568 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol 123, No. 3, September 2011 
Cooper 2006, Muvihill et al. 2008, Mattsson et al. 
2009). The breeding population of the waterthrush 
has remained stable and/or has slightly increased 
throughout its range (0.7%/yr, 1966-2007) (Sauer 
et al. 2008). Waterthrush occur along wooded 
freshwater streams throughout their non-breeding 
range (Eaton 1953). Master et al. (2002) suggest¬ 
ed they are also riparian specialists during the 
non-breeding season inhabiting streams with 
characteristics similar to that of the breeding 
season. 
We examined habitat use and behavioral 
strategies used by Louisiana Waterthrush during 
the non-breeding season. Our objectives were to 
quantify preferred habitat, describe on and off- 
stream habitat use. and ascertain if age affects 
habitat use. We made the following predictions: 
(1) waterthrush are riparian specialists occurring 
along headwater streams, and (2) they are 
territorial with older dominant individuals having 
smaller more resource-rich territories than youn¬ 
ger individuals. 
2009). Home ranges were quantified by collecting 
a minimum of 50 (Seaman et al. 1999) locations 
with a Garmin 72 or 76 handheld global 
positioning system (GPS) (Olathe, KS, USA), 
Individuals were located via homing and followed 
for 2 consecutive hours taking locations at 5-min 
intervals daily for the life of the transmitter 
(~ 12 days). Five-min sampling intervals were 
chosen so birds were able to switch between 
foraging substrates and/or could traverse their 
home range during this time. Accuracy of home 
range delineation increases at shorter time inter¬ 
vals. despite the possible autocorrelation that may 
exist between bird locations (de Solla et al. 1999). 
Ground and foraging substrates were quantified 
within 3-m radius plots at 12 randomly stratified 
bird locations based on the number of foraging 
observations within the different habitat types 
(stream and off stream), as well as six non-use 
locations created by randomly generated coordi¬ 
nates within individual home ranges. Plots were 
stratified by categorizing bird observations based 
METHODS 
Study Area .—This study was conducted along 
the Rio Sabana and a tributary at the northern 
boundary of the Caribbean National Forest in 
Sabana, Puerto Rico (18 46' N, 66 36' W). The 
Rio Sabana averages 9.3 m in width along this 
stretch with an average depth of 21 cm. The 
stream has riffles and rocks of varying sizes an 
average flow rate of 0.20-0.26 m/sec. and a 
mostly closed canopy of mature trees; some 
stretches are through a residential pan of Sabana. 
The amount of forest cover surrounding this 
portion of the watershed varies from contiguous 
forest to open areas with housing development. 
Fietd Procedures .—Waterthrush were captured 
between 15 January and 15 February using 
playback recordings and mist nets placed in 
highly used flight paths. Birds exhibited low 
response to playback, and most captures occurred 
with mist nets placed across streams. All individ¬ 
uals were banded with a unique color band 
combination (2 color bands and a USGS alumi¬ 
num band), and classified to age using plumage 
characteristics (Pyle 1997). No unhanded birds 
were observed alter the capture period was 
completed. All individuals (n = 22) were fitted 
wnh radio transmitters (0.36 g; Model LB-2N, 
Hoi oh, | Systems Ltd., ON. Canada) attached with 
a cotton string harness to ensure they would 
grade prior to spring migration (Hallworth et al. 
on the habitat type used and randomly selected in 
proportion to use based on the number of total 
observations. We visually quantified the percent 
cover of mud. leaf litter, water, and vegetation 
within each plot. Percent canopy cover was 
quantified using a densitometer. We measured 
prey abundance at the center of each vegetation 
plot within a 1-m radius by recording all 
arthropods observed for 90 sec and then recording 
new individuals while turning over litter for an 
additional 90 sec. All arthropods were identified 
to Order. All stretches of streams were used, and 
we were not able to compare use and non use 
areas along streams. We classified percent cover 
ot large rock (> 0.5-m diameter), medium rock 
f > 0.2—< ().5-m diameter), gravel (rock < 0.2-in 
diameter), leaf litter (LL), vegetation (Veg). mos*. 
roots and fallen coarse woody debris (Prop). anJ 
canopy cover (Canopy) along a 5-m stretch, it 
randomly stratified plot locations were within the 
stream. Width (m), flow rate (m/sec) and average 
depth (cm) of the stream were also quantified. 
Statistical Analyses .—We used home range 
tools for ArcGIS (Rodgers et al. 2007) to calculate 
fixed-kernel use distributions (UD) from GPS 
locations for each individual. We used least 
squares cross validation (LSCV) to ascertain the 
smoothing parameter value (Barg et al. 2005). L'se 
distributions ot 95% were considered an individ¬ 
ual's home range and UDs of 50% ^ cfC 
considered an individual’s core area (Barg et at. 
