We used 2 yearsˈ diel food weight data from stomachs of Lake Erie yellow perch Perca flavescens to make Elliott‐Persson (E–P) estimates of daily ration and to construct consumption response surface (CRS) regression models. The CRS models provided relatively accurate, low‐cost estimates of daily ration from only one to four fish collections – compared with nine for the E–P procedure. Besides reducing daily sampling effort, CRS models allowed flexibility in fish collection times, thus alleviating the need for 24‐h associations with sampling sites. Independent variables for CRS models included the food weight in fish stomachs from a particular collection (adjusted for water temperature) and the time of day when a fish collection was made. Accuracy of CRS model estimates of yellow perch daily ration relative to E–P estimates was evaluated with a jackknife‐like procedure. Effectiveness of CRS models arose from significant relationships that existed between food weights in yellow perch stomachs throughout certain periods of the day and average food weight in stomachs over 24 h. Similar relationships, and hence low‐cost estimation potential, are expected for other fish species with regular diel feeding activity. Because CRS models reduce time requirements in the field, they should allow more detailed coverage of important spatial and temporal variability in multiday food consumption studies for fishes. The CRS models also showed spatial robustness and a desirable capacity to estimate daily rations in years beyond those for which the models were developed.
