JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA Volume 71 • Part 1 • 1988 ©tie iSoyal §ociety of Western Australia To promote and foster science in Western Australia and counteract the effects of specialization President PATRON Her Majesty the Queen VICE-PATRON His Excellency Professor Gordon Reid Governor of Western Australia COUNCIL 1988-1989 J S Pate Ph D, D Sc, FAA, FRS Vice-Presidents M Candy M Sc, FRAS B Dell B Sc (Hons), Ph D Past President J T Tippett B Sc, Ph D Joint Hon Secretaries K W Dixon B Sc (Hons), Ph D L Thomas M Sc Hon Treasurer J Dodd B A, M Sc, Ph D Hon Librarian M A Triffitt B A, ALAA Hon Editor I Abbott B Sc (Hons), Ph D Hon Journal Manager J Backhouse B Sc (Hons), M Sc, Ph D Members J S Beard M A, B Sc, D Phil W A Cowling B Agric Sc (Hons), Ph D S J Hallam M A, FAHA L E Koch M Sc, Ph D K McNamara B Sc (Hons), Ph D J Majer B Sc, D1C, Cert Ed, Ph D V Sememuk B Sc (Hons), Ph D Journal of the Royal Society of Western Australia 71 (1), 1-6, 1988 Regeneration of Acacia and Kennedia from soil stored seed following an autumn fire in jarrah (Eucalyptus marginata) forest W L McCaw Department of Conservation & Land Management Research Centre, Manjimup. WA 6258 Manuscript received October 1987: accepted December 1987 Abstract Regeneration of species of Acacia and Kennedia following a fire in autumn was studied in the understorey of jarrah forest near Dwellingup. A total of 4 species of Acacia and 2 species of Kennedia were recorded in 46 plots assessed prior to burning. Six months after burning all except 1 species ( Acacia pulchella) were recorded in more plots than before the fire. Kennedia coccinea exhibited the greatest increase in distribution following burning. Total seedling numbers did not differ significantly between plots burnt at low intensity (< 300 kWm 4 and plots burnt at greater intensity, including full scorch of the overstorey. Between August 1980 and February 1983 simi- lar changes in plant density were recorded in fenced exclosures and adjacent unfenced areas at 5 sites indicating that grazing had no major impact on populations over this period. A broadscale survey of the 2 200 ha study area 1 year after the fire indicated that about 10% of the area carried dense thickets of A. pulchella and A. celastrifolia while a further 15% carried thickets of Acacia drummondii. Introduction Understoreys of native legumes are a distinctive feature of jarrah ( Eucalyptus marginata ) forest throughout the range of this forest type. Most of the common legumes are obligate seed re- generators which re-establish from seed stored in the top few centimetres of the soil profile (Shea et al 1979, Monk et al 1981). Hardseededness (Gill 1981) is common amongst the legumes with the result that there is little or no regeneration in the ab- sence of fire or mechanical disturbance of the soil. Germination of soil stored seed following fire is influenced by a number of factors (Gill 1981) including the abundance and vi- ability of seed, the depth of seed burial, the response of different species to heat and the extent of soil heating (Shea et al 1979, Monk et al 1981, Floyd 1976). The extent of soil heating de- pends on the physical characteristics and moisture content of the soil (Aston & Gill 1976) and the heating characteristics of the fire, in particular the quantity of fuel consumed (Knight 1981). Fires during summer and early autumn provide ideal conditions for soil heating because soils are dry and litter fuel is entirely consumed (Christensen et al 1981). In jarrah forest, regeneration of dense legume thickets has often been observed following summer fires of moderate to high intensity (Peet 1971 , Peet & Van Didden 1973, Shea et al 1979, Monk et al 1981) but has rarely been associated with the spring fires of low intensity which are periodically used to reduce ac- cumulations of fuel. Manipulation of the season and intensity of prescribed burning has been proposed as a means of promoting the regeneration of dense thickets of legumes for purposes including the maintenance of animal habitat (Christensen 1980), and the establishment of understoreys resistant to Phythophthora cinnamomi, the causal agent of jarrah dieback (Shea et al 1979). The factors which determine the regeneration and subsequent growth of native legume species need to be understood if such treatments are to be implemented on an op- erational scale. This paper examines the post-fire regeneration of species from 2 genera of native legumes, Acacia and Kennedia, which are commonly found on a wide range of sites in jarrah forest. As- pects examined are: regeneration in relation to species occur- rence prior to burning; comparison of the density of regener- ation in forest subject to 2 levels of crown scorch (an index of the intensity of fire); and the effect of grazing exclosures on changes in plant numbers during the 2.5 years following seedling germination. Study area The study was undertaken in the 2 200 ha Hakea forest block, 25 km SE of Dwellingup (34° 43'S, 116° 04'E). Six of the vegetation complexes described and mapped for the Darling System by Heddle et al (1980) occur within the area, as follows (Fig. 1); Dwellingup complex, medium-high rainfall (1074 ha); Yarragil complex, maximum swamp development (450 ha); Yarragil complex, minimum swamp development (139 ha); Murray-Bindoon complex (211 ha); Pindalup-Yarragil complex (172 ha); and Cooke complex (154 ha). A60438- 1 1 Journal of the Royal Society of Western Australia 71 (1), 1988 N 1 km Vegetation complexes correspond to the major landform units of Churchward & McArthur (1980), and typically combine several of the more specific site vegetation types defined by Havel (1975). The dominant vegetation on the lateritic uplands that com- prise about half the study area (Dwellingup and Cooke com- plexes) is open forest of jarrah and marri ( Eucalyptus calophylla), 20-30 m in height, with an understorey of low shrubs. The for- ests of the other vegetation complexes may also contain jarrah and marri but are generally less than 20 m in height and are dis- tinguished by the presence of other eucalypt species including yarri ( Eucalyptus patens) and wandoo ( Eucalyptus wandoo). The area of Yarragil complex in the southern part of the block is dominated by an extensive Banksia littoralis swamp. Fire history records available from 1938 onwards indicate that only small parts of Hakea block have been burnt by wildfires, but that the entire area was subject to fuel reduction burning in spring 1968 and again in spring 1974. Methods Pre-fire assessment The number of plants of Acacia and Kennedia species was counted in each of 46 permanently marked 20 x 20 m plots during January 1980. These plots were randomly located within the study area (Fig. 1). Contacts with each species were also re- corded from 50 point samples following the technique of Levy & Madden (1933). Two legume species common in the area ( Acacia preissiana and Bossiaea ornata ) were not included in the study because they commonly regenerate from rootstock after fire and it was considered that this might create difficulty in de- termining the true extent of regeneration from seed. Litter fuels samples were collected from four 1 m 2 quadrats in each plot for determination of oven dry weight. Understorey fuels were not sampled at each plot, but would probably have contributed an additional 1-2 t ha 1 to the total quantity of fuel on upland sites (Sneeuwjagt & Peet 1985). Samples of litter fuel and the 0-5 cm horizon of the soil were collected at a number of upland sites on the morning immedi- ately prior to burning of the study area. Samples were subsequently oven dried for moisture content determination. The fire The study area was burnt on 22 March 1980. The first wide- spread rain following the summer drought had fallen 2 days pre- viously with 3 mm being recorded in a portable rain gauge lo- cated at the eastern end of the study area. On the day of the fire, weather conditions were recorded at the study area and at a nearby fire lookout tower (Table 1). The area was ignited by an aircraft dropping incendiaries on a grid pattern; ignition commenced at 1500 hrs and continued for 2 hours, after which time fires continued to burn into the eve- ning until about 2100 hrs. Around the perimeter of the area the intensity of fires was generally low (<300 kWm” 1 ) but consider- ably higher intensities were observed from the aircraft within the centre of the area. Table 1 Summary of fuel, soil moisture and weather conditions at the Hakea study area on the day of burning (22 March 1980). Weather Maximum temperature (°C) 27 Minimum relative humidity (%) 30 In-forest wind speed at 1600 hr (kmh~L 3 Tower wind speed at 1600 hr (kmh _ 0 11 Tower wind direction ESE Fuel Litter fuel quantity (t/ha) mean 10.5 range 3.2-24.9 SE 0.8 Minimum litter fuel moisture content range (%) 9-10 Soil Soil Dryness Index" c 1600 Moisture content of 0-5 cm horizon (%) mean 4 range 3-7 A Refer to Mount (1972) and Burrows (1987). 9 Journal of the Royal Society of Western Australia 71 (1). 1988 Post-fire assessment Several months after the fire the height of crown scorch was measured on a sample (10-20) of dominant trees in each plot to provide an estimate of the intensity of fire. Based on the re- lationships reported by Burrows (1984), plots with scorch height <15 m were estimated to have burnt with an intensity <300 kWrrr 1 , which is regarded as low (Cheney 1981). Acacia and Kennedia regeneration within the 20 x 20 m plots was initially assessed in August 1980. The number of seedlings of each species was counted on four 10 m 2 circular quadrats which were located in the corners of each plot. This sampling procedure was adopted because the very high num- bers of seedlings in some plots made it impractical to count seed- ling numbers over the entire plot area. Plant densities before and after burning were expressed in terms of plant/m 2 . Point sampling was repeated in each plot 2.5 years after the fire. During early August 1980 grazing exclosures were con- structed at 6 sites (Fig. 1) where germination of A. drummondii and A. pulchella was widespread. At each site an area of about 0.5 ha was fenced with wire mesh to 1.5 m height, and seedling numbers counted on 40 permanently marked 1 m 2 quadrats within the exclosure and on 40 adjacent quadrats outside the exclosure. Quadrats were recounted in February 1983. Mean seedling numbers in 1981, and changes in seedling numbers be- tween 1981 and 1983 were compared between fenced and unfenced quadrats at each site using a t test. The broadscale distribution of Acacia and Kennedia species was surveyed 12 months after the fire. Survey points were lo- cated every 100 m along a series of transects which ran parallel at 400 m intervals across the study area. Plant numbers were counted on a 10 m 2 circular quadrant at each point. The distri- bution of each species was then plotted at 1:50 000 scale and overlayed onto a map of the vegetation complexes of Heddle et al (1980) for tallying of species occurrence by complex. Expected frequencies were calculated according to the hypoth- esis that species were present in equal proportion in each com- plex. Observed and expected frequencies were compared using the likelihood ratio (Zar 1974). Several species recorded only at a few points were not subject to this analysis. Results Four species of Acacia and 2 species of Kennedia were re- corded in plots prior to burning (Table 2). The most common species, Acacia drummondii, occurred in 17 plots while the re- mainder were found in 1-9 plots. Six months after the fire all species, except Acacia pulchella, occurred in more plots than before the fire; A. pulchella was re- corded in 9 plots before the fire but only 7 afterwards (Table 2). The Acacias typically regenerated within plots where they had been’recorded prior to burning and in 1-4 additional plots. Two species which had not been recorded before the fire, Acacia alata and Acacia extensa, each regenerated in 1 plot following the fire. Kennedia coccinea was considerably more widespread follow- ing the fire being recorded in 26 additional plots, while Kennedia prostrata regenerated in 5 additional plots (Table 2). In a few instances species did not apparently regenerate in all plots where they had been present prior to burning. A. drummondii . Acacia celastrifolia and K. coccinea each disap- peared from 1 plot after the fire, while A. pulchella disappeared from 3 plots. The relationship between the number of plants in plots before and after burning varied between species. The density of regen- eration of the 2 most common Acacias was strongly correlated to the preburn densities of these species within plots (A. drummondii r = 0.756, N = 21, P < 0.001; A. pulchella r = 0.718, N = 9, P< 0.01). A similar trend was apparent for several other species of Acacia although this could not be veri- fied statistically because they occurred in too few plots. On the other hand both species of Kennedia mostly regenerated in plots where they had not been recorded prior to burning and no such relationships were apparent. Numbers of seedlings in plots burnt at low intensity (<300 kWm *) were not significantly different from those ob- served in plots burnt at higher intensities, including a number subject to complete scorch of the overstorey canopy (Table 3). Table 2 Occurrence of legume species within plots before and 6 months after burning. Species Not recorded before or after burning No. of plots in Recorded before but not after burning which each species was: Not recorded before but recorded after burning Recorded both before and after burning A. alata R. Br. 45 0 1 0 A. celastrifolia Benth. 39 1 2 4 A. drummondii Lindley 25 1 4 16 A. extensa Lindley 45 0 1 0 A. pulchella R. Br. 36 3 1 6 A. urophylla Benth. ex Lindley 41 0 2 3 K. coccinea Vent. 15 1 26 4 K. prostrata R. Br. 40 0 5 1 A60438-2 3 Journal of the Royal Society of Western Australia 71 (1). 1988 Comparison of the density of seedling regeneration of intensity <300 kWm -1 ) and >15 m Table 3 all legume species in plots with levels of crown scorch <15 m (estimated fire or fully scorched (estimated fire intensity >300 kWm" 1 ). Variable Scorch height <15 m Scorch height >15 m Total no. of plots in class 15 31 No. of plots containing legume seedlings 15 30 No. of seedlings me an 1.32' 0.958' SD 1.110 1.653 range 0.025-3.725 0-6.500 'Not significantly different at 0.05 level. Table 4 Mean frequency (%) of legume species in 46 plots before and 2.5 years after burning determined by Levy point sampling. Species Before burning Mean frequency (%) 2.5 years after burning A. alata 0 0 A. celastrifolia 0.043 0.826 A. drummondii 0.348 3.391 A. extensa 0 0 A. pulchella 2.000 2.860 A. urophylla 0 0,261 K. coccinea 0.043 0.130 K. prostrata 0 0.043 No significant relationship was apparent between the number of seedlings regenerated and the quantity of litter fuel in the plot (Fig. 2). “ 2-50 Seedl ings m~2 At each of the 6 sites where exclosures had been fenced the density of Acacias in fenced and unfenced areas did not differ significantly at the time of the initial assessment in August 1981 (Table 5). Between August 1981 and February 1983 the density of plants declined to a similar extent in both fenced and unfenced quadrats at 4 of the sites. At the other 2 sites plant numbers increased over the same period: at site 1 plant num- bers in the unfenced quadrats increased; at site 3 plant numbers in the fenced quadrats increased. Only at site 3 did the change in plant numbers differ significantly between fenced and unfenced quadrats. Changes in plant density were of a similar magnitude at each site irrespective of the location within the study area. 0 25 s All of the species recorded in the plots were also encountered during the broadscale survey conducted 12 months after the fire at generally similar relative levels of frequency (Table 6). A further species ( Acacia divergens) was also encountered locally in swampy areas where no plots had been established. The hy- pothesis of uniform relative frequency in all vegetation com- plexes was sustained for K. prostrata and K. coccinea but not for any of the common Acacias (Table 6). m • 1 10 20 Fuel Quantity (tonnes ha' 1 ) Figure 2 Number of seedlings per m z in August 1980 in relation to the quantity of litter fuel in each plot at the time of burning. Pre- and postfire frequencies were calculated for each species from the Levy point sampling (Table 4). Within 2.5 years the fre- quency of contacts with each species had returned to prefire levels, and substantial increases were demonstrated for A. drummondii. A. celastrifolia . Acacia urophylla and K. coccinea. Data from the broadscale survey were used to estimate the proportion of the study area which had sufficient density of the larger legumes (A. pulchella and A. celastrifolia) to facilitate de- velopment of dense thickets. For the purposes of this estimate it was assumed that a density > 1 plant/m 2 of either species 1 year after the fire indicated potential for thicket development. On this basis plant densities were estimated to be sufficient for thicket development over about 10% of the study area. Using the same plant density criteria a further 15% of the area carried thickets of A. drummondii. Several of the other species grow to sufficient size but do not typically regenerate as thickets in jar- rah forest (A. urophplla) or are restricted to localized areas, usually in swamps (A. divergens. A. alata. A. extensa). 