Many pathogens are maintained by multiple host species and involve multiple strains with potentially different phenotypic characteristics. Disentangling transmission patterns in such systems is often challenging, yet investigating how different host species contribute to transmission is crucial to properly assess and manage disease risk. We aim to reveal transmission cycles of bacteria within the Borrelia burgdorferi species complex, which include Lyme disease agents. We characterized Borrelia genotypes found in 488 infected Ixodes ricinus nymphs collected in the Senart Forest located near Paris (France). These genotypes were compared to those observed in three sympatric species of small mammals and network analyses reveal four independent transmission cycles. Statistical modelling shows that two cycles involving chipmunks, an introduced species, and non-sampled host species such as birds, are responsible for the majority of tick infections. In contrast, the cycle involving native bank voles only accounts for a small proportion of infected ticks. Genotypes associated with the two primary transmission cycles were isolated from Lyme disease patients, confirming the epidemiological threat posed by these strains. Our work demonstrates that combining high-throughput sequence typing with networks tools and statistical modeling is a promising approach for characterizing transmission cycles of multi-host pathogens in complex ecological settings.