54 
Cunninghamia 
Vol. 1 (1): 1981 
SITES 
Figure 10. Total projective foliage canopy cover plotted for sites ordinated by first GOWER 
axis (moisture/drainage gradient). See Table 5 for site order. 
The percentage of the Agnes Banks flora recorded from Hawkesbury Sandstone 
areas and coastal sand dune areas is given in Table 7. Strong affinities with the 
Sandstone flora are indicated by the similarity with Dharug National Park where 
67% of the Agnes Banks flora also occurs. However, considering the large total 
number of species at Dharug, and the size of the area investigated (c. 12 000 ha), what 
is probably most surprising is that 33% of species at Agnes Banks were not recorded 
at Dharug. This is probably due to drainage and nutrient differences between the 
sandy soils developed on Hawkesbury Sandstone and the deep sands at Agnes Banks, 
and supports the general view that geology is the major determinant of plant dis¬ 
tribution in the Sydney area (Pidgeon, 1937). 
Coastal sand dune vegetation affinities are indicated in the 47% of the Agnes 
Banks flora found at Myall Lakes and the 25 % found at North Stradbroke Island, 
Queensland. Apart from coastal areas, the only similar sand deposit recorded for 
the Sydney district is on the Mellong Range near Putty, 60 km north of Agnes Banks. 
Forster et al. (1977) describe the vegetation there as Medium Open-forest of Eucalyptus 
sclerophylla and E. parramattensis. Although a full species list is not available, a 
combined list of 64 species included only 31% of the Agnes Banks flora. A full 
species list is likely to include a higher proportion than this. Banksia serratifolia 
has not been recorded. 
The drainage factor has long been used to explain the distributions of plant 
communities within coastal sand areas. Osborne & Robertson (1939) recognized 
dune Eucalyptus forest, dry heath, wet heath and swamp communities related to 
sand depth and drainage, each with floristic differences, but also with extensive 
ecotones between them. Numerical studies by Siddiqi et al. (1973) and Connor & 
Clifford (1971) showed that similar vegetation could be divided into floristic groups 
related to moisture/drainage conditions. Connor & Clifford found, however, that 
there was considerable species overlap in the groups formed. 
The numerical analyses of the vegetation at Agnes Banks support the des¬ 
criptions of the two major communities sampled, the Eucalyptus sclerophylla- 
Angophora bakeri-Banksia serrata woodland on dry well-drained sites, and the 
Eucalyptus parramattensis low open-woodland on poorly-drained sites, but emphasize 
the continuous gradient between them. Of the communities mapped, but only 
sampled with one site, the site ordination (Table 5) placed the only site in the Banksia 