4 Journal of the Royal Society of Western Australia 71 (1), 1988 Table 5 Mean numbers of A. drummondii and A. pulchella recorded in fenced and unfenced areas at 6 sites (40 lm 2 per fencing treatment at each site). See Fig. 1 for site locations. permanent quadrats Site Fencing treatment No. of plants m ^ in August 1980 Significance 1 Change in number m - ^ between August 1980 and February 1983 Significance^ 1 fenced 16.7 t = 0.095 0.10 t = 1.166 unfenced 16.3 NS + 1.08 NS 2 fenced 20.9 t = 0.421 0.75 t = 0.540 unfenced 17.8 NS -1.10 NS 3 fenced 24.3 t = 0.691 + 1.48 t = 2.116 unfenced 29.0 NS -0.56 p < 0.05 4 fenced 20.8 t 0.585 0.59 t = 1.39 unfenced 24.3 NS -0.97 NS 5 fenced 11.5 t = 0.850 1.60 t =0.860 unfenced 14.2 NS 1.00 NS 6 fenced 13.1 t = 0.700 - 1.75 t = 0.093 unfenced 15.6 NS 1.68 NS 1 Comparison of plant density in fenced and unfenced quadrats at August 1980. 2 Comparison of change in plant density in fenced and unfenced quadrats from August 1980 to February 1983. Table 6 Frequency of legume species within the 6 vegetation complexes represented in the Hakes study area, based on data from 537 sample points. Figures in parentheses indicate the frequency expected if representation was proportional to the area of each complex. Vegetation complex Area within study (ha) A.alata A . celastrifolia Observed and (expected) frequency within each complex for: A.diuergens A. drummondii A.extensa A. pulchella A.urophylla K. coccinea K. prostrata Dwellingup med-high rainfall 1074 0 6(13.1) 0 60(73.9) 0 28(30.6) 10(10.2) 49(49.0) 14(16.0) Yarragil (maximum swamp development) 450 0 12(5.5) 1 44(31.1) 0 7(12.9) 4(4.3) 20(20.7) 5(6.8) Yarragil (minimum swamp development) 139 0 1(1-7) 0 12(9.6) 5 9(4.0) 5(1.3) 7(6.4) 2(2.1) Murrary-Bindoon 211 1 1(2.6) 0 24(14.7) 0 2(6.1) 2(2.0) 12(9.7) 5(3.2) Pindalup-Yarragil 172 1 0(2.1) 0 2(11.8) 0 6(4.9) 0(1.6) 7(7.9) 5(2.6) Cooke 154 0 7(1.9) 0 10(10.8) 0 11(4.5) 0(1.4) 6(7.1) 2(2.3) Total 2 200 2 27 1 152 5 63 21 101 33 G Statistic — ND 24.54 ND 25.29 ND 18.69 12.49 1.27 3.43 Significance level — ND <0.001 ND <0.001 ND <0.005 <0.05 >0.05 >0.05 ND - not determined Discussion Six months after burning all of the legume species examined in this study, except A. pulchella, were recorded in a greater number of plots than before the fire. There were relatively few examples where species were recorded in plots before, but not after the fire, and these can probably be attributed to the sam- pling technique adopted for the post-fire assessment. The den- sity of Acacia populations has been observed to decline rapidly in the initial 4 years after fire (Monk et al 1981). Most, if not all species would have been less widespread at the time of the pre- fire assessment in 1980 than in the period soon after the 1974 fire. The most pronounced increase in apparent distribution fol- lowing burning was displayed by K. coccinea and it is likely that this species, being shortlived, had declined to a greater extent in the interfire period than had the longer lived, woody Acacias. Species richness and diversity in jarrah forest generally decline when the period between fires exceeds about 6 years due to the senescence of fireweeds and smaller herbaceous species (Bell & Koch 1980). 5 Journal of the Royal Society of Western Australia 71 (1), 1988 The intensity of fire, as indicated by crown scorch height, did not appear to influence the extent of germination. Peet (1971) found that the contribution of leguminous species to the understorey, based on projected foliage cover, only varied slightly between areas burnt at a wide range of intensity during a summer wildfire; legumes contributed 38% in defoliated for- est, 40% in fully scorched forest and 33% in forest subject to lesser crown damage. Legume contribution was, however, only 20% in similar forest which had been burnt in spring at low inten- sity. This is consistent with the hypothesis that germination of soil stored seed is governed primarily by fuel and soil moisture at the time of burning, rather than the acutal behaviour of the fire (Christensen & Kimber 1975). However where the intensity of fire is sufficient to seriously damage the forest canopy it is likely that understorey thicket development may be enhanced by the temporary reduction is competition. Peet (1971) re- ported that the total understorey cover (all species including leg- umes) did in fact increase proportionally with the level of dam- age to the canopy. Cochrane (1968) also reported that canopy characteristics had a primary regulating role on the regener- ation of understorey species following fire in dry sclerophyll for- est in Victoria. No significant influence of fuel quantity on legume germi- nation was established in this study, despite the probable im- portance of this factor (Christensen and Kimber 1975). Obser- vations following the fire indicated that a large proportion of litter fuel had been consumed, so that differences between the quantity measured prior to burning and the quantity actually consumed by the fire were not great. The most likely expla- nation for the lack of any clear relationship is that it was ob- scured by variation between plots in the quantity of seed, depth of burial and heat response characteristics of different species. A further factor potentially influencing germination response was rainfall before the fire. Although the amount of rain (3 mm) was small, it may have been sufficient to dampen the surface layer of soil and restrict penetration of heat to deeper levels (2- 4 cm) where much of the seed may be located (Shea et al 1979). The similar changes in numbers of plants in fenced and unfenced areas at 5 sites suggests that grazing did not have a major impact on legume population densities after August 1980. However, the impact of grazing on population densities prior to August, and possible effects on the height or form of in- dividual plants were not determined. Increases in number of plants at 2 sites after August 1980 would have been as a result of later germination. These results contrast with results re- ported by Shea et al (1979) and Christensen (pers. comm.) where heavy grazing of legumes has been observed particularly on small burnt areas and close to unburnt edges. The large size of the area burnt at Hakea and the fact that much of the sur- rounding forest had been burnt in the previous 12 months may have acted to disperse the potential grazing pressure. The 2 species of Kennedia were ubiquitous throughout the study area but the Acacias were not present with equal fre- quency in all of the mapped vegetation complexes, and several were restricted to a few locations. In some cases this reflects a consistent link between species occurrence and site character- istics. Several of the legumes encountered at Hakea were used as specific indicator species in the site classification scheme de- veloped by Havel (1975); these include A. alata (Type C), A. extensa (Type W), A. urophyUa (Types S, T, Q). A. drummondii is characteristic of Type O but was not selected as an indicator species. Each of the vegetation complexes mapped by Heddle et al (1980) comprise several of Havel’s site vegetation types and so the distribution of individual legume species would be linked to the occurrence of specific types within the complex. The limited presence of A. urophylla in the Yarragil (max. swamp development) complex would not be expected on the basis of known site vegetation occurrence of this species, but may simply reflect the limitations in the method used to combine the maps of species occurrence with those of the vegetation complexes. The 2 most common Acacias ( drummondii . pulchella), were recorded in all vegetation complexes but with unequal relative frequencies in each complex; further investi- gation of species/site relationships would be required to ad- equately explain this pattern of distribution. Acknowledgements I acknowledge the contribution of technical staff at the Dwellingup Research Centre of CALM who were involved in the establishment and monitoring of this study. Cadets from the Dwellingup Training School participated in the broadscale legume surveys. Ian Abbott. Grant Wardell-Johnson, Neil Burrows, Per Christensen and George Peet commented on an earlier draft, and Linda Simmonds typed the manuscript. References Aston A R & Gill A M 1976 Coupled soil moisture, heat and water vapour transfers under simulated fire conditions. Aust J Bot 14: 55-66. Bell D T & Koch J M 1980 Post fire succession in the nothern jarrah forest of Western Australia. Aust J Ecol 5: 9-14. Burrows N D 1984 Describing forest fires in Western Australia. W Aust Forests Dept Tech Pap 12. Burrows N D 1987 The soil dryness index for use in fire control in the south-west of Western Australia. Department of Conservation and Land Management WA Tech Rep 17. Cheney N P 1981 Fire behaviour. In: Fire and the Australian Biota (Gill A M, Groves R H & Noble I R). Aust Acad Sci, Canberra. Christensen P E S 1980 The biology of Bettongia penicillata Gray, 1837 and Macropus eugenii Desmarest, 1804 in relation to fire. W Aust Forests Dept Bull 91. Christensen P E & Kimber P C 1975 Effects of prescribed burning on the flora and fauna of south-west Australian forests. Proc Ecol Soc Aust 9: 85-106. Christensen P, Recher H & Hoare J 1981 Responses of open forest to fire regimes. In: Fire and the Australian Biota (Gill A M, Groves R H & Noble 1 R). Aust Acad Sci. Canberra. Churchward H M & McArthur W M 1980 Landform and soils of the Darling System. Western Australia. In: Atlas of Natural Resources. Darling System, Western Australia. Department of Conservation and Environment, Perth. Cochrane G R 1968 Fire ecology in south-eastern Australian sclerophyll forests. Proc Tall Timbers Fire Ecol Conf 8: 15-40. Floyd A G 1976 Effect of burning on regeneration from seeds in wet sclerophyll for- est. Aust For 39: 210-20. Gill A M 1981 Adaptive responses of Australian vascular plant species to fires. In: Fire and the Australian Biota (Gill A M, Groves R H & Noble I R) Aust Acad Sci, Canberra. Ffavel J J 1975 Site vegetation mapping in the northern jarrah forest (Darling Range). 1. Definition of site vegetation types. W Aust Forests Dept Bull 87. Heddle E M, Loneragan O W & Havel J J 1980 Vegetation complexes of the Darling System in Western Australia. In: Atlas of Natural Resources, Darling Systems. Western Australia. Department of Conservation and Environment, Perth. Knight I K 1981 A simple calorimeter for measuring the intensity of rural fires. Aust For Res 11: 173-7. Levy E B & Madden E A 1933 The point quadrat method of pasture analysis. NZ J Agric 46: 267-79. Monk D, Pate J S & Loneragan W A 1981 Biology of Acacia pulchella R. Br. with special reference to symbiotic nitrogen fixation. Aust J Bot 29: 579-92. Mount A B 1972 The derivation and testing of a soil dryness index using run-off data. Tasmanian For Comm Bull 4. Peet G B 1971 A study of scrub fuels in the jarrah forest of Western Australia. W Aust Forests Dept Res Pap 8. Peet G B & Van Didden G W 1973 Fire effects on understorey shrubs. W Aust Forests Dept Res Pap 8. Shea S R. McCormick J & Portlock C C 1979 The effect of fires on regeneration of leguminous species in the northern jarrah ( Eucalyptus marginata Sm.) forest of Western Australia. Aust J Ecol 4: 195-205. Sneeuwjagt R J & Peet G B 1985 Forest fire behaviour tables for Western Australia. Department of Conservation and Land Management, Perth. Zar J H 1974 Biostatistical analysis. Prentice Hall. New Jersey. 6 Journal of the Royal Society of Western Australia 71 (1), 7-13, 1988 Calc-alkaline lamprophyres from the Pilbara Block, Western Australia Nicholas M S Rock & Mark E Barley Department of Geology, University of Western Australia, Nedlands, WA 6009 Manuscript Received October 1987. accepted January 1988 Abstract A suite of calc-alkaline lamprophyre (dominantly spessartite) dykes and plugs forms a belt extending for some 250 km from Balfour Downs to north of Bamboo Creek in the eastern Pilbara. The lamprophyres show character- istic panidiomorphic texture and chemistry (eg. high F, Ba & K), and many are satellite to granitoid piutons. Avail- able data suggest that these lamprophyres, many of which were formerly recorded as “hornblende porphyries”, constitute basic members of a Proterozoic (cl 700-1800 Ma) calc-alkaline, lamprophyre-porphyry-granitoid in- trusive suite which mirrors younger suites worldwide. Lamprophyres are also known from the Shaw Batholith in the eastern Pilbara, and from near Roebourne in the western Pilbara. Many other minor intrusions recorded as hornblende porphyries”, “trachyandesites”, “diorites”, “mafic porphyries” and “diabases” may also be lamprophyres. Lamprophyres form a significant phase of Pilbara magmatism, and thus are important to syntheses of tectonism, magmatism and mineralization. Introduction Lamprophyres (notably the lamproite subgroup) are now known to host or be spatially associated with deposits of dia- mond (at Argyle, in the Kimberley Region of Western Australia: Jaques et al 1986) and of gold (at Wood’s Point, Victoria: Hills 1952). They can also be associated with carbonatites, which host deposits of phosphate and rare metals [eg. the Bow Hill lamprophyres of the east Kimberleys are contemporaneous with the Cummins Range carbonatite: Jaques et al 1985). Ac- tive exploration for lamprophyres is consequently being under- taken in Western Australia, and their very widespread distri- bution in this state (both in time and space) is becoming established: dykes, plugs and diatremes (ranging from Ar- chaean to Miocene in age) have already been discovered or documented in many areas (Jacques et al 1985; 1986- Rock et al 1987). This paper aims to reappraise: (1) all previously reported occurrences of “lamprophyres” in the Pilbara Block and (2) a few rocks formerly described under other names, which we also believe to be lamprophyres. It complements an extensive study of lamprophyres in the Yilgarn Block (Rock et al 1988). Petrological nomenclature follows Streckeisen (1979) and Rock (1984, 1987, 1988) throughout. As described in more detail by Rock et al (1987), the pub- lished literature, together with catalogues and computerized in- dexes to the rock collections of the Department of Geology of the University of Western Australia (UWA), the Geological Sur- vey of Western Australia (GSWA), CSIRO Division of Minerals and Geochemistry (Floreat Park), and several mining com- panies, have been searched for specimens described either as “lamprophyres”, or under names indicating possible lamprophyric affinities (eg. ‘mafic porphyries’ or ‘hornblende porphyries ). These searches revealed only two previously re- A60438-3 corded occurrences of “lamprophyres” from the Pilbara. Both of these were summarized by Miles (1945) in the first compi- lation of lamprophyres throughout Western Australia. Jaques et al (1985, 1986) added no information to Miles’ account in their otherwise comprehensive reviews of Australian alkaline rocks. Criteria for reappraising previously described rocks Streckeisen’s (1979) recommended list of lamprophyre field and petrographical characteristics can be expanded, into the fol- lowing set of criteria, which also consider rock chemistry (Rock 1987). They are listed in approximately decreasing order of sig- nificance, although it is the combination of as many as possible of these features which is diagnostic of lamprophyres: 1) Very high whole-rock K, Ba, Sr, Rb, Th, light rare earth el- ements (REE), P, F, C0 2 , S and S0 3 (1-3 orders of magnitude higher than in basaltic rocks), combined with moderate enrichments in Ti, Zr, Nb, basaltic levels of V, Cr, Co, and Ni, but near to below-basaltic levels of Y and heavy REE. 2 ) Abundance of primary biotite, amphibole, carbonates, zeolites, chlorite, epidote, fluorite, sulphates, etc; 3 ) Certain unusual chemical features of the constituent min- erals, including exceptionally high Ti contents in biotite, amphibole and clinopyroxene, and exceptionally high Ba in bio- tite or K-feldspar; 4 ) Textural features, such as panidiomorphism, battlemented biotites, lack of felsic phenocrysts, or the presence of leucocratic globular structures (ocelli); 5 ) Occurrence as commonly xeonolithic minor intrusions (dykes, sills, pipes, diatremes) associated with breccias, tuffs! pyroclastics, etc.; complex to bizarre intrusive forms are common. 