Other ecological interactions have been suggested as means for bacteria or gene exchange, e.g., host-parasite interactions or double Wolbachia infections [28, 36, 45]. However, in many other cases, opportunities for recombination are less obvious. Transduction involving vectors (e.g., plasmids, phages, or viruses) is a more likely manner of gene exchange. Good vector candidates

are bacteriophages, as these have been isolated from Wolbachia infected populations [60–62] and seem I-BET-762 mouse to be common in Wolbachia genomes [42, 63]. Phylogenetic analyses suggest that the bacteriophage WO is horizontally transferred between different Wolbachia strains, and is able to infect new Wolbachia hosts [60, 61, 64]. Other, free-living, bacteria might even be involved in phage-transfer. We also noted the presence of a bacteriophage in an individual of B. spec. I. The bacteriophage sequence, detected coincidentally with groEL primers, appeared similar to the sequence of the Wolbachia bacteriophage WOcauB1 from Cadra cautella (GenBank: AB161975; 12% p-distance) [65], and to part of the sequenced genome (located within the gene dnaA) of Wolbachia from Drosophila melanogaster (GenBank: AE017196; 11% p-distance). With strict vertical transmission, strong linkage disequilibrium between host mtDNA and Wolbachia would be expected. However, recombination may uncouple such associations, and could be a reason for the

observed lack of congruence between AMN-107 in vitro host mtDNA and Wolbachia STs. There are some signs of congruence, with related host strains (with identical COI sequences) sharing identical or closely related Wolbachia strains, but due to the high rate of recombination such associations are broken up rather quickly. Cardinium diversity For Cardinium, the two investigated genes showed highly similar phylogenies, giving no clear evidence for intergenic recombination. Also, no signs of intragenic recombination were found. There was however no congruence between Cardinium strains and associated host species: similar strains were

found in B. rubrioculus, B. sarothamni, and T. urticae. Only the strain infecting P. harti was clearly distinct from all other strains. The sharing of strains among different host species, and the occurrence 4-Aminobutyrate aminotransferase of divergent strains in one host population (FR21), suggest that horizontal transmission is also prevalent for Cardinium. Horizontal transmission seemed also to explain diversity patterns found for Cardinium infecting Cybaeus spiders [17]. Patterns of recombination and horizontal transfer should however be further studied including more genes. An MLST set for Cardinium is desirable, for reliable strain typing and for investigating patterns of recombination, horizontal transmission, or host manipulation. This requires the use of several independent P505-15 markers, sufficiently distant from each other within the genome.