Results

The output of our bioinformatics pipeline successfully recovered 94.03% of all Wolbachiaamplicons and yielded read count numbers between 4658 and 36548 per sample (summed over all seven Wolbachia genes; for details see File S1 and S2). This high coverage enabled us to identify a diversity of Wolbachia strains within and across hosts. Among 75Wolbachia -positive samples (59 scale insect samples and 16 associates), 68% were infected with a single Wolbachia strain and 32% were infected with more than one Wolbachia  strain (20 double and 4 triple infected). Among nine scale insect species with samples from more than one population, intra-species infection polymorphism was observed in Cryptes baccatus, Eriococcus confusus, Icerya purchasi, Parasaissetia nigra and Sphaerococcus ferrugineus (File S2). Among 29 scale insect species that were screened in this study, 6 species were always found co-infected (File S2). For example, two screened samples ofAkermes scrobiculatus were coinfected with strains belonging to Supergroup A (w Ake1) and Supergroup B (w Ake2) and two samples of Cystococcus echiniformis were coinfected with strains belonging to Supergroup A (w Sph1) and Supergroup F (w Cys1).
A total of 63 strains were identified and clustered into 31 strain groups belonging to three Wolbachia -supergroups (Figures 1, S2 and File S4). Most of the strains belong to Supergroups A (38) and B (21), but we also identified three strains groups from Supergroup F:w Cys1 and w Sph5 (respectively infecting Cystococcus echiniformis  and Sphaerococcus ferrugineus ), and w Sph3 (infecting two specimens of S . ferrugineus ). Based on the MLST database, these are the first Supergroup F strains reported in Australia. Although w Cys1 is placed within Supergroup F, it forms a unique clade compared to other reported Supergroup F strains (Figure S2).
The most diverse and abundant Wolbachia  strain group in our dataset is w Sph1, which includes 12 closely related strains and was detected in 23 samples belonging to eight scale insect, four wasp and one ant species (Figure 2, S3 and File S2, S4). Similar strains which are grouped within w Sph1 were reported before in several Australian ant species (MLST ST = 54, 19, 478 and 112) (Russell 2012). Based on the MLST database, it appears that this strain group has a cosmopolitan distribution (Oceania, North America, Europe, Asia and South Africa) and has already been reported in various insect groups (e.g., ST 19 in Coleoptera, Hymenoptera, Lepidoptera and Orthoptera). By contrast, some of the scale insect species are infected with uniqueWolbachia strains that were not observed in other scale insects or reported in any other insects (by searching both MLST and GenBank on October 1st2021), including infection of Apiomorpha variabilis withw Aphi1 and co-infection of Coccus hesperidium withw Coc1 and w Coc2.
ML trees based on a 947 bp alignment of scale insect genes (including COI, 28S and 18S), and based on 3065 bp aligned Wolbachia genes (including MLST, wsp, and 16S) are shown in Figure 2. In addition, an interactive figure of Wolbachia sharing among all host species, including associates, is provided in File S5. Evidence of numerous host shifting events can be seen in these figures. Both the Parafit test (ParaFitGlobal = 0.0008, P -value = 0.27) and the PACo test (m2XY = 49.3897, P -value = 0.2056) were non-significant. Therefore, there is no evidence of phylogenetic congruence between Wolbachia and their scale insect hosts, in the sense that non-independent evolution of the two groups is not supported by these tests.
Our Wolbachia sharing models revealed that incorporating phylogenetic distance substantially improved model fit (change in DIC = -5.88), and had a significant effect in the model (P<0.0001). The effect was highly non-linear, with high sharing probabilities at high relatedness that quickly dropped to near zero at greater phylogenetic distances (Figure 3). In contrast, incorporating geographic home range overlap slightly improved model fit (change in DIC = -2.76), but had no significant effect in the model (P=0.199). Inspecting the shape of the effect was not revealing. Furthermore, there was no significant effect of geographic distance between sampling locations (change in DIC > -2). Therefore, we do not interpret this effect as representing strong support for geographic effects onWolbachia sharing.
In searching for Wolbachia strains in pairs consisting of a scale insect and its direct associate, we found the same Wolbachiastrain groups in one out of five ants, one out of two flies, and three out of seven wasps. In the ant-Eriococcus sp1 pair that is co-infected with w Sph4 andw Apio2 strain groups, both the ant and the scale insect are infected with the same w Sph4 strain (w Sph4.4), but different w Apio2 strains (w Apio2.4 in the ant andw Apio2.5 in the scale insect, which differ by 4bp in gatB gene). The three wasp-scale insect pairs with similar Wolbachia strain groups have a single infection (wasp- Ascelis schraderi pair withw Sph1.11, wasp- Apiomorpha floralis pair with w Api 2.1 and wasp-Lachnodius eucalypti pair with w Lac). The fly-Icerya seychellarum pair share identical strains ofw Cal2.2. We did not find the same strain group in the single tested moth-scale insect and two beetle-scale insect pairs. As in scale insects overall, the most common strain group shared between scale insects and associates is the w Sph1 strain group (File S2).