Detection of Wolbachia infection and distribution in wild mosquitoes
In this study, the PCR-based Wolbachia screening method has a high positive detection rate with 86.3% of all sequenced amplicons having successful BLAST matches to Wolbachia. This suggests that the conventional PCR method is adequate for Wolbachia detection. Even if the study was to proceed without the additional DNA sequencing step, an observation of an amplicon band would likely indicate a true positive. Focusing on our results, Wolbachia is highly widespread across members of the Culicidae family. Here, we report infection in seven mosquito species that have not been previously described for harbouring Wolbachia. Overall, the percentage infection of screened individuals was 43.9% which was largely congruent with percentages reported in past studies from the Oriental region: 31% infection in Malaysia [81], 26.4% in Sri Lanka [39], and 61.6% in Thailand [82]. At the species level, past studies reported Wolbachia infection in 40% of all tested species in India [83], 18.2% in Sri Lanka [39], 51.7% in Taiwan [84], and between 28.1% and 37.8% in Thailand [82,85]. Our study showed that 51.2% of all tested species were infected with Wolbachia which is generally higher than most studies. This is likely attributed to the broad range of species tested in this study, including species from the genera Malaya, Verrallina, and Zeugnomyia [85]. It is also possible that infection prevalence may vary across geographical regions.
Wolbachia detection in three medically important mosquito genera, Culex, Anopheles, and Aedes, was highly consistent with past studies. These genera are responsible for the transmission of vector-borne diseases such as filariasis, malaria, and arboviral diseases [86]. Among the Culex mosquitoes, Wolbachia infection has been reported to be variable across its member species [39,46,82,84]. Similarly, infections were observed only in five out of 16 Culex species. We noticed moderately high Wolbachia infection in Culex quinquefasciatus (62.5%) which is a member of the Culex pipiens complex responsible for the transmission of filariasis worm disease in Singapore [86,87]. Surprisingly, between two closely related species, Culex pseudovishnui and Culex vishnui [88], no Wolbachia infection was observed in the latter which has been found to harbour Japanese encephalitis virus in the Southeast Asian region [89]. However, studies in India and Thailand showed a reverse pattern, with Wolbachia infection present in Cx. vishnui but not in Cx. pseudovishnui [39,85]. Although the two species are morphologically similar [53], in this study, DNA barcoding was conducted to aid morphological identification and thus, avoid any misidentification. This lends further support that infection prevalence may vary between populations which are distal geographically.
We did not detect Wolbachia in any of the wild-caught Anopheles species (18 individuals representing three species) examined in this study, many of which are potential malaria vectors [86]. This is largely consistent with previous reports published globally [39,90,91]. The absence of Wolbachia in Anopheles mosquitoes is thought to be due to the unsuitability of Anopheles reproductive tissues for Wolbachia establishment [84,85]. However, in recent years, there are reports Wolbachia detection in field Anopheles mosquitoes from West Africa [42,92,93] and Malaysia [94]. Knowledge of natural Wolbachia infections in Anopheles mosquitoes has implications on malaria control strategies [93], hence more wild-caught Anopheles samples should be screened to determine the infection status in Singapore more accurately.
Wolbachia was not detected in Aedes aegypti, the primary vector of dengue in the Southeast Asian region [87]. Conversely, Wolbachia infection was moderately high in the secondary vector Aedes albopictus. This pattern is highly consistent with past studies which reported an absence of infection in wild Ae. aegypti [21,95], but found stable infection in wild Ae. albopictus [96]. Although Ae. aegypti and Ae. albopictus belong to the same subgenus Stegomyia, and occupy similar ecological niches [97], they are rarely found in the same locality which was likewise observed in this study [43,98,99]. This could imply a certain degree of competitive exclusion between the two species, preventing them from occupying the same space. There is evidence showing that symbionts may influence host’s resource acquisition and specificity which ultimately lead to competitive exclusion between closely related host species with differing symbiont infections [100,101]. However, research on Wolbachia-induced competitive exclusion is scarce except for a few studies on heterogonic gall wasps [102], grasshoppers [103], and gall-inducing aphids [104]. Given the widespread influence of Wolbachia, future research can explore potential cases of Wolbachia-induced competitive exclusion between closely related species which will have a huge implication on understanding symbiosis and speciation.
