Intra- and inter-individual genetic diversity in G. f. fuscipes Wolbachia lineages
Based on groEL, G. f. fuscipes Wolbachia lineages are diverse (Tables
2), with groEL haplotype diversity varying only slightly (Hd = 0.905-1.0) regardless of scale (e.g., dataset, lineage, group subdivision or collection site, Figure
4, See Additional file
1: Table S1). When not superinfected (infected only with Group 1), groEL haplotype diversity is still high (Hd = 0.938, Table
2). There are 102 and 21 unique groEL haplotypes in the complete and in the conservative dataset, respectively. Of the 21 unique groEL haplotypes, 13 were found in many individuals and sites (See Additional file
1: Table S4 and Figure S4).
Finding Wolbachia strain diversity in wild host populations is not unprecedented, but it has not been reported within individuals at this geographic scale. In European cherry fruit flies, Rhagoletis, multiple strains were found using the wsp gene (Wolbachia surface protein), but multiple strains within an individual were not reported
. Similarly in planthoppers, Perkinsiella, a number of B and F supergroup strains were identified using wsp, with F strains inferred to be potential recombinants
. While these and other studies identified high levels of Wolbachia diversity among individuals
, our study found similar high levels of sequence variation at a much smaller scale, within individuals (See Additional file
1: Figure S2).
We suggest that the unique Glossina life history traits facilitate the identification of a transient phase of Wolbachia infection dynamics. Tsetse flies have a viviparous reproductive biology, where one oocyte matures and a single larva is nourished in an intrauterine environment. Females reproduce over their 3–4 month life span, producing 8–10 progeny. The low Wolbachia densities in G. f. fuscipes may reflect this viviparous biology of tsetse, since the few oocytes present in tsetse females may not necessitate retention of high Wolbachia densities that are required in oviparous females. However, Wolbachia densities in G. f. fuscipes were significantly lower than in the laboratory line G. morsitans morsitans, a species with similar life history traits. Environmental influences on Wolbachia densities in natural populations may be relevant and should be tested in other natural Glossina populations. Since G. f. fuscipes Wolbachia densities are very low, if a new Wolbachia variant arises within an individual, it is more likely to be observed and to have a proportionally larger impact on the overall genetic diversity in an individual than in a high-density Wolbachia infection. Thus, the peculiar tsetse life history may indirectly shape Wolbachia diversity within an individual host and allow the identification of variants that would otherwise not be detected.
Origin of G. f. fuscipes Wolbachia infections
The observed patterns and levels of genetic diversity of the two supergroups and their co-occurrence with any one host mtDNA haplotype suggests that the origin of G. f. fuscipes Wolbachia (hereafter wGff refers to any Wolbachia strain found in G. f. fuscipes) infections is complex and different from Wolbachia infection patterns reported in other studies. In insect populations that have undergone a selective sweep due to CI, Wolbachia infections are often associated with a single mtDNA haplotype from one or a few females
[16, 48]. In G. f. fuscipes, wGff are associated with at least 26 host mtDNA haplotypes (Figure
4, See Additional file
1: Table S1) with only 15 of these host haplotypes carrying the observed wGff sequence diversity. In addition, these mtDNA haplotypes are found in all wGff groups. These observations could suggest that the infection in this tsetse species is ancient with unprecedented horizontal and imperfect transmission. Although this scenario is possible, it is a less likely explanation, because even with horizontal transmission we would expect to see a geographic break in the Wolbachia groEL haplotypes, as we see for the host mtDNA haplotypes. An alternative hypothesis is that multiple females with different mtDNA haplotypes were initially infected. Although tests of the association between infection and mixed region host haplotypes were not significant (Randomization test, p = 0.17), all but four of these host mtDNA haplotypes are found in, but not only in, the region where northern and southern host mtDNA haplotypes co-occur (See Additional file
1: Figure S4, Table S1). This suggests that the wGff Group 1 infection in Uganda may have started in the region where we observe mixed host haplotypes. Since these haplotypes are found throughout Uganda, the wGff infection may have spread from there via host dispersal and subsequent gene flow. Our genetic data support this hypothesis as wGff differentiation between fly populations with the two mtDNA haplogroups is low (Table
3). Furthermore, wGff prevalence is associated with host genetic groups defined by microsatellite loci
While our data are suggestive of a mixed region origin, the mechanism is unclear. Maternal transmission of Wolbachia implies independent infection of each host mtDNA haplotype. Thus our data suggest that wGff in Uganda were repeatedly infected with Wolbachia (See Additional file
1: Table S1), a condition also supported by simulation studies as an initial transient phase in Wolbachia establishment in a new species
[13, 14], but never before observed empirically. Moreover, it is unlikely that flies dispersing from the mixed region are the sole source of Wolbachia infections in Uganda, as we find wGff with unique host mtDNA haplotypes (KK; Figure
1) from a distinct western Uganda tsetse group defined by nuclear microsatellite data (
, See Additional file
1: Table S2). This suggests a second infection potentially from western Uganda. Interestingly, tsetse in Kakoga (KK) and in the Lake Victoria region (Figure
1) carrying rare mtDNA haplotypes are Wolbachia-infected, suggesting a relatively recent infection with a closely related wGff. Since these are rare host mtDNA haplotypes, too few wGff sequences are available to test this hypothesis.
