218 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 2. June 2012 
were more common on Star Island. Appledore 
Island is larger and has more trees and taller 
vegetation. Species more common on Appledore 
Island were primarily area-sensitive, forest-breed¬ 
ing species. Suomala at al. (2010) present 
complete habitat and species abundance data for 
Appledore and Star islands. 
The differences in species distribution patterns 
between Appledore and Star islands (Appendix) 
provided an opportunity to compare mass gain, 
recapture rate, and stopover time between the 
islands in relation to species distribution differ¬ 
ences. We hypothesized that if habitat use during 
migratory stopover is related to foraging oppor¬ 
tunities we should expect higher mass gain on the 
island where they are more abundant 
The proximity of Star and Appledore islands 
and their approximate north-south alignment 
suggest potential patterns of movement between 
the islands. Morning flights for habitat exploration 
prior to settling at a stopover site (Hutto 1985, 
Moore ct al. 1990, Wiedner el al. 1992) should 
lead to habitat shifts from lower quality habitat on 
one island to more suitable habitat on the other 
island; those species more numerous on a given 
island should be most likely to crossover to that 
island. Alternatively, onward migration during 
morning flights (Bingman 1980, Wiedner et al. 
1992), would lead to seasonally directed same-day 
movement between the islands. 
Our objectives were to: (J) compare recapture 
rate, stopover time, mass gain, and fat class 
between the islands in relation to differences in 
relative species abundance between the islands; 
and (2) examine potential movement of migrants 
between the islands in relation to these factors. 
We hypothesized that differences in frequency 
and duration of stopover and mass gain among 
migrants would correspond to differences in 
species distribution between the islands. We 
predicted that birds would move between Star 
Island and Appledore Island in a seasonally- 
appropriate direction during morning flights and 
be captured on both islands. 
METHODS 
Study Sues.—O ut study sites were on Star and 
^ P ' e ^^ e J slands in th e I^es of Shoals 
(42 58 N, 70 36 W) in the Gulf of Maine. This 
group of nine small islands and several ledges is 
AnnSj °T / nearest P olnt of the mainland. 
ppledore Island. Maine is the largest of the 
islands at 33.6 ha and Star Island, New Hampshire 
is 13.4 ha. The minimum shore-to-shore distance 
between Star and Appledore islands is ~0.7 km 
and the study sites were 1.6 km apart. 
Bird Surveys .—Mist netting was conducted 
simultaneously on both islands in 1999 and 2000 
during spring (I I May-8 Jun 1999. 10 May-8 Jun 
2000) and fall (16 Aug-30 Sep 1999 and 2000) 
migration. Nets (6 X or 12 X 2.6 m. 4 shelves. 
30 mm mesh) were operated from sunrise to sunset, 
and checked every 30 min. Up to five nets were 
operated on Star Island and up to 10 nets were 
operated on Appledore Island. Nets were placed in 
similar, representative habitat types for each island 
(habitat data in Suomala et al. 2010). Captured 
birds were brought to a central location on each 
island and banded with USGS aluminum hands. 
'The following data were recorded for each bird 
captured: age, sex, degree of skull pneumatization 
(scale of 0-3) during fall, unflattened wing chord 
(0.5 mm), fat class (scale of 0-4), tail length 
(0.5 mm), tarsus length (0.1 mm), and mass 
(0.01 g). Birds recaptured on subsequent days were 
treated the same as initial captures. Both banding 
stations used identical protocols following Morris 
et al. (1994, 1996). 
1 he fat classification system initially followed 
Morris et al. (1996) but was condensed (6 to 2 
categories) during analysis. This provided an 
approximate separation of birds with some fat 
present (categories I -4, 1 = fat lining lurcula to 
4 - lat extending over pectoralis; presumably 
sufficient tat for continued flight.) and birds 
without lat (categories 0 and 0.5, none to a trace). 
Data Analyses .—The analyses included only 
those species that do not breed on the Isles ol 
Shoals; thus, all individuals were considered 
stopover migrants. We used Systat Version 10 
(SPSS Inc. 2000) for statistical analyses. We did 
not use Bonferroni corrections in the analyses to 
reduce the chance that real differences or relevant 
patterns would be missed (Jones et al. 2002. 
Gotelli and Ellison 2004). There is a risk of Type I 
error, hut there is also a risk of missing relev ani 
patterns when Bonferroni corrections are used 
(Gotelli and Ellison 2004). overlooking important 
biological differences. 
Species recapture percentages were analyzed 
foi each sampling period to learn if there were 
differences between seasons and years. We did 
not statistically adjust recapture analyses for the 
different number of nets between the islands 
because the Star Island nets occupied a much 
higher percentage of available habitat than nets on 
