Summary

Description

In 2001, a national marsh bird monitoring protocol was developed for National Wildlife Refuges across North America. The protocol has been unproved each year with feedback from refuges that used it to monitor secretive marsh birds. However, few participants are incorporating methods to evaluate observer bias, or evaluate the differences in vocalization probability between surveys conducted during different times of day. Between 18 May and 14 July 2006 we conducted 4 replicate marsh bird surveys on 7 routes at St. Vincent National Wildlife Refuge, an island off the northwest coast of Florida. We conducted doubleblind multipleobserver surveys, paired morning and evening surveys, and paired morning and night surveys to examine the factors that influence detection probability. Observer detection probability was similar for all species detected clapper rail, 77; common moorhen, 69; and least bittern, 78. Observer detection probability was higher for clapper rails during the callbroadcast period of the survey and higher for common moorhen and least bittern during the passive period of the survey. Vocalization probability was highest in morning for all species detected. However, only common moorhen and least bittern had significantly more birds detected on morning surveys when compared to evening surveys P 0.026, and P 0.064 respectively, and only clapper rail had significantly more birds detected on morning surveys when compared to night surveys P 0.014.