Additionally, given the frequent artificial Wolbachia infection into Ae. aegypti for bio-control purposes [105–109], our findings could suggest that infected Ae. aegypti might not be stably maintained in the wild. This can be advantageous for vector population suppression as the cytoplasmic-incompatibility effect of any artificially introduced Wolbachia strain will likely be fully manifested in the uninfected native population [21]. However, this also implies that such a bio-control method may have low long-term effectiveness if the infection cannot be naturally sustained in the wild population. The detection of natural Wolbachia infection in wild Ae. aegypti, therefore, has a huge implication on vector control programmes [21]. Not only does it inform the selection of suitable Wolbachia strain prior to its field-release, but it can also be used to gauge the long-term effectiveness of the vector control programme.
Interestingly, the sex of the mosquitoes had an effect on Wolbachia infection status. This could be an artefact of the various Wolbachia-induced reproductive phenotypes such as parthenogenesis and male-killingresulting in offsprings which are largely female [15]. Over multiple generations with vertical Wolbachia transmission, one would observe an increasing proportion of females that are infected. Hence, this phenomenon could be a consequence of Wolbachia’s reproductive manipulation and vertical transmission.
While we were unable to statistically test for the effects of locality on infection status due to uneven and small sample sizes of the respective species across different localities, our results suggest that mosquitoes found in localities across Singapore have roughly equal chances of harbouring Wolbachia. This also suggests that underlying physiological factors and phylogenetic relatedness in mosquitoes contributed more to the Wolbachia infections than the habitat which they are found in.
The reproductive effect of Wolbachia can be masked or enhanced by other reproductive endosymbionts such as Cardinium, Rickettsia, and Spiroplasma [7,26–29]. Unfortunately, we were unable to detect those endosymbionts due to a high degree of false positives using PCR-based screening methods [Additional file 1]. This is likely attributed to primers which are not optimised for screening mosquito-specific endosymbionts [110–112]. As a result, co-infections of various reproductive endosymbionts were not identified among wild mosquitoes which would have provided greater insights into the synergistic effects of co-infections on mosquito evolution. There is, hence, a need to develop and optimise alternative screening methods, such as multilocus sequence typing (MLST) techniques, especially for the detection of Cardinium, Rickettsia, and Spiroplasma in mosquitoes.
Tissue tropism of Wolbachia infection in mosquitoes
Wolbachia was detected mainly in the reproductive tissues which corroborates with results from studies across multiple insect groups [15,84,113],suggesting that Wolbachia is mainly vertically transmitted. Interestingly, through the course of this study, there was a significant variation in the size of reproductive traits (testis and ovary length) across and within species. These reproductive traits did not vary significantly with Wolbachia infection status, even after accounting for phylogenetic relatedness [see Additional file 2].
Infection in the gut and leg tissues was detected, albeit infrequently. This is not surprising as previous reports have detected Wolbachia in those tissues [34–36,114]. Interestingly, the nucleotide sequences from gut and leg infections tend to be shorter in length. Considering that Wolbachia is unlikely to survive extracellularly for a long duration [35], the small amplicon size suggests potential horizontal integration of the Wolbachia genome into the host genome for a few species. This phenomenon has been observed in several Wolbachia hosts [115,116], and mosquito species such as Aedes aegypti and Culex quinquefasciatus [117,118]. For instance, a recent study showed that horizontal integration of Wolbachia genome into the host genome can have sex determination and evolution implications. This is evident in the common pillbug Armadillidium vulgare, resulting in the formation of a new sex chromosome [119]. Researchers have also proposed that horizontal gene transfer between endosymbiont and host can result in evolutionary innovation where new functional genes arise for both host and bacteria [117,118].
Future research should explore the relative importance of each transmission method with relation to host-endosymbiont ecology and evolution. Such tissue-specific screening methods can be used in other arthropods especially when the mode of transmission is not clear. Currently, most Wolbachia screening is conducted on grounded specimens or specimens in their entirety [39–41]. By doing so, researchers would be unable to determine tissue tropism of Wolbachia infection which could provided clues to its mode of transmission. In this context, adopting tissue-specific screening methods can seek to verify or refute the assumption that Wolbachia is transmitted vertically which is common in literature [15,30].
