508 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 3. September 2012 
TABLE I. Support for models predicting daily survival rates for microhabitat at Red-crested Cardinal nests at Estancia 
La Matilde, Argentina (2005-2008). 
Model 
Deviance 
AAIC, 
k 
Wt 
S (above + date)’*' 
448.3 
0.00 
3 
0.716 
S (above) 
452.2 
2.01 
2 
0.252 
S (horizontal) 
457.4 
7.07 
2 
0.029 
S (date) 
458.6 
8.27 
2 
0.021 
S (site) 
459.5 
9.61 
4 
0.009 
S (tree) 
459.8 
9.51 
3 
0.006 
S (age) 
460.5 
10.23 
2 
0.004 
S (below) 
461.4 
11.04 
2 
0.002 
so 
463.4 
11.07 
1 
0.000 
u " AIC, value of ihe top model = 
= 454.35. 
Deviance difference between each model and the saturated model in -2 log likelihood: AAIC, = difference between each model and the lop model in 
Akaike's Information Criterion corrected for small samples (AIC,); k number of parameters in the model; ti, - Akaike weight, a measure of each model’sielati'e 
support within the set ot candidate models. S(.) is the general model that assumes constant DSR among nests and over time, Stbelowl. Sthorizontal). and Siahc'c 1 
are the models for the microhabitat cover below, horizontal, and above nests, respectively. S(agc) is the model where DSR has a linear relationship with age of the 
nest. S(tree) is the model including the tree species Where the nest was built. Stsitel is the model including the nest location within the forest. Stdatel is the mod; 
where DSR has a linear relation with date of season. 
covariates and with the logit-link function 
(Dinsmore et al. 2002). The list of candidate 
models was based on combinations of factors that 
a priori may affect Red-crested Cardinal nest 
survival. We ranked and compared models using 
AAICc (estimated as ihe relative difference 
between the top ranked model and each other 
model). We considered models with AAICc < 2 
to be equally parsimonious (Burnham and 
Anderson 2002. 2004). We also examined 
whether the suspected effect of vegetation cover 
on nest survival was consistent across tree 
species (interaction tree X cover). We report 
parameter estimates from the single best model 
when the top model was strongly supported (tr, > 
0.70) (Burnham and Anderson 2002). We report 
95% confidence intervals for each parameter 
based on the unconditional variances. We 
obtained daily survival estimates from the 
logistic-regression equation of the best-supported 
model. Survival probabilities were the result of 
daily survival rate over the assumed duration of 
breeding cycle (27 days). Reported values are 
means ± SE. 
RESULTS 
Thirty-four nests produced at least one fiei 
ling (resulting in an overall apparent nest-survi 
rate of 0.32) while the rest of the nests w 
depredated (68%. n = 72 nests). The total num 
ot success!ul nests did not differ between seast 
(Chi-square: fa = 2.82. F = 0.24). Nests w 
bmlt in Tala (n - 68 nests. 64%) and Coroni 
(/? 38 nests, 36%) at a height of 3.6 ± (). | 
(range = 1.5-6.8 m). Microhabitat nest cover 
(i.e., leaves and branches covering the nest) was 
42.1 ± 1.9% for horizontal measures. 51.1 ± 
1.8% for above, and 29.4 ± 1.6% for below 
measures. Sixty-four nests (60%) were in the 
border of the chains of forest, 23 nests (22%) were 
in the small isolated forest patches, and 19 nests 
(18%) were in the center of the continuous chains 
of forest. 
We monitored nests over a 131-day interval 
(from 16 Oct to 24 Feb) for 1.262 exposure days. 
Nest cover above and date within season were 
important covariates in modeling DSR (Table 1). 
Models including horizontal and below nest 
cover, age of the nest, tree species, and site had 
a lower AlCc value than the null model, but did 
not contribute significantly to the best model 
(Table I). The AICc values of the rest of the 
covariates were higher than the AICc's null model 
and they were rejected. Interaction tree X cover 
showed that effect of vegetation cover on nest 
survival was consistent across both tree species. 
The best fitted model contained the additive 
effects of nest cover above and date throughout 
the season (wi = 0.72; Table 1). DSR improved 
with increasing nest cover above and when 
decreasing the date throughout the season (Ta¬ 
ble 2; Fig. 1). 
DISCUSSION 
Vegetation cover surrounding the nest and date 
ol reproduction within the breeding season had a 
significant effect on predation risk of Red-crested 
Cardinal nests. These results suggest increased 