The inclusion of G. f. fuscipes from a distant geographic location (DRC) with extremely divergent mtDNA from the Ugandan flies and from a colony population of a different subspecies (G. f. quanzensis)
[49, 50] infected by the two Wolbachia lineages (Figure
2) allows us to discuss three possible scenarios for the origin of the Wobachia infection(s). First, infection by both Wolbachia lineages was widespread and pre-dates the sub-species divergence. This would lead to a correlation between Wolbachia and host mtDNA divergence
[16, 51], as we see in the maternally transmitted obligate symbiont, Wigglesworthia in the Ugandan G. f. fuscipes from the same regions
. We do not see this in wGff. Second, Wolbachia is shared between geographically disparate samples because there is extensive dispersal (and subsequent gene flow) among G. f. fuscipes populations. In such a case, we would expect to see wGff haplotypes associated with at least one widespread host haplotype. We find wGff associated with the most common host mtDNA haplotypes, but these host haplotypes are not widespread (Figure
4). Further, wGff is not associated with host genetic groups defined by mtDNA variation, but with those defined by nuclear variation, whose patterns likely originated via genetic drift, not gene flow from geographically distant populations
[26, 28, 29]. Third, it is possible that there were multiple independent infections in G. f. quanzensis in DRC and in G. f. fuscipes in Uganda. Although our data appear to only support the last of these hypotheses, our sampling design does not permit us to specifically test any of these hypotheses. However, these hypotheses warrant investigation to understand Wolbachia infection dynamics in this species, as it can shed light on the general evolutionary dynamics of Wolbachia infections, which are not possible to address in other systems that do not have the viviparous life history traits of Glossina species.
Relevance to CI
In the presence of CI, Wolbachia is expected to be associated with few high frequency host mtDNA haplotypes
. In combination with data from
, we found Wolbachia associated with 26 host mtDNA haplotypes. Of the approximately 40 mtDNA haplotypes found in Ugandan G. f. fuscipes[28, 29], more than half are infected with wGff Group 1, and only three with wGff Group 2. Indeed, the host mtDNA haplotypes infected with Group 1 are some of the most common in Uganda (Figure
4), but the low sample size for some host mtDNA haplotypes, due to low infection density, makes it difficult to draw inferences about this pattern. Interestingly, in nearly all of the high frequency mtDNA haplotypes
[28, 29], wGff-infected individuals are more common than those that are not infected. Although these differences are not all significant, it suggests some fitness advantage for infected flies (Figure
4), consistent with occurrence of CI in G. f. fuscipes.
Unexpectedly, our data found wGff associated with rare mtDNA haplotypes, a result also supported by Wolbachia prevalence data (See Additional file
1: Figure S4,
): in 365 flies with known host mtDNA haplotype, wGff was associated with 12 extremely rare mtDNA haplotypes. CI-causing Wolbachia are expected to have higher fitness, driving associated host haplotypes to high frequency, as seen in some of the common groEL haplotypes in our dataset (See Additional file
1: Figure S4). Contrary to the typical observation of a single female driving an infection in insect populations, theoretical studies suggest that before CI can sweep Wolbachia through a population, multiple independent infections must occur
[13, 14]. It is possible that sweeps occur in other insects too rapidly to observe these multiple, independent infections and we may have captured Wolbachia, even in rare groEL haplotypes, due to the unique host life history.
Our genetic data do not provide evidence that bidirectional CI has shaped genetic variability in G. f. fuscipes[28, 29]. Although Group 2 is primarily limited to southern host mtDNA haplotypes (Figure
3, See Additional file
1: Table S2), we found that this association was not significant (p = 0.16). Furthermore, two superinfected individuals have northern host mtDNA haplotypes (JN18, JN6; groEL haplotype 37). This result is unexpected, if we assume solely maternal Wolbachia transmission, and suggests that Group 2 infections have either independently arisen in the northern host mtDNA haplotype lineage, or there is some horizontal transfer of Group 2 infections from southern G. f. fuscipes to individuals found in the northern mtDNA haplogroup. Horizontal transfer among different insect species must occur for Wolbachia to infect novel hosts, but horizontal transfer among different host species with closely related Wolbachia has rarely been empirically documented
. Since either horizontal transmission or independent infections appear to be common in G. f. fuscipes, genetic data may not be the ideal method to detect any form of CI as these processes may obscure host genetic patterns induced by Wolbachia. Furthermore, very few of our samples seem to be infected with Group 2 Wolbachia, potentially reducing our power to detect patterns. In contrast, we see some evidence that wGff is associated with the most common haplotypes (Figure
4), suggesting a potential fitness advantage of wGff. Thus, we suggest that it is crucial to examine transmission efficiency and perform laboratory mating experiments before excluding the possibility of bidirectional CI in G. f. fuscipes.