Use of geometric forms to estimate volume of invertebrates in ecological studies of dietary overlap

Abstract

Studies of diet overlap of insectivores, especially lizards, have been important for the development of much ecological theory. However, measures of overlap, usually based on dietary volumes, vary widely among studies. Most researchers estimated the volumes of prey in stomach contents subjectively or attempted to reconstruct the volume of individual prey items from linear measurements of length (L), width (W) and/or height (H), or from mass. We studied the relationship between indices of volume based on indirect measures and direct measures of volume by fluid displacement for a collection of arthropods that simulated the diet of a generalist lizard. For each individual arthropod, we calculated the proportional error [(measured volume - estimated volume)/measured volume)] for each estimation method. For individual large insects, there were significant differences between orders in the proportional error for all methods. The methods based on linear measurements had only weak relationships (r2 ≤ 0.4 in all cases) with measured volume. Mass had a stronger (R2 = 0.9), but curvilinear relationship with measured volume. For the combined masses of all arthropods in each diet category from stomach contents of four species of lizards, there were also significant differences in the proportional error between arthropod orders for all methods (P ≤ 0.018 in all cases), except for mass (P = 0.126). Because most indices of overlap are based on proportions, errors in estimation for any one category affect the degree of overlap for all other categories found in that stomach. Cluster analysis of different indices based on the same data indicated differences of up to 10% in proportional overlap. Such large differences for identical diets mean that it is extremely difficult to compare studies that used different methods of volume estimation. We conclude that the use of general formulas to reconstruct the volumes of arthropods are not appropriate for most studies of diet overlap and that extreme care should be exercised in attempts to perform meta-analyses on studies that purportedly measured prey volumes.