1. Non-native mussels have increased water clarity in many lakes and streams in North America and Europe. Diel variation in catchability of some fish species has been linked to visibility during survey trawls (used to measure escapement). 2. Water clarity increased in nearshore areas of western Lake Erie by the early 1990s, following passage of legislation in 1972 to improve water quality (e.g. reduce phosphorus loading) and the invasion of dreissenid mussels (Dreissena spp.) beginning in 1987. 3. We hypothesised that increased water clarity in Lake Erie resulted in decreased catchability of young-of-year (age-0) yellow perch (Perca flavescens Mitchill) during daylight compared to during night. We used a two-tiered modelling approach to test this hypothesis on the ratio (R) of catch per hour (CPH) during night to CPH during daylight in bottom trawl surveys conducted during 1961-2005. 4. First, we examined seven a priori models. The first model, the 'null' model, represented no change in R over time. Three more models tested whether the timing of the change in R was associated with passage of water quality legislation only, dreissenids only (two-period models) and both legislation and dreissenids (three-period models). Three additional models included a 3-year lag before the effects of legislation, dreissenids or both occurred. Secondly, all possible two- and three-period models with a minimum of 2 years per time period were explored a posteriori. The a posteriori procedure determined the temporal transitions to higher R that were best supported by the data, without regard to a priori hypotheses. 5. Night CPH was greater than daylight CPH in 3 of 11 years during 1961-72, in 10 of 15 years during 1973-87, and in 14 of 18 years during 1988-2005. During 1991-2005 night CPH exceeded daylight CPH in all years except one, and night CPH was more than twice daylight CPH in 10 years during this period. 6. The best a priori model had two periods, with a break between 1990 and 1991, corresponding to 3 years after the dreissenid invasion. Similarly, the best two- and three-period a posteriori models both had breaks between 1990 and 1991. The results supported our hypothesis that age-0 yellow perch exhibited a transition to lower catchability during daylight compared to night, and the timing of the transition coincided with the establishment of dreissenid mussels. 7. The most plausible mechanism for our results was increased visibility of the trawl during daylight, resulting in increased avoidance of the trawl. These results have potential applications wherever non-native mussels have increased water clarity.