Journal of the Royal Society of Western Australia, Vol. 67, Parts 3 and 4, 1985, p. 137-148. Dune vegetation of the Swan Coastal Plain, Western Australia by I. D. Cresswell* and P. B. Bridgewater** School of Environmental & Life Sciences, Murdoch University, Perth, W.A. *Now at-Division of Groundwater Research, CSIRO, Wembley, W:A. **Now at-Bureau of Flora and Fauna, Canberra, A.C.T. Manuscript received 19 June 1984; accepted 19 1'.1arch 1985. Abstract A study of vegetation on the aeolian deposits of the Swan Coastal Plain, using classifica­ tion techniques, revealed 49 vegetation units. These units were agglomerated into plant communities and complexes, on the basis of floristic composition. A Conspectus to the vegetation classification is provided. Vegetation in this region is generally considered to be species-rich. Vegetation on the oldest <Bassendean) dune system was most species-rich. The complexity of vegetation pattern (veget<ttion texture) was greatest on the intermediate aged (Spearwood) dune system. Vegetation on the Quindalup dune system was both species-poor and had the simplest texture relative to the other two dune systems. It is postulated that species-richness and vegetation texture change dynamically through time,_ as the dune systems become more stabilised and leached. A key to the described communities is provided to allow researchers, conservation managers, teachers and other interested persons to allocate vegetation of particular bushland sites to a plant community described in the Conspectus, and thus increase knowledge of the vegetation variation in bushland remnants across this region. Introduction successfully in this vegetation, and the late Prof R. Ti.ixen (pers. com.) remarked, on seeing some of the Much of the work on vegetation description in raw data from this study, that this must surely be Western Australia has used a structural or physio­ one of the best areas in the world to apply classi­ gnomic approach, e.g. Diels 0906), Speck (1952), fication techniques! Seddon (1972). Recent work (e.g. Heddle 0979), Heddle et al. 0980)) utilised floristic attributes, within a structural framework, based on the work of Methods and results Havel 0968) who invoked the concept of site­ vegetation types using a quantitive ordination-type (a) Data collection and analysis approach to the analysis of floristically based data. In the present study, over 400 separate samples of Heddle 0979) emphasised what she saw as a vegetation were collected from about 100 sites on the confusion of vegetation pattern, caused by the aeolian derived dune systems of the Swan Coastal apparently continuous nature of vegetation change. Plain <Fig. 1). Each sample was a quadrat of 10m2, She states: from which all vascular plant species were listed. "For the vast majority of the area investigated in the Each species was allocated a value on the Braun­ Perth Region the pattern of vegetation has been detected Blanquet 0964) cover/abundance scale. on the localized scale but not over larger areas. Possibly one of the reasons was the difficulty encountered by earlier workers in classifying large areas of vegetation. This As an anonymous reviewer indicated, choice of was due to the fact that the vegetation is a complex, any technique is largely a matter of convenience and predominantly continuous population pattern, with both purpose, a point with which we concur. In the individual species and groups of species having dissimilar distribution patterns, vegetation in the field being con­ present study we have opted for a classification sequently mixtures in varying proportions of continuity technique, with the aim of identifying and clarifying and discontinuity. This multi-dimensional nature of the the main vegetation variation, despite the apparent vegetation is by no means a feature unique to Western Australia." confusion caused by high levels of species richness. We believe that vegetation of the Swan Coastal Initially vegetation samples were allocated to Plain (Fig. 1) is amenable to treatment by classi­ groups using the cluster analysis technique of Carlson ficatory techniques; the sharply changing nature of ( 1972). These were then further refined using com­ the soil systems and a highly seasonal climate are puter program VEGCLASS <Bridgewater and environmental controls which favour sharp, rather Morales, 1982). VEGCLASS is a computer-aided than diffuse, boundaries in vegetation. system of tabular synthesis, which simplifies analysis The very high species richness of the Coastal Plain of large data sets. These analyses revealed 49 vegeta­ vegetation can, however, obscure these boundaries tion units, which were arranged in a hierarchical and produce an illusion of vegetation complexity. A classification based on floristic relationships, follow­ great many species occur spasmodically in space and ing the schema used by Pen (1983). Details of these time, reacting to seasonal perturbations and irregular vegetation units appear in the following conspectus, environmental features such as fire. Other reasons and full floristic tables are included in Cresswell for high species-richness are advanced by Lamont <1982). Vegetation from Rottnest Island, and other et al. 0984). Previous work by Bridgewater (1982) offshore islands, has some special features which are showed that a classification strategy could be used not treated in this paper. 