Analyst and Observer Variability in Density Estimates from Spot Mapping

We studied variation in density estimates from spot mapping that was attributable to analysts and observers, using expert birders with little or no prior experience with spot mapping. Three observers independently spot mapped one 42-ha plot (Markwood) in mixed-conifer forest, and four others independently spot mapped another (Teakettle). All observers analyzed all maps. Consistency among analysts and observers in estimating the numbers of territories of breeding species on each plot was generally poor. Across all combinations of analysts and maps, 71% of all ANOVAs had significant analyst and/or observer effects. Observer effects were generally greater than analyst effects. When observers analyzed their own species maps, CVs of individual species ranged from 0% to 173% (mean 41%) at Markwood and from 0% to 188% (mean 60%) at Teakettle. As in similar studies, mean CVs from pooled totals of all species were less than those from individual species and were within the range of variation found by other researchers. Based on the range of CVs observed among species in this study, the number of sample plots needed to detect a statistical difference in density of a given species between samples is probably prohibitive for most studies. Instead, practitioners need to design studies to control observer and analyst variability.