514 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol 124. No. 3. September 2012 
mortality rates for nests in each 5-day age class, 
and nest survival probability for each reproductive 
period. Two nests parasitized by Common 
Cuckoos were excluded from analysis because 
they did not meet the requirements for analysis 
using the Mayfield Method. 
I measured three nest-site characteristics fol¬ 
lowing nest predation or Hedging to avoid 
disturbing nest sites and parental behavior during 
the breeding period. Measurements of character¬ 
istics of depredated nests were not affected by 
predator activity as >90% of the depredated nests 
were not affected by predators as the nests were 
completely intact. I measured nest height from 
the ground to the bottom of the nest using a 
measuring pole. Bull-headed Shrikes build an 
open-cup nest and branches attached to the side 
and/or below support the nest. I counted branches 
that were attached to at least one point of the nest, 
It was difficult to evaluate whether a branch 
supported a nest, and I defined supporting 
branches as those attached to at least one point 
of the nest. The number of thorny branches was 
counted in the same way. 1 defined the number of 
branches as the sum of the numbers of thorny and 
non-thomy branches, as I wanted to examine both 
the effects of the density of branches and the 
effect of thorns. 
Potential predators were Carrion Crows (Cor- 
vus corone). Large-billed Crows (C. macro- 
rhynchos). Japanese rat snakes ( Elaphe climaco- 
phora). Japanese weasels ( Mustek/ itatsi ). rats 
(Rattns spp.). mice ( Apodemus spp.), and domes¬ 
tic cats {Pelts catus). I examined evidence of 
predators at depredated nests because predation 
was not directly observed in most cases. I 
assigned predation to mammals if the nest 
structure was damaged and to birds or reptiles if 
the nest was intact (Yosef 1994. 2001). 
Data were analyzed using a Generalized Line 
Mixed Model (GLMM) with a binomial err 
structure and a logit link function (success = 
predation = 0). Two nests in which a fraction , 
nestlings Hedged were excluded from lab 
analysis because partial predation might huv 
occurred. A tull model included three fixed term 
i.e.. nest height, number of branches, and numb, 
of thorny branches and one random term, i.e 
research year. Likelihood ratio tests were used I 
test each fixed term. The effect of the breedin 
period during which predation had occurred o 
nest-sue characteristics was analyzed using 
Generalized Linear Model (GLM) with a binomu 
error structure and a logit link function (the period 
that predation occurred; egg period = I. nestling 
period = 0). Likelihood ratio tests were used to 
test each of the fixed terms. The R 2.12.1 
statistical package (R Development Core Team 
2010) was used for all analyses. 
RESULTS 
Thirty-eight nests were observed during the 2- 
year study period (13 nests in 2008 and 25 nests ia 
2009). Fourteen nests (36.8%) were successful 
and 16 (42,1%) were depredated. Nest predation 
was equally likely during either the egg or 
nestling stage: eight of 30 nests in the egg stage 
were depredated, compared to eight of 22 nests in 
the nestling stage (// - 0.065. P = 0.80). Twelve 
ot the depredated nests (75%) were completely 
intact after predation and four were damaged 
Four nests were deserted and two were lost to 
brood parasitism. Two nests may have been 
partially depredated, although the actual reason 
is unclear. The nest survival probability was 52% 
(325 nest-days, 13 losses) during the incubation 
period and 67% (280 nest-days. 7 losses) during 
the nestling period. 
Neither nest height (Fig. 1A) nor number of 
thorny branches (Fig. 1C) was significantly cor¬ 
related with nest success (Table I). There was no 
etlecl ol nest-site characteristics on the breeding 
stage in which predation occurred (GLM. nest 
height: x' - 0.13, P - 0.72; number of branches: 
X“ = 0.32, P = 0.57; number of thorny branches: 
X' = 1 11. P = 0.29). However, the number of 
branches was negatively correlated with the 
probability of nest predation (Table 1: Fig. IB). 
The number of branches attached to the nest 
ranged Irom three to 21. Depredated nests were 
attached to approximately eight branches (» = 16. 
range = 3-18). whereas successful nests were on 
~14 branches (n = 14. range = 8-21). 
DISCUSSION 
Nest predation (16 of 24. 66.7%) was the main 
cause ol nest tailure of Bull-headed Shrikes in this 
study area. I found that high-density of branches 
attached to a nest reduced the risk of nest 
predation of Bull-headed Shrikes. There are two 
possible explanations for this effect. First, branch¬ 
es may impede predators from preying on nests, 
because a predator with a large body may find it 
dilticult to move through the branches. Second, 
numerous branches may serve as cover that 
conceals the nest from predators. It appeared that 
