Symbiosis is a continuum of long-term interactions ranging from mutualism to parasitism, according to the balance between costs and benefits for the protagonists. The density of endosymbiont density is in both cases a key factor that determines both the transmission of symbiont and the host extended phenotype and is thus tightly regulated within hosts. However, the evolutionary and molecular mechanisms underlying bacterial density regulation are currently poorly understood. In this context, the symbiosis between the fruit fly and its intracellular bacteria Wolbachia (wMelPop strain) is particularly interesting to study. Although vertically-transmitted, the symbiont is pathogenic, and a positive correlation between virulence and wMelPop density is observed. In addition, the number of repeats of a bacterial genomic region -Octomom- varies between individuals, but most likely also within them, and is positively correlated to the Wolbachia density. Such genetic heterogeneity within the host could promote conflicts between partners by increasing within-host competition between symbiont genotypes through a process analogous to the tragedy of the common. To characterize the determinisms at play in the regulation of bacterial density, we first introgressed wMelPop in different genetic backgrounds of D. melanogaster. We found different density levels and Octomom copy numbers in each host lineage, suggesting a host influence on density regulation through Octomom copy number selection. To confirm this hypothesis, we performed new replicated introgressions on the two Drosophila populations that exhibited the most extreme density levels. However, we found no evidence of host influence on density regulation. Instead, we found instability in infection patterns across generations, which rather suggests an influence of drift. Moreover, using reciprocal crosses with the two extreme lineages, we confirmed the absence of host regulation on density levels and Octomom copy number, and a strong influence of drift. We then discuss how drift, both on the symbiont population during transmission and on the host population, could limit the efficiency of selection in such a symbiotic system, and the consequences of drift on the regulation of density and composition of bacterial populations.