Diversity and host-specificity of Wolbachia strains
Not only does the wsp molecular marker allow successful detection of Wolbachia infection across numerous taxa, it also enables strain genotyping and evolutionary comparison between detected Wolbachia strains [60]. In this study, Wolbachia wsp sequences were clustered into twelve putative Wolbachia strains falling within the supergroup A or B. This is consistent with previous studies that looked at Wolbachia infections in mosquitoes [39,80,85]. Each mosquito host species was only infected by strains belonging to A or B, with the exception of Aedes albopictus which harboured both. Infection of more than one strain (superinfection of wild Ae. albopictus with Wolbachia supergroup A and B) has been previously reported, and this phenomenon was commonly observed to be fixed in those examined populations due to strong cytoplasmic incompatibility effects [120,121]. This suggests stable vertical transmission of both strains in Ae. albopictus. Additionally, only four out of twelve putative strains were identified to previously typed Wolbachia strains reported by Zhou et al. [60] and Ruang-Areerate et al. [80] – Wol 1, Wol 2, Wol 3, and Wol 8 were identified as wPip, wAlbB, wCra, and wRi strain respectively.
Host specificity is thought to be a characteristic of the ancestral Wolbachia strain, with host flexibility reported mainly in Wolbachia supergroup A and B [122]. In our study, we found a combination of specialists and generalists with more counts of the former. A study of mosquitoes from Taiwan showed a similar pattern [84]. In bees, a mixture of Wolbachia supergroup A host-specific and host flexible strains in the population has also been reported [49]. While our estimates of specialists and generalists might vary with greater sampling effort, the higher numbers of specialists observed can be explained by the process of reciprocal selection between host and endosymbiont over evolutionary time [123]. This is also known as the “Red Queen” dynamics, where the endosymbiont constantly adapts to its host to ensure continued establishment in the same host [124]. An alternative strategy of being a generalist can also be maintained in a population. It ensures survivorship in an environment where resources (i.e. hosts) are rarely found [123]. However, there are generally more instances of host specialists than generalists across numerous parasitic and endosymbiotic taxa [125–127].
The standardised phylogenetic host specificity scores revealed that host flexibility among generalists varied greatly. Understanding Wolbachia host specificity has huge implications especially for the optimisation of Wolbachia biocontrol strategy. Not only should researchers select strains which can effectively limit pathogen replication [128], they should also select strains for their host specificity. This would not be possible without the screening of a wide variety of species or closely related species which was achieved in this study. A host-specific strain will decrease the likelihood of infection host shift to non-target species, thereby minimising the strategy’s overall ecological risk.
Evolutionary relationship between mosquito and Wolbachia
Host-Wolbachia relationships are often understudied and limited to a few taxa [52]. Studies have shown that the evolutionary associations between Wolbachia and their insect hosts do vary across taxa [49–52,129]. Likewise, our exploratory analyses of the mosquito hosts and their Wolbachia support such a complex relationship, with neither co-speciation nor host shifting fully accounting for the evolutionary association in these lineages.
Based on the tanglegram, a broad association pattern between mosquitoes and Wolbachia strains was observed (Fig. 3). Aedes mosquitoes tend to be associated with Wolbachia supergroup A, while other species, particularly of the genus Culex, were largely associated with Wolbachia supergroup B. This showed that closely related Wolbachia strains are likely to establish themselves in related hosts. There might have been radiation of Wolbachia in these clades after their respective initial establishments. Nevertheless, the observed variations on host-endosymbiont associations make the mosquito-Wolbachia association pattern still questionable.
