Abstract
Obligate avian brood parasites show dramatic variation in the degree to which they are host specialists or host generalists. The screaming cowbird Molothrus rufoaxillaris is one of the most specialized brood parasites, using a single host, the bay-winged cowbird (Agelaioides badius) over most of its range. Coevolutionary theory predicts increasing host specificity the longer the parasite interacts with a particular avian community, as hosts evolve defences that the parasite cannot counteract. According to this view, host specificity can be maintained if screaming cowbirds avoid parasitizing potentially suitable hosts that have developed effective defences against parasitic females or eggs. Specialization may also be favoured, even in the absence of host defences, if the parasite's reproductive success in alternative hosts is lower than that in the main host. We experimentally tested these hypotheses using as alternative hosts two suitable but unparasitized species: house wrens (Troglodytes aedon) and chalk-browed mockingbirds (Mimus saturninus). We assessed host defences against parasitic females and eggs, and reproductive success of the parasite in current and alternative hosts. Alternative hosts did not discriminate against screaming cowbird females or eggs. Egg survival and hatching success were similarly high in current and alternative hosts, but the survival of parasitic chicks was significantly lower in alternative hosts. Our results indicate that screaming cowbirds have the potential to colonize novel hosts, but higher reproductive success in the current host may favour host fidelity.