724 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124, No. 4. December 2012 
s'l E '■ Descriptives,at i stics for variables used in models to estimate the number of singing male Cerulean Warblers 
o 10 L 7 k r ri ' ' W ' 2006 - Habi “' VariahlCS lhe ° f «* >"=* defined by 
or lu-km buffer represented by that vegetation type. ’ 
Variable 
Number singing males 
Bottomland forest in 250 m 
Developed land in 250 m 
Upland forest in 250 m 
Bottomland forest in 10 km 
Developed land in 10 km 
Upland forest in 10 km 
Forest in 10 km 
Mean * SD 
4.7 ± 5.66 
6.07 ± 8.05 
0.27 ± 0.45 
68.57 ± 18.92 
0.67 ± 1.19 
0.27 ± 0.34 
80.27 ± 16.13 
80.87 ± 15.35 
Min Max 
0.0 
18.0 
0.0 
36.5 
0.0 
3.2 
11.5 
96.0 
0.0 
6.6 
0.0 
1.5 
25.8 
96.8 
.30.7 
97.3 
a 10-km buffer in north Missouri to 97.3% fores 
in southeastern Missouri (Table 1 ). 
We found support for both local and landscape 
effects on Cerulean Warbler abundance. The 
model with both landscape and local effects had 
overwhelming support (w, = 0.984; Table 2). 
Model lit was good based on the £-fold validation 
procedure with a mean Pearson correlation of 0.84 
(range = 0.67-0.96) bclween predicted and 
observed values and the overdispersion parameter 
was dose to I (c = 0.83). Abundance of Cerulean 
Warblers mcreased 390.7. 8.7, and 4.1 times over 
the observed range of forest within 10 km 
bottomland forest within 250 m. and upland forest 
within 250 m. respectively (Fig. 2). The 95% 
confidence mtervals for the effects, bottomland 
and upland forest within 250 m. overlapped zero 
e a), which indicated some uncertainty in 
these effects. However, the weight of evidence for 
he most supported model with the effects 
bottomland and upland forest, was 61.5 times 
that of the second most supported model without 
those effects (wj/w 2 ; Table 3). The model coeffi¬ 
cient for (he percent of developed land within 
250 m was relatively large and negative compared 
to other coefficients (Table 3) but. because 
peicent developed land only ranged from 0 to 
1-5%, the overall effect on density was smaller 
than other effects and the confidence interval was 
large. 
DISCUSSION 
Wc found strong support for our hypothesis that 
abundance was affected by both local and 
landscape-scale habitat composition. The greatest 
Ufeei on abundance was the amount of forest 
within a 10-km buffer, followed by the effect of 
riparian forest and upland forest within a 250-m 
buffer. The strength of the effect for the amount of 
orest in the landscape emphasizes rhe importance 
of extensive forest areas for the Cerulean Warbler 
and the potential negative effects of fragmenta¬ 
tion, edge, or imerspersion of non-forest land 
uses. Counts of Cerulean Warbler were predicted 
table 2 
for small 
II samples (AICc), differencTi^AIcJ !yom"he2 2 X I? 8 likelihood - Akaik «-*'-s Information Criteria adjusted 
Parameter (o) for models estimating n T T*' , (A) * Akaike "eights Or,*. and overdispirsion 
Missouri, 1999-2006. nmber of s '"8'"S Cerulean Warblers in 5-km river segments in 
-50 m + upland forest 250 m + forest 10 km 
Intemepi. dcve.oped 10 km + bottomland fomst 
10 km + upland forest 10 km 
*^ + ***« 250 m + talomland forest 
-50 m + upland forest 250 in 
Intercept 
560.42 
578.61 
573.14 
584.81 
A 
H 7 
c 
0.00 
0.984 
0.83 
8.26 
0.016 
0.73 
37.82 
0.000 
0.55 
49.49 
0.000 
0.50 
,lJ foresl - and "Plund deciduous foresi, respecmclv 
ireAi 
