Tremendous effort is expended in counting migrant raptors in North America,
where over 1,500 count sites exist. The Biological Resources Division of t
he United States Geological Survey is investigating whether migration count
s can serve to monitor population trends. To that end, we investigated the
effects of changes in survey duration on power to detect trends using vario
us sample sizes, lengths of surveys, and sampling frequencies within a seas
on based on raptor migration count data collected from 7 sites. As expected
, power to detect trends increased as sample size (number of years of count
s) increased; its magnitude depended on level of trend and variation in ann
ual counts. Except for cases with extremely low or high power, 5 additional
survey years increased power 20%-50%. Changing survey length affected powe
r, but results varied and depended on the migration pattern of species at s
pecific sites. Power to detect trends did not change appreciably when count
s were conducted only during 90 days of peak migration (for all species com
bined). However, when annual counts were based on 30 or 60 consecutive days
of peak migration, we found decreases in power for most site-species combi
nations. Changing the number of days counted during the week resulted in va
riable changes in power. Our results suggest the number of count days withi
n weeks necessary for sufficient monitoring should be evaluated on a site-
and species-specific basis. We determined that a coefficient of variation o
f 30% or less in annual counts is needed to detect a 3% average annual decr
ease in counts over 25 years with at least 80% power. This requirement was
met at one or more sites for 14 of 20 raptor species. Power to detect incre
asing trends exceeded power to detect decreasing trends, which is unfortuna
te from a conservation perspective. Nevertheless, if count data are collect
ed consistently, large-scale population trends for some raptor species may
be monitored effectively using migration counts.