434 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol 123, No. 3. September 2011 
TABLE 1. Inter-taxon comparisons of all four parameters, /'-values for Mann-Whitney U-test are shown numerically, 
with significant results in bold. Results of Isler et al.'s (1998) diagnosability criterion are given as abbreviations D = 
diagnosable or ND = not diagnosable. Blank cells provide no inter-taxon comparison. 
subgularis 
(« = 9) 
subgularis 
epia 
mangoliensis 
number of 
‘whoops’ 
‘whoop’ 
duration 
0.000 
0.000 0.000 
song duration 
song pace 
0.009 
0.000 
| 0.600 1 0.000 
epia 
D 
■ ND 
| 0.008 0.036 
(n = 12) 
ND 
D 
0.002 0.004 
mangoliensis (n = 
ND 
ND 
ND 
nd] 
14) 
ND 
D 
■m nd 
ND | 
per Mann-Whitney £/-test, whereas songs were 
not diagnosable across geographic regions using 
the Isler criterion. Thus there may be incipient— 
potentially clinal—vocal differentiation in epia 
across Sulawesi, but the differences are not as 
pronounced as towards other taxa. 
Discriminant analysis resulted in a canonical 
function plot that showed subgularis vocalizations 
clustering together as a spatial agglomeration 
separate from mangoliensis and epia with the 
latter two displaying partial spatial overlap 
(Fig. 4). Cross-validated discriminant analysis 
predicted 100% of subgularis individuals to be 
members of that taxon, while only 75% and 86% 
of epia and mangoliensis. respectively, were 
predicted to be members of their own taxon on 
the basis of the four vocal parameters examined. 
DISCUSSION 
Vocal Leapfrog Pattern.—The three taxa of 
Maroon-chinned Fruit Dove occur in a line from 
Sulawesi in the west, across the islands of 
Banggai and Peleng in the Banggai Archipelago 
in the middle, to Taliabu and Mangole of the Sula 
Archipelago in the east (Fig. I). Previous song 
descriptions of this pigeon were restricted to epia 
from Sulawesi, which is the most commonly 
encountered of the three subspecies. We provide 
the First documentation of the songs of mango¬ 
liensis from the Sula Islands and of subgularis 
from the Banggai Islands, and demonstrate 
significant vocal differences among all three taxa 
(Table 1). We also show the terminal taxa on the 
west (epia) and the east ( mangoliensis) are 
vocally closer to each other than either is to its 
geographically adjacent neighbor subgularis 
(Fig. 4: Table I ). The song of subgularis differs 
from its eastern and western neighbors in its rapid 
pace and the relatively large number of 'whoop' 
elements within a single song bout (Table 1). Its 
acoustical impression on the human ear is 
diflerent from that of the more similar-sounding 
epia and mangoliensis songs. Omission of song 
duration as a vocal parameter in our analysis 
would have accounted lor an even stronger 
diagnosability score for the song of subgularis 
compared to the other two taxa. This is especially 
true considering the similarity of song durations 
between mangoliensis and subgularis is an 
artefact of the interplay oI many “whoop" notes 
delivered at a high pace ( subgularis ) versus fewer 
"'whoop" notes delivered at a lower pace 
( mangoliensis ). 
The Sulawesi taxon epia has a much larger 
Nonb versus Cental 
values are indicated by bold print. ND = not diagnosable. diagnosability criterion. Significant 
Number of Mean numhei of 
__ ___recordings ‘whoop’ notes 
Mean duration of 
'whoop' notes 
Mean song 
duration 
Mean vocal pace 
epia Central ~l 657 ± 098 
/'-value (Mann-Whitney U) N/A 8 ’ 8 ? fj 59 
kler et al.’s (,998) diagnosability N /A 
0.13 ± 0.03 
1.45 ± 0.24 
4.56 + 0.40 
0.15 ± 0.03 
2.69 ± 1.00 
3.36 ± 0.31 
0.343 
ND 
0.018 
ND 
0.003 
ND 
