I investigated variation in fish species composition in marsh ponds of Point Pelee National Park between 1940 and 2003. The effects of interannual variation (time between samples of the fish assemblage), breaches of the barrier beach (connecting the marshes to Lake Erie), and variation in sampling methods used to quantify composition, explained 24%, 11% and 6%, respectively, of the total variation in species similarity among samples. Indicating there was significant natural turnover in species composition and that breaches were important agents of change. I further investigated the effect of sampling gear type and effort on spatiotemporal variation in the composition of fish assemblages to determine the most efficient methods for future sampling of fishes at Point Pelee. Notwithstanding that sampling method improvements are needed to measure temporal variation in fish assemblages, over long time periods, natural change and disturbance effects are a greater influence on species composition.