Bartonella infections from wild mice and voles (Apodemus flavicollis, Mi. oeconomus, Microtus arvalis and Myodes glareolus) were sampled from a forest and old-field habitats of eastern Poland; a complex network of Bartonella isolates, referrable to B. taylorii, B. grahamii, B. birtlesii and B. doshiae, was identified by the sequencing of a gltA fragment, comparable to previous studies of Bartonella diversity in rodents. Nested clade analysis showed that isolates could be assigned to zero- and one-step clades which correlated with host identity and were probably the result of clonal expansion; however, sequencing of other housekeeping genes (rpoB, ribC, ftsZ, groEl) and the 16S RNA gene revealed a more complex situation with clear evidence of numerous recombinant events in which one or both Bartonella parents could be identified. Recombination within gltA was found to have generated two distinct variant clades, one a hybrid between B. taylorii and B. doshiae, the other between B. taylorii and B. grahamii. These recombinant events characterised the differences between the two-step and higher clades within the total nested cladogram, involved all four species of Bartonella identified in this work and appear to have played a dominant role in the evolution of Bartonella diversity. It is clear, therefore, that housekeeping gene phylogenies are not robust indicators of Bartonella diversity, especially when only a single gene (gltA or 16S RNA) is used. Bartonella clades infecting Microtus were most frequently involved in recombination and were most frequently tip clades within the cladogram. The role of Microtus in influencing the frequency of Bartonella recombination remains unknown.

The impact of host identity and habitat type on the density of hard ticks (Ixodes ricinus and Dermacentor reticulatus) infections on rodents in forest and abandoned field habitats in NE Poland was investigated. Ixodes ricinus was most abundant in the forest system, but D. reticulatus, although rarer, was most abundant in the field system. Environmental humidity and the much lower density of rodents probably limited the abundance of I. ricinus larvae in the field system, although this tick was still common on Microtus oeconomus from around small ponds. Nymphs of I. ricinus were comparatively rare in the forest, probably because of infection of non-rodent hosts. Dermacentor reticulatus nymphs on the other hand were very much more common in the ears of Microtus species than would have been predicted based on larval densities. The impact of habitat change (in this case successional change following field abandonment) on tick densities is emphasised, and the role of Apodemus as an epidemiological bridge, linking woodland and field habitats, is highlighted.

SUMMARYThe epidemiology ofBartonellaspecies infectingApodemus flavicollisandMyodes glareolusin a forest in Eastern Poland was followed for 2 years using mark-recapture. Infections could be acquired in any month, but prevalence, and probability of infection, peaked in the summer. There were significant differences in the pattern of infections between the two species. Both hosts were primarily infected as juveniles, but the probability of infection was highest forA. flavicollis, which, evidence suggests, experienced longer-lasting infections with a wider range ofBartonellagenotypes. There was no evidence of increased host mortality associated withBartonella,although the infection did affect the probability of recapture. Animals could become re-infected, generally by differentBartonellagenotypes. Several longer lasting, poorly resolved infections ofA. flavicollisinvolved more than 1 genotype, and may have resulted from sequential infections. Of 22Bartonella gltAgenotypes collected, only 2 (bothB. grahamii) were shared between mice and voles; all others were specific either toA. flavicollisor toM. glareolus,and had their nearest relatives infectingMicrotusspecies in neighbouring fields. This heterogeneity in the patterns ofBartonellainfections in wild rodents emphasizes the need to consider variation between both, host species andBartonellagenotypes in ecological and epidemiological studies.