Unrealistically high costs of rejecting artificial model eggs in cuckoo Cuculus canorus hosts

Most studies that have tested the egg-recognition and egg-rejection ability of European cuckoo Cuculus canorus hosts have used artificial model eggs that are much harder than real Cuckoo eggs. Here we evaluate whether the use of such models overestimates the costs of egg rejection by hosts. We tested 17 potential cuckoo host species in south-eastern Spain with both artificial hard cuckoo-egg models and real eggs taken from a population of house sparrows Passer domesticus breeding in captivity. The puncture resistance of sparrow eggs, measured in the laboratory, was more similar to that of real cuckoo eggs than was the resistance of artificial models, although sparrow eggs were less resistant than real cuckoo eggs. Potential host species with a grasp index greater than 200 mm(2) did not suffer high rejection costs when rejecting hard models, probably because they are grasp ejectors. However, all species with a smaller grasp index suffered high costs when rejecting hard artificial models. For these species the frequency and magnitude of costs were significantly higher when rejecting artificial hard models than when rejecting real eggs. For some species the breakage of real eggs was quite difficult (they needed 97 to more than 4000 pecks in video-recorded ejections), and sometimes the birds suffered rejection costs. These results show that realistic estimates of the frequency and magnitude of rejection costs for hosts with small bills cannot be obtained by using artificial models, and also that for a variety of medium-sized puncture ejector species the costs when rejecting real eggs may be low.