6 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124, No. 1, March 2012 
in position is that the birds moved, perhaps 
following their food source. This net westward 
directional movement is less likely to be an 
artifact due to shading or weather-related varia¬ 
tion because it was consistent among the three 
birds tracked. 
Northern Black Swift roost sites and roosting 
behavior in South American wintering areas are 
unknown. Waterfalls, caves, and dripping rock 
faces serve as roosting and breeding sites in 
breeding areas. There is only one documented 
observation of a roost site for Northern Black 
Swifts in South America, discovered during fall 
1992 and 1993, on the walls of a steep gorge 
along the Rio Caucu in the foothills of Colombia. 
Black Swifts roosted consecutively at dusk for a 
week in a compact group, mainly with White- 
collared Swifts (Slreptoproche zonaris ), clinging 
to the volcanic rock of a 40-m cliff overlooking 
the river, indicating that rocky river banks are 
used as roosting sites during migration (Stiles and 
Negret 1994). Similar .sites may be used in 
wintering areas if available, but this information 
is completely lacking. Non-brccding Common 
Swifts (Apus apus) are known to ‘roost* aerially in 
breeding areas and it is believed they spend 
~9 months of the year continuously on the wing. 
Non-breeding birds may fly continuously for 
several years (Backman and Alerstam 2001, 
Tarburton and Kaiser 2001), Common Swifts are 
also known to occasionally roost at night by 
hanging on the foliage of trees ( Holmgren 2004). 
Any of these scenarios is possible for Northern 
Black Swifts in wintering areas. Foraging activities 
of the Northern Black Swift in wintering areas are 
unknown but kernel densities indicate the swifts 
range over a large area in winter, suggesting the 
birds spend a lot of time on the wing. 
The capability ot geolocators for tracking small 
birds is still being explored and the potential is 
great. However, the devices are not without 
limitations. A major obstacle for success is that 
once the devices are deployed, the bird must be 
recaptured after a complete migration cycle has 
occurred to obtain data. Our study indicates the 
suitability of the Northern Black Swift for 
geolocator deployment and recapture, primarily 
due to this species* strong nesting colony fidelity 
and abihty to carry small devices for lone periods 
Calculation of latitude is unreliable around 
equmox and near the equator because lengt 
day and mght is equal. The accuracy of calcul 
day length is especially affected for terrestrial 
species hy shading factors that alter recorded light 
levels such as cloudy weather, foliage, and 
topographic shading of roost sites, resulting in 
latitude uncertainties. Fudickar el al. (2011) found 
the devices had an error of 201 ± 43 km for 
latitude and 12 ± 3 km for longitude (±95% Cl) 
for stationary geolocators (n = 30) in forested 
habitat. The apparent retreat of bird #554 from 
Colorado to the Pacific Ocean south of Baja 
California (Fig. 1) during spring migration is the 
result of one data point and the accuracy of this 
fix is questionable. It may or may not represent an 
actual movement by the bird and is possibly the 
result of an extended period of shading. We did 
not eliminate this position fix since the total day 
length did not drastically differ from the other day 
lengths of that time period. 
Accurate longitudinal information can be ascer¬ 
tained as this is not affected by equinox and we 
successfully used longitude near the autumnal 
equinox to indicate when the birds arrived at their 
wintering location. Black Swift breeding require¬ 
ments, such as nesting behind waterfalls in deeply 
shaded niches in steep and narrow canyons, or in 
caves where the performance of geolocators is 
often compromised by darkness, resulted in some 
unusable data during the breeding season. Docu¬ 
mented Black Swift nocturnal roosting behavior 
during migration is limited and indicates this could 
be a factor influencing the effectiveness of 
geolocators for tracking this species. If winter 
nocturnal roost sites are similar to those docu¬ 
mented in migration and al breeding sires, this will 
also influence the accuracy of the data collected by 
geolocators. Despite these limitations, geolocators 
far surpass hand recovery information or depen¬ 
dence on sporadic sightings to identify migratory 
paths and winter distribution of the Northern Black 
Swift. 
Understanding the theory behind geolocation is 
extremely important for interpreting and using the 
data collected to produce maps showing animal 
movements (Hill 1994). Once the theory is 
understood, knowledge of the behavior of the 
animal being studied and of weather patterns in 
the area where the animal was tracked can be used 
to provide insight into movement patterns. The 
mapped winter range of Black Swifts is an area 
that typically experiences high cloud cover. Thus, 
a significant number of the location fixes are most 
likely shifted to the north artificially because of 
cloud cover in the winter range as compared to the 
