504 
THE WILSON JOURNAL OF ORNITHOLOGY . Vol 123. No. 3. September 2011 
TABLE 1. Influence of nest substrate, concealment, 
based on Cox proportional hazard regression model. 
nest height, and year on 
White-throated Thrush nest survival. 
Variable 
Parameter estimate r SE 
V-value 
P-valuc 
Hazard ratio 
Nest substrate 
Concealment 
Nest height 
Year 
0.7513 ± 0.3037 
-0.3185 ± 0.1024 
0.5226 ± 0.1598 
0.3700 ± 0.1426 
6.1202 
9.6710 
10.6957 
6.7373 
0.013 
0.002 
0.001 
0.009 
2.120 
0.727 
1.686 
1.448 
Wc randomly selected one of the nesting 
attempts for use in analyses when a nest was 
used more than once during a year. A few nesting 
attempts were excluded from analyses if uncer¬ 
tainty about the nest initiation date was too great: 
for example, if eggs were detected 1 day and not 
again and no parent was ever seen. The nestling 
period for both species is known to be 14 days 
(Cohen and LindeU 2004; C. A. Lindell, unpubf 
data) and we calculated an average incubation 
period of 12 days for both species based on data 
from the present study. We calculated a range of 
Posable laying and hatching dates for each 
nesting attempt based on contents from nest 
checks. Nest initiation dates were grouped into 
7-day intervals for each year. The first interval 
started w.th the earliest nesting attempt. For 
example, the earliest laying dale occurred on 17 
March and the first interval was 15-21 March 
The second interval was 22-28 March. Intervals 
were created until all nests were included. 
We used Cox proportional hazard regression 
models to examine relationships between land- 
cover (coffee. pasture, road), year (1999-2002) 
nest height, concealment, nest substrate (ground 
or vegetation), nest initiation interval (when first 
egg was laid), and survival of nests. This type of 
diST m0del dOCS n °‘ assume a Particular 
distribution or constant mortality (Allison 2010) 
and nests that were observed for longer periods of 
ime are weighted more heavily. We initially 
conducted univariate analyses to test the potential 
effects of land-cover, year, nest height, conceal¬ 
ment, nest substrate, and nest initiation interval c>n 
nest survival. PROC PHREC, was used for 
continuous variables while PROC L1FETEST 
rooV Se A d i. f0r Ca ^; rical variables (SAS Institute 
2003, Allison 2010). Those variables with />- 
vS f, ^ inc,uded farther multi- 
rr ac,ion eor ww >- 
h S and land-cover, nest height, 
and the interaction between these two variables 
lor Clay-colored Thrushes. Our final models 
included only variables that were significant at 
f* s 0.05. Variables in the final models met the 
assumption of proportionality (Allison 2010). 
Sample sizes at the Rio Negro plantation were 
limited and Irom only 1 year. We did not test lor 
the influence of particular variables on nesr 
success but calculated daily mortality rates 
(DMRs) for nests of both species at this site using 
the Mayfield Method (Mayfield 1975. Johnson 
1979), having calculated exposure days for 
attempts with successful, failed, and uncertain 
outcomes following Manolis et a). (2000). We 
also calculated DMRs for categories of nests from 
Las Alturas for those variables that were included 
in final Cox regression models. Nest success rates 
were calculated by subtracting DMRs from I to 
obtain the daily survival rate, and then raising this 
rate to the number of days in the incubation and 
nestling periods, i.e., 26 for both species (May- 
field 1975). 
RESULTS 
We found 351 White-throated Thrush nesting 
attempts at Las Alturas over the 4-year period in 
abandoned coffee (n - 179), pasture (» = 96). 
roads (n = 64). and forest fragments (n = 12 > 
Ninety Clay-colored Thrush nesting attempts were 
observed in coffee {n = 66). pasture (n = 22), and 
roads (n = 2). Numbers of nesting attempts 
included in the analyses are lower because 
attempts missing data for a variable being tested 
were not included. 
Nest Success .—The final model for White- 
throated Thrush included data from 317 White- 
throated Thrush nesting attempts and, for Clay- 
colored Thrushes, 88 attempts. Substrate, con¬ 
cealment. nest height, and year significantly 
influenced White-throated Thrush nest survival 
(Table 1). Hazard ratios indicate that: (I) nesting 
attempts in vegetation had a 112% greater rate of 
failure than those on the ground, (2) nesting 
