van Etten & Fox: Vegetation classification - Hamerlsey Ranges 
zone of transition with surrounding communities, whilst 
others clearly intergrade (spatially) with other 
communities. These two contrasting types of transition 
have often been described as ecotone and ecoclines 
respectively (van Leeuwen 1966; van der Maarel 1990). 
Hypothetically, transition types form a continuum 
between these two extremes (van der Maarel 1976), 
although Hobbs (1986) recognised 5 discrete types of 
transition in the field, each with their own characteristic 
pattern of species turnover and controlling factors. 
Narrow ecotones (i.e. particularly rapid spatial 
transitions between communities) are evident in the 
landscape as vegetation boundaries. Essentially two 
types of narrow ecotones have been recognised: structural 
boundaries where more-or-less only the dominant species 
of the upper stratum change dramatically (e.g. many 
savanna - forest boundaries) and floristic boundaries 
where there is an overall change in species composition 
over a short distance (Wiens et al. 1985; Hobbs 1986). 
Selective disturbance has been attributed as a cause of 
the first type of ecotone (e.g. the elimination of trees in 
part of the transition zone by disturbance), whereas the 
second is seen as the response of two distinct and 
relatively homogeneous environments meeting (van der 
Maarel 1976; 1990). The differential profile of this study 
suggests that transitions between HG communities on 
hill slopes and HG on upper pediments are ecoclines 
implying a gradual change in controlling environmental 
factors, although with considerable point to point 
variation in these factors. Such a pattern would be 
expected given the gradual change of slope moving from 
upland to pediment. The T. melvillei HG on bajada slopes 
(community 5) is spatially distinct from communities 
upslope and downslope. Transitions between these 
communities can therefore be described as narrow 
ecotones although each seems to of a different type. The 
transition with T. basedowii HG upslope seems to be more 
of a floristic boundary as there is a rapid turnover of 
species, whilst the transition with mulga woodland 
downslope (community 6B) is likely to be a structural 
boundary (given little change in species composition 
apart from the dominant species). Van Leeuwen et al. 
(1995) who studied the mulga woodlands of the study 
area, also recognised this distinctive T. melvillei HG 
community and postulated it was the result of recent fire 
within mulga woodland with hummock grass 
understorey on alluvial slopes. Indeed observations of 
burnt mulga stems in this community support this 
hypothesis. Van Leeuwen et al. (1995) felt that given 
sufficient time (without fire), this community would 
return to a mulga-hummock grass mix such as 
community 6B. In other words, these two communities (5 
& 6B) may be successional stages of the same community. 
Van Etten (1988) found structural boundaries between 
mulga woodlands with hummock grass understoreys 
and hummock grassland elsewhere in the Hamersley 
Ranges and hypothesised that, on gradual slopes at least, 
they were the result of differences in fire regimes: 
hummock grasslands burn on a regular basis, whilst 
mulga communities rarely burn and contain many fire 
sensitive species. Mulga itself is a fire sensitive species 
and, although it has a limited capacity to regenerate from 
rootstock, it is vulnerable to repeated fires (Fox 1985; 
Hodgkinson 2001). Fire then is a possible cause of these 
structural boundaries between communities 5 and 6B. 
Long-term monitoring of boundaries is occurring in the 
study area to confirm the dynamic nature of these 
boundaries (A.N. Start, pers. comm.). The floristic 
boundary upslope of the T. melvillei - mulga communities 
suggest a distinct environmental change along the gentle 
pediment and bajada slopes. It is hypothesised that this 
change corresponds to the change from Tertiary, 
consolidated, colluvium deposits of the pediments with 
their typical ironstone pavements to more recent 
alluvium deposits typical of bajada and alluvial fans. 
Transitions between communities further downslope 
could not be identified because of inadequate sampling 
in the transition zones. Clearly there is a need to sample 
more intensively along catenae and to replicate studies 
over a variety of topographical gradients elsewhere in 
the study area to clarify transition types between 
communities and to test hypotheses generated here 
relating to spatial change between communities. The 
single catena studied here, however, demonstrates the 
usefulness of transects in describing vegetation patterns 
in real space, and that transitions between communities 
identified along the toposequence vary considerably in 
terms of width and other characteristics. Austin (1989) 
and Austin & Smith (1989) recommended the study of 
transition zones as an important adjunct to ordinations 
and classifications to clarify the spatial features of 
transitions. 
In the south-western portion, geology and landform 
are considerably more complex compared to the 
ironstone ranges and intervening broad valleys 
elsewhere in the study area. Geological formations, 
which influence surface soils, comprise a range of 
volcanics (varying from ancient Archaean granites and 
gneisses to lower Proterozoic basalts to dolerite 
intrusions of various ages) and calcareous sedimentary 
deposits (Proterozoic sandstones and dolomites to 
Tertiary limestone/calcrete). Although three 
communities were recognised as existing on these 
substrates, they were relatively heterogeneous. As time 
and access constraints did not enable the full topographic 
and geological variation to be adequately sampled, there 
remains a strong likelihood that more communities can 
be defined in the southern half of the study area. One 
possible community is A. victoriae - A. bivenosa open 
shrubland (with chenopods) on calcareous flats. Mulga 
communities with understoreys different from that 
described here (e.g. dominated by T. pungens and 
Eremophila fraseri) are known for volcanic uplands of the 
area. Greater sampling is recommended for this portion 
of the Hamersley Ranges. Furthermore, a number of 
known landform types were not sampled in the study 
area as they occurred in small patches/narrow bands 
(and therefore were beyond the scale of the study) or 
were not encountered. These included scarps and cliff 
faces, patches of cracking clays, gorges and narrow 
drainage lines through pediments, all of which contain a 
fairly discrete suite of species and, frequently, species 
restricted to those land forms. 
The classification presented here therefore should not 
be regarded as a comprehensive or an ultimate account 
of plant communities of the Hamersley Ranges. Rather it 
is a preliminary classification scheme on which to build 
and modify as more land around the study area is 
sampled. Further sampling is also needed to confirm 
those communities which were shown to differ according 
77 
