Where and when animals forage depends on the spatio-temporal distribution and catchability of their prey. In dynamic environments, animals can repeatedly target areas that provide predictable availability of prey or may search for ephemeral conditions of high prey availability. However, how foraging behaviour is initiated in response to static versus dynamic environmental conditions is difficult to study, since both environmental data sources are often lacking. In this study, central-place foraging Sandwich terns were tracked using GPS loggers during foraging. Hidden Markov models showed that the probability of switching between transit and foraging was most strongly affected by the static variable sediment type. Wave period (a dynamic variable related to weather), salinity (a dynamic variable) and water depth (another static variable) affected the transition probability to a lesser extent. Cloud cover, wind speed and current speed were only included in lower ranked models. Air and water temperature were not included in any model. Consistent with the greater importance of static versus dynamic abiotic conditions, consistency between foraging trips of the same individual varied irrespective of tidal, diurnal or seasonal cycles, although trips made close in time within a season were slightly more similar than trips with a larger time gap. We suggest that Sandwich terns target broad areas with coarser sediments, where sandeels (Ammodytidae) are more common, and that weather variables may be related to prey visibility. Our study suggests that even in highly dynamic environments, static environmental variables may more strongly affect foraging behaviour of coastal seabirds than dynamic variables.