Analysis of feeding preference experiments

Area of Science:

• Ecology
• Marine Biology
• Animal Behavior

Background:

• Consumer feeding preference studies often overlook natural food changes during trials.
• Existing statistical methods inadequately control for autogenic food changes without consumers.
• Incorrect statistical approaches can inflate the significance of observed feeding preferences.

Purpose of the Study:

• To highlight the inadequacy of current statistical methods in feeding preference studies.
• To propose an appropriate statistical analysis for preference tests involving autogenic food changes.
• To demonstrate the impact of a revised statistical approach on interpreting feeding behavior.

Main Methods:

• Critique of statistical analyses in published feeding preference studies.
• Proposal of a t-test for analyzing between-food differences in mass loss.
• Comparison of experimental (with consumer) and control (without consumer) data.
• Application of the t-test to a dataset on sea urchin feeding on brown algae.

Main Results:

• Standard statistical practices in feeding preference studies often suppress variance, overstating significance.
• A t-test on mass loss differences between experimental and control groups is statistically appropriate.
• Low replication in control groups hinders the detection of significant factors, such as simulated grazing.

Conclusions:

• Accurate statistical analysis is crucial for reliable feeding preference data.
• The proposed t-test method provides a more rigorous evaluation of consumer feeding behavior.
• Simulated grazing significantly reduces the attractiveness of brown algae to sea urchins, a finding obscured by inadequate statistical methods.