The role of gastric evacuation experiments in quantifying the feeding rates of predatory fish

Abstract

1. The foundation of the method for estimating feeding using gastric evacuation studies lies in the input = output rule. More complex models may have specialized uses, for example where there are marked short-term fluctuations in feeding. 2. There is a lack of standardization in experimental design, data analysis and terminology which hampers comparative studies of gastric evacuation and often leads to unnecessary confusion. 3. Most gastric evacuation models are based on single-meal experimental designs, even though it is well documented that a second meal can alter the evacuation pattern of the first meal. 4. Assessment of the best-fitting gastric evacuation model is usually arrived at by choosing the model that gives the highest r2. This does not necessarily indicate whether one model offers significant improvements over another, and it does not indicate whether the resulting feeding model can be applied in practice to give improved estimates of feeding for natural fish populations as a whole. 5. Censoring and other biases can distort the results of gastric evacuation experiments so that the observed model of gastric evacuation becomes increasingly misrepresentative of the functional gastric evacuation relationship. Most published studies have not attempted to accommodate for these effects and are likely to overemphasize the curvature of evacuation and under-estimate evacuation at low levels of stomach fullness. 6. The acceptance of a power model over a linear model results in the need for accurate estimates of the mean and variance of stomach contents of fish in the wild. There do not appear to have been any attempts to determine whether this can be achieved with sufficient accuracy for the power model to give more reliable estimates of feeding than the simpler linear model. 7. There are many opportunities for bias in estimating the stomach contents of fish in the field, many of which do not appear to have been given a great deal of attention in most studies. 8. One of the biggest sources of potential error in the process of estimating feeding can be generated by errors and biases in the estimate of population size. Errors in the estimate of population size directly influence the estimate of total consumption of fish populations and, in turn, the size of future populations. 9. There is a need for models based on feeding patterns more akin to those occurring naturally, rather than on experimental designs based almost entirely on single meals containing only one type of prey.