137 [Eill] QUINDALUP 1::--::~--J SPEARWOOD i:::TJ BASSENDEAN 0~ ~6 6 KILOMETRES SCALE 1:20 000 ~ Journal of the Royal Society of Western Australia, VoL 67, Parts 3 and 4, 1985. (b) Conspectus of vegetation units Tables 1-3 are summaries of species occurrences Vegetation units identifed in the study are described in the vegetation units. Identifying species for each in this section, from floristic and structural view­ unit are those with presences of 4 or 5 in each points. A key to the vegetation units is included as table. Tables 4 and 5 summarise the congruence an appendix. This key should enable workers in the between these vegetation units and the structural field to identify particular vegetation sites within the formations of Specht et a!. 0974). region described. A. Stirlingia latifolia-Oxylobium capitatum The base unit in the hierarchy is the community, complex analagous to the association of the Ziirich-Montpellier System (Bridgewater 1981). Communities are A.l. Dasypogon bromeliifolius-Lyginia divided into sub-communities and variants, and barbata community aggregated to form complexes. In most cases sub­ A.l.a. Scaevola paludosa sub-community communities are named after a distinguishing species. Where a sub-community has the same species comple­ A.l.a.i. M onotaxis grandiflora variant ment as that for the community the epithet "typical" Usually restricted to the tops of dune ridges in is used. Use of the community-complex nomencla­ the Karrakatta soil association. Occasionally ture follows that of Pen ( 1983) and should not be Xylomelum occidentale occurs as extensive patches, confused with the use of complex by Heddle et al. more or less replacing Banksia species as dominants. (1980). A.l.a.ii. Acacia willdenowiana variant For the lowest level of hierarchical division some distributional and ecological notes are included. Dis­ Localised on dune · slopes at the junction between tribution of the vegetation units across the Swan Karrakatta and Bassendean soil associations. Coastal Plain is shown in Figure 2. Species nomen­ A.l.b. Hardenbergia compton/ana sub­ clature is that of Marchant (in prep.). Specimens of community most species ·named are lodged with the Western Australian Herbarium. Terminology of the geo­ A.l.b.i. Leucopogon propinquus variant morphic units used in the text follow Bettenay Localised on the upper slopes of dune ridges in et a!. 0960). the eastern sector of the Karrakatta soil association. Dune QUINDALU p I System SPEARWOOD BASS EN DEAN Soil p Association QUINDALU COTTESLOE KARRAKATTA BASSENDEAN/SOUTHERN RIV ER Average Species (11) (22) (37) (40) Richness 8 3{29) {38)A.I.b. c 2{19) D. I (12) 8.1.a (42) C.l.a (21) D.2.b. (13) A.l.a (47) A2.b (431 5 D.2.a(17) C. I b.(17) >= A l.c. (42) - - A2.a. (23) ~ ~ IT: D.2.c.(13) D4 (II) > A3(44) r- 8 2 (22) ~ E(7) => -- 0 B I. b. (24) gs--> F(3) __1._1_ 1.2/1.3/J 1/J.2/K/M/N HI H.2 H.3 ~ z~ L 1/L3 ~B L2. f5i= G.I/G.2/G 3 "-;!: n:w Q(') ~w .:> n:owz iii:'i :r>­ a_Ww>: Figure 2.-Distributions of plant communities across the Swan Coastal Plain. Tl!e length and position of the li.nes is an indication of the breadth of the community distribution across the soil associations. Numbers in parentheses are trw average species richness for vegetation samples which make up the communities. 139 Journal"of the Royal Society of Western Australia, Vol. 67, Parts 3 and 4, 1985. A.l.b.ii. Allocasuarina fraseriana variant B.3. Dryandra nivea-Lechenaultia linarioides Common on dune slopes of the Karrakatta soil community association. Occurs on ridge topes in the Cottesloe soil associa- A.l.c. Patersonia occidentalis community t;on, with limestone cap rock slightly exposed, or near to the surface. In some localities Agonis Restricted to swales and lower dune slope in the fiexuosa occurs in the tree layer. Bassendean, Southern River and eastern Karrakatta soil associations. Excessive disturbance to this com­ C. Dryandra sessilis-Calothamnus quadrifidus munity by mechanical activity can cause a shrub complex community dominated by Adenanthos cygnorum to C.l. Dryandra nivea-Phyllallfhus calycinus develop. Many road cuttings and housing develop­ community ments in the eastern Bassendean system clearly show this. C.I.a. Hakea prostrata sub-community A.2. Allocasuarina humilis-Synaphea Confined to the western parts of Cottesloe soil spinulosa community association, on shallow soils with large proportion of A.2.a. Acacia pulchella-Conostephium limestone rock outcrop.