Proliferative kidney disease (PKD) is a high-mortality pathology affecting freshwater salmonid populations in Europe and North America. Its causative agent, the myxozoan Tetracapsuloides bryosalmonae, has a complex life-cycle exploiting freshwater bryozoans (mainly Fredericella sultana) and salmonids as hosts. PKD has recently increased in incidence and severity, causing remarkable declines in fish catches. In addition, environmental change is feared to cause PKD outbreaks in regions at higher latitude and altitude, as warmer water temperatures exacerbate disease development and fish mortality. In this perspective, this Thesis develops an integrated approach, involving field and modelling work, for the prediction of the incidence of PKD in river basins. In particular, the Thesis develops a novel spatially-explicit model of PKD epidemiology in riverine host metacommunities. The model, summarizing the current knowledge on disease transmission modes and parasite's life-cycle, accounts for both local population and disease dynamics of bryozoans and fish, as well as hydrodynamic dispersion of parasite spores and hosts along the river network. Model validation was attained through an integrated study of PKD in a prealpine Swiss river, where data on fish abundance, disease prevalence, concentration of primary hosts' and parasite's DNA in environmental samples (eDNA) and water temperatures were gathered at multiple locations within the catchment. In this context, a new method for predicting the spatial distribution of bryozoan density based on eDNA samples was developed. As water temperature is crucial to PKD severity, a deterministic, spatially-explicit water temperature model was formulated and tested on the case-study basin. The model, based on water and energy budgets at the reach scale, considers the effects of the spatial heterogeneity in environmental drivers, allowing the evaluation of gradients of daily mean water temperature along the river network. Stability and sensitivity analyses performed on the local epidemiological model proved that the introduction of T. bryosalmonae in a disease-free community is very likely to trigger a PKD outbreak. Simulation experiments conducted on synthetic river network replicas showed that network connectivity engenders high PKD prevalence at downstream sites, while the speed of invasion fronts in disease-free environments is amplified by climate change. Moreover, patchily located bryozoan hotspots proved able to sustain the infection at the whole river scale. Validation on the case study confirmed the reliability of the epidemiological model, and identified the prediction of bryozoan distribution as a key direction for future research. In this regard, the developed eDNA transport model possibly opens avenues for the modeling of species distributions in freshwater ecosystems, also beyond the case of PKD. The spatially-explicit temperature model was shown to outperform traditional models based on local heat budgets. Such a tool is instrumental to several ecohydrological applications, from the identification of river reaches at highest PKD-related risk, to the assessment of habitat suitability of fish or other freshwater species. Overall, this Thesis bridges the fields of ecology, epidemiology, hydrology and mathematical modelling in order to produce an integrated study of PKD, in the perspective of grasping the factors allowing disease persistence and devising mitigation strategies.