200 
D.J. Kitchener, N. Cooper, I. Maryanto 
Function 2 
Function 2 
Figure 3 Plots of Functions 1 and 2 from Canonical Variate Analysis (DFA) of male and female adult Myotis 
combined, based on the following three island groups in the Lesser Sundas: Alor, Wetar and Yamdena. The 
NSW specimen (arrowed) was unallocated (see text for clarification) for (a) five skull and dental characters 
and (b) five external characters. Locality codes as for Figure 2 captions. 
width, M 2 breadth, least interorbital breadth, C'C' 
width) selected from the above analysis and using 
the Western Australia and Queensland/Northem 
Territory Groups only. The Solomon Island sample 
(N=2) and the Papua New Guinea/Seram sample 
(N=3) were too small to include as groups - so 
these latter specimens were included in this 
analysis as unallocated specimens. This analysis 
extracted a highly significant function with 97.1% 
of individuals classified to their correct group. 
Only one specimen was misclassified between 
these these two groups: one from the Western 
Australian Group was misclassified to the 
Queensland/NT Group (Figure 5a). On this 
Function the two Solomon Island specimens 
allocated to the Western Australian Group while 
the PNG/Seram individuals were intermediate. 
The characters loading heavily (>0.5) on Function 1 
and which were influential in discriminating 
between the Western Australian Group and the 
Queensland/NT Group were anteorbital foramen 
width, dentary length and C'C 1 width (Table 5a). 
The Western Australian Croup - externals 
A DFA was run on all characters for the island 
groupings identified in the earlier DFA analysis on 
skull, dentary and dental characters (WA, Qld/NT, 
Solomons, PNG/Seram). This latter analysis 
extracted two significant Functions. Function 1 
explained 77.6% of the variance, and Function 2 
20.6% (Figure 4b). A total of 89.5% of individuals 
were classified to their correct group. 
Misclassifications were as follows: one Western 
Australian Group individual to the Queensland/ 
NT Group and three of this latter group to the 
Western Australian Group. A DFA was again run 
using a reduced set of five characters (ear length, 
forearm length, digit 4 metacarpal length, digit 4 
phalanx 2 length and digit 5 metacarpal length) 
selected from the above analysis and using the 
Western Australia and Queensland/Northern 
Territory Groups only. The Solomon Island sample 
(N=2) and the Papua New Guinea sample (N=2) 
were too small to include as groups - so these were 
included in this analysis as unallocated specimens. 
This analysis extracted a highly significant 
Function with 100% of individuals classified to 
their correct group. On this Function both the two 
Solomon Island specimens and the Papua New 
Guinea specimens allocated to the Queensland/NT 
Group (Figure 5b). The characters loading heavily 
(>0.5) on Function 1, and which were influential in 
