ECOLOGY OF RAINFOREST AND SCLEROPHYLLOUS COMMUNITIES 
73 
In the valley floors soils are highly variable 
depending upon the nature of the parent mate¬ 
rial. Composition of the river and creek beds 
ranges from fine alluvial silts and clays to sands 
or skeletal layers of rocks and rubble from land¬ 
slides; consequently some areas show a high con¬ 
tent of organic matter while others are almost 
pure sand. 
Land use and fire history. Aboriginal hunter- 
gathers lived in the region before Europeans 
colonized the area following the arrival of 
graziers in the 1840s (Land Conservation Coun¬ 
cil 1985). Accessible sclerophyll and grassy 
forest was cleared for improved pasture to sup¬ 
port sheep and cattle grazing. Much land on poor 
soils has been abandoned and allowed to revert 
to shrubland. particularly in times of economic 
hardship. 
As in most parts of Australia, the Glenaladale 
region has a long history' of fire. Numerous fires 
have occurred this century, and in the most 
recent catastrophic fire in February 1965 exten¬ 
sive areas of the park were burnt. The fire passed 
over Woolshed Creek as a crown fire leaving the 
gully relatively undamaged; thus the rainforest 
of this gully epitomises the classic fireshadow 
(seeFig. 5A). A. W. Howitt, who in 1876 was the 
first European to visit this region, also noted this 
effect, commenting on “the cool moist atmo¬ 
sphere in the gorges while the bushfires rage 
overhead" (Howitt 1904). However, there is 
some evidence of spot firing and damage to trees 
due to flaming debris and rocks rolling down 
into the gully from the burning slopes above 
(Frankenberg 1965, Waller 1965). 
The rainforest in Bull Creek was less fortunate 
as a south-westerly change blew the fire directly 
up the lower section of the creek, burning the 
vegetation in the gully and severely damaging it. 
The rainforest higher up the creek, however, was 
protected due to the changed direction of the 
valley and remained untouched (Waller 1965). 
METHODS 
Recording of vegetation. Vegetation was assessed 
in 1985 on the Braun Blanquet cover scale in 8 x 
4 m quadrats, a size shown by minimal area 
quadrats to contain more than 85% of the spe¬ 
cies within a distinct community type. A total of 
260 quadrats were surveyed as closely as prac¬ 
ticable on a 75 m grid pattern, with some con¬ 
centration on sites showing obvious topographic 
or floristic differences. 
Where possible, all plant species were identi¬ 
fied and recorded on site. Material that required 
further examination w r as collected and identi¬ 
fied at Melbourne University and the National 
Herbarium of Victoria. Unless stated otherwise, 
nomenclature for vascular plant species follows 
Forbes & Ross (1988), for bryophytes Scott et al. 
(1976). and for lichens Filsen & Rodgers (1979). 
Nomenclature for vegetation in the study area is 
difficult due to the unusual plant associations, 
and so descriptive names have been used; e.g. 
damp sclerophyll (sensu Gullan ct al. 1981). 
Where possible the structural classifications 
follow Specht (1970), but the term rainforest is 
employed in preference to closed forest due to 
the distinctly different species compositions of 
these communities. 
Vegetation analysis. The floristic data (presence- 
absence) w'ere assessed using both classificatory 
and ordination programs in the CSIRO TAXON 
library^ (Ross et al. 1983). The program 
MACINF (an update of MULTBET; Lance & 
Williams 1967) provided an hierarchical, poly- 
thetic-agglomerative classification. MACINF 
measures the dissimilarity of individual quad¬ 
rats by the Shannon-type information stat¬ 
istic, and the fusion of these individuals into 
groups is based on a centroid sorting strategy. 
The diagnostic program GCOM (an update of 
GROUPER; Lance et al. 1968) was also used to 
identify the contribution of each species to the 
groupings at each level of the hierarchy. Initially 
a twenty group level was arbitrarily selected but 
the dendrogram w r as subjectively terminated at 
the six group level as further subdivisions were 
considered to be either minor variants or not 
meaningful. To complement the classification, 
the group centroids of the dissimilarity matrices 
produced by MACINF w'ere ordinated by prin¬ 
cipal coordinate analysis (Gower 1966) using the 
program PCOA. To display the ordination of 
group centroids the three dimensional pattern 
was showm by using the minimum spanning 
ordination (MSO) technique of Gillison (1978). 
This technique displays the pattern of group cen¬ 
troids along the third axis in terms of a series of 
circles of graded size, with the largest and small¬ 
est circles representing the two groups which are 
farthest apart on the axis. 
Insolation and moisture. The potential insola¬ 
tion at each site was determined from slope and 
aspect data obtained at each quadrat using the 
relationship given by Clifford (1951) for latitude 
38°S. These values were corrected for any shad¬ 
ing affecting the insolation due to the surround- 
