Parasite species drives microbiome composition
Our results suggest that the bat fly families Streblidae and Nycteribiidae may have different primary symbionts in Brazil (Figure 2).Arsenophonus has been previously identified as a primary symbiont of bat flies (Morse, Dick, et al., 2012; Morse et al., 2013; Trowbridge et al., 2006; Wilkinson et al., 2016). The high relative abundance and prevalence of Arsenophonus in all streblid bat flies sampled in our study is consistent with the hypothesis that Arsenophonusacts as the primary symbiont in the family Streblidae. However, it is unlikely that Arsenophonus is acting as the primary symbiont of the nycteribiid species that we sampled, given its absence or low relative abundance. Only one previous study which examinedArsenophonus as the primary symbiont of bat flies included representatives of New World Nycteribiidae (Morse et al., 2013). Using cloning and Sanger sequencing, those authors detected anArsenophonus variant in two individuals of Basilia boardmani (Nycteribiinae), which is restricted to North America. The strain detected was distantly related to the Arsenophonussymbionts found in all other examined members of the subfamily Nycteribiinae and similar to those found in triatomine bugs. In combination with the evidence presented here, these findings suggest that Brazilian Basilia spp. rely on a different endosymbiont than both the Streblidae and North American and Old World members of Nycteribiidae.
In contrast to what we found in streblids, nycteribiid species were dominated by Wolbachia and Bartonella (Figure 2), both of which have been previously detected in bat flies (Morse, et al., 2012; Wilkinson et al., 2016). Wolbachia is a primary symbiont in some insects (Dedeine et al., 2005; Hosokawa et al., 2010; Nikoh et al., 2014), but acts facultatively in others (Pontes & Dale, 2006), and also sometimes functions as a reproductive parasite, causing mitochondrial DNA sweeps as it increases its own prevalence in a population (Cariou et al., 2017; Jiggins, 2003; Lack et al., 2011; Speer et al., 2019).Bartonella is an intracellular pathogen found in many mammalian groups, including humans (Breitschwerdt & Kordick, 2000).Bartonella is common in bats and bat flies (Stuckey et al., 2017), with bat flies potentially acting as a vector to bats (Morse, et al., 2012). It may be that Wolbachia and Bartonella both have high relative abundance in nycteribiid bat flies because they are acting as primary or facultative symbionts, or because they are acting as a reproductive parasite and pathogen, respectively. If the latter is true, it may be that one of the less abundant bacteria, likeCandidatus Phlomobacter11Standard nomenclature for a candidate genus., Mycoplasma , or Rickettsia , is the primary symbiont or another microbe not detected by 16S rRNA sequencing (e.g. fungi; (Gibson & Hunter, 2010).
In addition to family-level differences between bat flies, microbiome composition was parasite species-specific (Table 2; Figure 3). Host-specificity of arthropod microbiomes has been found previously in tsetse flies (Glossinidae; Aksoy et al., 2014), which are also members of the Hippoboscoidea. That microbiome composition shows such a strong signal of parasite species identity indicates that even non-maternally inherited bacteria may be maintained through life history traits (e.g., host bat associations, microclimatic preference). The impact of bat species identity on the parasite microbiome is difficult to parse given the hierarchical nature of the association of bat fly species with specific bats species. Where bat species and bat sex seem to have a large impact on the bat fly microbiome when considered as individual variables (Table 2), these effects disappear when the interaction between parasite species and each host bat variable is taken into account (Table 3). We did not have information on bat roosts, but this has been previously shown to have a large impact on bat fly abundance and prevalence (Hiller et al., 2020). As indicated by our network analyses of landscape variables, it may be that PERMANOVA is good at identifying variables that have a strong impact on variation (i.e., parasite species), but is not the right approach for teasing apart the impact of variables of small effect in this type of complex system (i.e., host bat and landscape).