did deviate from independence across all taxa (χ 2 MH = 41.73, df = 1, P < 0.001) and individual Chi squared tests for independence for each taxon showed signi cant deviation from independence at the Bonferroni corrected α = 0.00833 in G. pallidipes (P < 0.001) and G. p. gambiensis (P < 0.001) (Supplementary Table 6). The prevalence of coinfection of Sodalis and Trypanosoma in wild tsetse populations varied with tsetse taxon and location. No coinfection was found in many taxa and many locations. The co-infection was found only in G. f. fuscipes (2.73%), G. m. morsitans (15.72%) and G. pallidipes (9.22%) in east, central and southern Africa ( Figure 3B, Table 6 and Supplementary Table 6).

Impact of co infection on Trypanosoma and Sodalis density
Attempts to assess the density of Trypanosoma and Sodalis under single (S -/T + ) and (S + /T -) or double infection (S + /T + ) was conducted using qPCR with primers mentioned in Supplementary Table 2

Discussion
The implementation of the SIT in the context of an area-wide integrated pest management strategy to eradicate tsetse ies relies on the release of sterile males in the targeted area. This was successful in eradicating a population of G. austeni from Unguja Island of Zanzibar 37 and signi cant progress was made in the eradication programme implemented against G. p. gambiensis in the Niayes area of Senegal 38 . However, as both male and female tsetse ies are vectors of Trypanosoma species, the release of large numbers of sterile male ies bears a potential risk of temporarily increasing disease transmission during the initial release phase of an SIT programme 39 . Therefore, mitigating measures are required to reduce or eliminate this potential risk, especially in areas where sleeping sickness (HAT) is endemic. To date, to mitigate such risks, sterile males are offered two-three blood meals mixed with the trypanocidal drug isometamidium chloride, before being released which reduces the risk of Trypanosoma transmission signi cantly but does not eliminate it 40,41 . In addition, other approaches were proposed to minimize such risks such as paratransgenesis 42,43 and combining paratransgenesis with SIT 44 .
Several previous studies reported a potential positive correlation between Sodalis and Trypanosoma infections 28,30,32,36,45−48 , leading to the hypothesis that Sodalis might facilitate the establishment of Trypanosoma infections in the tsetse midgut 23,26,27 . This would be possible through the production of a chitinase enzyme encoded by Sodalis which, through its chitinolytic activity, enhances the permeability of the peritrophic membrane in the tsetse midgut, making it easier for the Trypanosoma to cross the barrier and establish an infection in the y's ectoperitrophic space 49,50 . This process results in the accumulation of N-acetyl-D-glucosamine, which would further facilitate the establishment of a Trypanosoma infection by inhibiting the activity of the anti-parasitic tsetse lectins 51 . Although these theories have never been experimentally proven, it is supported by the positive correlation between Sodalis and Trypanosoma infections in various studies 45,46,48 . Moreover, the positive correlation between Sodalis and Trypanosoma infections might be affected by the genotype of Sodalis and the Trypanosoma taxon 28 . These previous reports contrast with the results of Trappeniers et al 24 , who demonstrated no difference in Trypanosoma infection in tsetse ies infected or not with Sodalis and other previous report indicating the lack on correlation between Sodalis and Trypanosoma infection [34][35][36] .
The dynamics of Sodalis infections in tsetse colonies has been well studied and previous studies indicate that Sodalis is more frequently present in colonized tsetse ies than in wild tsetse populations 36,52 . The prevalence of Sodalis was 80 and 100% in colonized G. m. morsitans and G. p. gambiensis, respectively 52,53 , which is higher than the prevalence of Sodalis reported in wild populations of these tsetse species. This seems to indicate that the colonizing and rearing process of tsetse ies favours the transmission and spread of Sodalis infections. Recently, colonies of G. pallidipes, G. p. gambiensis, G. f. fuscipes, G. m. morsitans, G. m. centralis and G. m. submorsitans maintained at the FAO/IAEA Insect Pest Control Laboratory were screened for Sodalis infections and showed a prevalence of 100%, only G. brevipalpis had a lower prevalence of 95% (data not shown). Similarly, the prevalence of other symbionts such as Wolbachia was 100% for colonized species such as G. m. morsitans, G. m. centralis, and G. swynnertoni 48 . Taken into consideration that mass-rearing conditions enhances Sodalis infections and that Sodalis infections might facilitate the establishment of a Trypanosoma infection in the midgut, sterile male tsetse ies that are derived from colonies might be effective vectors of Trypanosoma species and, therefore, might increase the Trypanosoma transmission. It is therefore important that the managers and planners of SIT programmes are aware which tsetse species show a positive correlation between Sodalis and Trypanosoma infections to be able to take the necessary mitigating actions.
Many previous studies have examined the prevalence of Sodalis and Trypanosoma species in wild tsetse populations 30,32,35,45,54 , but this study presents for the rst time the prevalence of Sodalis and Trypanosoma species on a continental scale. In addition, the methods used were standardized and all carried out in one laboratory that avoided discrepancies in the results due to different handling of tsetse samples, DNA extraction, and PCR methods as observed in these other studies that were done in different laboratories. Our results indicate that Sodalis and Trypanosoma prevalence varied with tsetse species and geographical location (with an overall trypanosome prevalence of 23,5%), which agrees with many previous studies 55 .
