Journal of the Royal Society of Western Australia, 87(2), June 2004 
DISTANCE ALONG TRANSECT (%) 
Figure 4. Differential profile showing changes in floristic composition (SSH score) with distance along a topographic profile of 35 sites 
located along a ridge top to valley floor catena at Marandoo. The approximate position of communities are indicated in both geographic 
space (horizontal axis) and ordination space (vertical axis). 
Australia (Griffin 1990) and the Chihuahuan Desert of 
North America (Wierenga et al 1987; Plumb 1991). 
Keighery el al (2000) found certain Acacia species (such 
as A. sclerospcrma, A. tetragonophylla and A. linophylla) 
widespread and often dominant across a range of 
vegetation types on red sands of the southern Carnarvon 
Basin. Similarly A. aneura dominates many recognisable 
plant communities in the Murchison River catchment 
(Curry et al 1994). 
Nine types of hummock grassland communities are 
identified on upland and piedmont slopes. These 
communities have very uneven species distributions as 
the hummock grasses (mostly either one or two of T. 
wiseana, T.basedowii, T. brizoides or T. melvillei) are clear 
dominants at most sites. Such species were highly 
influential in determining floristic gradients and major 
groupings even following severe (log) transformations. 
Despite this, the sub-dominant and uncommon tree and 
shrub species were important in determining 
communities within the vegetation subtypes. 
A major gradient in species composition identified 
corresponds to the topographic sequence from hill or 
mountain top to valley floor (i.e. catena) typical of the 
northern and eastern parts of the study area where the 
topography consists of a series of well defined ranges 
and ridges of ironstone formations with intervening 
broad valley systems. Many of the major species of the 
study area are associated with this floristic gradient (e.g. 
A. aneura , T. basedowii, E. leucophloia ). A sequence of 
communities was also recognised along such catenae. A 
catenary pattern of relatively distinct communities has 
been described many times for arid, mountainous areas 
(Wierenga et al 1987; Cornelius et al 1991; Parker 1991; 
Cowlishaw & Davies 1997) and has been described as a 
characteristic feature of such regions (Ayyad 1981). The 
general catenary sequence found in the northern part of 
the study area can be summarised as follows: i) 
hummock grassland (HG) with E. kingstnillii on mountain 
tops; ii) HG with E. leucophloia (one of three communities 
depending on substrate and soil) on upland slopes; iii) 
HG with E. gamophylla on pediment slopes; iv) HG with 
sparse A. aneura (mulga) on bajada slopes and outwash 
areas; v) mulga woodland with tussock and other grasses 
on alluvial flats. Large drainage basins have an 
additional community dominated by E. victrix and 
tussock grass below the mulga woodlands, whereas 
broad drainage lines, where present, have woodland of 
E. xerothermica with tussock grass understorey and major 
streams have tall woodland dominated by E. victrix and 
E. camaldulensis. It is clear from the descriptions given 
that several communities are strongly associated with 
particular geomorphological features. There is also 
evidence of a secondary differentiation (within 
landforms) by geological substrate; floristically distinct 
communities were found on uplands of the Marra 
Mamba Iron formation, Brockman Iron formation and 
volcanics, for instance. Relationships between vegetation 
patterns and environmental variables will be further 
explored in a subsequent paper. Vegetation studies of the 
greenstone and ironstone ranges of the Eastern 
Goldfields (Gibson et al 1997; Gibson & Lyons 2001) have 
demonstrated the importance of soil type in determining 
plant communities, with much of the edaphic variation 
linked to catenary patterns in landform and surface 
geology. 
Although several communities are floristically distinct, 
others appear to be transitional, either in spatial terms 
(e.g. ecotones or ecoclines) or in temporal terms (different 
serai or successional stages). Differential profiles enable 
one to simultaneously characterise the nature of 
transition zones in both spatial and floristic terms. In this 
way they have an advantage over ordinations which 
show patterns based on floristic similarities only. 
Differential profiles describe beta diversity (i.e. species 
turnover) along topographic gradients (Whittaker 1965; 
Whittaker 1972). The differential profile of a typical 
catena of the study area (Marandoo) shows that some 
communities are relatively discrete with only a narrow 
76 
