Validity of using artificial nests to assess duck-nest success

Artificial nests have been used in numerous studies of nest success because they can provide adequate sample sizes and be placed in accordance with experimental designs. However, the critical assumption that a strong linear relation exists between success of artificial and natural nests has rarely been tested. In 1993-94, we estimated nest success for upland-nesting ducks across northern Montana using data for artificial (n = 953) and natural (n = 1,318) nests. We collected data at 16 research sites that had a wide range of success for artificial (0.08-0.80) and natural (0.23-0.80) nests. Our data from 2 habitat types and 2 years produced a weak but significant linear association between success of artificial and natural nests (r(2) = 0.20, regression slope = 0.47, P = 0.04, n = 16). Success of both nest types varied by habitat and year. For nests in planted cover, slopes of regression equations predicting success of natural nests from success of artificial nests varied by year (P = 0.02; 1993: r(2) = 0.57, P = 0.04; 1994: r(2) = 0.87, P = 0.01). In 1993-94, the amount of concealing cover at a research site was related to success of artificial nests but not natural nests, which may indicate that different types of predators were attracted to each nest type and used different methods to find each nest type (olfaction vs. vision). Data from artificial nests could not be used to make valid inferences about success of natural nests in different habitats or years.