High Trypanosoma prevalence (> 30%) was also found in G. m. morsitans and G. pallidipes from central and east Africa which is in agreement with the high prevalence of Trypanosoma species found in G. m. Our study showed that the prevalence of different Trypanosoma species and or subspecies can be different in different tsetse taxa. In G. tachinoides in Ghana, the Trypanosoma vivax infection was high (>10%) as well as the infections of the Trypanozoon (Tz) and Tsg group and the mixed infections of Tv-Tsg. However, the prevalence of T. congolense was very low. These results are in agreement with the prevalence of T. brucei s.l (11%) and T. congolense forest type (2.6%) reported in the same tsetse species in Cameroon. However, the same study reported a prevalence of 13.7% of T. congolense savannah type 35 , which was not observed in our study. Our results agree with the results of Lefrançois et al., 59  Likewise, the prevalence of Sodalis infection varied signi cantly with tsetse taxon and location and the highest prevalence was found in G. m. morsitans and G. pallidipes. Our results agree with the high prevalence of Sodalis reported in G. pallidipes (~50%) in one location in Kenya regardless of the y age 33 ; however, the same study reported low Sodalis prevalence in another location. In another study in Kenya, Wamwiri et al., 32 reported moderate Sodalis prevalence in G. pallidipes (16%) and low prevalence in G. austeni (3.7%), which is in agreement with our results. On other hand, our results are different from the low prevalence (< 8%) found in G. m. morsitans and G. pallidipes in Zambia 36 . In another study in Zambia, Sodalis prevalence in G. m. centralis, was reported to be 15.9% with no signi cant difference between inter-site prevalence 55 . In our study, the prevalence of Sodalis in G. brevipalpis was low (< 2.3%) which contradicts the high prevalence (93.7%) found in this species in Zambia 36 . In the Democratic Republic of the Congo, the global prevalence of Sodalis in Glossina fuscipes quanzensis midgut averaged 15.5%, but in certain locations the prevalence exceeded 40% 63 . In Nigeria, Sodalis prevalence in G. p. palpalis and G. tachinoides was 35.7% 64 which is higher that the prevalence reported in our study for both species.
The data from our study indicate that the Trypanosoma and Sodalis infections were very low or absent in some tsetse taxa from certain locations such as G. austeni in Eswatini for Trypanosoma and Sodalis infections and several species in west Africa for Sodalis infections. The lack of Sodalis and Trypanosoma infection in these samples might be due to (i) low number of tested samples (ii) the use of the DNA extracted from the whole body of tsetse adults (iii) the possibility of the collected samples being infected with different strains/genotypes that might not be detected with the primers used and (iv) the infection of Sodalis and Trypanosoma are under the detection limit of the used PCR. It is important to note that due to the high number of samples tested in our study, the nested PCR to detect low infection level was excluded for technical reasons. The rst reason might apply for G. austeni in Eswatini (n=30) where we can only state with 95% con dence that the true rate of Sodalis or Trypanosoma infection is 10% or less, following the method of Couey and Chew 65 .
Our results indicate signi cant deviation from independence (correlation) of Sodalis and Trypanosoma infections in G. medicorum, G. p. gambiensis and G. pallidipes. However, the lack of detection of any tsetse adult with co-infection of Sodalis and Trypanosoma in G. medicorum, and G. p. gambiensis might indicate a negative correlation. Such negative trend might be supported by the lower density of Sodalis in the ies with co-infection (S + /T +) compared to these with Sodalis infection only (S + /T − ). On other hand the lack of impact of Sodalis infection on Trypanosoma density does not support the negative trend and agreed with the results of Trappeniers et al., 24 reported on colonized ies. This results also agreed with previous results reporting the absence of direct correlation between the presence of Sodalis and the acquisition of a Trypanosoma infection 66 . However, an inverse correlation was reported between Sodalis and the vector competence where the presence of Sodalis in both midgut and proboscis of G. p. gambiensis was associated with its status as a poor vector, whereas it is not found in the proboscis of G. m. morsitans (major vector). It is worth noting that all previous studies of Sodalis infection in G. p. gambiensis and its interaction with Trypanosoma infection was carried out with ies reared under laboratory conditions 28, 29,67 . The correlation between Sodalis and Trypanosoma infection in G. pallidipes is positive, evidenced with the relative high number (n = 170) of tsetse with co-infection. This positive correlation was also found in G. pallidipes from Kenya although with too few ies with co-infection to enable us to draw a de nite conclusion 32 . Although co-infections were found in G. m. morsitans and G. f. fuscipes in some locations, the global correlation was missing. This is in agreement with the positive correlation found between Sodalis and Trypanosoma infection in G. m. centralis in Zambia, in which there was a 6.2 fold increase in the likelihood of a y being infected with Trypanosoma if Sodalis was present 55 . More studies are needed to enhance the potential control interventions mediated by endosymbionts to reduce parasitic infections 64 .
The results of this study clearly indicate that the interaction between Sodalis and Trypanosoma infection is complex, species-speci c and remains unclear and requires further investigation. The prevalence results indicate that Sodalis and Trypanosoma infections are not independent in some species, such as G. p. gambiensis and G. medicorum in west Africa and G. pallidipes in central and east Africa, and this requires further investigations to clarify this relationship. In case of a positive correlation between Sodalis and Trypanosoma infection in these species, additional measures should be taken when implementing the SIT to reduce the Sodalis density in the sterile males released in the targeted area to maximize the safe implementation of the SIT. These measures might include the mixing of Sodalis phage(s) 29,68 with the blood meals to feed the mass-reared ies to reduce the Sodalis density in these ies. This certainly would require the isolation and propagation of the Sodalis phage(s). In addition, the blood meal offered to the males before release can be supplemented with one or more of the following antimicrobial products to reduce Sodalis density, i.e. streptozotocin 23 , indolicidin and OaBAC 5 mini 69 , or with the trypanocidal drug isometamidium chloride. The use of the Sodalis phage as well as these antimicrobial products requires further studies to 1) develop methods to isolate the phage, 2) determine the conditions (e.g. suitable concentration) for its use, and 3) determine the impact on Sodalis density, tsetse productivity and survival. The quanti cation of Sodalis and Trypanosoma density showed that in G. m morsitans and G. pallidipes, Sodalis infection does not have an impact on Trypanosoma infection indicating no additional measures need to be taken during the implementation of SIT against these species.

