Morphological variation in Hipposideros bicolor 
Table 1 Multiple regressions on sex and localities 
(Peninsula Malaysia, Java, Sumbawa, Selaru, 
Timor, Roti, Savu and Sumba islands) of 
Hipposideros bicolor for (a) cranial, dentary and 
dental characters and (b) external body 
characters. F values are presented for the 
main effects only; there were no significant 
(P<0.05) interactions. For explanation of 
character codes see Materials and Methods 
section. Significance levels are *, P<0.05; ” 
P<0.01; and *” P<0.001. 
Table la 
Character 
Sex 
Locality 
GSL 
5.668’ 
9.054”* 
MW 
2.069 
2.067 
RL 
4.901’ 
3.936” 
CW 
0.005 
3.839” 
LIB 
2.802 
10.957’” 
CB 
0.176 
4.709” 
TBL 
0.063 
0.941 
ILD 
0.232 
16.740*” 
CL 
1.139 
0.384 
M 3 M 3 W 
5.792* 
9.936’” 
CB 
2.116 
10.239’” 
P 3 B 
0.065 
4.266” 
M’M 3 L 
0.008 
0.044 
M 3 B 
4.073 
3.580” 
C’M 3 L 
0.279 
6.521”’ 
DL 
0.655 
10.843’” 
I,M 3 L 
1.673 
6.254’” 
d.f. 
1,41 
6,41 
Table 1(b) 
Character 
Sex 
Locality 
SVL 
0.492 
2.039 
TVL 
6.897’ 
3.571” 
EL 
0.123 
4.110” 
TIL 
5.877’ 
1.441 
FA 
2.805 
3.826” 
D2MC 
0.269 
4.050” 
D3MC 
2.859 
5.351”’ 
D3P1 
2.116 
11.586*” 
D3P2 
0.556 
1.034 
D4MC 
2.513 
5.141*” 
D4P1 
5.675* 
7.026*” 
D4P2 
0.018 
0.977 
D5MC 
0.751 
4.521” 
D5P1 
5.718’ 
5.093”’ 
D5P2 
3.116 
2.207 
ANB 
1.826 
1.592 
PNB 
2.231 
3.715” 
d.f. 
1,55 
6,55 
181 
either cranial, dentary, dental or external body 
characters. However, the following characters were 
significantly sexually dimorphic at 0.05> P> 0.01: 
greatest skull length; rostrum length; outside M 3 - 
M 3 width; tail to vent length; tibia length and both 
digit 4 and digit 5 phalanx 1 length. However, 
considering the number of interactions being 
tested, these were only weakly significant 
interactions (Table la, b). 
Most cranial, dentary, dental and external body 
characters differed significantly at (P< 0.01) with 
locality except: mastoid width; tympanic bulla 
length; cochlea length; upper molar row length; 
snout to vent length; tibia length; digits 3, 4 and 5 
phalanx 2 length and anterior noseleaf breadth 
(Table la, b). This indicated the existence of 
considerable morphological variation among the 
populations. This variation is also apparent in the 
descriptive statistics presented in Table 2 for 
groupings of these island populations. 
Discriminant Function Analysis (DFA) 
Analyses of cranial, dentary and dental 
characters were run separately from those of 
external body characters. Both sexes were 
combined in these analyses. Characters excluded 
from the analyses were: zygomatic width 
(frequently damaged and some missing cases); 
greatest skull length; rostrum length; outside M 3 - 
M 3 width; tail to vent length; tibia length; digit 4 
phalanx 1 length and digit 5 phalanx 1 length 
(sexually dimorphic). 
Initially, DFA were run for all characters, but 
because many of the groupings of populations had 
fewer individuals than there were characters 
measured, DFA in all cases were also run with a 
subset of five characters. These five characters 
were selected to provide values that minimised 
Wilk's lambda (How et ah 1996). In all cases the 
reduced set of five characters provided groupings 
of individuals in discriminant function space that 
were similar to those based on the complete set of 
characters. For this reason, only the DFA based on 
five characters are detailed below. 
Cranial, dentary and dental characters 
The DFA based on five characters (lachrymal 
breadth, least interorbital breadth, dentary length, 
P 3 breadth and I 1 M 3 length) and using as groups 
each island and the Peninsula Malaysia population, 
extracted three significant functions which 
indicated the presence of four putative groups as 
follows: Selaru; Timor; Roti/Savu; Peninsula 
Malaysia - with the few Java and Sumba specimens 
not clearly allocated (Figure 2a, b). The relatively 
high proportion (89.7%) of individuals that were 
classified to their correct locality indicated that 
these groups were robust. 
The DFA based on the above four groups, with 