7 Journal of the Royal Society of Western Australia 71 (1), 1988 The usefulness of characteristic (4) may be reduced in the Pilbara, because most Precambrian rocks have suffered low- grade metamorphism, which readily destroys textural charac- teristics and makes true meta-lamprophyres petrographically similar to metamorphosed intermediate and basic rocks. Relict lamprophyric textural idiosyncracies, such as “battlemented” biotites, consistently euhedral pseudormorphs after biotite, amphibole or pyroxene, and apatite phenocrysts, are neverthe- less observable in the Pilbara rocks described as lamprophyres below. In assessing whole-rock chemical data, we recognise that metasomatism (including K-metasomatism) is widespread in the Pilbara Block (eg. Barley 1984, et al 1984), and could have in- duced some of the lamprophyre characteristics in originally non- lamprophyric rocks, or destroyed the original character of true lamprophyres. We have therefore relied less on high K contents than the combination of high Ba, Rb, Sr, La and Ce, with high V, Cr, Co and Ni, while fully taking into account the petrogra- phy. Fortunately, sufficient data are also available for associ- ated igneous rock-types (both Archaean and Proterozoic) to allow more objective, comparative assessments. The chemistry of the rocks here claimed as lamprophyres is so distinctly differ- ent from all these associated types that their identity is unques- tionable. Figure 3 confirms the chemical similarity of the Pilbara lamprophyres to global averages. A reappraisal of Pilbara lamprophyre occurrences Old Fortune Copper Mine, Roebourne, Western Pilbara Miles (1945: 4-5) recorded a “typical mica lamprophyre (ker- santite)” with a “rude schistosity”, occurring as a dyke in quartz-gabbro. The relevant GSWA sample (12579; section no. 95103) fully confirms Miles’ identification, carrying chlorite pseudomorphs after abundant, ‘battlemented’ biotite phenocrysts highly characteristic of mica-lamprophyres (Rock 1984). Unfortunately, this particular rock is too chloritized, silicified and carbonated to be worth analyzing, and we know of no other available material from this locality to follow the identi- fication further. Shaw Batholith A suite of ultramafic lamprophyre dykes was discovered dur- ing a Department of Geology (UWA) research program on this, one of the major Archaean granitoid batholiths of the Pilbara (Fig. 1; Bettenay et al 1981). These dykes, composed predominantly of sodic titanaugite and alkali to sodic-calcic amphiboles (arfvedsonite, kataphorite, etc.) are detailed in a companion paper (Bettenay et al 1988). Eastern Pilbara Miles’ (1945) only other reported lamprophyre, from Bam- boo Creek, is a rock with “scattered clear feldspar phenocrysts up to 2 mm diameter”, which is “not typical of the family” but is “in all probability an altered form of original mica lamprophyre — near kersantite”. Re-examination of the thin section (GSWA 12545/ s 1868) indicates that this rock is a metamorphosed dolerite (epidiorite), and shows no lamprophyric characteristics. However, some rocks occurring in the same general area, previously described as ‘hornblende porphyrites’ (Maitland 1905, Finnucane 1935, Noldardt & Wyatt 1962), and as ‘hornblende porphyries’ or ‘trachyandesites’ (Hickman 1978, Barley 1980, Lewis & Davy 1981, Hickman et al 1983), are un- doubtedly lamprophyres. Indeed, lamprophyres have now been annotated as such on the Geological Survey of Western Australia’s recently published Balfour Downs 1:250 000 sheet (Williams 1987). They form part of a suite of dykes and plugs, which extends for some 250 km from Balfour Downs to north of Bamboo Creek (Fig. 1). A suite of lamprophyres, intermediate porphyries and granitoids from near Bamboo Creek has now been dated at cl800 Ma by a Pb-Pb whole-rock isochron (Barley et al in prep.). The most abundant lamprophyres, mapped as “hornblende porphyries” on the GSWA’s earlier Nullagine and Yarrie 1:250 000 sheets (Hickman 1978, Hickman et al 1983) contain 15 to 30 per cent of panidiomorphic, dark green-brown hornblende and, more rarely, phlogopitic biotite phenocrysts, in a fine-grained, variably sericitized and carbonated, feldspathic groundmass carrying small apatite and magnetite euhedra. The hornblende may carry clinopyroxene cores, and is rhythmically zoned, chloritized or carbonated (Fig. 2). New data for three lamprophyre plugs — two satellite to the Bridget Adamellite (Hickman 1978), and one part of a composite spessartite-quartz monzonite plug near Bamboo Creek — are compared in Table 1 with analyses of the Bridget Adamellite (from Barley 1980). Other rocks in the eastern Pilbara that have broadly similar chemical composition are Archaean intermediate volcanics (Hickman 1983, Barley et al 1984). Although the Si0 2 , MgO, Ni, and V contents of these rocks overlap those of the lamprophyres, the lamprophyres have consistently higher Ba (>800 ppm), Rb (>80ppm) and Sr (>650 ppm) than a suite of relatively unaltered Archaean intermediate volcanics (Barley 1980) with Ba <650 ppm, Rb <50 ppm and Sr <600 ppm. Lewis and Davy (1981) employed the term ‘trachyandesite’ as a chemical description of hornblende-phyric dykes and plugs intruding the Mt Edgar Batholith, whilst noting that these rocks had higher K?0 than Le Maitre’s (1976) average trachyandesite. The term is now inappropriate in terms of the revised IUGS chemical classification of igneous rocks (Le Bas et al 1986), as it implies alkaline affinities for calc-alkaline rocks, and does not recognize their textural, minor or trace element characteristics. The term ‘porphyry’ is equally inadequate. In- stead, the exotic chemistry of these rocks (eg. reported Ba con- tents of 800-2000 ppm, and F contents of 800-1100 ppm, coupled with moderate Rb, Ce, etc.), their lack of feldspar phenocrysts, and their panidiomorphic mafic phenocrysts, indi- cate that they should be termed spessartites (calc-alkaline lamprophyres dominated by hornblende-plagioclase). The spessartites commonly appear as plugs and dykes satel- lite to intermediate hornblende-plagioclase porphyries and hornblende-bearing monzonites to quartz monzonites, such as the Bridget Adamellite (Fig. 1, Hickman 1978, Barley 1980). Similar porphyry plugs and small granitoid intrusions are associ- ated with lamprophyres in the Mt Edgar Batholith (Mt Edgar it- self is a small, hornblende-plagioclase porphyry to quartz- monzonite plug) and near Bamboo Creek (Hickman 1978). Petrography and chemical analyses of the Bridget Adamellite (Table 1, Barley 1980) and hornblende-plagioclase porphyries from the Mt Edgar Batholith (Lewis & Davy 1981) indicate that although containing abundant modal quartz (up to 20 per cent by volume; visual estimate) most of these granitoids should be called monzonites or quartz monzonites rather than adamellites (using the terminology of Streckeisen 1976). 8 20 ° 119 ° + + + + V V V V V V V V 120 ° 121 ° Phanerozoic sedimentary rocks Proterozoic volcanic and sedimentary rocks Archaean granitoids Archaean greenstone terrain O Lamprophyre sample location Figure 1 Locality sketch-map of the eastern Pilbara Block, showing general locations of known lamprophyres. Information largely taken from Geoloqical Survey of WA 1:250 000 Series maps. Figure 2 Photomicrograph of spessartite lamprophyre forming a plug satellite to the Bridget Adamellite (sample 86393). Field of view 8 mm; crossed polars. Note twinned, panidiomorphic hornblendes (Hb) one with a core of clinopyroxene (Cpx). Sr K Fb Ba TLi l\b G& P Zr Hf Sn Ti Y Yb Sc Cr N Pearce plot (Rock/MORB) Figure 3 MORB-normalized multi element plots ("spidergrams") of eastern Pilbara suite lamprophyres compared with average global calc-alkaline lamprophyres (pattern labelled “global CAL") Data from Table 1 and Rock (1987). Elements are arranged such that their incompatibilities and immobilities increase towards the centre of the diagram (Pearce 1983). Journal of the Royal Society of Western Australia 71 (1). 1988 New ana Table 1 lyses of confirmed lamprophyres and associated rocks from the Pilbara Lamprophyres Bridget adamellite Bamboo Area of Bridget adamellite BC106.3 UWA 86392 UWA 86393* UWA 86389 UWA 86390 UWA 86391 Si0 2 58.79 58.89 58.44 63.54 63.82 64.76 ai 2 o 3 15.24 14.02 14.08 14.67 14.53 14.84 Fe 2 °3 8.16 3.39 3.44 3.67 2.96 2.46 FeO NA 4.96 4.72 2.54 3.08 2.78 MgO 3.36 4.11 4.41 2.35 2.38 2.33 CaO 5.66 5.97 6.11 4.19 4.22 4.31 Na 2 0 3.42 3.91 3.02 3.51 3.49 3.51 k 2 o 3.47 3.70 3.44 4.15 4.11 3.43 h 2 o+ 0.86 0.90 0.73 0.38 0.51 0.81 Ti0 2 0.70 0.50 0.53 0.45 0.44 0.38 P 2 0 5 0.29 0.30 0.36 0.23 0.24 0.24 MnO 0.15 0.19 0.12 0.13 0.11 0.08 C0 9 NA 0.05 0.33 0.19 0.30 0.23 Total 100.10 100.89 99.73 100.00 100.19 100.16 Trace elements (ppm), in order of atomic number V 152 197 169 NA NA NA Ni 21 42 44 NA NA NA Cu 8 37 100 26 45 28 Zn 68 84 97 74 74 58 Rb 140 90 104 132 129 123 Sr 675 715 690 670 670 623 Y 11 16 19 20 16 15 Zr 86 102 103 123 125 121 Nb 6 5 8 NA NA NA Sn 2 NA 1 NA NA NA Ba 840 1014 970 900 970 670 La 11 11 22 NA NA NA Ce 76 86 92 NA NA NA Pb 30 92 85 NA NA NA Th 21 15 18 NA NA NA CIPW weight % norms (analyses recalculated to 100% free of H 2 0 and C0 2 qz 8.64 4.38 9.17 14.86 15.57 18.68 ab 29.33 33.11 25.83 29.87 29.66 29.92 or 20.78 21.88 20.55 24.66 24.39 20.42 an 16.20 9.78 14.86 12.08 11.89 14.71 di 8.49. 14.41 9.22 5.03 4.66 3.10 hy 10.94 10.81 13.92 8.83 8.96 8.92 mt 3.60 3.88 3.82 2.84 2.79 2.43 il 1.35 0.95 1.02 0.86 0.84 0.73 ap 0.68 0.70 0.84 0.54 0.56 0.56 cc 0.00 0.11 0.76 0.43 0.69 0.53 Total 100.00 100.01 100.00 100.00 100.00 100.00 DI+ 58.75 59.37 55.56 69.39 69.62 69.02 tThornton-Tuttle differentiation index NA = not analysed *see photomicrograph in Fig. 2. Data from Barley (1980), supplemented with new trace element determinations using conventional XRF techniques (courtesy of Dr R Chang, UWA). Journal of the Royal Society of Western Australia 71 (1). 1988 0 2 4 6 %k 2 o 8 Figure 4 Variation diagrams illustrating coherent trends for geochemically homologous elements in the inferred Proterozoic lamprophyre-porphyry minor intrusive suite of the eastern Pilbara. Data from Table 1, Lewis & Davy (1981) and Barley (1980 unpubl; details available on request). Lamprophyre dykes and plugs are from near Bamboo Creek and the Bridget Adamellite, ‘andesite dykes’ are hornblende-plagioclase porphyry dykes from the Mt Edgar Batholith, and porphyries are silicic porphyries from the Mt Edgar Batholith. Limited available data, illustrated in Figure 4, suggest that the spessaritites, hornblende-plagioclase porphyries from the Mt Edgar Batholith (‘andesite dykes’ of Lewis & Davy 1981) and the Bridget Adamellite form a fairly coherent geochemical suite. Note, however, that whilst the hornblende-plagioclase porphyries overlap the granitoids in composition, the spessartites are more basic than any of the known plutonic rocks. Silicic porphyries from the Mt Edgar Batholith are also plotted on Figure 4. Some of these are spatially related to lamprophyre and quartz monzonite intrusions, plot on exten- sions of the lamprophyre-quartz monzonite trend, and may also be genetically related. Williams (1987, in press) records plugs of “microcline-biotite trachyte or lamprophyre” from the vicinity of Balfour Downs (Fig. 1), thus extending the belt of presumed Proterozoic lamprophyric magmatism in the eastern Pilbara to at least 250 km. These rocks intrude the Fortescue and Hamersley Groups, and range in composition from trachyte, with phenocrysts of albite and microcline set in a fine-grained feld- spathic groundmass (also coarser grained albite-microcline granite), to minette (calc-alkaline lamprophyre dominated by biotite-orthoclase), containing biotite phenocrysts in a variably altered feldspathic groundmass with small apatite and magnet- ite euhedra. The relationship between the minettes and the spessartites further north is as yet uncertain, although such rocks commonly coexist in regional dyke-swarms, where they may even be heteromorphic (Rock 1984, Rock et al 1986b). Conclusions The eastern Pilbara spessartites appear to represent mafic end- members of a previously unrecognized lamprophyre-porphyry- minor intrusive suite, typical of examples accompanying other calc-alkaline granitoids of many ages on all six continents (Rock 1984, 1987). Unfortunately, detailed data as yet are few, and some of the available trace element analyses are only semi- quantitative (Lewis & Davy 1981). Features which nevertheless ally these minor intrusions with better-substantiated lamprophyre-porphyry associations elsewhere include the following: (a) the chemical gradation in Figure 4 {cf Barnes et al 1986, Rock et al 1986b); (b) the occurrence of lamprophyres as plugs satellite to granitoids, notably plugs adjacent to the Bridget Adamellite and similar granitoids near Bamboo Creek {cf Rock et al 1986a, b); (c) the occurrence of plutonic chemical equivalents to the inter- mediate and felsic but not to the lamprophyric dykes {cf Rock et al 1988a). More detailed work is therefore in progress on this minor in- trusive suite, to determine the full range of lamprophyres and associated granitoids, their petrogenesis, and their tectonic significance. 12 Journal of the Royal Society of Western Australia 71 (1), 1988 Acknowledgements. We thank the Geological Survey of Western Australia (es- pecially Ian Williams. John Lewis and Will Libby) for access to records, thin sections and rock specimens. Rod Marston for pointing out the lamprophyric affinities of satel- lite plugs to the Bridget Adamellite during supervision of MEB's Ph D project, Tim Blake for providing the speciment of lamprophyre from near Bamboo Creek, and Ray Chang (UWA) for the analyses. David Groves is also thanked for constructive criti- cism of this manuscript. References Barley M E 1980 Evolution of Archaean calc-alkaline volcanics. PhD Thesis, Univ W Aust. Barley M E 1984 Volcanism and hydrothermal alteration of the Warrawoona Group, East Pilbara. Univ W Aust Geology Dept & Extension Publ 9: 23-36. Barley M E. Borley G D. Sylvester G C, Groves D 1 & Rogers N 1984 Archaean calc- alkaline volcanism in the Pilbara Block. Precambrian Res 24: 285-319. Barley M E, Blake T S & McNaughton N J (in prep) A Proterozoic calc-alkaline lamprophyre-porphyry-granitoid association from the eastern Pilbara Block. Barnes R P. Rock N M S & Gaskarth. J W 1986 Late Caledonian dyke-swarms in southern Scotland: new field, petrological and geochemical data from the Wigtown Peninsula, Galloway. Geol J 21: 101-125. Bettenay L F. Bickle M J. Boulter C A, Groves D I. Morant P, Blake T S & James B A 1981 Evolution of the Shaw Batholith-an Archaean granitoid-gneiss dome in the eastern Pilbara, Western Australia. Geol Soc Aust Spec Publ 7:361-372. Bettenay L F. Rock N M S & Mather P J 1988 Archaean-Proterozoic ultramafic lamprophyres from the Shaw Batholith, Pilbara Block, Western Australia. Precambrian Res (submitted). Finucane K J 1936 The Nullagine conglomerates. Areal Geology and Geophysical Survey of Northern Australia Report 4. Hickman A H 1978 Nullagine, Western Australia, Sheet SF/51-5. Geol Surv W Aust 1:2 500 000 Series Explan Notes. Hickman A H 1983 Geology of the Pilbara Block and its environs. Bull Geol Surv W Aust 127. Hickman A H, Chin R J & Gibson D L 1983 Yarrie, Western Australia, Sheet SF/51- 1. Geol Surv W Aust 1:250 000 Series Explan Notes. Hills E S 1952 The Wood’s Point dyke swarm, Victoria. Sir Douglas Mawson Anniver- sary Volume, University of Adelaide: 87-100. Jaques. A L, Creaser R A, Ferguson J & Smith C B 1985 A review of the alkaline rocks of Australia. Trans Geol Soc S Afr 88:311-335. Jaques A L, Lewis J D & Smith C B 1986 The kimberlites and lamproites of Western Australia. Bull Geol Surv W Aust 132. Le Bas M J, Maitre R W, Streckeisen A & Zanettin B 1986 A chemical classification of volcanic rocks based on the total alkali-silica diagram. J Petrol 27: 745-750. Le Maitre R W 1976 The chemical variability of some common igneous rocks. J Pet- rol 17: 589-637. Lewis J D & Davy R 1981 Felsic dykes of the Mt Edgar Batholith. Geol Surv W Aust Ann Rep 1980: 103-108. Maitland A G 1905 Further report on the geological features and mineral resources of the Pilbara Goldfield. Bull Geol Surv West Aust 20. Miles K R 1945 Some Western Australian lamprophyres. J R Soc W Aust 31: 1-15. Noldart A J and Wyatt J D 1962 The geology of portion of the Pilbara Goldfield covering the Marble Bar and Nullagine 4-mile map sheets. Bull Geol Surv W Aust 115. Pearce J A 1983 Role of the sub-continental lithosphere in magma genesis at active continental margins; In: Continental Basalts and Mantle Xenoliths (ed C J Hawkesworth & M J Norry). Shiva, Orpington. Rock N M S 1984 Nature and origin of calc-alkaline lamprophyres: minettes, vogesites, kersantites and spessartites. Trans R Soc Edinburgh: Earth Sciences 74: 193-227. Rock N M S 1987 The nature and origin of lamprophyres: an overview. In: Alkaline Igneous Rocks (ed J G Fitton & B G J Upton) Spec Publ Geol Soc London 30: 191-226. Rock N M S 1988 Kimberlites as varieties of lamprophyres. Proc 4th Int, Kimberlite Conf, Spec Publ Geol Soc Aust (in press). Rock N M S, Cooper C & Gaskarth J W 1986a Late Caledonian subvolcanic vents and associated dykes in the Kirkcudbright area, Galloway, SW Scotland. Proc Yorks Geol Soc 46: 29-38. Rock N M S, Gaskarth J W & Rundle C C 1986b Late Caledonian dyke-swarms in southern Scotland: a regional zone of primitive K-rich lamprophyres and associ- ated vents. J Geol 94: 505-522. Rock N M S, Duller P, Haszeldine R S & Groves, D I L987 Lamprophyres as gold ex- ploration targets: some preliminary observations and speculations. Univ W Aust Geology Dept & Extension Publ 11: 271-286. Rock N M S, Gaskarth J W, Henney P J & Shand P 1988a Late Caledonian dyke- swarms of northern Britain: some tectonic and genetic implications of their distri- bution and chemistry. Can Mineral 26:3-22. Rock N M S, Halberg J A, & Mather P J 1988b Lamprophyres in the gold fields of the Yilgarn Block, Western Australia. Univ W Aust Geology Dept & Extension Publ 12 (245-275). Streckeisen A 1976 To each plutonic rock its proper name. Earth Sci Rev; 12; 1-33. Streckeisen A 1979 Recommendations of the IUGS Subcommission on the systematics of igneous rocks: classification and nomenclature of volcanic rocks, lamprophyres, carbonatites and melilitic rocks. Geology 7: 331-335. Williams I R 1987 Balfour Downs. Western Australia, Sheet SF/51-9. Geol Surv W Aust 1:250 000 Series (map only). Williams 1 R (in press) Balfour Downs, Western Australia, Sheet SF/51-9. Geol Surv W Aust 1:250 000 Series Explan Notes. 