Technique-dependent biases in determination of diet composition: An example with Ring-Billed Gulls

We compared the diet of Ring-billed Gulls during the nestling period using two simultaneous sampling methods: pellets regurgitated by adults and direct observations of chick provisioning. We also compared the dietary composition of courtship feeds and adult spontaneous regurgitations to that of chick provisions and found no difference between the diets of adults and chicks. The two sampling methods suggested strikingly different diets at the same colony location. Pellets were predominantly composed of plant material, birds and scavenged human refuse (67% of individual items identified), while insects (19%) and fish (11%) made up the rest of the identifiable food remains. In contrast, most Ring-billed Gull chicks were fed either earthworms (50%) or whole fish (44%). Our observations of chick diet lacked insects possibly because we did not record feeding data during the evening (due to logistical limitations). Pellet contents were biased by the over-representation of indigestible hard parts of some food types (e.g., plant, human refuse) and lack of hard parts of others (e.g., earthworms). Such technique-dependent biases led to a non-random sample of food remains from gulls that fed on garbage and in farm fields, representing approximately seven percent of the population. Given the importance of diet studies, there is a need for more systematic, controlled studies to calibrate sampling techniques to actual animal diets.