524 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol 123, No. 3, September 2011 
movement; the narrower the latitude bar the more 
reliable the assessment. We ascertained overall 
route fidelity by examining the frequency distri¬ 
bution of the banding-to-recapture distance for 
each owl. We explored general regional differ¬ 
ences in route fidelity by analyzing fidelity 
separately for owls banded in the Great Lakes 
Basin, the Appalachian Mountains, and the 
Atlantic seaboard. These regions are defined 
inexactly (Fig. 1), but are sufficient for making 
broad comparisons in Northern Saw-whet Owl 
movements among regions. We used Chi-square 
contingency tests to ascertain if differences in 
fidelity exist among regions or among regions and 
all owls. 
Migration Direction .—We generated rose dia¬ 
grams (Kovach Computing Services 2010) to 
calculate the mean azimuth and angular distribu¬ 
tion of all banding-to-recapture vectors of North¬ 
ern Saw-whet Owls captured at different stations 
in the same migration season. Additional rose 
diagrams were generated that considered owls 
recaptured >100 or >500 km from banding 
location to assess the possible influence of owls 
being recaptured disproportionately among prox¬ 
imate stations (thereby influencing overall angular 
distribution). We analyzed regional differences in 
directionality by isolating owls banded around the 
Great Lakes Basin, in the Appalachian Mountains, 
or along the Atlantic seaboard and recaptured 
>100 km from banding location (Fig. 1). We 
compared the mean directionality among these 
groupings using pair-wise Watson-Williams 
F-tests described in Fisher (1993). 
Age-differentiated Migration.—'We tested 
whether spatial differences exist between adult 
and juvenile movement patterns. We aggregated 
banding events into 01 latitude bars and calcu¬ 
lated the age ratio of the owls within each bar. 
Linear regressions were used to assess the strength 
of the relationship between age ratio and latitude 
bar. This was done across all years (1999-2008) 
and independently for each year to reveal 
differences in movement patterns between irrup¬ 
tion and non-irruption years. 
We examined spatial differences in migration 
by performing a surface interpolation of the age 
ratio of Northern Saw-whet Owls at banding 
stations with >50 banding events. The interpola- 
u f d inverse-distance weighting (ESRI 
2008) of age ratios at banding stations within a 
f seaf ch radius around each predicted raster cell 
(power - 2). This search neighborhood restricts 
the influence of distant stations while being 
sufficiently inclusive to interpolate the entire 
surface. This was done for all owls and separately 
for irruption and non-irruption years. We com¬ 
pared mean banding latitude by age class of all 
migrating Northern Saw-whet Owls using Wil- 
coxon Rank-Sum tests. This was done with pooled 
data (1999-2008) and separately by years. 
We examined whether proportions of adults 
versus juveniles differed among years using Chi* 
square contingency tests, followed by a post-hoc 
analysis of means for proportions to identify 
which years were significant deviants (SAS 
Institute Inc. 2010). 
RESULTS 
We reviewed information on 81,584 Northern 
Saw-whet Owls banded in eastern North America 
during fall migration (I Sep to 31 Dec) 1999-2008. 
Banding information was provided by 356 banding 
stations, 132 of which reported >50 banding 
events. Twenty stations reported 58% of all banding 
events (Fig. I). Forty-five percent of the 81,184 
banding events (of 81,584 total) with assigned age 
were adults and 55% were juveniles. There were 
2,184 owls recaptured during fall migration, 
Seventy-three (3.3%) of these were recaptured 
>1.000 km from the original banding site. 
Migration Timing .—There was a clear trend of 
northern banding events occurring earlier in the 
migration season than southern banding events 
(Fig. 2). The mean ± SD banding day at each 
latitude bar was 3.8 ± 2.7 days later than the bar 
immediately north of it. Mean banding day was 
not significantly different at any latitude bar 
among juveniles, adults, and all Northern Saw- 
whet Owls (x 2 22 = 0.03. P > 0.95). The average 
difference between adult and juvenile mean 
banding day at each latitude was 1.2 days (range 
= 0.17-2.88). Surface interpolation of predicted 
mean banding day revealed a similar north-south 
trend with earlier means occurring consistently 
farther north than later means (Fig. 3). The 
earliest mean banding days were predicted for 
eastern Ontario and Quebec, and progressed 
gradually southward. The latest mean banding 
days were in Virginia, Delaware, West Virginia, 
and Indiana, although the interpolation was not 
performed farther south due to the lack of banding 
stations and tew records in that region. 
Northern Saw-whet Owls moved -10.5 km per 
night on average (Fig. 4). The line-of-best-fit of 
mean migration speed was drawn using owls 