13 Journal of the Royal Society of Western Australia 71 (1), 15-22, 1988 Widespread regeneration failure of Persoonia elliptica (Proteaceae) in the northern Jarrah forest of Western Australia Ian Abbott & Paul Van Heurck Research Centre, Department of Conservation and Land Management, Hayman Road, Como WA 6152 Manuscript received August 1987: accepted February 1988 Abstract Persoonia elliptica R.Br. is a widely distributed but scattered understorey tree of the northern jarrah forest of Western Australia. Stem diameters (at breast height) of 10-30 cm are common but most populations are deficient in trees <10 cm in diameter. The reason for this was studied. Fruit production is annual and abundant. Although seed viability is high, none could be induced to germinate in laboratory or forest even after physical and chemical treatment. In the forest, the passage of low intensity fire over leaf litter causes abundant germination of dispersed seeds, but few seedlings survive because of browsing by vertebrates, assumed to be kangaroos and wallabies. Height growth of seedlings and lignotuberous seedlings, and diameter growth of trees is slow: we estimate that diameter (breast height) of 10 cm is attained, on average, in c90 years and that recruitment of P. elliptica over much of the northern jarrah forest ceased between 1870 and 1904 (95% confidence limits). A speculative model relating known changes in fire frequency and plausible changes in the abundance of kanga- roos and wallabies since permanent European occupation of south western Australia in 1829 is proposed to ac- count for the present stand structure of P. elliptica. Introduction The failure of regeneration of trees in forests and woodlands has been frequently recognized. Regeneration failure occurs when seedlings fail to attain the size of saplings, giving rise to a population with discontinuous age classes (Jones 1945). Poss- ible causes of regeneration failure include low seed set, excess- ive predation of seeds before and after dispersal, low seed vi- ability, poor germination, and high mortality of seedlings. However, a major cause is the palatability of seedlings to ani- mals, eg. insects and molluscs (Watt 1923), rabbits (Watt 1919, 1923; Lange & Graham 1983), rodents (Watt 1923, Wardle 1959, Pigott 1969), sheep (Crisp & Lange 1976, Lange & Willcocks 1980, Pigott 1983), goats (Coblentz 1978, Clark & Clark 1981), and kangaroos (Hall, Specht & Eardley 1964). Other factors including temperature, rainfall, or light, have sometimes been implicated (Wardle 1959, 1963a, b, 1978; Hall et al 1964; Pigott 1969). Widespread or chronic regeneration failure has not been re- ported from natural eucalypt forests and woodlands of Aus- tralia, although in some years there may be extensive removal of seeds and fruits by ants and mammals or browsing by mam- mals (Cunningham 1960, Gilbert 1961, Needham 1960, Statham 1983). Despite this, natural regeneration is usually prompt and adequate. To ensure impartiality, selection of referees and consideration of their reports was undertaken by Dr N G Marchant at the request of the Hon Editor. Persoonia elliptica R.Br is an understorey tree of the northern jarrah ( Eucalyptus marginata Donn ex Smith) forest, attaining maximum height of cl m and maximum diameter at breast height (130 cm, DBH) of c36 cm. It is widely distributed there (Churchill 1959). For convenience, we recognize four stages in the life of P. elliptica: seedlings (plants up to one year old), lignotuberous seedlings (plants >1 yr old but <15 cm tall), sap- lings (plants 15-130 cm tall) and trees (plants >130 cm tall). Our observations since 1979 throughout the northern jarrah for- est have shown that saplings and trees with DBH<10 cm are seldom present although seedlings and lignotuberous seedlings occur in most populations. In this paper we document the present population structure (frequency of differently sized individuals) of P. elliptica and examine factors relevant to explaining the observed failure of regeneration. These include supply and viability of seed, suc- cess of germination and establishment of seedlings. Measure- ments of growth of seedlings and trees are used to estimate when the regeneration failure began. Finally, we speculate about ecological changes that may have been responsible for the regeneration failure. Population Structure In this Section, we document the diameter structure of nine populations of P. elliptica and provide brief notes about the oc- currence of seedlings and lignotuberous seedlings. 15 Journal of the Royal Society of Western Australia 71 (1), 1988 Methods Stands of jarrah forest were chosen mainly on the basis of contrasting rainfall zone and fire and logging history (Table 1). Most were of 15 ha, with one 3 ha (No. 9) and another 125 ha (No. 6). Six of the stands were typical of the northern jarrah for- est in that they obviously lacked small diameter trees of P. elliptica. Three stands (Nos. 7-9) were chosen because of atypically high representation of saplings and small trees of P. elliptica. These three stands were the only ones found by us that had this unusual representation of small trees and saplings. A complete enumeration of diameter was made in popu- lations where there were fewer than 30 trees; in larger popu- lations only the first 30-50 trees were measured. Diameters were measured over bark at 130 cm above ground level (DBH). The diameter of trees which were forked below, or damaged at breast height, was measured 50 cm or 100 cm above ground level and the DBH calculated from a regression equation relat- ing stem diameter to its height of measurement above ground level. The diameter of the largest stem was measured on trees with several stems growing from the same lignotuber. Saplings were specially searched for and the occurrence of seedlings and lignotuberous seedlings was noted. Results In the nine populations studied (Fig. 1), the modal DBH classes were 8-10 cm, 12-14 cm, 16-18 cm, 18-20 cm and 24- 26 cm (each once), 10-12 cm (twice) and 14-16 cm (thrice). Populations 1-6 had a structure typical of most of the northern jarrah forest, with a deficiency of DBH classes 0-6 cm and lim- ited representation of DBH classes 6-8 and 8-10 cm. Very few trees in any of the populations had DBH values exceeding 30 cm. In populations for which complete enumerations of trees were made, density was variable: 0.2 individuals ha^ 1 (No. 6), c3-4 ha 1 (Nos. 1, 2, 4) and c20 ha 1 (No. 9). The spatial pattern of trees in No. 2 is random, with mean distance between trees of 16 m (Abbott 1984a). Seedlings and lignotuberous seedlings were found in all popu- lations except Nos. 6 and 8 but seemed to be most abundant in populations 1, 2, 5 and 9, where they were found readily. There was no obvious association between the presence of seedlings and fire and logging history of the forest stands (Table 1, Fig. 1). It is important to record that seedlings and lignotuberous seedlings of P. elliptica are very rare in comparison to those of other tree species in the jarrah forest. Saplings were found in populations 7 and 9. Annual Production and Dispersal of Fruits In this Section we quantify the phenology of fruit production and consider fruit fall and seed dispersal. Methods The presence of fruit on the tree or on the ground beneath the canopy was recorded when trees were measured (in Spring) for DBH (as in the previous section). Detailed observations of flowering times were made in population 2 but casual obser- vations were also made elsewhere. In population 2, five easily accessible branches on each of five trees were marked and the numbers of fruits present counted regularly over two years. In this stand and in population 5, the distance of seedlings from the edge of the canopy of the nearest reproductive tree was measured. Results and Discussion Flowering took place in December, and fruits (drupes) at- tained full size (cl 7 x 10 mm) by the following May. Many of the immature fruits aborted between March and May. Trees with DBH <11 cm were found not to have ever fruited. There were large annual differences in the numbers of fruits produced, which were greatest in July 1981 and least in July 1983 (Fig. 2). Most fruit had fallen to the ground by August of each year. All fruit fell beneath the canopy because of the large terminal velocity of the fruit (Abbott 1984a). Table 1 Characteristics of study areas. Locality Grid Rainfall Features reference* zone (mm) 1 Ashendon BN 71.2.9 900-1000 State forest. High quality forest. Cut-over 1914, 1930.** 2 Ashendon BN 71.2.8 900-1000 State Forest. High quality virgin forest. 3 Chandler BQ 69.8.2 1100-1200 State Forest. High quality forest. Cut-over 1890, unburnt since 1937. 4 Mundlimup BT 63.5.6 1200 1300 State Forest. High quality forest. Cut-over 1872, 1928. ** 5 Loc. 990 BT 61.9.8 1200-1300 Reserve. High quality virgin forest. 6 Yarragil DD 705.7 1100-1200 State Forest. High quality forest. Cut-over 1933. Near southern edge of range of P. elliptica in the jarrah forest. Unburnt since 1973. 7 Sawyers AP 72.2.3 900-1000 State Forest. High quality forest. Cut-over 1870s. ** 8 Loc. 6203 AM 81.3.9 600-700 Reserve. Low quality virgin forest. 9 Mt Helena AN 70.6.9 1000-1100 Private property. Low quality forest, last cut-over cl950 and last burnt 1965. 'CALM 1:50 000 maps (publicly available). “Fire regime of periodic low intensity (- 300 kW m'b fires since the 1950s. 16 Percentage of trees in DBH classes Journal of the Royal Society of Western Australia 71 (1). 1988 Figure 1 Population structure of Persoonia elliptica. Abscissa shows presence of seedlings or lignotuberous seedlings as s, of saplings as S and DBH class from 0.1-2, 2.1-4 .... 34.1-36 cm. 17 Journal of the Royal Society of Western Australia 71 (1), 1988 Figure 2 Annual variation in production and shedding of fruit of Persoonia elliptica in population 2. One branch of one tree died between October 1982 and March 1983, Although seeds can be found at any time on the ground be- neath the canopy of P. elliptica, we have seldom observed seedlings there. For example, in population 2, the mean dis- tance of seedlings from the edge of the canopy of the nearest P. elliptica tree was 28.6 m (range 2-66 m, N = 40). In population 5, the mean distance was 5.5 m (N = 44). In this population the average distance between the bole and the edge of the canopy of the same tree was 2.4 m (N = 24). These data indicate that fruits or seeds are removed from the tree, or from the ground under the tree, or both, by some disper- sal agent. Vertebrates rather than invertebrates are responsible (Abbott & Van Heurck 1985). Common Brushtail Possums ( Trichosurus uulpecula) and Western Brush Wallabies ( Macropus irma) have been observed to take fruits (S. Davies, pers. comm.). We have also found seed in the faeces of Common Brushtail Possums, Western Grey Kangaroos (M. fuliginosus) and Emus ( Dromaius nouaehollandiae). Two common frugivor- ous birds, the Grey Currawong ( Strepera versicolor) and the Australian Raven (Corvus coronoides), may also eat drupes but we have no direct observations. The fruit is also palatable to hu- mans and may have been eaten by Aborigines. Viability and Germination of Seed Here we examine whether the deficiency of trees of P. elliptica with diameter <10 cm results from most seeds either being inviable or germinating poorly. Methods Viability of seeds with endosperm present was checked with tetrazolium chloride (Colbry et al 1961). The following treatments to germinate seeds (collected from beneath the canopies of many trees) were tried: (a) Seeds were kept at 18°C with full light or 12 hour light/12 hour dark in the laboratory. Part of the sample was then placed on the soil in cages amongst population 2 and kept under obser- vation for two years. (b) Seeds were soaked in distilled water and then placed in the forest as in (a). (c) Seeds were kept at 4°C for several weeks and then treated as in (a). (d) Seeds were filed at both ends to reveal the endosperm and then treated as in (a). (e) Seeds were boiled in water for 15 minutes and then treated as in (a). (f) Digestion by birds was simulated by scarifying seeds for 3 hours or treating with acid (as in Glyphis et al 1981); seeds were then placed in the forest as in (a). (g) Seeds were treated with gibberellic acid (method of Bachelard 1967) and then placed in the forest as in (a). (h) Seeds were placed between hessian sacks (method of Crossland 1981). (i) Seeds were placed in a small area (1.5 m x 1.5 m) of litter which was then ignited. Seeds were then placed in the forest as in (a). (j) Seeds from several prolifically fruiting trees in a paddock at Mt Helena were placed in an adjacent fenced enclosure contain- ing a remnant of native vegetation. (k) Seeds were placed in cages in the forest in various treat- ments of shading, litter type and depth, and trenching (Abbott 1984b). (l) Seeds from populations 5 and 9 were placed in cages on the forest floor in population 2. (m) Fresh drupes were offered to captive emus in order to assess whether passage through the gut facilitated germination. Results and Discussion Drupes collected from four populations yielded the following percentage viability: 81% (N = 52), 100% (N = 7), 73% (N = 51) and 48% (N = 23). Abbott (1984b) quoted 84% vi- ability for another sample. In all cases there was only one seed per drupe. A sample of 154 seeds collected from under a large P. elliptica tree yielded only 7.8% that were obviously damaged by fungi and insects. The presumption is that the remainder were fit for germination. None of the experimental treatments applied induced seed to germinate, in agreement with Kullman (1981). There is clearly a dormancy mechanism but its precise nature remains unresolved. Our observation (preceding section), that nearly all seedlings are found far from fruiting trees, implicates passage through the gut of vertebrates as the first step in breaking this dormancy. The emus refused to eat the drupes; it is difficult to introduce new types of food into the diet of captive emus (S. Davies, pers. comm.). 18 Journal of the Royal Society of Western Australia 71 (1), 1988 Establishment and Survival of Seedlings Opportunistic observations made in jarrah forest stands sub- ject to periodic low intensity fire (<300 kW m l ) in spring showed abundant germination of P. elliptica seedlings the follow- ing winter. Inspection of these stands more than one year after such fires showed that very few seedlings remained. For ex- ample, of 19 seedlings tagged in October 1982 after fire in Sep- tember 1981 near population 2, only five were alive by Febru- ary 1983. The other 14 plants had either disappeared or had only their stems remaining, indicating that they had not died be- cause of drought but had died after being browsed. The type of leaf loss was not ascribable to invertebrates but consistent with browsing by vertebrates, probably kangaroos and wallabies. If the seedlings had been killed by drought, their shrivelled re- mains should have been evident. We therefore designed an ex- periment to assess the effect of browsing by vertebrates on seedling survival. Methods In Chandler Block, 2 km from population 3, a forest stand burnt in spring 1982 was searched in September 1983 for P. elliptica seedlings and 49 were located and marked. Twenty were caged (12 mm mesh) and the rest were left as controls. Survival was assessed in November 1983 and March 1984. In August 1981, 14 one year-old seedlings in population 2 were tagged and their survival was checked in August 1982 and August 1983. In September 1981, 41 one year-old seedlings in population 5 were tagged and checked in August of 1982 and 1983. Results and Discussion The survival of 6 months-old (November) seedlings differed little between those caged (88.2%) and those uncaged (78.6%). However, after the first summer (March), at 9 months age, there was a significant difference in survival (85% caged, 21% control, P<0.05). Several of the uncaged seedlings showed clear evi- dence of having been killed by browsing. Because fresh faeces of kangaroos and wallabies (but not of emus) were present, we assume that the first two were responsible. Evidently during the first summer these seedlings are highly palatable to these vertebrates. After the first year, however, survival of uncaged seedlings is generally high. Of the 14 one year-old seedlings marked in population 2, 13 were still alive one year later and 12 were alive after a further year. Of the 40 one year-old seedlings marked in population 5, 37 were still alive one year later and 35 remained alive after a further year. Hence annual mortality of the seedlings after the first year of life averaged 6.9%. These data suggest that by the second summer the lignotuberous seedling is no longer attractive food for browsing vertebrates. The apparent contradiction between our observations of abundant germination after fire in the forest and our unsuccess- ful attempt to germinate seed after an experimental fire ((i), in the previous section) may be explained by assuming that the lat- ter seed had not passed through the gut of a vertebrate. In such a case the seed would still have been dormant. Fire alone is in- sufficient for germination. Rate of Growth Here we examine height growth of seedlings, lignotuberous seedlings and saplings, and diameter growth of trees. Methods Rate of growth of seedling, lignotuberous seedlings and sap- lings was determined from annual measurements of height above ground level. Tree growth was determined from annual measurements of DBH, rather than by counting of growth rings, as these were difficult to discern. Results and Discussion Nine months-old seedlings near population 3, caged to pre- vent browsing, attained a mean height of 5.2 cm (Table 2). Those seedlings 15 months-old (and caged) had achieved a mean height of 6.4 cm (Table 2). Caged plants showed an aver- age annual height increment of 0.2 (range -2 to 1) and 0.6 (range 0-2) cm yr- 1 in 1981-82 and 1982-83 respectively. Uncaged plants showed an average annual height increment of (-0.4 (range -6 to 4) cm yr- 1 (Table 2). Thus, as expected, browsing retarded the height growth of those seedlings and lignotuberous seedlings that survived browsing. Uncaged saplings showed an average annual increment in height of 5.9 cm (range 0-14 cm). Table 2 Growth in height of Persoonia elliptica. Population No. (as in Table 1) Age (years) Measured Height (cm) Mean Range N Seedlings and lignotuberous 2 seedlings 1 Sep 1981’ . 