136 
Ganl. Bull. Singapore 70 (Suppl. 1) 2018 
Table 3. A summary of the multivariate linear regression model performed on environmental 
variables at Nee Soon against the raw scope of the first two DC A axes. Listed here only the 
significant environmental variables. 
RCA 
FE 
OA 
TEM 
ORP 
DO 
PH 
DEP 
SI 
BSG 
BRO 
MPW 
R squares 
0.312 
0.321 
0.280 
0.222 
0.159 
0.120 
0.123 
0.093 
0.063 
0.063 
0.055 
0.054 
p-value 
0.000 
0.000 
0.000 
0.000 
0.006 
0.018 
0.020 
0.041 
0.128 
0.169 
0.210 
0.224 
Table 4. A summary of eigenvalues, percentage (%) of total variation represented and p value 
of permutation test for the first four axes resulting from the CCA for odonates at Nee Soon 
freshwater swamp forest. 
Axis 
Eigenvalue 
Percentage % 
Permutation (p ) 
1 
0.6294 
26.24 
0.003 
2 
0.4757 
19.83 
0.018 
3 
0.3258 
13.58 
0.091 
4 
0.2211 
9.216 
0.315 
association included Euphaea impar, Prodasineura interrupta, Libellago aurantiaca, 
Vestalis amethystina, Vestalis amoena and Amphicnemis gracilis. All these are typical 
forest damselfly species. Environmental variables that best described this group were 
stream water with high DO, stream channel with high amount of in-stream macrophytes 
and woody debris, stream substrate with high proportion of silt, and with low water 
temperature and high ORP. The third group of sites are mostly found at headwater 
areas of various stream branches; dominant species in this group are Drepanosticta 
quadrata, Prodasineura notostigma, Orchithemis pruinans, Prodasineura collaris, 
Libellago hyalina, Copera marginipes. Environmental variables associated with 
this group were locations away from forest edges and open areas, with high riparian 
canopy cover, stream banks with significant adventitious roots into water, shallow 
water depth with low pH. The ordination points for this group of sites were spread out, 
with small areas of overlap. The majority of survey sites were in the second and third 
group, which covers most of the Nee Soon catchment. In terms of species composition 
and associated environmental variables, the three groups were very distinct from each 
other. 
Principal components analysis (PCA) was performed to analysis the 23 
environmental variables (Fig.8. The results were used to summarise environmental 
conditions for Nee Soon odonates. The significant axes were determined by percentage 
of eigenvalues which are expected to be above a random model (Broken Stick) 
curve (Fig. 9). The first four principal components of the PCA explained 63% of the 
variation in all the environmental variables (Table 5). PC 1 explained the 26% of the 
total environmental variation, was positively correlated to distance from forest edges 
