Artificial bird nests, external validity, and bias in ecological field studies

We review an increasingly commonly used experimental approach in avian ecological studies that typifies the difficulties of maximizing two major components of experimental design: internal and external validity. Avian ecologists estimate reproductive success by measuring the rate at which nests in specified treatment groups fledge or fail. Because real bird nests are time-consuming and difficult to find and are rarely dispersed across treatments in sufficient numbers for powerful statistical analyses, investigators often utilize artificial nests, which are human-constructed surrogates baited with eggs and distributed in the field across treatments of interest. The primary advantage of artificial nests is that investigators can control the distribution and numbers of nests. The underlying assumption of similar predation rates on artificial and real nests is almost never tested. In the few studies that directly compare predation of artificial and real nests, however, external validity (or the degree to which results of an experiment can be generalized to subjects other than the experimental sample) is poor: predation of artificial nests differs from that of real nests in unpredictable and inconsistent directions. A primary reason for discrepancies may be that real and artificial nests attract different predators. We conclude that past protocols for use of artificial nests should be abandoned unless investigators carefully demonstrate minimal bias between measures of predation on real and artificial nests.