Quantification of avian parental behavior: What are the minimum necessary sample times?

Methods used to study parental behavior are largely idiosyncratic and are rarely tested to verify that they yield reliable data. In particular, the minimum number and duration of observations used to describe behavior accurately is rarely, if ever, justified. Using data on parental behavior of Eastern Kingbirds (Tyrannus tyrannus) collected by video in Oregon, we evaluated three questions regarding sampling methods: (1) Is within-pair behavior sufficiently consistent among days so that observations from a single session are representative of parental behavior? (2) Does a 1-h sampling window fairly represent parental behavior or is a longer period needed to accurately describe behavior? (3) Assuming that a 1-h period is sufficient, does the first hour immediately after disturbance to start recordings yield reliable data? We found that among-pair differences in parental behavior were consistent over different days and that a single hour of observation was representative of parental behavior over a longer (2-3 h) period of observation. However, our data also suggested that the first hour yielded less representative data than the remaining 2-3 h; nonetheless, consistent among-pair differences were still detectable. The shortcoming of data collected in the first hour appeared to be due to a lingering effect of human disturbance from setting up recording devices rather than a time-of-day effect. We suggest that researchers test the appropriateness of their methods, but our kingbird data suggest that observations over relatively brief periods can yield useful and reliable data.