Benson, Vegetation of Agnes Banks 
51 
By inspection, however, a division can be drawn, probably between sites 1 and 17, 
a division which almost entirely agrees with the first division of the MULTBET site 
classification. The habitats indicated by the MULTBET species groups are given 
in Table 6. 
The second site axis (accounting for 8.2% of variation) has burnt sites at one 
end, and sites with highest projective foliage cover at the other. Species positively 
correlated are Bossiaeaheterophylla (0.70), Acacia brownei (0.50) and Cyathochaeta dia- 
ndra (0.59). The first two species resprout vegetatively after fire although they are leg¬ 
umes, which generally regenerate rapidly from seed. Cyathochaeta diandra is reported 
by Siddiqi et al. (1976) as the most abundant species in the earliest stages of regrowth 
after fire in coastal ground-water heath in Bouddi National Park. Negatively 
correlated are Pteridium esculentwn (—0.61), Entolasia stricta (—0.58) and Persoonia 
nutans ( — 0.55). Pteridium esculentum, a rhizomatous fern, is common after fire 
but removed by regular burning. Entolasia stricta may appear rapidly after burning 
but persists for a long time afterwards. From field observations, only Persoonia 
nutans seems to be restricted to sites with a dense projective foliage canopy cover 
that are not frequently burnt. 
The third axis (accounting for 8.1 % of variation) again has burnt sites at one 
end, but is not related to projective foliage cover. Species positively correlated are 
Hypochoeris radicata (0.50), an exotic species which may indicate disturbance, 
Dilhvynia floribunda (0.47) which regenerates from seed after fire and Eragrostis 
brownii (0.46). Species negatively correlated are Hypolaena fastigiata (-0.67), 
Lepyrodia scariosa (-0.58) and Olax stricta (-0.58). Siddiqi et al. (1976) included 
Hypolaena fastigiata in a list of low-growing species which grew well in the early 
stages of regrowth after a fire and recorded Lepyrodia scariosa within a year of burning. 
The first two axes of the species ordination account for 26.1 and 19.3% of the 
variance in the vegetation data respectively. The first axis is again related to a 
moisture/drainage gradient (Figure 8) and confirms the importance of the character 
in determining the type of vegetation. The second axis appears to be related to the 
site nroiective foliage cover. Sites positively correlated with this axis are sites 20 
( 4 % projective foliage cover), 21 (4%), 5 (8%), 11 (4%) and 2 (16%). Site 6, with 
thehighest projective foliage cover of all sites (52%), is negatively correlated with 
this axis. Members of species-group 6 are concentrated at the high canopy cover 
end of the axis (Figure 8). This group includes Pteridium esculentum, which had 
a correlation with high canopy cover along site axis 2, and Hardenbergia violacea 
and Grevillea mucronulata. The latter two species generally colonize disturbed 
sites particularly where the canopy has been destroyed, yet here appear to be con¬ 
fined to sites where disturbance has been minimal and projective foliage cover is 
high Axis 3 (6.1 % of variation) is possibly related to fire; there is a positive corre¬ 
lation with burnt sites 19 (0.61) and 22 (0.57). Species at the positive end of the 
axis are Eragrostis brownii, Entolasia stricta, Amperea xiphoclada, Bossiaea hetero¬ 
phylla and Olax stricta. 
Quantitative Analysis 
The separation of the sites into two communities is illustrated by Figure 9, 
which shows percentage basal areas for the dominant tree species. Angophora 
bakeri and Eucalyptus'parramattensis, the two main dominants, are very rarely 
found together, and are confined to the dry and wet communities respectively. 
Banksia serrata is most common at the very dry end, but Eucalyptus sclerophylla 
does not appear to be influenced by variations in moisture/drainage and occurs in 
many situations. 
Figure 10 shows the projective foliage cover change along the gradient. The 
dry sites although showing marked fluctuations have a high projective foliage cover 
(mean [transformed data] = 24% + 15.8, - 12.8) and confirm the use of the structural 
formation woodland (Specht, 1970) while the wet sites (mean [transformed data] — 
6% + 5.1, -3.6) are in open-woodland. 
