466 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 123. No. 3. September 2011 
usecHn LtfZvZ P “ '7” °f * Sm8 ‘ e Junco song from Arixnna, USA, illustrating the measumn-s 
77 ZL"! ; f7 g Z rc|M " s -f ««• «■** syllable/trilled unit and its thmt unique notes 
Shi, taZcv mt “nT unique syllable/,rilled unit and its three unique notes (el -3). (d) peat frequency, fe, 
SuhLZ sons,'” ‘ g ’ hand " ' J ' h ^ ™« •»* has six syllables, two of which ere Unique. 
,he nnS a 22' H 7™" 7 Sy “ C ln ,h " I-*** “ample, the number of unique mils is the same as 
haveaZhZXu 'nTr f,!’ « *■*«* “ <*<*■ J« »I™ syllable. This pZeular song doe, no, 
' P g D ‘ r 0-1 l uslr;l| e sample pliruses (continuous utterances of unrepeated, dissimilar syllables]. 
syllables, number of unique syllables, mean 
number of repetitions per syllable, number of 
trills* mean number of repeated syllables per trill, 
number of unique notes per trill, number of 
phrases, mean of number of noles per phrase, peak 
frequency (frequency of the maximum ampli¬ 
tude), lowest frequency, highest frequency, and 
bandwidth (highest frequency minus lowest fre¬ 
quency). We calculated the mean for each song 
feature for each vocalizing individual, and a 
single mean value represented each variable for 
an individual bird. All song measurements were 
performed manually in RavenPro 1.3 (Charif et al. 
2008) using a Hamming window and a fast 
Fourier transformation (FFT) length of 1,024, 
providing a spectral resolution of 43 Hz. 
Statistical Analyses .—Wc used two comple- 
ZTc? m f th0ds 10 “*y» differences in song 
character's, JCS among juncos from the three 
gions. ( ) Kruskal-Wallis one-way analysis of 
variance with post-hoc comparisons based on 
pairwise Mann-Whitney (/-tests and the student- 
ized range (Sokal and Rohlf 1995), and (2) linear 
discriminant (unction analysis (DFA) to classify 
individuals to regions based on song features. The 
latter analysis may be biased towards correctly 
assigning groups, and we used a jackknifed 
classification matrix as a more conservative 
grouping method. All analyses were conducted 
in Program R (R Development Core Team 20091. 
Significance thresholds were adjusted to 0.0038 
following a Bonterroni correction for multiple 
comparisons. Results are presented as means ± 
SE unless labeled otherwise. 
RESULTS 
Eleven of the 13 features of Yellow-eyed Junco 
song differed among the three regions (all. f 2 
11.15, P < 0.0037; Table 1). The only features 
that did not differ among regions were the number 
