Smith and Paton • SONGBIRDS SHIFT AUTUMN MIGRATION 
561 
to examine the relative likelihood of each model 
compared to the highest ranked model, and the 
relative importance {w + (/)) of each explanatory 
variable by summing w,- across all candidate 
models in which that variable occurs (Burnham 
and Anderson 2002). We set a significance level 
of P < 0.05 for all statistical tests. 
RESULTS 
Mean autumn temperatures in Kingston. Rhode 
Island increased by 0.03 ±0.01 C per year 
between 1960 and 2007 (F,. 4 6 = 10.4. P < 0.01, 
r = 0.19). Year was included in the highest 
ranked models explaining variation in mean 
capture dates for seven of 11 long-distance 
migrants (Table 1) and seven of eight short- 
distance migrants (Table 2). Models including 
year only were the highest ranked for four long¬ 
distance migrants (Table 1), and four short- 
distance migrants (Table 2). Models including 
only mean autumn temperature were the highest 
ranked for four species: American Redstart 
(Setophaga ruticilla), Blue-winged Warbler ( Ver- 
mivora cyanoptera), Veery {Cathunts fuscescens), 
and Swamp Sparrow ( Melospiza georgiana) 
(Tables I, 2). However, multiple models were 
also plausible for all species except Hermit 
Thrush {Caiharus guttatus), Ruby-crowned King¬ 
let ( Regulus calendula), and Swamp Sparrow (Aj 
< 2; Tables 1, 2). These other models included 
annual capture rate for eight long-distance 
migrants (Table 1) and four short-distance mi¬ 
grants (Table 2). Annual capture rale alone was 
die highest ranked model explaining variation in 
mean capture date for the Ovenbird (Seiurus 
ourocapilla) (Table 1). 
Relative importance values revealed that year 
was the most likely of the three variables to 
explain variation in mean capture date for 14 
species (Table 3). Mean autumn temperature had 
comparatively high relative importance for one 
Rmg-distance migrant and three short-distance 
migrants (Table 3). Annual capture rate was of 
high relative importance for Yellow-rumped 
Warbler ( Dendroica coronata), and followed year 
closely in terms of relative importance values for 
Northern Waterthrush (Parkesia noveboracensis ) 
(Table 3). There was a significant positive linear 
trend for mean capture date from 1960 to 2007 for 
seven of 19 species (Figs. 1, 2) with significant 
relationships for three of 11 long-distance mi¬ 
grants and four of eight short-distance migrants 
(Pigs. 1, 2). Five species had a nonlinear rela- 
TABLE 3. Relative importance of three variables that 
explain variation in mean Julian capture date of migratory 
birds at Kingston Wildlife Research Station, Rhode Island, 
between 1960 and 2007. Relative importance values are 
calculated as the sum of w, from all models in which a 
given variable occurs. 
Annual 
Species 
Year 
Temperature" 
capture rate” 
Long-distance migrants 
American Redstart 
0.359 
0.839 
0.280 
Black-and-white Warbler 
0.999 
0.347 
0.325 
Blackpol! Warbler 
0.999 
0.701 
0.200 
Blue-winged Warbler 
0.402 
0.589 
0.387 
Common Yellowthroat* 
0.949 
0.294 
0.334 
Gray Catbird* 1 
0.995 
0.218 
0.515 
Northern Waterthrush 
0.653 
0.307 
0.562 
Ovenbird 
0.395 
0.369 
0.590 
Red-eyed Viren 
0.952 
0.375 
0.244 
Veery* 
0.865 
0.331 
0.212 
Wood Thrush 
0.564 
0.403 
0.398 
Short-distance migrants 
Eastern Towhee 
0.999 
0.251 
0.308 
Dark-eyed Junco 
0.631 
0.355 
0.380 
Hermit Thrush 
0.997 
0.774 
0.221 
Ruby-crowned Kinglet 0 
0.993 
0.229 
0.214 
Song Sparrow 
0.582 
0.603 
0.363 
Swamp Sparrow 
0.401 
0.690 
0.255 
White-throated Sparrow* 
1.000 
0.746 
0.212 
Yellow-rumped Warbler 
0.900 
0.659 
0.758 
“ Mean maximum daily temperature between 7 August and 31 October. 
*' Calculated as the number of birds captured per 100 net hrs between 7 
August and 31 October. 
1 Quadratic function of year used in AIC,. analyses. 
d Cubic function or year used in AIC, analyses. 
tionship between mean capture date and year 
(Fig. 3). 
Mean autumn temperature had the highest 
relative importance for two long-distance mi¬ 
grants and two short-distance migrants (Table 3). 
However, relative importance of mean autumn 
temperature was similar to that for year for Song 
Sparrow {Melospiza melodia), and distinctions 
between relative importance values for the three 
variables were small for Blue-winged Warbler 
(Table 3). Annual capture rate had the highest 
relative importance for Ovenbird, and year and 
mean autumn temperature were similar in relative 
importance (Table 3). 
DISCUSSION 
We documented significant long-term delays in 
timing of the migration period for birds at a coastal 
stopover site in southern New England; 38% of 
long-distance migrants (n = 8) and 50% of short- 
