Johnson et al. • MIGRATION PATH OF RUSTY BLACKBIRDS 
699 
synsacrum harness (Rappole and Tipton 1991) of 
5-inin tubular Teflon ribbon (Bally Ribbon Mills. 
Bally, PA. USA). Harness si/e was calculated 
using allometric equations (Naef-Daenzer 2007) 
and adjusted in the field for individual fit. The 
geolocator and attachment harness weighed 2.0 g. 
or 3.4% of a Rusty Blackbird's body mass ( v - 
57.8 g. n = 16 birds). We monitored each 
instrumented bird over the nesting period for 
evidence of poor fit or abnormal behavior. We 
located three birds fitted with geolocators and 
recaptured them at their nest sites in June 2010 
following intensive, repeated surveys of the study 
area (Matsuoka et al. 2010). 
The MklOS measured light-level data at 60-sec 
intervals and stored maximum values every 
10 min. Geolocators were calibrated during a 5- 
day period that preceded deployment. We esti¬ 
mated the average sun elevation angle that 
corresponded to our chosen light threshold from 
clean sunrise/sunset events during the calibration 
period (Fox 2010). We analyzed the data on 
ambient light levels from the geolocators to 
identify daily locations using BasTrak software 
with the single threshold method (Fox 2010). We 
visually inspected the transitions at each sunrise 
and sunset, and removed any that appeared to be 
influenced by shading events and other sources ot 
interference. Latitude was estimated from the 
length of each solar day/night and longitude from 
the liming of each solar noon/midnight. We did 
not estimate latitude from data recorded 15 days 
before and after spring and autumn equinoxes 
because latitude estimates during equal day and 
night lengths are highly inaccurate (Stutchbury 
et al. 2009. Fox 2010). 
Rust> Blackbirds are believed to move during 
the non-breeding season in response to changes in 
weather and availability of foraging habitats and 
food resources (Hamel and Ozdenerol 2009. 
Luscier et al. 2010). We estimated core areas 
used by each individual during the non-breeding 
season, rather than a single averaged location. We 
used Home Range Tools (Rodgers et al. 2007) and 
ArcGIS 9.3 (ESRI 2009) to estimate kernel 
probability densities of home ranges during 
autumn stopover and winter periods. We used a 
fixed-kernel parameter with least-squares cross 
validation for calculating the smoothing parame¬ 
ter (h). This has performed well for simple 
mixtures (< 16 components) of points with the 
precision of fitted surfaces approaching an asymp¬ 
tote at a sample size of 50 locations (Seaman and 
Powell 1996, Seaman et al. 1999). Variances of the 
,v and v coordinates were unequal, and we rescaled 
data after Rodgers et al. (2007). The grid size was 
1 km. We conservatively defined each bird's core 
area for stopover and wintering periods using the 
50% density contour (Seaman et al. 1999). We 
used landcover data from the North American Land 
Change Monitoring System (2005) to characterize 
the general habitats within core ranges. We 
identified long-distance movements from breeding 
and wintering areas as migration events and 
delineated stopover areas as clusters of locations 
recorded during 2:3 days. We used the center of 
core ranges to estimate distances among breeding, 
stopover, and wintering locations, and summed the 
individual flight segments to estimate the total 
distance traveled. 
RESULTS 
All 17 birds fitted with geolocators were alive 
and behaving normally when we last observed 
them in breeding areas as late as mid-July 2009. 4 
to 6 weeks following attachment. Individuals with 
geolocators were associated with 13 nests; two 
were abandoned (I pair re-nested and successfully 
Hedged young), three were depredated or flooded, 
and eight fledged young. We recaptured two 
females and one male in June 2010, but did not 
observe any of the remaining 14 birds (82%) in 
either the 2010 or 2011 breeding seasons. The 
original mean (± SD) weight of the three birds 
that relumed ( v = 57.7 ± 0.8 g) was similar to 
birds (n = 13) with geolocators that did not return 
(,v = 57.9 ± 3.8 g). We found harnesses to be 
loose fitting upon recapture and to have worn 
away the surrounding feathers on the synsacrum 
and inner thighs of each of the three birds. 
However, each of the birds behaved normally in 
breeding areas in 2010, and all three fledged 
young from separate nests. 
The distance between the filtered geolocator 
locations and known nest sites during the breeding 
season averaged 145 km (range = 128-175 km, 
it — 3). The routes and timing of migration were 
similar for all three birds (Fig. I). The three birds 
departed breeding areas during 7-9 September 
and used a series of stopover sites from 19 October 
to 29 November spanning the Prairie Potholes, 
Badlands and Prairies, and Eastern Tall Grass 
Prairie regions that included southern Saskatch¬ 
ewan. North Dakota, South Dakota, and Iowa. 
Each bird stopped in either North or South Dakota 
or both (Fig. 1). Mean habitat composition of core 
