Population dynamics is the study of the abundance of a species at different life stages of a species, the interactions between these life stages and sometimes the interactions with other species. Stage-structured modelling is a popular approach for population dynamics studies. This approach examines populations based on their ecology and allows the incorporation of complex dynamic processes. Model outputs are sensitive to the parameter values. It then becomes crucial to accommodate and quantify parameter uncertainty. This is of particular importance when the population of interest is exploited and the risk of over-exploitaion or extinction needs to be assessed. When studying real world examples of populations exploited by fisheries, several additional problems often arise such as: multiple and heterogeneous sources of information (e.g. datasets collected at different spatial and temporal scales), missing observations, life stages of interest not directly observable. The Bayesian framework allows all of these issues to be handled within the general framework. Thus has proven its particular value in studying the dynamics of exploited populations. Indeed, unknown quantities have associated probability distributions reflecting their uncertainty. Dealing with variations in the interactions/processes between life stages or limited and indirect ecological data is also facilitated by Bayesian modelling. In this study, I examined a large Atlantic salmon population located in the Foyle catchment (Ireland). This population has been exploited for several centuries and particularly during the 20th century. This study focused on the period from 1959 to present for which most monitoring data is available from the Loughs Agency (formerly the Foyle Fisheries Comission). The Loughs Agency is responsible for the management of the salmon population. The aim of the agency is “to manage [the] fisheries towards maximum sustainable exploitation for commercial and recreational purposes”. In order to do so, it is important to understand the regulatory mechanisms occurring in the population in order to (i) estimate the number of fish returning to river, i.e. pre-fishery abundances (PFAs), and (ii) to derive standard reference points for assessing the population status with regards to its sustainable exploitation. To this end, a state-space model is implemented within a Bayesian framework. A life stage and spatially structured dynamic model describes the lifecycle of the Main components of Atlantic salmon in the Foyle catchment. Several empirical datasets related to the abundances of the stages at different scales of space and time, over a period of 50 years are brought together. Observations and process errors are taken into account ultimately allowing PFAs to be estimated. A retrospective analysis was also carried out providing insights on the historical status of the population and its exploitation. Geo unit specific abundances of the different states and their associated uncertainty are estimated. The main state of interest is the pre-fishery abundance (PFA), during the time-series considered (1959-2008) the salmon population reached its apex in the mid 1960’s. This was followed by a steep decrease until the mid 1970’s. From then to present, the population followed a slow declining trend with a slight recovery in the mid 1980’s. This decline is shown to be mostly due to a decline of the 0+ juvenile to returning adult survival which is accentuated some years by some overfishing.