6.4 3-10 14 2 2 Aug 1982’ 6.6 4-11 13 2 3 Aug 1983* 7.2 4-11 12 Near 3 <1 Mar 1984’ 5.2 2-7 17 5 2 Aug 1982 6.9 3-10 47 5 3 Aug 1983 6.5 3-10 33 Saplings 9 7 Sep 1982 62.2 20-120 12 9 7 Sep 1983 68.1 22-130 12 caged to prevent browsing by vertebrates 19 Journal of the Royal Society of Western Australia 71 (1). 1988 Table 3 Growth in diameter at breast height of Persoonia elliptica. Population No. (as in Table 1) Period DBF! range Mean DBH increment (cm yr -1 ) Range 95% confidence interval N 1 Feb 1982-Feb 1983 11.7-25.1 0.12 0-0.4 0.03-0.21 10 1 May 1981 -May 1982 15.4-34.7 0.20 0-0.4 0.08-0.32 6 1 Apr 1982 May 1983 10.2-22.8 0.11 0-0.2 0.05-0.17 10 2 Feb 1981 -Feb 1982 11.6-25.1 0.03 0 - 0.1 0-0.06 9 4 May 1981-Apr 1982 7.3-21.3 0.11 0-0.4 0.08-0.14 41 4 Apr 1982-May 1983 7.5-31.7 0.10 0-0.4 0.06-0.14 38 9 Sep 1982-Sep 1983 2.5-23.5 0.14 0-0.6 0.06-0.22 17 Diameter increment of trees, averaged over all populations, was 0.11 cm yr- 1 , with 95% confidence limits of ±0.02 (Table 3). Values as high as 0.4 or 0.6 cm yr -1 were sometimes re- corded but of the 131 increments measured, 40.5% were zero. We calculate that the average P. elliptica tree with DBH of 8, 10, 12 or 20 cm should respectively be 73 years (95% confidence interval of 62-89 years), 91 (77-111) years, 109 (92-133) years or 182 (154-222) years old. That is, establishment took place in 1908 (95% confidence interval of 1892-1919), 1890 (1870- 1904). 1872 (1848-1889) or 1799 (1759-1827). Height and diameter growth is much slower than that of Banksia grandis Willd., which is also an understorey tree species of the northern jarrah forest (Abbott 1985). Effect of Fire on Survival and Growth Because P. elliptica develops a lignotuber (woody swelling at the base of the stem) in the second year of life, typical low inten- sity fires should rarely kill plants two or more years old. We ob- served that leaves and shoots are killed by these fires but new shoots resprout from the lignotuber within a few weeks. Mean height of 3-year old lignotuberous seedlings in popu- lation 5 in August 1983 was 6.5 cm (Table 2). A patchy low in- tensity prescribed fire in spring 1983 killed shoots, but by Febru- ary 1984 mean height was 4.0 cm (N = 36). A sample of lignotuberous seedlings in unburnt patches had a mean height of 7.0 cm (N = 19). Percentage survival from August 1983 to Feb- ruary 1984 was 100% for the unburnt seedlings and 84% for those burnt. In other stands, several small trees completely scorched by fire recovered quickly by growing new branches. Many of the larger P. elliptica trees in the northern jarrah forest carry fire scars on the lower stem, evidence of their ability to recover from even intense fire. General Discussion Three distinct hypotheses can be offered to explain failure in regeneration of a tree species: 1 No seed is available; 2 Seed is available but is not viable; 3 Seed is available and viable but (a) The disperser of the seed has become rare or extinct, (eg. Temple 1977); (b) Extreme climatic events or di- gestion of seed by a vertebrate are needed to break the dormancy of the seed; or (c) Some relevant eco- logical process has changed in nature or frequency over the period of interest. Our data allow rejection of the first two hypotheses. Because seedlings and lignotuberous seedlings can occur well away from the nearest reproductive trees, there is obviously no lack of dis- persal agents. We therefore reject hypothesis 3a and the part of hypothesis 3b involving vertebrate digestion. The calculated rate of diameter growth implies that the missing diameter classes <8 cm represent a period of 70-80 years. Local weather records show great variability in annual rainfall over the period 1900-1980 with drought years (eg. 1914. 1940, 1959, 1977) and very wet years (1915, 1917, 1926, 1945, 1955, 1964). Most possible combinations of temperature/rainfall should therefore have occurred during this period and at least one of these should have resulted in massive breaking of dormancy at least once (assuming that these factors are indeed causative). Thus hypothesis 3b is not favoured although difficult to reject with certainty. Because historical records are incomplete or lacking, it is not easy to define ecological changes that may have occurred over the last 70 years (hypothesis 3c). Using the experimental and observational data collected to identify ecological process(es), we constructed a plausible history of events. Large quantities of fruit are produced annually by larger trees of P. elliptica. The seed has a high viability but has a dormancy period broken effectively by passage through the vertebrate gut. Abundant germination is observed in the winter following a low intensity spring fire. Seedlings are highly palatable to brows- ing (probably by kangaroos and wallabies) during the first sum- mer and few survive. We therefore identify fire (for germination) and herbivory (causing mortality of seedlings) as the relevant ecological processes affecting regeneration of P. elliptica. How have both processes interacted in the past? We have re- liable information about fire (Abbott and Loneragan 1983) but only anecdotes about past densities of kangaroos and wallabies in jarrah forest. The main points of our speculative model are summarized in Fig. 3. We now proceed to the detailed argument. Before the disappearance of Aborigines from the region in the 1850s, the fire regime in the northern jarrah forest was probably one of frequent, low to moderate intensity fires (Abbott & Loneragan 1983). The density of kangaroos and wallabies should have been relatively low because of hunting pressure from Aborigines and the Dingo Canis familaris dingo (Whitehouse 1977, Abbott 1980). We therefore expect that ger- mination occurred regularly and that seedling survival was high, resulting in abundant recruitment of P. elliptica. 20 Journal of the Royal Society of Western Australia 71 (1), 1988 human activity Aborigines present settlement begins Aborigines removed from forest logging begins frequency: high low-medium high fire regime intensity: low-moderate higher high season: summer summer summer fire excluded frequent low spring Figure 3 Speculative model relating historical changes in fire regime and hunting pressure by Aborigines to germination and establishment of Persoonia elliptica. Following the disappearance of forest-dwelling Aborigines until the beginning of jarrah logging in the 1870s, the frequency of fires probably declined, resulting in an increase in fire inten- sity. The density of kangaroos and wallabies should have slowly increased, corresponding to the decrease in numbers of Aborigi- nes (but probably no change in density of the Dingo). Both fac- tors should have resulted in declining recruitment of P. elliptica. From the 1870s until the introduction of the Forests Act in 1919, fires once again became frequent because of adjacent agricultural clearing and abundant logging debris on the forest floor. Fires probably also increased in intensity for the same reason. Thus, the increasing length of time without Aboriginal occupation and the intensive farming of land adjacent to the for- est (and consequent European trapping of the Dingo because it preyed on sheep) should have allowed densities of kangaroos and wallabies to increase to very high levels, effectively sup- pressing recruitment of P. elliptica. Except as noted below, we do not think that hunting of wallabies and kangaroos by Euro- peans was of great importance because of the large extent of forest, the long distances between settlements and the small population of Europeans present. From about 1930 until about 1955, fire was excluded from the northern jarrah forest except for annual burning of 20 m wide firebreaks around 100-200 ha compartments of forest. Such fire exclusion should have reduced the supply of young, nutrient rich plant tissue as food for kangaroos and wallabies, leading to a decline in their numbers. Little recruitment of P. elliptica should have occurred in that period. From 1955 to the present, the northern jarrah forest was again subjected to fre- quent (5-7 years) low intensity fires. The densities of kangaroos and wallabies were observed to increase following this change in fire policy (J. Havel, pers. comm.) and the density of kangaroos is still relatively high (Short et al 1983). Recruitment of P. elliptica would have remained low. Although the Dingo in forest areas was replaced by the Fox ( Vulpes vuples) by the 1930s, the latter rarely preys on kangaroos and wallabies (Brunner et al 1975) and so is not relevant to the ecology of P. elliptica. The ageing of P. elliptica on the basis of current diameter growth indicates that, on average, the <10 cm diameter classes represent a period of c90 years, ie. regeneration began to fail towards the end of the nineteenth century. The virtual absence of trees of DBH <8 cm (equivalent to c70 years age) suggests that recruitment of P. elliptica over most of the northern jarrah forest had ceased by about 1910. How then do we explain the structure of populations 7-9 (Fig. 1) in which trees with DBH <6 cm are relatively well rep- resented? These three populations occur in the extreme north- eastern sector of the northern jarrah forest and have been close to European settlement (farms, orchards, towns) since the 1870s. We suggest that in and around these stands, hunting pressure from Europeans has been relatively high and consist- ent during the past 100 years. This depressed the population densities of kangaroos and wallabies and allowed some recruit- ment of P. elliptica. 21 Journal of the Royal Society of Western Australia 71 ( 1 ), 1988 Although the model outlined above is speculative, particu- larly when compared to the elegant historical study of Peterken and Tubbs (1965), parts of it are testable. The seedlings of P. elliptico that have been caged in population 2 and near popu- lation 3 should attain DBH of 8 cm within 70-80 years. An index of the abundance of kangaroos and wallabies, based on counts of faecal pellets in various forest stands (Hill 1981) could be cor- related with the density of P. elliptica seedlings. Acknowledgments We thank S. Davies for kindly allowing us to study the Persoonia population (No. 9) on his property, and for attempting to feed drupes to his captive emus. P. Christensen. S. Davies. J. Havel, O. Loneragan and R. Underwood provided helpful criticism. References Abbott I 1980 Aboriginal man as an exterminator of wallaby and kangaroo populations on islands round Australia. Oecologia 44: 347-354. Abbott I 1984a Comparisons of spatial pattern, structure, and tree composition be- tween virgin and cut-over jarrah forest in Western Australia. For Ecol Manage 9: 101-126. Abbott I 1984b Emergence, early survival and growth of seedlings of six tree species in Mediterranean forest of Western Australia. For Ecol Manage 9: 51-66. Abbott 1 1985 Rate of growth of Banksia grandis Willd. (Proteaceae) in Western Aus- tralian Forest. Aust J Bot 33: 381-391. Abbott I & Loneragan O 1983 Influence of fire on growth rate, mortality, and butt damage in Mediterranean forest of Western Australia. For Ecol Manage 6: 139-153. Abbott I & Van Heurck P 1985 Comparison of insects and vertebrates as removers of seed and fruit in a Western Australian forest. Aust J Ecol 10: 165-168. Bachelard E P 1967 Effects of gibberellic acid, kinetin, and light on the germination of dormant seeds of some eucalypt species. Aust J Bot 15: 393-401. Brunner H. Lloyd J W & Coman B J 1975 Fox scat analysis in a forest park in south eastern Australia. Aust Wildl Res 2: 147-154. Churchill D M 1959 The Tertiary and Quaternary vegetation and climate in relation to the living flora in South Western Australia. PhD thesis. Botany Dept, Univ W Aust. Clark D A & Clark D B 1981 Effects of seed dispersal by animals on the regeneration of Bursera graueolens (Burseraceae) on Santa Fe Island. Galapagos. Oecologia 49: 73-75. Coblentz B E 1978 The effects of feral goats (Capra hircus) on island ecosystems. Biol Cons 13: 279-282. Colbry V L. Swofford. T F & Moore R P 1961 Tests for germination in the laboratory. Seeds. Yearbook of Agriculture, 433-443. USDA. Crisp M D & Lange R T 1976 Age structure, distribution and survival under grazing of the arid zone shrub Acacia burkittii. Oikos 27: 86-92. Crossland T 1981 Germination of Sandalwood seed. Ann Rep Mulga Res Centre W Aust Inst Technol, Bentley, 49-50. Cunningham T M 1960 Seed and seedling survival of Eucalyptus regnans and the natural regeneration of second-growth stands. Appita 13: 124-131. Gilbert J M 1961 The effects of browsing by native animals on the establishment of seedlings of Eucalyptus regnans in the Florentine Valley, Tasmania. Aust For 25: 116-121. Glyphis J P. Milton S J & Siegfried W R 1981 Dispersal of Acacia cyclops by birds. Oecologia 48: 138-141. Hall E A A, Specht R L & Eardley C M 1964 Regeneration of the vegetation on Koonamore Vegetation Reserve, 1926-1962. Aust J Bot 12: 205-264. Hill G J E 1981 A study of grey kangaroo density using pellet counts. Aust Wildl Res 8: 237-243. Jones E W 1945 The structure and reproduction of the virgin forest of the north tem- perate zone. New Phytol 44: 130-148. Kullmann W J 1981 Seed germination records of Western Australian plants. Kings Park Res Notes No. 7. Lange R T & Graham C R 1983 Rabbits and the failure of regeneration in Australian arid zone Acacia. Aust J Ecol 8: 377-381 Lange R T & Willcocks M C 1980 Experiments on the capacity of present sheep flocks to extinguish some tree populations of the South Australian arid zone. J Arid Envir 3: 223-229. Needham R J 1960 Problems associated with regeneration of Eucalyptus gigantea in the Surrey Hills area. Appita 13: 136-140. Peterken G F & Tubbs C R 1965 Woodland regeneration in the New Forest, Hamp- shire, since 1650. J Appl Ecol 2: 159-170. Pigott C D 1969 The status of Tilia cordata and T. platyphyllos on the Derbyshire lime- stone. J Ecol 57: 491-504. Pigott C D 1983 Regeneration of oak-birch woodland following exclusion of sheep. J Ecol 71: 629-646. Short J. Caughley G. Grice D & Brown B. 1983 The distribution and abundance of kangaroos in relation to environment in Western Australia. Aust Wildl Res 10: 435-451. Statham H L 1983 Browsing damage in Tasmanian forest areas and effects of 1080 poisoning. Bull For Comm Tas No 7. Temple S A 1977 Plant-animal mutualism: coevolution with Dodo leads to near ex- tinction of plant. Science 197: 885-886. Wardle P 1959 The regeneration of Fraxinus excelsior in woods with a field layer of Mercurialis perennis. J Ecol 47: 483-497. Wardle P 1963 Vegetation studies on Secretary Island, Fiordland. Part 5: Population structure and growth of Rimu (Dacrydium cuppressinum) . N Z J Bot 1: 208-214. Wardle P 1963b The regeneration gap of New Zealand Gymnosperms. N Z J Bot 1: 301-315. Wardle P 1978 Regeneration status of some New Zealand conifers, with particular reference to Libocedrus bidwillii in Westland Natural Park. N Z J Bot 16: 471-477. Watt A S 1919 On the causes of failure of natural regeneration in British oakwoods. J Ecol 7: 173-203. Watt A S 1923 On the ecology of British beechwoods with special reference to their regeneration. J Ecol 11: 1-48. Whitehouse S J O 1977 The diet of the dingo in Western Australia. Aust Wildl Res 4: 145-150. JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA CONTENTS OF VOLUME 68 PART I (Published 30 June 1986) Page Destruction of australites by aborigines in part of the Eastern Goldfields, Western Australia W H Cleverly & I Cleverly 1 The effect of wave action on the shell morphology of Littorina unifasciata Gray. G Basingthwaighte & W Foulds 9 Ecology of the large indigenous earthworm Megascolex imparicystis in re- lation to agriculture near Lancelin, Western Australia. I Abbott, J S Ross & C A Parker 13 A further find from the Youndegin meteorite shower. J R De Laeter & D J Hosie 17 Host distribution, potassium content, water relations and control of two co- occurring mistletoe species. B Lamont 21 Contents Volume 67 27 PART 2 (Published 29 October 1986) The Whitfords Cusp — its geomorphology, stratigraphy and age structure. V Semeniuk & D J Searle 29 A biology of the desert fringe — Presidential address — 1984. S J J F Davies 37 PART 3 (Published 29 October 1986) Terminology for geomorphic units and habitats along the tropical coast of Western Australia. V Semeniuk 53 PART 4 (Published 22 May 1987) Australites from Hampton Hill Station, Western Australia. W H Cleverly gj High-temperature retrograde adjustments in some Precambrian granulite- facies rocks at Albany, Western Australia. N C N Stephenson 95 JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA CONTENTS OF VOLUME 69 PART 1 (Published 22 May 1987) Page Apodibius serventyi sp. nov., a new clawless water-bear (Invertebrata : Tardigrada) from Western Australia