Journal of Field Ornithology (Mar 2022)
Capture height biases for birds in mist-nets vary by taxon, season, and foraging guild in northern California
Abstract
Mist-netting is a widely used technique for capturing birds to estimate abundance, demography, and population trends. Investigators in most studies use primarily or only ground-level mist-nets that extend < 3 m vertically. Therefore, data analyses require assumptions that the degree to which birds of interest occupy unsampled vertical space is unrelated to variables of interest, e.g., age structure, sex ratio, arrival and departure dates, and probability of recapture. Despite the widespread use of analyses based on these assumptions, they have rarely been tested. We analyzed capture data from paired ground-level and elevated (~3–5 m above the ground) mist-nets from 1993 to 2020 at Coyote Creek Field Station in Milpitas, CA, USA. Because capture height biases are driven by behavior, we expected that they may vary by net location, foraging guild, capture history, age, sex, and season. We built binomial models in a Bayesian framework to analyze the effects of these variables. Of 43 taxa, 13 were biased toward capture in elevated nets and seven toward capture in ground-level nets. These biases showed little variation among three net locations. Capture height biases were largely consistent with the documented heights of different foraging guilds. In one taxon, recaptured birds were more likely to be captured in elevated nets, possibly because of net avoidance or differences between overwintering and transient individuals. Only one taxon each exhibited either a sex or age effect on capture height. We found seasonal patterns in capture height for five taxa including residents and short- and long-distance migrants. Our results demonstrate that capture height biases were present at our research site and that the standard practice of deploying only ground-level nets may bias data in ways not generally recognized. Further study of these biases could improve mist-netting methods and increase the value of bird banding data.
