With roughly 200,000 species of which at least 90% are endemic to the Island, Madagascar is one of the Earth's richest biodiversity hotspots (Paulian, 1987; Myers et al., 2000; Vences et al., 2009). For Malagasy dung beetles, the level of endemism reaches 96% (Miraldo et al., 2011). Like for many other organisms that have evolved on the island since its isolation from other land masses since the late Cretaceous era (Noonan and Chippindale, 2006), the high level of species radiation in dung beetles has been attributed to different eco-evolutionary mechanisms. Strong environmental gradients and ecosystem heterogeneity across the island have often been suggested as some of the main drivers of the Island's exceptional biodiversity (Vences et al., 2009; Miraldo and Hanski, 2014). In contrast, little is known about the potential influence of biotic factors, such as the presence of endosymbionts, in the diversification of Malagasy species.

The genus Nanos includes 42 species of dung beetles endemic to Madagascar (Montreuil et al., 2014). This clade is considered one of the most recent and ecologically successful dung beetle lineages in Madagascar (Wirta, 2009; Wirta et al., 2010). The phylogeny of this clade was recently revisited by combining morphological and molecular data from both mitochondrial and nuclear markers (Wirta, 2009; Montreuil et al., 2014). This work improved our understanding of the phylogenetic relationship between many morphologically similar Nanos species, but also highlighted potential introgression between species, especially within the N. clypeatus group (Wirta et al., 2008; Wirta, 2009; Montreuil et al., 2014). Furthermore, Wirta (2009) showed that the mitochondrial DNA of the six species from the N. clypeatus group was less variable than that of the rest of the Nanos species.

Maternally inherited endosymbiotic bacteria act as a selfish entities that can modify their host reproductive strategies to their own benefit (O'Neill et al., 1997; Dyson and Hurst, 2004; Engelstädter and Hurst, 2007). Text-book examples include the association between Wolbachia strains inducing cytoplasmic incompatibility (CI) in their hosts. Due to CI, the zygotes of uninfected female hosts mated to Wolbachia-infected males die, while zygotes from Wolbachia-infected females thrive in the environment (Carrington et al., 2011; Kriesner et al., 2013; Turelli et al., 2018). In these conditions, Wolbachia can obscure studies of mitochondrial variation in the host population. The mitochondrial selective sweep associated with the invasion of the inherited endosymbiotic bacteria can indeed reduce the mitochondrial diversity observed in infected lineages when compared to uninfected ones, and confuse conclusions from phylogenetic studies (Jiggins, 2003; Charlat et al., 2009; Jäckel et al., 2013; Kriesner et al., 2013; Mazur et al., 2016; Cariou et al., 2017). Furthermore, studies have suggested that the induction of CI can also select for premating isolation between matrilines carrying different infection status (Telschow et al., 2005a, 2007). The selection pressure for the evolution of mate-choice strategies toward individuals carrying the same infection status can lead to strong premating isolation, which in time will support reproductive isolation between the host lineages, and lead to speciation (Telschow et al., 2005a, 2007; Cariou et al., 2017).

Here we suggest that infections with Wolbachia strains might play a role in the loss of mitochondrial diversity observed in the N. clypeatus group, which includes six dung beetle species. We also hypothesize that incompatible Wolbachia infections could keep the different taxa as clear evolutionary units despite clear genetic introgression patterns between the N. dubitatus and N. vieittei species. We discuss the potential implications of these findings for the evolution and diversity of the dung beetles in Madagascar.

Our study is the first to report on Wolbachia infections in Malagasy dung beetles. We detected Wolbachia infection in 56 specimens (47%) from four of the six dung beetle species screened (67%). This is consistent with current estimates proposing that over 40% of worldwide insect species (Stouthamer et al., 1999), and not so different from lower infection rate estimates detected in other families and species of Coleoptera (Kawasaki et al., 2016; Kajtoch and Kotaskova, 2018; Kolasa et al., 2018). Five of the Wolbachia variants we characterized belong to the Wolbachia A-supergroup, and are genetically very similar to each other (at most 42 polymorphic sites across 1,800 bp sequenced). The last infection remains to be characterized, or even confirmed. These results suggest the island is potentially not only a biodiversity hotspot for dung beetles, but also a diversity hotspot for the endosymbiotic bacteria they carry.

