The demographic history of a population is important for conservation and evolution, but this history is unknown for many populations. Methods that use genomic data have been developed to infer demography, but they can be challenging to implement and interpret, particularly for large populations. Thus, understanding if and when genetic estimates of demography correspond to true population history is important for assessing the performance of these genetic methods. Here, we used double-digest restriction-site associated DNA (ddRAD) sequencing data from archived collections of larval summer flounder (Paralichthys dentatus, n=279) from three cohorts (1994-1995, 1997-1998 and 2008-2009) along the U.S. East coast to examine how contemporary effective population size and genetic diversity responded to changes in abundance in a natural population. Despite little to no detectable change in genetic diversity, coalescent-based demographic modelling from site frequency spectra revealed that summer flounder effective population size declined dramatically in the early 1980s. The timing and direction of change corresponded well with the observed decline in spawning stock census abundance in the late 1980s from independent fish surveys. Census abundance subsequently recovered and achieved the prebottleneck size. Effective population size also grew following the bottleneck. Our results for summer flounder demonstrate that genetic sampling and site frequency spectra can be useful for detecting population dynamics, even in species with large effective sizes.