Conclusion
Sodalis and Trypanosoma infection varied with tsetse taxon and location. There is a signi cant positive correlation between Sodalis and Trypanosoma infection in G. medicorum, G. p. gambiensis and G. pallidipes; however, no signi cant correlation was found in other tsetse taxa and locations. The results of this study will enable the decision makers of SIT projects to better plan and take the necessary measures to ne-tune and optimize SIT e ciency and safety.   77 . A total of 6860 tsetse ies, belonging to ten tsetse species, were collected for this study ( Table 1). The majority of the samples were collected in Burkina Faso (2274), Kenya (1008), Senegal (547) and South Africa (526). As the distribution of most tsetse species is allopatric (only few species are sympatric), not all tsetse species were collected from each country. Following collection, y samples were preserved in 95% ethanol or propylene glycol and shipped to the FAO/IAEA Insect Pest Control Laboratory (IPCL) in Seibersdorf, Austria and stored at -20 °C until analysis. Total DNA was extracted from individual whole y bodies using the DNeasy tissue kit (QIAGEN Inc., Valencia, CA) following the supplier's instructions. The DNA quality and concentration were measured by spectrophotometry (Synergy H1 Multi-Mode Reader, BioTek, Instruments, Inc., USA) and subsequently kept at 4°C until screened for Sodalis and Trypanosoma infections. To verify the quality of the extracted DNA, a set of speci c primers amplifying the Glossina spp. microsatellite GpCAG133 sequence (Supplementary Table 2) and only the successful samples were included in the analysis 21,78 .

