Does the presence of an observer affect a bird's occurrence rate or singing rate during a point count?

The approach or presence of an observer may affect the behavior of nearby birds, rendering them either more or less detectable than when no observer is present due to a change in singing rates. To test whether there are systematic detection biases associated with the presence of an observer during point count bird surveys, we compared the occurrence and singing rates of birds during a 10‐min period immediately preceding the time when an observer arrived to conduct a count and during the formal count itself by extracting song information from autonomous sound recorders. We obtained recordings of 36 species of birds detected at ≥5 locations in one of three vegetation types, including burned conifer forest, green conifer/riparian streamside forest, and riparian bottomland/marshland. We found that species richness and both the probability of occurrence and singing rate for any of the species recorded were unaffected by the presence of an observer. In addition, the probability of occurrence did not differ significantly among four 2.5‐min recording sessions during 10‐min counts when an observer was present. Thus, the presence of an observer did not appear to introduce any detectable systematic bias that would make bird lists or unadjusted occurrence rates inaccurate on that basis alone. In addition, rates of bird occurrence across 2.5‐min temporal subsets of a 10‐min count did not vary in a systematic way that would violate the assumption of equal occupancy across adjacent time periods as sometimes used to build detection histories in occupancy modeling.