C E Morgan & C A Nicholls 1 Australites from the Kimberley region, Western Australia B Mason 5 Salmonella infections and animal condition in the mainland and Bald Island populations of the quokka (Setonix brachyurus: Marsupialia) R P Hart, S D Bradshaw & J B Iveson 7 The effects of urbanization on the ant fauna of the Swan Coastal Plain near Perth, Western Australia J D Majer & K R Brown 13 Air pollution components in Perth F Lax, W A Robertson & B V Garkaklis 1 9 PART 2 (Published 16 July 1987) Mangroves of the Dampier Archipelago, Western Australia V Semeniuk & P A S Wurm 29 PART 3 (Published 23 September 1987) Grazing pressure by the tammar (Macropus eugenii Desm.) on the veg- etation of Garden Island, Western Australia, and the potential impact on food reserves of a controlled burning regime D T Bell, J C Moredoundt & W A Loneragan 89 Wetlands of the Darling System — A geomorphic approach to habitat classification C A Semeniuk 95 Rb-Sr Geochronology of granitoids from Mount Mulgine, Western Australia J R de Laeter & J L Baxter 113 PART 4 (Published 4 February 1988) Aspects of variation in histology and cytology of the external nasal gland of Australian lizards H Saint Girons & S D Bradshaw 117 Northern Sandplain Kwongan: regeneration following fire, juvenile period and flowering phenology P G van der Moezel, W A Loneragan & D T Bell 123 Northern Sandplain Kwongan: community biomass and selected species re- sponse to fire J C Delfs, J S Pate & D T Bell 133 Northern Sandplain Kwongan: effect of fire on Hakea obliqua and Beaufortia elegans population structure D T Bell, P G van der Moezel, J C Delfs & W A Loneragan 139 JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA CONTENTS OF VOLUME 70 PART 1 (Published 26 February 1988) Page Cainozoic stratigraphy of the Yeelirrie area, northeastern Yilgarn Block, Western Australia D K Glassford 1 The Bridport Calcilutite V Semeniuk & D J Searle 25 PART 2 (Published 2 May 1988) Geochronology of the Mons Cupri Archean volcanic centre, Pilbara Block, Western Australia G C Sylvester & JRde Laeter 29 Origin of limestone lenses in Perth Basin yellow sand, Southwestern Australia V Semeniuk & D K Glassford 35 The unconformity in the Kelly Belt, East Pilbara Craton R C Horwitz 49 Addendum J R De Laeter & J L Baxter 55 PART 3 (Published 8 November 1988) Floristic reconnaissance of the northern portion of the Gregory National Park, Northern Territory, Australia D M J S Bowman, B A Wilson & P L Wilson 57 Consainguineous wetlands and their distribution in the Darling System, Southwestern Australia C A Semeniuk 69 PART 4 (Published 1 December 1988) Geomorphology, stratigraphy and Holocene history of the Rockingham — Becher plain, Southwestern Australia D J Searle, V Semeniuk & P J Woods 39 A census of the larger fungi of Western Australia Part II R N Hilton 111 Petrology and origin of beach sand along the Rottnest Shelf coast, Southwestern Australia D J Searle & V Semeniuk 119 INSTRUCTIONS TO AUTHORS The Journal publishes (after refereeing) • papers dealing with original research done in Western Australia into any branch of the natural sciences; • papers concerning some biological or geological aspect of Western Australia; • authoritative overviews of any subject in the natural sciences, integrating research already largely published in the more specialized national or international journals, and interpreting such studies with the general membership of the Society in mind; • analyses of controversial issues of great scientific moment in Western Australia. Prospective authors of papers in the last two categories should consult the Hon Editor for further advice. Contributions should be sent to The Honorary Editor, Royal Society of Western Australia, Western Australian Mu- seum, Francis Street, Perth, Western Australia, 6000. Publication in the Society’s Journal is available to all categories of members and to non-members residing outside Western Australia. Where all authors of a paper live in Western Australia at least one author must be a member of the Society. Papers by non-members living outside the State must be communicated through an Ordinary or an Honorary Member. Submission of a paper is taken to mean that the results have not been published or are not being considered for publication elsewhere. Free reprints are not provided. Reprints may be ordered at cost, provided that orders are submitted with the return galley proofs. Authors are solely responsible for the accuracy of all information in their papers, and for any opinion they express. Manuscripts. The original and two copies must be submitted. They should be typed on opaque white paper with double- spacing throughout and a 3 cm margin on the left-hand side. All pages should be numbered consecutively, including those carrying tables and captions to illustrations, which appear after the text. Illustrations, both line drawings and photographs, are to be numbered as figures in a common sequence, and each must be referred to in the text. In composite figures, made up of several photographs or diagrams, each of these should be designated by a letter (eg Figure 2B). To avoid risk of damage to original figures, authors may retain these until after the paper is accepted. The copies of the figures accompanying the manu- script must be of good quality. Authors are advised to use the most recent issue of the Journal as a guide to the general format of their papers. Words to be placed in italics should be underlined. To facilitate editing, papers must be accompanied by a table of contents, on a separate sheet, showing the status of all headings. Following acceptance and editing of the manuscript, authors with access to word processors/computers should amend their manuscript and forward to the Hon Editor their disk (noting the computer and word-processing program used) and one copu of the paper. References must be set out as follows: Paper Jackson A 1931 The Oligochaeta of South-Western Australia. J R Soc W Aust 17:17-136. Twigg L, Majer J D & Kotula R 1983 the influence of fluoroacetate producing plants upon seed selection by seed harvesting ants. Mulga Res Centre W Aust Inst Technol, Bentley, Ann Rep 6:75-80. Book Jacobs M R 1955 Growth Habits of the Eucalypts. For Timb. Bur, Canberra. Chapter in book Dell J 1983. The Importance of the Darling Scarp to Fauna. In: Scarp Symposium (ed J D Majer) W Aust Inst Technol, Bentley, 17-27. The Title should begin with a keyword. The Abstract should not be an expanded title, but should include the main substance of the paper in a condensed form. The metric system (SI units) must be used. Taxonomic papers must follow the appropriate international Code of Nomenclature, and geological papers must adhere to the International Stratigraphic Guide. Spelling should follow the Concise Oxford Dictionary. Authors should maintain a proper balance between length and substance, and papers longer than 10 000 words would need to be of exceptional importance to be considered for publication. Authors will be charged page costs (currently $30 per page) if papers exceed 8 printed pages. Short papers (2-4 printed pages) are particularly sought as these often ensure full use of the 32 pages available in each part. Illustrations. These should be prepared to fit single or double column widths. Illustrations must include all necessary letter- ing, and be suitable for direct photographic reduction. No lettering should be smaller than 1 mm on reduction. To avoid un- necessary handling of the original illustrations, which are best prepared between 1.5 and 2 times the required size, authors are advised to supply extra prints already reduced. Additional printing costs, such as those for folding maps or colour blocks, will be charged to authors. Supplementary Publications. Extensive sets of data, such as large tables or long appendices, may be classed as Sup- plementary Publications and not printed with the paper. Supplementary Publications will be lodged with the Society’s Library (C/- Western Australian Museum, Perth, WA 6000) and with the National Library of Australia (Manuscript Section, Parkes Place, Barton, ACT 2600) and photocopies may be obtained from either institution upon payment of a fee. JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA CONTENTS VOLUME 71 PART 1 1988 Page Regeneration of Acacia and Kennedia from soil stored seed following an autumn fire in jarrah (Eucalyptus marginata) forest W L McCaw 1 Calc-alkaline lamprophyres from the Pilbara Block, Western Australia N M S Rock & M E Barley 7 Widespread regeneration failure of Persoonia elliptica (Proteaceae) in the northern jarrah forest of Western Australia I Abbott & P Van Heurck 15 Edited by I Abbott Registered by Australia Post — Publication No. WBG 0351 No claim for non-receipt of the Journal will be entertained unless it is received within 12 months after publication of Part 4 of each Volume The Royal Society of Western Australia, Western Australian Museum, Perth Circulation of this Journal exceeds 600 copies. Nearly 100 of these are distributed to institutions and societies elsewhere in Australia. A further 200 copies circulate to more than 40 countries. The Society also has over 350 personal members, most of whom are scientists working in Western Australia. The Journal is indexed and abstracted internationally. A60438/4/88 GARRY L. DUFFIELD, Government Printer, Western Australia VOLUME 71 PART 1 JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA Volume 71 • Parts : ISSN 0035-922X iHoyal Society of tUestern Australia To promote and foster science in Western Australia and counteract the effects of specialization PATRON Her Majesty the Queen VICE-PATRON His Excellency Professor Gordon Reid Governor of Western Australia COUNCIL 1988-1989 President J S Pate Ph D, D Sc, FAA, FRS Vice-Presidents M Candy M Sc, FRAS B Dell B Sc (Hons), Ph D Past President J T Tippett B Sc, Ph D Joint Hon Secretaries K W Dixon B Sc (Hons), Ph D L Thomas M Sc Hon Treasurer J Dodd B A, M Sc, Ph D Hon Librarian M A Triffitt B A, ALAA Hon Editor I Abbott B Sc (Hons), Ph D Hon Journal Manager J Backhouse B Sc (Hons), M Sc, Ph D Members J S Beard M A, B Sc, D Phil W A Cowling B Agric Sc (Hons), Ph D S J Hallam M A, FAHA L E Koch M Sc, Ph D K McNamara B Sc (Hons), Ph D J Majer B Sc, Die, Cert Ed, Ph D V Semeniuk B Sc (Hons), Ph D Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989, 23-47 The Quindalup Dunes: the regional system, physical framework and vegetation habitats V Semeniuk 1 , I D Cresswell 2 & P A S Wurm 3 1 21 Glenmere Road, Warwick W A 6024. 2 PO Box 1076, Subiaco W A 6008. 3 PO Box 186, North Perth W A 6006. Manuscript received March 1987; accepted February 1988 Abstract The Quindalup Dunes contain a variety of aeolian landforms developed by regional climatic, geomorphic and sedimentologic factors, as well as local coastal/strandline processes, and vegetative and pedogenic pro- cesses. Superimposed on these are factors of distance from the strandline (which determines the degree of wind effects), soil salinity, and height above water table (which is related to height above sealevel). These factors and processes have resulted in a range of geomorphic units, habitats and vegetation responses that can be recognized at various scales of reference. A consistent terminology for geomorphic units and habitats has been developed in order to compare tracts of Quindalup Dunes along the various sectors of the southern west coast of Western Australia. The large to medium scale geomorphic units include parabolic dunes, chaots (chaotic dune terrain), shore-parallel ridges, blowouts, un- dulating plain and flats. The small scale geomorphic units, that essentially equate with the basic vegetation habi- tats, are subdivisions of the larger scale geomorphic units; these include various slope units; crests and de- pressions of the larger scale units. Each of the five sectors of the south west coast of Western Australia contains distinct associations of geomorphic units and vegetation habitats that reflect a difference in the regional sedimentological and geomorphic setting, as well as gradients in climate and other edaphic factors. Changes in dominant habitat types from sector to sector, together with the climatic gradient, favour smaller scale heterogeneous distribution in the structure and floristics of the vegetation units along the extent of the Quindalup Dunes. Our analysis of the distri- bution of reserves for flora and fauna within the Quindalup Dunes indicates that the regional variety of landforms and vegetation habitats is not adequately represented. In particular, there is no reservation of Quindalup landforms and habitats representative of Geographe Bay, the barrier dunes of Leschenault-Preston Sector, and the cuspate beachridge plain exemplified by Point Becher. Introduction The Holocene coastal dune zone of the Swan Coastal Plain is generally a relatively narrow assemblage of landforms formally termed the Quindalup Dunes (McArthur & Bettenay 1960). The zone extends from Dunsborough in the south to Dongara in the north along the south west coast of Western Australia. Within this zone various authors have described and mapped veg- etation, or landforms related to vegetation. Smith (1973) provided a guide to the flora of the coastal habi- tats and subdivided the flora of dunes into 4 types: foredune vegetation; mobile dune vegetation; stabilized dune vegetation; and tall closed dune scrub. Smith (1985) later subdivided the dune vegetation into 3 types, apparently excising the tall closed dune scrub from the classification. A number of later authors, particularly in unpublished government reports, adopted the subdivision of Smith (1973) as the basic vegetation units of the Quindalup Dunes. Speck (1952) and Seddon (1972), on the other hand, provided maps of the dune zone, generally treating the vegetation complexes within the Quindalup Dunes as a single unit. Heddle (1979), Heddle et al (1980) and Beard (1976, 1981) similarly mapped the flora of the zone, treating the unit essentially as a homogeneous system although recognizing at least 2 alliances, namely a strand and foredune alliance, and a mobile and stable dune alliance. McArthur & Bartle (1980a, b) described various stages of Quindalup dune landform evolution (Ql, Q2, Q3 and Q4) and its stabilization by vegetation, and related vegetation assemblages to these landforms. More recently Cresswell & Bridgewater (1985), utilizing a floristic and landform/soil ap- proach, incorporated the Quindalup Dune System into their overall treatment of vegetation of the Swan Coastal Plain. They subdivided the Quindalup Dune vegetation into eight units, which were then related to geomorphic location. However, the Quindalup Dunes present a much more vari- able system of habitats than perhaps has been appreciated. The variability in habitat is due to: processes and stages of landform development as determined by regional factors; processes and stages of landform development as determined by local geomorphic history; relative position of landform units with re- spect to the strandline; relative relief of the various small scale landform units with respect to water table; degree of develop- ment of soils and calcrete; and soilwater salinity. These interacting factors have developed a wide range of small scale geomorphic units each with its own distinct relief, slope, soil cover, and location relative to sea effects. Because geomorphic processes, and the resultant geomorphic units (vari- able in time, space, intensity of development, and scale), are the A63378-1 23 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. fundamental determinants of habitats, a geomorphic approach should provide a useful framework for vegetation studies. The approach adopted here is to describe the geomorphic units and habitats at various scales of reference in the Quindalup Dunes, and to relate them to coastal and aeolian processes and veg- etation. This is a direct application of the coastal sector analysis of Searle & Semeniuk (1985) and landform analysis to veg- etation investigations. There is a recurring suite of geomorphic units peculiar to each of the 5 coastal sectors of Searle & Semeniuk (1985) and thus the sector approach provides a framework to identifying and understanding the differences within the Quindalup Dunes along their extent. The approach is useful for analysing the Quindalup Dune vegetation regionally and, in combination with the subcontinental gradient of climate and variability in species pool, helps to explain the complicated pattern of habitats and vegetation assemblage within the Quindalup Dune System. Such information is important for studies in conservation, coastal management or assessment of regional significance of vegetation and landforms within a given sector of the Quindalup Dunes. This paper reports on the first stage of investigations of the vegetation of the Quindalup Dunes. The objectives of the paper are to describe the regional setting, regional variety and the local variability of the physical features of the Quindalup Dunes throughout their full extent along the Swan Coastal Plain so that the physio-chemical framework, geomorphic units and habitats can be identified. As such this paper provides information on geomorphology and vegetation habitats of the Quindalup Dunes to a level not previously reported and provides the basis for more detailed studies of the habitats and vegetation of the Quindalup Dunes in the future. Methods The results of this paper are based on intensive fieldwork, re- connaissance field surveys, aerial photograph studies, low alti- tude aerial surveys and literature review. Fieldwork involved study of geomorphology, stratigraphy, soils and vegetation by surface mapping, coring/trenching and sample collection. The maps produced in this paper are the result of field work sup- plemented by ground truthing. Aerial photographs and ground surveys were also used to identify the range of geomorphic units in the region. Sites of intensive fieldwork include (Fig. 1): Geographe Bay area, Leschenault Peninsula-Myalup area, Yalgorup National Park-Mandurah area, Pt Becher- Rockingham Plain area, Trigg Island-Whitfords-Two Rocks area, and areas around Lancelin, Cervantes, Jurien Bay, Green Head-Leeman, and Dongara. Intensive fieldwork was sup- plemented by reconnaissance surveys and low altitude aerial surveys over the remainder of the southwest coast between Geographe Bay and Dongara. Examination of aerial photo- graphs, utilizing black/white and colour photographs, was also undertaken. The approach of Semeniuk (1986) was used in mapping and naming of geomorphic/habitat units. This method involves identifying units observable and mappable at a given scale of reference. The scales of reference, slightly modified after Semeniuk (1986), are: regional scale (100km x 100km frame of reference); large scale(10km x 10km frame of reference); medium scale(lkm x 1km frame of reference); and small scale (100m x 100m frame of reference). The same geomorphic unit may be observable at several scales of refer- ence. For example, parabolic dunes may be mappable at re- gional and large scales. At medium scale, different types of parabolic dunes such as fretted or attenuated, may be identifi- able. At small scale only geomorphic sub-units of the parabolic dune may be mappable, such as the crest or bowl. Figure 1 Distribution of Quindalup Dunes and location of study sites in Southwestern Australia. The stratigraphy of the Quindalup Dune terrain was deter- mined by drilling, trenching and augering, and from information in the literature (Searle 1978, Semeniuk 1983, Woods 1983, Searle & Semeniuk 1985, Semeniuk & Searle 1985a, b 1986). Soil profiles were documented in numerous locations in each of the study sites. The soils were described in terms of humus con- tent, structure, fabric, texture and composition. Soil samples for analyses of salt content were collected along 2-4 transects for each sector. Samples were collected in summer from the shore to the hinterland in transects parallel to the dominant summer wind direction. Sample sites were generally spaced 100m apart for distances 0.5-2. 