A critical review of methods of studying fish feeding based on analysis of stomach contents: application to elasmobranch fishes

Abstract

Using real data sets of elasmobranch fishes as examples, this paper presents a critical review of selected methods and statistical approaches used in fish feeding studies and makes recommendations on the application of such methodology. The percent index of relative importance is proposed as a standardized measure in dietary analyses, and a three-dimensional graphical representation of the diet is introduced. Multiway contingency table (log-linear) analysis is recommended to test for dietary variations. Caution is advised when using rank correlation to study dietary overlap and parametric tests when stomach content data do not satisfy parametric assumptions. Sampling gear type, experimental design, and statistical tests can affect results on diel feeding chronology, and stomach content weights do not suffice to interpret diel feeding chronology. On the basis of sampling requirements and model assumptions, the Diana and Olson-Mullen methods appear to be the most appropriate approaches for estimating daily ration in sharks. Use of resampling techniques is highly desirable because they provide a measure of the error in daily ration estimates. Using several criteria to evaluate the best-fitting model of gastric evacuation in fishes is also strongly advocated. Overall, increased consolidation of methods and analyses is recommended to facilitate comparative studies.

Influence of sample size and sampling frequency on the quantitative dietary descriptions of a predatory fish in the Benguela ecosystem

“…In addition, none of the studies so reviewed contained any estimates of the precision of samples used. Cumulative diversity curves were first employed by Pielou (1966), and they have since been recommended and described in many studies (Elliot 1971, Hurturbia 1972, Baltz and Morejohn 1977, Cailliet 1977, Hoffman 1979, Calliet et al 1986, Duffy and Jackson 1986, Cortés 1997, Griffiths 1997. In the present study, the diversity asymptote was mostly (90%) attained before all stomachs were included, thus indicating that most …”

“…Given that snoek diet changes ontogenetically (Griffiths 2002), and that juveniles were not sampled consistently, only fish >70 cm FL were used for the analysis. For all statistical comparisons of prey composition, prey items contributing < 5% were combined to form a so-called "other" group, to reduce the likelihood of Type 1 errors (Cortés 1997).…”

Food habits of the rough rayRaja radula(Chondrichthyes: Rajidae) from the Gulf of Gabès (central Mediterranean Sea)

“…The IRI values were converted to a percentage to facilitate comparisons between prey items (Cortés 1997).…”

Diet composition and feeding habits of the eyespot skate, Atlantoraja cyclophora (Elasmobranchii: Arhynchobatidae), off Uruguay and northern Argentina

“…Diet and data analysis. Percentage by frequency of occurrence (%O), number (%N), weight (%W) and index of relative importance (%IRI; Pinkas et al, 1971;Cortés, 1997) for each prey were used to describe the diet of A. cyclophora. To determine whether a sufficient number of individuals was sampled to conduct statistical analyses, the order of stomachs was randomised 100 times and the mean cumulative Shannon diversity index was plotted as a function of sample size.…”