Symbionts are classified as parasites, commensals, or mutualists based on their individual-level effects on host traits such as mortality or reproduction rates. However, the classification of symbiosis might change when the focal interaction is investigated at the host population scale, for multiple generations, or under different environmental contexts. We investigated the individual- and population-level effects of a microsporidian, Ordospora pajunii, and a yeast, Metschnikowia bicuspidata, on the key freshwater grazer Daphnia dentifera. Specifically, we tested whether the microsporidian, which has low individual-level virulence, can switch to mutualism and protect its host when the highly virulent yeast is in the environment. We tracked population dynamics of Daphnia and infection prevalence in populations exposed to each symbiont separately, both symbionts together, or unexposed controls. We found that O. pajunii infection reduced M. bicuspidata outbreaks, but we also found that the seemingly mildly virulent O. pajunii strongly reduced host population size, while the highly virulent M. bicuspidata had no detectable negative effects on host population size. The harmfulness of O. pajunii was most likely due to transgenerational virulence - reduced survival of unexposed offspring of exposed mothers. The lack of adverse effects, and short-term positive effects, of M. bicuspidata on Daphnia populations were consistent with a hydra effect. Our study shows that investigating symbioses at the population-level, with consideration of transgenerational effects and ecological context, can change the qualitative conclusions regarding the impacts of symbionts, and that studies that fail to do so risk focusing mitigation and control measures on the wrong symbionts.