560 
THE WILSON JOURNAL OF ORNITHOLOGY . Vol. 123. No. 3. September 2011 
TABLE 2. Model selection results for the highest ranked candidate AIC r models that explain mean capture dale of 
eight short-distance migrants at Kingston Wildlife Research Station, Rhode Island, between I960 and 2007. Models are 
ranked based on Akaike weights (wy) and only models with A, < 2.0 are shown. 
Model’ K" n' 
RSS 
logttZ ,)' 1 
AIQ 
A, 
"tty 
Eastern Towhee (Pipilo erythrophrhalmus) 
Y = year 3 39 (714) 
3,830.83 
-89.45 
185.59 
0.0 
0.52 
Y = year + caprate 4 39 (714) 
3,743.66 
-89.00 
187.18 
1.6 
0.23 
22 
Dark-eyed Junco (Junco hyemalis) 
Y = year 3 34 (690) 
710.93 
-51.68 
110.17 
0.0 
0.35 
Y = caprate 3 34 (690) 
743.87 
-52.45 
111.71 
1.5 
0.16 
22 
Y = temp 3 34 ( 690 ) 
744.13 
-52.46 
111.72 
1.6 
0.16 
2.2 
Y - year + caprate 4 34 (690) 
Hermit Thrush ( Catharus guttatus) 
697.84 
-51.37 
112.11 
1.9 
0.13 
26 
Y = year + temp 4 40 (680) 
Ruby-crowned Kinglet (Regulus calendula) 
398.49 
-45.98 
101.09 
0.0 
0.61 
Y = year + year 2 4 39 (638) 
Song Sparrow (Melospiza melodia) 
304.74 
-40.09 
89.36 
0.0 
0.60 
Y = year 3 37 ( 554) 
3,362.43 
-83.43 
173.58 
0.0 
0.27 
Y - temp 3 37 ( 554) 
3,424.31 
-83.76 
174.25 
0.7 
0.19 
1.4 
Y - temp + caprate 4 37 ( 554 ) 
3,202.23 
-82.52 
174.30 
0.7 
0.19 
1.4 
Y - year + temp 4 37 (554) 
Swamp Sparrow (A/. georgiana) 
3,222.47 
-82.64 
174.53 
0.9 
0.17 
1.6 
Y = temp 3 34 ( 417) 
White-throated Sparrow ( Zonotrichia albicollis) 
326.83 
-38.47 
83.74 
0.0 
0.50 
Y = year + year 2 + temp 5 45 (2.728) 
Yellow-rumped Warbler (Dendroica coronaia) 
349.64 
-46.13 
103.80 
0.0 
0.59 
Y = year + temp + caprate 5 42 (3,044) 
Y = year + caprate 4 42 (3,044) 
Y = year + temp 4 42 (3.044) 
321.83 
355.66 
358.20 
-42.76 
-44.86 
-45.01 
97.19 
98.81 
99.10 
0.0 
1.6 
1.9 
0.46 
0.20 
0.18 
2.3 
2.6 
,, . ' ■ * u *P ,urc uaIe - “mp = mean maximum daily temperature 
<hc h number of b,rds captured per 100 net hours between 7 August and 31 
c Number of estimatable regression parameters in model ,n,-i,„i.„.. _ 
ate calculated as 
-•-t,-parameters in model including intercept and vm-mne 
C Calculated l 196(1 “ nd 2007 is in P aren,heses - 
Evidence ratio where w, = the highest ranked model. ^ 
mean autumn temperature and annual capture rate 
for each species were linearly related to year: 
thus, we assumed linear relationships between 
mean autumn temperature or annual capture rate 
and mean capture date. 
We used linear regression to assess the 
relationship between year and mean autumn 
temperature (including 1995-1997). We used 
linear regression to test for long-term trends in 
mean capture dates over time for species for 
which a linear function best described the 
relationship between year and mean capture date 
in the AIQ. analyses. We examined residual plots 
to confirm that data met normality and variance 
assumptions. 
We assessed the relative importance of year 
(using the model selected in the analysis), mean 
autumn temperature, and annual capture rates for 
explaining variation in mean capture date. We 
used mean arrival dates in our analysis, rather 
than first arrival dates based on research by Mills 
(2005) and Miller-Rushing et al. (2008) (see also 
Van Buskirk et al. 2009). We evaluated a 
candidate model set for each species that included 
all combinations of the predictor variables, We 
calculated AIC,. and w,- and used differences in 
A1Q. (A, - AIC c/ - - AIC ( . mi,,) to rank candidate 
models and considered models with A, < 2 to be 
substantially supported (Burnham and Anderson 
2002). We also calculated evidence ratios (wjwj) 