Each strain shows a very different prevalence within the respective host species. Studies have shown that variations in the prevalence of Wolbachia are not rare in insects (Sazama et al., 2019), and these patterns of infection can be the results of the host, the symbiont, or both (Charlat et al., 2005; Hornett et al., 2006). The successful spread of the Wolbachia strain wNvie_A in all N. pseudoviettei specimens, and almost all N. viettei specimens may suggest that the strain is beneficial to its host. As mentioned earlier, CI-inducing strains might rapidly spread in their host populations (Turelli et al., 2018), as the infected females produce viable offspring from mating with either infected or uninfected males, while survival of the offspring from uninfected females is restricted to the females mated to uninfected males (O'Neill et al., 1997; Telschow et al., 2005b). It is thus possible that wNvie_A is a CI-inducing strain. However, Wolbachia can also spread across their host populations through enhancing other aspect of its host's fitness. Recent studies have shown that Wolbachia can promote the survival of its host against pathogens such as viruses (Hedges et al., 2008; Teixeira et al., 2008; Osborne et al., 2012). Many dung beetles lay their eggs directly in patches of dung or carrions, or in underground chambers filled with these materials (Viljanen, 2009a,b). The Nanos beetles are all rollers, rolling away samples of dung or carrions on which their larvae later feed and develop (Hanski et al., 2008; Rahagalala et al., 2009; Viljanen, 2009b; Viljanen et al., 2010). These environments are often rich in pathogenetic micro-organisms, including bacteria, viruses, and fungi, that may threaten the survival of the offspring. Whether the common strain wNvie_A, or any strain described in this study, can protect their dung beetle hosts against pathogens from their environment remains to be investigated.

In the tropical butterfly Hypolimnas bolina, Duplouy et al. (2010) have previously shown that geographical variations in the prevalence of a Wolbachia strain across the South Pacific populations of the butterfly results from temporal variations in the history of the symbiosis on the different islands. Populations showing a low infection rate have thus more recently acquired the infection than populations showing a higher infection rate (Duplouy et al., 2010). The strain variant wNvie_A is found in almost all N. veittei and N. pseudoveittei specimens, and in association with a wide diversity of mitotypes in these species, we thus suggest that these host-symbiont associations might be rather old. The other five strains are much rarer in their respective host populations. This pattern might either suggest that those strains are new to their respective hosts and restricted to few mitotypes in the populations, or that the strains are being purge from their host populations, due to potential costs of the symbiont (Charlat et al., 2009; Duplouy et al., 2010).

Nanos viettei is a taxonomic unit that is, based on both the mitochondrial and nuclear DNA, very well-separated from all other species from the N. clypeatus group (Montreuil et al., 2014). Wirta (2009) calculated that the split between this species and the other five species in the group occurred at most 7.2 Mya. All strain variants characterized in these dung-beetle species are very similar to each other, but their phylogenetic tree is not congruent with that of their host. Such incongruence has been reported in many host-symbiont associations (Jäckel et al., 2013; Sontowski et al., 2015; Kolasa et al., 2019), including Wolbachia-infected insects of Madagascar (Linares et al., 2009; Zouache et al., 2011). Although it is possible that some of our Nanos dung-beetle species inherited their Wolbachia infections from a common ancestor, while others have lost or gained new infections during species radiation, our data also does not suggest radiation of Wolbachia in parallel to the radiation of N. clypeatus group.

