Interpreting Selectivity Indices Calculated from Field Data or Conditions of Prey Replacement

Publisher Website
Google Scholar
Abstract
Indices of prey selection are compared empirically to clarify the kind of information each can provide and when the indices can be used appropriately. Emphasis is given to Chesson's alpha (α) and Strauss' Linear index (Li) because they are commonly used and represent two mathematically dissimilar approaches to measuring selection. Neutral selection for α and Li is not equivalent. For Li, neutral selection for a prey type occurs when selectivity equals the weighted (by relative abundance) average selectivity for all other prey types. For α, neutral selection equals the unweighted average for all prey types, which is equivalent to 1 divided by the number of prey types. Li and α respond differently to rare prey types. This is especially evident in field samples having large differences in relative prey abundances. Li values decrease as abundance of a prey type decreases in the environment, in contrast, α values often increase greatly as the proportional abundance of a prey type in the environment decreases. Consequently, α values, but not Li values, are often strongly influenced by the values for very rare prey, α values are sensitive to changes in the number of prey types, but not to changes in prey proportion. When the number of prey types is constant, α may be the most meaningful indicator of prey type preference. In contrast, Li values are sensitive to changes in prey proportions in the environment. When prey proportions are constant among experiments, Li is appropriate for measuring changes in the impact of a predator on a prey type and changes in selectivity relative to the average prey item. If number of prey types and prey proportions vary among experiments, Vanderploeg and Scavia's W or E* may be most appropriate.