Variable circular-plot (VCP) counts are statistically more sound than point counts because they are adjusted for the probability of detecting birds at different distances and under different conditions. However, many ornithologists use point counts rather than VCP counts because they believe that assumptions of the VCP method are almost always violated, leading to poor results, and because earlier field tests using ad hoc analysis methods gave variable and relatively poor results. We conducted the first field test of the VCP method where the exact density of a forest bird was known as part of re-establishing the 'Oma'o Myadestes obscurus in former range. All 'Oma'o in the new population were intensively monitored by radio telemetry so that the number present during four VCP censuses was known. Excluding the first census, when three of the four detections were of the same individual, differences in VCP density estimates ranged from -34% to +24% (mean 0%) even though =18 'Oma'o were detected per survey. We review critical assumptions of the VCP method and make recommendations for data analysis based on our experience with the method on Pacific islands.