The ParaFit analysis showed weak support for congruency between host and endosymbiont phylogenies. Among the 18 host-Wolbachia associations, only the link between Mansonia indiana and Wol 3 showed a significant association (Fig. 3). This was interesting considering that Wol 3 was largely host flexible. Given that this was the only significant association, it is worth carrying out further genus-specific study on Mansonia spp. to elucidate coevolutionary patterns within a group of closely related mosquito species. Perhaps, the degree to which Wolbachia coevolves with its mosquito host vary across different taxa levels [74]. The analyses thus far suggest that mosquito-Wolbachia associations are likely random at higher taxonomic levels with occasions of mosquito-Wolbachia co-speciation at finer phylogenetic resolution (i.e. similar to patterns seen in diffuse coevolution).
Referring to the event-based analysis performed in Jane 4.0 (Fig. 4), co-speciation events were infrequent as compared to other evolutionary events. We noticed a greater proportion of host shifts and numerous losses. Interestingly, the least cost coevolutionary reconstruction indicated multiple consecutive host shifts occurring near the tips of the cladogram. This suggests that co-speciation does not fully explain the evolutionary associations between mosquito hosts and Wolbachia. Instead, recent host shifting through horizontal transmission seems to promote Wolbachia diversification. This lends greater support that horizontal transmission between distantly related species is possible [32,33,130].
Furthermore, losses, which represent endosymbiont extinction events that occurred upon host speciation, seemed to dominate the evolutionary history of Wolbachia. Extinction events are believed to be frequent in host-endosymbiont systems [123], due to either evolution of resistance in the host or declining host population size which results in the inability for highly specialised endosymbionts to establish themselves [131,132]. Additionally, losses could potentially influence endosymbiont evolution through the creation of vacant niches [131]. The observed losses followed by host shifts in the mosquito-Wolbachia relationship are possible consequences of vacant niche exploitation by generalists. Perhaps, this enabled successful endosymbiont invasion due to minimal intra-strain competition. Therefore, horizontal Wolbachia transmission and losses may play a bigger role in accounting for Wolbachia diversity than previously expected.
As this was an exploratory study, we were unable to determine the exact mechanism behind Wolbachia’s diversity and evolutionary association. The presence of numerous specialists could be a sign of mosquito-Wolbachia coevolution since coevolution is fundamentally reciprocal selection between host and endosymbiont which gives rise to micro-evolutionary changes [133]. The numerous host shifts and losses might have, however, blurred the effects of vertical transmission over a long evolutionary period [52]. Thus, co-speciation might have occurred within smaller clades of Wolbachia and mosquitoes, but at higher taxa levels, horizontal transmission and loss events are more likely the prominent force driving Wolbachia evolution.
Strengths, limitations, and future direction
Three distinct methods were employed to explore evolutionary associations given their respective strengths and limitations: (i) The tanglegram allows for clear visualisation of host-endosymbiont association without taking into account any evolutionary relationships, but there are calls for careful interpretation as the degree of entanglement may not necessarily represent phylogeny congruency [134]. (ii) The global ParaFit test seeks to address this limitation by testing for global congruency in an unbiased, statistical approach [74]. (iii) The event-based method enables evaluation of potential evolutionary events that might have occurred throughout the endosymbiont’s evolutionary history such as co-speciation, duplication, and host shifting. This last method, however, cannot fully differentiate a topological congruence from an evolutionary event [135]. Without the time of divergence for both symbiont and host, a co-phylogenetic pattern may be better explained by ecological factors (as compared to co-speciation) given that bacterial lineages often evolve faster than the hosts [136,137], and the high likelihood of host shifts among closely related species [133].
The Wolbachia wsp gene has been shown to provide phylogenies with a good resolution [60], and this study provides an exploratory snapshot of the evolutionary associations between mosquito hosts and their Wolbachia endosymbionts. Of course, this is a potential caveat since only a single gene was used to construct the respective phylogenetic trees. To obtain a more accurate phylogeny, future studies can adopt MLST [17,51], or whole-genome shotgun sequencing in their methods [52]. The former could potentially characterise putative Wolbachia strains that cannot be distinguished with wsp gene primers.
Notwithstanding the limitations, the employment of various analytical methods allows for a well-rounded and comprehensive examination of the evolutionary association between Wolbachia and mosquito hosts which are lacking in current literature. Future studies interested in the evolution of medically important vector species could narrow their scope on the Aedini tribe which will provide greater statistical power for the examination of mosquito-endosymbiont association.