Although mostly vertically transmitted from the mother to the offspring, Wolbachia is also known to transfer horizontally between host species (Vavre et al., 1999; Bailly-Bechet et al., 2017). Such transfer of infection may occur during hybridization events between species. For example, in Drosophila mauritania, different populations bear the mitotypes of another fruitfly species, D. simulans, and variants of the Wolbachia strain wMau, most likely due to transfer of both cytoplasmic entities during recent introgression events (Rousset and Solignac, 1995). However, other ecological links are likely to also support the horizontal transfer of symbiotic bacteria between host species. Species with similar resources, such as larval host plant, or species sharing parasites, such as parasitoid or mite infections, may thus also share similar Wolbachia infections (Stahlhut et al., 2010). Ecological studies on the Nanos species have shown that N. veittei mate with other Nanos species in the laboratory (Viljanen, 2009a,b). The survival of the offspring has however never been investigated. Furthermore, previous phylogenetic studies do not suggest introgression between the populations of N. pseudoveittei and N. viettei (Miraldo and Hanski, 2014). In contrast, the two species co-occur in the northern limit of N. viettei geographical range (Viljanen, 2009a; Miraldo et al., 2011; Montreuil et al., 2014). There, these two species, which are generalists in their choice of dung, may enter in contact and share dung patches or parasites. These ecological links may have facilitated the horizontal transfer of Wolbachia between the two species (Kolasa et al., 2019). Unfortunately, little is known about the parasitoid community associated to the Malagasy dung beetles, and no study has ever investigated the potential for dung as a suitable environment for symbiont transfer between carrion insects.

Strikingly, some of the rare strain variants characterized in this study are found in association with specimens showing clear genetic patterns of introgression. Strain variant wNvie2_A, a close variant of wNvie_A, is found in two N. viettei that carry the mitochondrial DNA from N. dubitatus. These particular specimens were collected from a region where extensive hybridization is known to happen between the two sympatric host species (Viljanen, 2009a; Wirta, 2009). Previous studies described these introgressed individuals as clear units within the N. viettei species that were however separated by unknown processes (Miraldo and Hanski, 2014). We suggest that sympatry of the two species may have facilitated the transfer of the Wolbachia between specimens. However, the divergence between strain variants wNvie_A and wNvie2_A may have led these specimens to become incompatible potentially due to CI. Such postzygotic incompatibility would have, with time, led to radiation of the introgressed lineage carrying wNvie2_A from the other N. viettei, wNvie_A-infected lineages. Simultaneously, because wNvie2_A-infected N. viettei individuals are morphologically similar to N. viettei, they might rarely mate with the locally occurring N. dubitatus specimens. Again, with time, prezygotic isolation may have also isolated this lineage from the other N. dubitatus lineages. The strain wNdub_A, a second close variant of wNvie_A, is found in N. dubitatus specimens. Both strains and their respective host species do not occur in sympatry in Madagascar. Although wNdub_A was identified in association to a rather wide diversity of mitotypes in N. dubitatus, our current sample size is unfortunately too small to further speculate on the potential phenotypic effects this strain might have on its hosts, or on the origin of the strain in N. dubitatus.

Despite the fact that our restricted sample size is preventing a more comprehensive assessment of prevalence rates and any evidence of geographical variation between collection locations, our results still support the idea that Wolbachia might play a key role in the ecology and evolutionary biology of the endemic Malagasy dung beetles. Viljanen (2009b) showed that some Nanos species can mate in the laboratory, thus it is possible for future studies to confirm whether the different strains of Wolbachia induce CI in these insects. Only then will we be able to confirm whether Wolbachia play a key role in designing host species boundaries, or can in any ways enhance their host fitness and contribute to niche colonization and rapid geographical expansion of these ecological successful dung beetle species.

The datasets generated for this study can be found in GenBank, Accession codes are provided in the manuscript.

The study only include work on DNA extracts from samples collected under the Academy of Finland projects to late professor Ilkka Hanski. All samples were collected prior to 2014, no Nagoya protocol was then required.

AD designed the research and wrote the manuscript. AM and AD collected the data, analyzed the data, and revised previous versions of the manuscript.