0km along a transect. At each sample site 5 replicate surface soils were collected. In the laboratory soils were dispersed in an aliquot of distilled water to leach out sol- uble salts and the salinity of the resultant solution was deter- mined by conductivity meter. The salinity results were then con- verted to mg salt/cm 3 of soil. The organic and humus content of soil was determined by heating soils to >450°C to expel carbon as C0 2 . 24 Journal of the Royal Society of Western Australia. Vol. 71, Parts 2 & 3. 1989. Regional setting Geomorphology The Quindalup Dune System extends along the modern shoreline of the Swan Coastal Plain, spanning a distance from Geographe Bay in the south to Dongara in the north (Gentilli & Fairbridge 1951). It varies from a narrow unit to a locally exten- sive system up to 10km wide. Topographically the system varies from a low-relief, subdued coastal plain 3-5m above MSL, to steep aeolian highlands with a relative relief of c 30 m. The Quindalup system generally adjoins the older Spearwood Dune system (to landward) but in many locations the Quindalup Dunes are encroaching over the Spearwood Dune terrain. Climate The coastal dune belt of the Swan Coastal Plain spans a cli- mate gradient from humid in the south at Geographe Bay to semi-arid at Dongara (Gentilli 1972). Such wide climate vari- ation should influence coastal processes and hence the develop- ment of different habitats, dynamics of habitats, availability of water, and consequently the resultant vegetation. The climatic parameters considered most important to development of habi- tats and to maintenance of vegetation are wind, rainfall, evapor- ation and temperature. These factors interact to develop a variability in habitat and plant response both regionally and locally. For instance, onshore wind gradually changes in inten- sity and direction from Geographe Bay to Dongara (Searle & Semeniuk 1985). As a result, blowouts and parabolic dunes in- crease in number, and the parabolic dunes extend further inland and become more northerly aligned from south to north along the coast. Rainfall, evaporation and temperature also are criti- cal to dune development and vegetation response, and again from south to north there is an increase in aridity reflected by landforms and vegetation. Rainfall, varying from c 500mm/yr in the north to >800mm/yr in the south, determines the amount of moisture available in the vadose zone and the extent to which the groundwater table is salinized. Climate data for selected lo- cations are summarized in Table 1. Stratigraphy The sand of the Quindalup Dunes stratigraphically is referred to as Safety Bay Sand. The unit is juxtaposed against the older more landward Spearwood Dune System or Yoongarillup Plain (MacAthur & Bettenay 1960, MacArthur & Bartle 1980a), the underlying materials of which are stratigraphically referred to as Tamala Limestone (Playford et al 1976). The Safety Bay Sand also may adjoin and overlie a Holocene seagrass sedimentary unit, the Becher Sand (Semeniuk & Searle 1985b), or an estuar- ine sedimentary unit, the Leschenault Formation (Semeniuk Table 1 Climate data for selected localities along the coast of Southwestern Australia 1 Climatic Data SECTOR LOCALITY Annual Rain Annual 2 Mean daily rain- days evapo- temperature fall per ration (°C) in (mm) annum (mm) summer (January) Mean max. temperature (°C) in summer (January) Mean daily temperature (°C) in winter (July) Mean min. Wind 3 in Summer temperature (°C) in winter (July) 1 Busselton 838 137 1200 21.2 28.8 12.4 8.1 landbreeze/seabreeze system mainly 0-20km/ hr; emanating mainly from SSE & NW respectively 2 Bunbury 881 122 1300 21.9 27.4 12.9 9.1 landbreeze /seabreeze system mainly 0-20km/hr, & Mandurah 897 121 1500 23.1 28.7 13.4 9.4 but up to 20-40km/hr in the northern areas mainly from SE-NE 3 Fremantle 775 128 1900 23.0 27.7 13.9 10.2 & SW respectively 4 Lancelin 627 126 2000 23.0 28.7 14.6 10.1 landbreeze / seabreeze system mainly 20-40km/hr & Jurien Bay 519 100 2200 23.0 29.3 14.5 9.6 mainly from SE-E & S-SW respectively 5 Geraldton 477 88 2500 25.1 31.6 14.3 9.2 1 Data from Bureau of Meteorology 1975 2 Estimated from evaporation map (Bureau of Meteorology 1980) 3 See Searle & Semeniuk 1985a. Only summer wind is considered important in developing onshore aeolian landforms. Onshore wind in winter is usually mild, although periodically punctuated by storms accompanied by rain. 25 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. 1983). The Safety Bay Sand may have one of several types of stratigraphic relationships with the adjoining stratigraphic units (Playford et al 1976, Searle 1978, Semeniuk 1983, Semeniuk & Searle 1985b, Searle & Woods 1987). These relationships are (Fig. 2): Type 1, a sheet of Safety Bay Sand overlies seagrass bank sedimentary deposits, but is detached from Tamala Limestone; Type 2, a ribbon to prism of Safety Bay Sand abuts and encroaches upon Tamala Limestone; Type 3, a ribbon or shoestring of Safety Bay Sand is perched upon and overlies Tamala Limestone in a near coastal setting; Type 4, lenses, ribbons and shoestrings of isolated and detached Safety Bay Sand sedimentary deposits are perched upon and overlie Tamala Limestone; and Type 5, a shoestring of Safety Bay Sand overlies estuarine sedimentary deposits but is detached from Tamala limestone. The significance of these types of large scale stratigraphic con- tacts is that each provides a separate geomorphic and hydrologic setting for the Quindalup Dunes. A low relief sand plain situated 2 to 3 to 5m above the water table and overlying a sand aquifer, for instance, provides a different setting to a high relief dune terrain situated 10-30m above the water table and underlain by limestone that has a calcrete capstone. The various stratigraphic types and their distribution with respect to the coastal sectors of Searle & Semeniuk (1985) are illustrated in Figure 2. There also is a variety of stratigraphic features internal to the Safety Bay Sand; these include: 1) beach/beachridge sheet, overlain by aeolian sand sheets and lenses, which are cross- layered to root-structured to homogeneous (Semeniuk & John- son 1982); and 2) aeolian sand sheets, lenses and wedges, cross- layered to root-structured to homogeneous, with intercalated soil sheets and local development of calcrete sheet (Semeniuk 1983, Semeniuk & Meagher 1981a, Semeniuk & Searle 1985a). Figure 2 The range of stratigraphic relationships of Safety Bay Sand with adjoining units. 26 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. The variety of internal stratigraphic features and lithologies, such as calcretes and buried or intercalated soil sheets, has some relationship to vegetation in that it influences moisture re- tention and either facilitates or retards vadose water migration. The inter-relationships of calcrete and vegetation have been previously discussed in Semeniuk & Meagher (1981a) and Semeniuk & Searle (1985a). Soils The soils of the Quindalup Dunes have been described by McArthur & Bettenay (1960), McArthur & Bartle (1980a, b), Semeniuk & Meagher (1981) and Semeniuk & Searle (1985a). Soils are mostly arenosols (sandy soils) and are developed as pedogenic overprints on a quartzose calcareous, medium to fine grained sand (usually a quartz skeletal lithoclast grainstone, in the terms of Dunham 1962). Degree of soil development is indi- cated by the amount of humus developed, the degree to which carbonate grains have been leached and the extent of root struc- turing and bioturbation. The most common soils, listed here in order of developmental maturity and area abundance, are: thin (<20cm), weakly humified quartzose calcareous sandy soil; thin (<50cm), humic quartzose calcareous sandy soil; thick (l-2m), humic quartzose calcareous sandy soil; thick (l-2m), humic quartz sandy soil. The soils generally form sheets over the dune terrain and have their thickest development in swales and lowlands. Sub- sequent landward migration of aeolian sediments may bury a soil profile. Details of soil profiles and soil features as they relate to habitats and vegetation will be presented in future publications. The results of soil salinity transects are shown in Fig. 3. Gen- erally soil salinity, although initially highest near the shore, does not gradually increase to landward. However, the occurrence of humic soils increases the capacity of soils to retain salt and re- sults in locally high values of salt content. The mean salt content of the soils increases regionally from south to north in response to more intense (salt bearing) onshore winds, more evaporation and the decreased effect of leaching rainfall. However, at any area locally the salt content of soils also increases due to humus content. In northern areas where deflation flats have been formed close to a water table, indurated carbonate crusts may also be developed on the aeolian sands. These crusts are wide-spread sheets hundreds of square metres in area and up to 30cm thick. These crusts result in localized limestone-like pavement habitats. Vegetation The vegetation of the Quindalup Dunes occurs predominantly as a system parallel to the coast (Beard 1976, 1981). Because the Quindalup Dunes occur over a wide climatic gradient in a north-south direction, and may exhibit an east-west variability in landforms and habitats, it may be expected to ex- hibit regional variation. A gradient is evident in the structure and floristics of the vegetation from south to north. In the southern sectors low forest, woodlands and scrub are the dominant types with Agonis flexuosa, Eucalyptus gomphocephala and Acacia spp. as the main overstory species. In the northern sectors the dominant structural units are scrub and heath, with several species of Acacia and Melaleuca as the main overstory species. Although regional changes in vegetation structure and floristics should be gradual, in response to a north-south climatic gradient, this is not strictly the case. The dominant vegetation habitats developed in each of the five sectors identified by Searle & Semeniuk (1985) are distinct from adjoining sectors, and therefore there is not a simple recurring pattern of similar habitats along the entire length of the southwestern coast. Thus the changes in dominant habitat types between the sectors and the climatic gradient along the length of the coast interact to de- velop a heterogeneous distribution in the structure and floristics of the vegetation units at regional and large scales. At the local scale vegetation also is strongly related to habitat features. Therefore as the habitat types change so does vegetation struc- ture and floristics. Within any given area there will be vegetation response due to factors such as distance from ocean, soil devel- opment, position in the landscape and fire history. Furthermore, soil and landscape factors are also related to climate, and altogether produce distinct regional and local patterns in the vegetation of the Quindalup Dune System. Terminology for geomorphic/habitat units It is important to compare similar geomorphic units or habi- tats if patterns of vegetation distribution are to be understood at the regional scale through to the local scale. Accordingly, it is necessary to apply a consistent set of geomorphic/habitat terms throughout the Quindalup Dunes. In terms of vegetation habitats it is necessary only to note the resulting shape of the land surface, and apply a non-genetic term to describe it. This approach is the basis for our choice of non-genetic terminology. Criteria adopted in this paper to de- scribe and name dune landforms are: dune geometry (eg para- bolic); relief (eg high, medium, low, undulating, flat); continuity (continuous vs disrupted); and alignment relative to shore (par- allel, oblique, transverse) Geomorphology A number of authors have described and classified dune landforms (Goldsmith 1985, Cooper 1967, McKee 1979, 1982, Breed & Grow 1979, Davies 1980, Mainguet 1984, Hesp 1984a, Tinley 1985, etc). Most of the terms in these works are non-genetic and based on geometric criteria ^frid, as such, are adopted here. However some terms have been coined in this paper because there were inadequate terms in the literature for the type of landform encountered in this study, or because the only terms available were genetic. The new terms for dune landforms coined in this paper are chaots, conical hill residuals, and shore-transverse ridges. Defi- nitions of terms used are provided in Table 2. Parabolic dunes also have been subdivided into crescentic, attenuated and fret- ted types. The fretted category is a new form described in this paper. Some of the subdivisions of parabolic dune systems into components of arms, inner face, advancing face and bowl, for purposes of distinguishing habitats for vegetation, also are new. The other terms used are established in the literature but are modified by descriptors to denote features such as relief, conti- nuity and alignment in relationship to the shore. Illustrations of large and medium scale units are provided in Fig. 4. Some of the new terms in this paper are equivalent, either fully or in part, to those in the literature. This applies to the terms shore-transverse ridges, bowls, and conical hill residuals. Shore-transverse ridge is equivalent to longitudinal dune of Thom (1965), and may be partly equivalent to wind rift dune of Hack (1941) and Mabbut (1977). However, shore-transverse ridge is preferred because the term does not carry a genetic con- notation, as wind rift does, and does not imply an orientation with respect to wind direction and origin that is associated with the term longitudinal. The term bowl incorporates the term dune slack of Ranwell (1972), but the former term is preferred because it does not carry implication of wind deflation down to a non-erodable surface such as rock, shingle or wet sand. The 27 A. _Q O k_ 0 Q. < / ) E co k— O) 0.7- 0 . 6 - 0.5- 0.4- 0.3- 0 . 2 - 0 . 1 - 0.5- 0.4- 0.3- 0 . 2 - 0 . 1 - 0.7- 0 . 6 - 0.5- 0.4- 0.3- 0 . 2 - 0 . 1 - 0.4- 0.3- 0 . 2 - 0 . 1 - 0.7- 0 . 6 - 0.5- 0.4 • 0.3- 0.2 0 . 1 ' Effect of ' Humic soil sample proximity to sea Dongara 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r to 0.90 1 0.23 400 600 800 1000 1200. 1400 1600 1800 2000 Whitfords Effect of proximity to sea —i 1 r — i 1 1 1 r 1 1 1 1 1 1 1 r—i 1 1 — 200 400 600 800 1000 1200 1400 1600 1800 2000 Becher Point i i i i i i i i i i t i 200 400 600 800 1000 1200 Leschenault Peninsula 1 1 1 1 1 r— 1400 1600 1800 “I 1 1 1 1 1 1 1 1 I ' T- 200 400 600 800 1000 1200 Humic soil sample Busselton — i — i — i — i — i 1 — i — i — i — i — i — i — 200 400 600 800 1000 1200 Distance along transect (metres) B Mean and Standard Deviation of salt concentration in soil 7.0- 6 . 0 - 5.0- 4.0- N = 35 Figure 3 A Salt concentration in soils along transects from shore to hinterland, parallel to dominant wind direction, for each sector (replicate sample n -5). B Bulk mean salt concentration of coastal dune soils in a subcontinental perspective from Busselton to Dongara. 