Effective population size (Ne) is a quantity of central importance in evolutionary biology and population genetics, but often notoriously challenging to estimate. Analyses of Ne are further complicated by the many interpretations of the concept and the alternative approaches to quantify Ne utilising widely different properties of the data. On the other hand, alternative methods are informative for different time scales such that a set of complementary methods should allow piecing together the entire continuum of Ne from a few generations before the present to the distant past. To test this in practice, we inferred the continuum of Ne for 45 nine-spined stickleback populations (Pungitius pungitius) using whole-genome data. We found that the marine populations had the largest historical and recent Ne, followed by coastal and other freshwater populations. We identified the impact of both recent and historical gene flow on the Ne estimates obtained from different methods and found that simple summary statistics are informative in comprehending the events in the very recent past. Overall, our analyses showed that the coalescence-based trajectories of Ne in the recent past and the LD-based estimates of near-contemporary Ne are incongruent, though in some cases the incongruence might be explained by specific demographic events. Despite still lacking accuracy and resolution for the very recent past, the sequentially Markovian coalescent-based methods seem to provide the most meaningful interpretation of the real-life Ne varying across time.