Aedes aegypti mosquito eggs can remain quiescent for many months before hatching, allowing populations to persist through unfavorable conditions. Aedes aegypti infected with the Wolbachia strain wMel have been released in tropical and subtropical regions for dengue control. wMel reduces the viability of quiescent eggs, but this physiological cost might be expected to evolve in natural mosquito populations that frequently experience stressful conditions. We therefore compared the costs of wMel infection for quiescent egg viability in field-derived and laboratory populations. Quiescent egg viability was highly variable in wMel-infected populations, with greater costs of wMel in field-derived populations. In contrast, there was little variation between matched field-derived and long-term laboratory populations lacking wMel, suggesting that laboratory adaptation does not influence this trait and that differences are due to wMel infection. Comparisons of populations collected a year apart show a decline in costs under laboratory rearing conditions involving a rapid turnover of mosquito generations; this pattern was consistent across populations despite their origin, suggesting adaptation of mosquitoes to the wMel infection under laboratory conditions. Reciprocal crossing experiments confirm that differences in quiescent egg viability were mainly due to the genetic background and not Wolbachia alone. wMel-infected mosquitoes hatching from long-term quiescent eggs showed partial loss of cytoplasmic incompatibility and female infertility, highlighting additional costs of long-term quiescence. Our study provides the first evidence for a shift in Wolbachia phenotypic effects following deliberate field release and establishment and it highlights interactions between Wolbachia infections and local adaptation. The unexpected changes in fitness costs observed here suggest potential tradeoffs with undescribed fitness benefits of the wMel infection.