462 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol 124. No. .?. September 2012 
FIG. 7. Four acoustic features with the heaviest (and significant) loadings on the discriminant function separating 
syllables sung by Woody Island and Esperance birds. 
DISCUSSION 
I conclude (1) the acoustic properties of songs 
of Zosterops lateralis gimldi occupying Woody 
Island differ significantly from the song traits of a 
ro 
o 
'c 
o 
c 
CO 
O 
FIG. 8. Comparison of Silvereyes' syllables recon 
on Woody Island versus Esperance mainland site. Val 
tor canonical variate (ordinate) from linear discrimin 
analysis of six features measured on each syllable ; 
averaged across syllables for each bird. Boxplots sh 
means (dotted lines), 25th and 75th percentiles (box), F 
and 90th percentiles (whiskers) and outliers (solid point 
conspecific mainland population near Esperance. 
15 km distant. (2) the island song features are less 
variable than those of the mainland sample, and 
(3) individuals within populations differ from one 
another. The differences between the two popu¬ 
lations. as well as between individuals, are not 
discrete but instead show considerable overlap in 
acoustic structure. The kinds of constituent 
syllables in songs were highly variable in form 
and apparently may be continuously generated us 
innovations as a bird continues singing, although 
discrete but extremely large repertoires could be 
revealed with much larger samples of songs. 
There appears to be a minor number of syllable 
types, based on my examination of song syllables, 
which can be found in subsequent songs ot an 
individual. However, what I judged to be a 
seemingly unbounded repertoire of syllables 
emerging in a singing bird presents problems for 
comparisons between individuals or populations. 
A continuous production of syllable innovations 
does not entirely preclude comparisons between 
populations, but does require longitudinal 
