Hallworth et al. • LOUISIANA WATERTHRUSH HABITAT USE 
569 
2005). The areal overlap of home range and core 
areas was quantified using Hawth's analysis tools 
tor ArcGIS (Beyer 2004). 
Akaike information criterion (A1C), corrected 
for small sample sizes (AIC r ; Burnham and 
Anderson 2001), was used to evaluate models 
comparing habitat variables between use and non¬ 
use areas. We chose seven biologically relevant 
variables on the basis of field observations which 
included percent muddy substrate (Mud), percent 
leaf litter (LL). percent prop roots and fallen 
coarse woody debris (Fallen), percent canopy 
cover (Canopy), percent standing/running water 
(Water), prey availability (Prey), and percent 
vegetation cover (Veg). Kullback-Leibler infor¬ 
mation and Akaike's information criterion, cor¬ 
rected for small sample sizes (AICc) (Burham and 
Anderson 2001) were used to evaluate the models. 
The lowest AAIC r value indicates the most 
parsimonious model. Thus, the model with the 
lowest AAIC,. value indicates goodness-of-fit to 
the data while minimizing the number of 
parameters in the model. Binary logistic regres¬ 
sions were calculated with JMP 8.0 (SAS Institute 
Inc. 2010). 
Home range sizes were normally distributed 
despite low sample sizes (ASY: Shapiro-Wilks’ = 
0.872, df = 13, P = 0.069; SY; Shapiro-Wilks’ = 
0.972, df = 6, P = 0.907). Core area size was log 
transformed to meet the assumptions of normality. 
Home range and core area size were compared 
between age classes and location (Rio Sabana or 
its tributary) with a two-way ANOVA. A Chi- 
square contingency table was used to ascertain if 
number of overlapping individuals was contingent 
upon age of the individual. Variable importance 
weights (IWi) were calculated by summing the 
candidate model weights that included the vari¬ 
able (Burnham and Anderson 2001). Least square 
regression was used to examine the relationship 
between prey availability and individual body 
condition. Body condition was obtained from the 
residuals of a linear regression of mass and the 
first principal component from a principal com¬ 
ponent analysis using morphometric data to 
account for body size (Jakob et al. 1996). A 
Kniskal-Wallis test was used to identify differ¬ 
ences in body condition between years due to 
small sample sizes within each year. Kolmogorov- 
Smirnov tests were used to compare body 
condition and habitat use between age classes. 
Pearson correlations were used to identity the 
relationship between habitat use and individual 
body condition. A Chi-square contingency table 
was used to examine differences in habitat use 
between after-second-year (ASY) and second- 
year (SY) individuals. Arcsine square root trans¬ 
formations were used to normalize data collected 
as percentages. Prey abundance was log trans¬ 
formed to meet the assumptions of normality. We 
used a = 0.05 for all statistical tests and data are 
presented as means ± SE. 
RESULTS 
Twenty-two waterthrush were tracked during 
the 3 years, eight in 2005, seven in 2006, and 
seven in 2007. Fourteen were ASYs. six were 
SYs, and age of two could not be reliably 
classified. Males and females have similar 
plumage and were not distinguishable. 
The number of location points obtained per 
individual ranged from 30 to 206 with a mean of 
91.2 ± 10.6. Home range size varied widely from 
0.38 to 8.80 ha. The average home range and core 
area sizes were 3.03 ± 0.54 ha and 0.71 ±0.14 ha, 
respectively. ASY (2.15 ± 0.44 ha) individuals 
had smaller home ranges than SYs (4.65 ± 
1.24 ha: F u „ - 8.00, P = 0.013). Core areas 
ranged from 0.07 to 2.14 ha. Older individuals 
(ASYs) had smaller core areas than younger (SY) 
individuals (F M * * 8.53. P = 0.011). Neither 
home range (F U8 =■ 2.55, P = 0.131) or core area 
(f 1 18 , p = 0.181) size differed between the Rio 
Sabana and its tributary when accounting for the 
age of the individual. The extent of home range 
overlap varied between 0 and 71 .5% with a mean 
overlap of 20.5 ± 5.03%. The home ranges of 
ASYs (24.6 ± 6.67%) had more overlap than the 
home ranges of SYs (4.10 ± 2.20%; / = 2.31, di 
= 17, P = 0.03). The mean percent overlap of 
core areas was 6.45 ± 2.66% and did not differ 
between age classes (t = 0,92, df = 17, P = 
0.37). Neither the number of overlapping home 
ranges (X 2 3 = 3.75, P > 0.05) nor core areas OC 2 
= 2.43, P > 0,05) differed between ASYs and 
SYs. The density of waterthrush during the study 
ranged from three to five individuals/km and was 
estimated more accurately in our study through 
use of radio transmitters. 
Waterthrush foraged mostly along streams (64 
± 7%) followed by muddy substrate (27 ± 6%). 
A small proportion of the time foraging was in 
housing developments (7 ± 5%) or on nearby 
roads (2 ± 1%). After-second year individuals 
spent more time foraging along streams and 
muddy substrates than expected and avoided 