Trypanosoma prevalence and genotyping
Polymerase chain reaction (PCR), following the method of Njiru et al. 79 that used the primers ITS1-CF and ITS1-BR (Supplementary Table 2

Analysis of the Trypanosoma and Sodalis density
Samples showing Trypanosoma infection with Sodalis (Sod + /Tryp + ) and samples not infected with Trypanosoma but infected with Sodalis (Sod + /Tryp -) were evaluated with quantitative PCR (qPCR) to assess the impact of Trypanosoma infection (regardless the Trypanosoma type) on Sodalis density. The qPCR was performed using a CFX96 Real Time PCR Detection System (Bio-Rad). The iC gene was ampli ed with the following primers: sodqPCR-FliCF and sodqPCR-FliCR 81 (Supplementary Table 2 Table 2) were used to assess the Trypanosoma density in the tested samples. The DNA from all selected samples was diluted to a nal concentration of 4 ng/μl and 5 μl of the diluted DNA was used for qPCR to determine Sodalis and Trypanosoma DNA density normalized to the housekeeping β-tubulin gene. The ampli cation mixture contained 5 μl of DNA template, 200 nM of each primer, and 7.5 μl iQ TM SYBER Green Supermix (Bio-Rad). qPCR cycling conditions for Sodalis were as follows: initial denaturation at 95 °C for 2 min; 39 cycles of 95 °C for 5 s, 55 °C for 30 s, one step at 95 °C for 5 s and a melting curve constructed from 65°C to 95 °C in increments of 0.5 °C for 5 s. The same conditions were used for Trypanosoma except the annealing temperature was at 60 °C. The analysis of the Sodalis, Trypanosoma and Tubulin densities was based only on qPCR data with the expected melting curve at 81.5°C, 85.5°C and 86°C, respectively.

Data analysis
The prevalence data were recorded and analyzed with the general linear model (GLM) 82 . The prevalence of Sodalis, Trypanosoma species and each Trypanosoma species and co-infection were tested for differences between the tsetse taxa and between countries. For each country, the prevalence was assessed again for differences between the localities where the ies were collected and between the tsetse species present in each country. In the absence of Sodalis or Trypanosoma infection, the probability of infection was calculated following the method of Couey and Chew 65 . Trypanosoma prevalence between taxa was compared between species by a pairwise comparison of proportions with a Bonferroni correction and Benjamini-Hochberg correction. The analyses were executed in R v 4.0.5 83 To analyse the qPCR data, normalized density of Trypanosoma and Sodalis against the house keeping gene (tubulin) was extracted from the CFX Maestro software. Samples giving a valid density (not N/A) for both Trypanosoma and Sodalis were retained for further statistical analysis in R. Similarities in the structure of Sodalis and Trypanosoma (single and multiple) infection and the role of different factors such as countries and tsetse taxa, were assessed using the matrix display and metric multidimensional scaling (mMDS) plot with bootstrap averages in PRIMER version 7+. The bootstrap averages plots were displayed with a Bray and Curtis matrix based on the square-root transformation of the Sodalis and Trypanosoma (single and multiple) infection abundance data 91

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.