Identifying horizontal transfers, recombination events and test of endosymbiont similarity:
To test for horizontal transmission of endosymbionts across the soil arthropod hosts, two sets of analyses were done. The first was a visual estimation of the horizontal transfer obtained by comparing host and endosymbiont phylogenies. The second was a quantitative estimation of the correlation between the pairwise distance matrices of hosts and their corresponding endosymbionts. The Spearman method (r ) of Mantel test for correlation (Legendre & Legendre, 2012) was determined in R v1.2.5. A total of 5 different correlations were computed with their corresponding host sequences: all the Wolbachia infected samples (33), only A supergroup Wolbachia infected samples (16), only B supergroup Wolbachia infected samples (15), onlyCardinium infected samples (8) and Arsenophonus infected samples (7).
To identify and visualize recombination events within theWolbachia concatenated MLST sequences (2079bp), SplitsTree v4.14.8 (Huson, Kloepper, & Bryant, 2008) was used to construct phylogenetic network using uncorrected p -distance and Neighbor-net method (Bryant & Moulton, 2004). To statistically evaluate presence of recombination, Φ test (Bruen, Philippe, & Bryant, 2006) was used in SplitsTree v4.14.8. These identified events were then evaluated using RDP4 v4.97 (Martin, Murrell, Golden, Khoosal, & Muhire, 2015) which has several suits of programs to detect and identify recombination events (like RDP, GENECONV, MaxChi, SiScan, BootScan, Chimaera and 3Seq). A recombination event was considered significant if it was shown to be statistically significant by at least three or more detection methods. Recombination breakpoints were also identified with RDP v4.97.
To test whether the endosymbionts are largely being transferred within the community, and thereby have less sequence divergence than expected, we used the model by Baldo et al. (2008) to test the similarity of endosymbionts within community. Briefly, the mean pairwise distance of endosymbionts presents in soil arthropod community were compared with an equivalent number of pairwise distances randomly selected from the database for 10000 iterations. These iterations were computed to give a null distribution for comparison with the soil endosymbiont sequence data by Wilcox rank-sum test with continuity correction and 95% confidence interval (performed in R v1.2.5). From our sampling, we found 17 unique ST’s for Wolbachia supergroup A and 15 forWolbachia supergroup B infections. Whereas, in the PubMLST database there were 228 unique ST’s for A supergroup and 252 ST’s for B supergroup (last checked May 2019). For the 9 samples infected withCardinium and 8 for Arsenophonus , 248 and 228 homologues, respectively, were obtained from NCBI (last checked May, 2020). Pairwise distance was calculated using MEGA7 (Kumar et al ., 2015) and then corrected with Jukes and Cantor model in DNAsp v5.10.01 (Librado & Rozas, 2009). Density plots for endosymbiont divergence from the soil arthropods and the databases were plotted in R (http://R-project.org).