28 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3. 1989. Table 2 Definition of terms Geomorphic term Definition* Comments Shoreline dune ribbon shore-parallel, low relief dune complex with ribbon- shaped plan and cross-sectional geometry; the complex abuts/adjoins the hinterland a regional to large scale feature Barrier dunes shore-parallel, low or high relief, narrow dune complex forming a barrier to lagoon or estuary a regional to large scale feature Cuspate beachridge plain low relief accretionary plain of parallel sand ridges; coastal margin of plain is cuspate; plain is result of coalescence of adjoining cusps a regional to large scale feature Cuspate foreland, or cusp isolated, accretionary sediment body; triangular in plan; composed of low relief to high relief dune complex a regional to large scale feature; the term cuspate foreland refers to the largest type of cusp (see Bates & Jackson 1980); the term cusp in the literature usually refers to small beach cusps but there is no scale restriction to the term; in this paper there also is no size implication in the use of the term Perched dunes shoreward encroaching dune complex of parabolic dunes and sand sheets perched upon and transgressing the upland hinterland limestone terrain; irregular plan and cross-sectional geometry a regional to large scale feature Foredunes shoestring deposit of sand developed by aeolion processes usually as a low ridge immediately landward of beach and seaward of the first high relief-medium relief dune complexes further to landward; may comprise a narrow belt on the sheltered coast of cuspate forelands subdivision of foredune morphology is provided by Hesp (1984a) Parabolic dunes sand dune, u-shaped to spatulate in plan, convex in downwind direction. Three types are recognized based on plan geometry: crescentic, which is a short to elongate u-shape; attenuated which is markedly elongate to the extent that the dune form consists mostly of parallel arms; and fretted, where the arms of the dune have developed subsidiary smaller blow-outs and parabolic dunes. Parabolic dunes are subdivided into components of arms, bowl, advancing face, inner face, conical hill residuals attenuated parabolic dunes are termed hairpin (Bates & Jackson 1980) and fretted parabolic dunes are in part synonymous with compound imbricated parabolic dunes of Tinley (1985) Chaots a chaotic system of sand hills, mostly conical in shape and of various sizes and relief, and associated, mostly circular depressions. The entire chaot system may be low relief, medium relief or high relief. The chaot system itself may be sheet-form or ridge form this geomorphic term is new as defined in this paper; however the term incorporates erosional as well as accretionary dune forms and is intended to be descriptive not genetic 29 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. Table 2 Definition of terms (continued) Geomorphic term Definition* Comments Shore-parallel ridges a system of linear parallel sand ridges usually of similar relief (ie low or high) with intervening linear depressions. The ridges may be subdivided into continuous and discontinuous (disrupted) types these low ridges are also termed “beachridges” in this paper and by others (Woods & Searle 1983) Shore-transverse ridges a system of high relief ridges developed transverse to the shore; in specific areas there may be a system of adjacent parallel transverse ridges the term may be in part synonymous with wind rift dunes of Hack 1941 & Mabbut 1977; these dune forms are not longitudinal dunes Blowout small to large trough shaped depression or scour formed by wind erosion. Blowouts are subdivided into the components of floor, walls, and, where vestiges of the original terrain remain, conical hill residuals. The transported sand may form local conical hills or parabolic dunes Bates & Jackson 1980 use the term ’’blowout” to refer to the eroded terrain and to the adjoining accumulation of sand, where recognizable, derived from the depression. However in this paper if the derived sand assumes a recognizable parabolic form it is termed a parabolic dune Undulating plain low relief plain with broad, gentle undulations A large to medium to small scale feature Flat medium to small scale geomorphic feature com prised of flat terrain the flat is not obviously linked to any parabolic dune, in which case it would be a bowl; see bowl Conical hill residual small to medium scale conical hill left as an erosional residual as the surrounding terrain is scoured away, either in a blowout or within a migrating parabolic dune the conical hill residual usually has a capping of tenacious vegetation which has determined why the landform remains as a residual Bowl flat or slightly concave floor of the inner portion of a parabolic dune; the bowl is contained by the arms and inner face of the parabolic dune bowls and flats are similar except that bowls are confined by parabolic dune arms; the term is partly synonymous with dune slack, Ranwell 1972 Coppice dune a conical mound or hummock of sand accumulated around vegetation (Cooper 1967) these dunes are not common in the study area Barchans and Barchanoid ridges intergradational spectrum from small isolated crescentic dunes oriented transverse to wind direction, with a gently convex windward face, concave leeward face and horns pointing downwind (barchans), to ridges, transverse to wind direction, with incipient barchan geometry these dunes are not common in the study area and usually occur as small-medium scale units on crests and margins of active parabolic dunes Transverse dunes linear, strongly asymmetric dune ridge oriented transverse to wind direction with gently sloping windward face and steep leeward face (Bates & Jackson 1980); grades into barchanoid ridge these dunes are not common in the study area and usually occur as small-medium scale units on crests and margins of active parabolic dunes Wetland Wet, waterlogged or inundated flats within the dune terrain the wetland types are not discussed further in this paper * of dune geometry regardless of whether dune is mobile, bare, or fixed 30 Journal of the Royal Society of Western Australia. Vol. 71, Parts 2 & 3, 1989. term bowl is intended to describe the geometric form of the con- cave centre of a parabolic dune system regardless of whether or not its floor lies at a distinct stratigraphic or hydrologic interface. The term flat thus also is partly equivalent to dune slack. The term conical hill residual is preferred to remnant dune of Davies (1980) because it conveys description of a conical shaped rem- nant . The descriptive term chaot is applied to a system of dunes whose surface is a chaotic system of conical landforms. The chaotic terrain may be the result of erosion producing conical hill residuals, or the result of erosion and accretion, where ac- cretion has resulted in acute coppice dunes, parabolic dune fronts, and sand sheets. In these terrains with the chaotic system composed of a closely related aggregate of landforms of various origins it is difficult to separate accretionary and erosional components for classification. The term chaot thus is intended to address this situation descriptively. The use of terminology such as “vegetated dune” or “fixed dune” as a geomorphic unit as distinct from ’’bare and mobile dunes” or “non-vegetated dunes” is widespread in the literature (Goldsmith 1985) and in the past also has been applied to the coastal dunes of Western Australia (Semeniuk & Meagher 1981b). This terminology for geomorphic units based on ab- sence or presence of vegetation has been rejected in this paper. Dune landforms are described and termed according to their ge- ometry, relief and configuration, and the presence of vegetation should have no part in primary terminology of geomorphology and habitats. Accordingly, dunes such as parabolics that vary from actively-mobile bare forms, to recently-fixed vegetated forms, through to older, fixed vegetated forms, may comprise similar components and geometry, and as such are simply termed parabolic dunes. The absence, presence, or range of vegetation cover is viewed as varying stages of plant coloniz- ation on a dune habitat. The description and nomenclature of the various stages of plant colonization through time is viewed as the realm of plant ecology and not geomorphology. Habitats Vegetation responds to variability in landform and edaphic features at the small scale, and it is the small scale geomorphic unit that forms the basic habitat unit. Thus the range of small scale features of the various large to medium scale geomorphic units are treated as the habitat types of the Quindalup Dunes. However, geomorphology alone does not determine whether a given terrain is a suitable habitat for the various species of veg- etation. Other features of the habitat, such as seaward aspect, landward aspect, north {sun-facing) aspect, height relative to the water table, salinity of vadose and phreatic water, and extent of soil development also need to be considered. The smallest scale of geomorphic features provides the basis upon which to overlay the other edaphic features. A63378-2 31 Journal of the Royal Society of Western Australia. Vol. 71. Parts 2 & 3, 1989. The amount of detail required to describe habitats for veg- etation necessitated subdividing the landforms of the coastal dunes into smaller and smaller units to arrive at the scale at which the vegetation responds. Accordingly, there is a range of geomorphic terminology applicable at the various scalar frames of reference (Table 3). The range of smaller scale variability in geomorphology can result in a mosaic of small scale vegetation responses. Thus shore-parallel, low, continuous ridges provide one type of habitat distribution, and shore-parallel, low, dis- rupted ridges provides a more complex habitat system. In this paper, the term habitat is used interchangeably with small scale geomorphic unit and differentiation of habitats be- yond this scale was not undertaken. The basic geomorphic and habitat unit, however, will be differentiated on other edaphic cri- teria in future habitat and vegetation studies. Geomorphic processes and the development of coastal dune morphology Whereas the approach adopted here to describe and name geomorphic units of the coastal dunes in the first instance has been based on non-genetic precepts, it is nonetheless worth- while to describe briefly the origin and genetic inter- relationships of aeolian landforms to provide an understanding of their temporal and spatial relationships. The importance of recognizing the genetic category to which a geomorphic unit belongs is that habitats can be broadly viewed in terms of their Table 3 Geomorphic units present at each scale of mapping Regional scale Large scale Medium scale Small scale • Shoreline dune ribbon • Foredunes • Foredunes • Hill or ridge slope • Barrier dunes • Parabolic dunes • Parabolic dunes • Linear depression crescentic crescentic • Cuspate beachridge plains attenuated attenuated • Circular depression fretted fretted • Cuspate forelands, or cusps • Chaots • Components of parabolic • Conical slope • Perched dunes low relief dunes • Crests medium relief arms high relief bowl inner face • Shore parallel ridges advancing face conical hill residual • Shore transverse ridges • Chaots • Blowout low relief medium relief • Undulating plain high relief • Shore parallel ridges continuous low relief continuous high relief disrupted low relief disrupted high relief • Shore transverse ridge ridge slope swale crest • Barchans • Barchanoid ridges • Transverse dune • Coppice dune • Flat • Blowout • Undulating plain 32 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. longevity, stability and dynamics which then can be related to in- terpretations about the stage of succession that a vegetation complex has achieved, recognizing that other processes such as fire also might influence vegetation succession. The variety of coastal dune landforms can be categorized into 4 genetic/process-related types: accretionary; erosional; mo- bile; and degraded. These types may be expressed at the large, medium or small scales. In addition the various types may be re- lated in time and space {eg an eroding dune supplies sand to a mobile dune which, once mobilized, uses the store of sand in its migration. There also may be smaller scale overprinting of one dune form on another. For example, a medium scale system of shore-parallel, continuous, low ridges (accretionary dune) may be overprinted by small scale blowouts (erosional form) to de- velop small scale shore- transverse parabolic dunes (mobile dune). Thus the original shore-parallel system of continuous low ridges is transformed into a system of shore-parallel disrupted low ridges. This sequence has been described previously, in part, by Ranwell (1972). The main large to medium scale geomorphic units occurring within the four genetic coastal dune landform categories are: • accretionary types: foredunes; coppice dunes; shore- parallel ridges (beachridges); chaots; • erosional types: blowouts; flats; • mobile types: parabolic dunes; barchans; barchanoid ridges; transverse dunes; • degraded types: chaots; undulating plain; shore- transverse ridges. Further accretion, erosion, degradration, or migration of these basic landform types results in the proliferation of medium and small scale geomorphic units. The relationships and evol- ution of one landform into another are diagrammatically illus- trated and described in Fig. 5. Figure 5 Inter-relationships between the various geomorphic units of the Quindalup Dunes, and the process by which one landform develops into another. 33 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Geomorphology and habitats of the Quindalup Dunes The coast adjoining the Swan Coastal Plain has been divided into five sectors by Searle & Semeniuk (1985). Each is dis- tinguished by a unique combination of modern onshore and off- shore geomorphology, coastal processes and Holocene sedi- ment accumulations. The sectors from south to north are: Geographe Bay Sector; Leschenault-Preston Sector; Cape Bouvard-Trigg Island Sector; Whitfords- Lancelin Sector; and Wedge Island-Dongara Sector. Each sector has a distinct array of Quindalup Dunes based on the criteria of: 1) total external geometry (eg beachridge plains, cuspate forelands or cusps, using the definition in Bates and Jackson (1980) that a cuspate foreland is the largest type of ’’cusp”); 2) internal array of landforms (linear depressions (swales) and ridges, parabolic dunes, blowouts); 3) dynamics of landform (prograding strandline, actively migrating parabolic dunes, vertically degrading ridges); and 4) relative relief. The Quindalup Dunes within each sector are described below in terms of both geomorphology and vegetation habitats. At the regional scale the different shapes of the Quindalup Dune ter- rain can be recognized as 1) shoreline dune ribbon, 2) barrier dune, 3) cuspate beachridge plains, 4) cuspate forelands, or cusps, 5) perched dunes. Representative maps showing distri- bution of typical landforms/habitats within each sector are pro- vided in Figs 6-11. A summary of the essential geomorphic fea- tures of the Quindalup Dunes in each sector is provided in Fig. 12. A summary of the main habitats encountered in each sector is p'fm/ided in Table 4. Table 4 Dominant habitats occurring in each of the sectors Sector Dominant regional scale units Dominant medium scale geomorphic units Dominant habitat types 1 E-W oriented, low, linear barrier of, and a N-E oriented shoreline ribbon of parabolic dune undulating plain shore-parallel ridges • undulating plain • linear depessions • ridge slopes, seaward aspect • ridge crests and landward 2 N-S oriented linear high barrier dune and northern section of shoreline ribbon of parabolic dunes attenuated parabolic dunes blowouts shore-transverse ridges • bowl • inner face • advancing face • crest • conical hill residuals • blowout floor • blowout wall • ridge slopes, north and south aspect • ridge crests • linear depressions 3 extensive cuspate low beach ridge plain attenuated parabolic dunes • bowl • inner face • advancing face • crest • conical hill residuals chaots • crests of chaots • slopes of chaots • circular depressions 4 large scale cuspate forelands (or cusps) and perched parabolic dunes fretted parabolic dunes • bowls • inner face • advancing face • crest • conical hill residuals 5 large scale cuspate forelands (or cusps) and perched parabolic dunes attenuated parabolic dunes • bowl • inner face • advancing face • crest • conical hill residuals fretted parabolic dunes • bowls • inner face • advancing face • crest • conical hill residuals shore-parallel ridge systems • linear depressions • ridge slopes, seaward & landward aspects • ridge crests 34 Journal of the Royal Society of Western Australia. Vol. 71, Parts 2 & 3. 1989. LEGEND Large and Medium Scale units FOREDUNE 0 O □ « 2 z * 2 « (/) H o < 1 a 1b 1 c Crescentic Attenuated Fretted AA \ A A / A_A o UJ < > o. c n LU LU CL O 0 Q 1 E <0 — 2a Low relief i- o 2b Medium relief 5 O 2c High relief Jo 4a 4b 3a 3b 3c Low continuous High continuous Low disrupted Bowl Inner face Advancing face Conical hill residual UNDIFFERENTIATED FLOOR WALL UNDULATING PLAIN FLAT COPPICE DUNE SHORE-TRANSVERSE RIDGE 3d High disrupted Small Scale units CREST Other units Undiff. Quindalup Dunes 5 o ° 3 o O O c HILL OR RIDGE SLOPE LINEAR DEPRESSION CIRCULAR DEPRESSION CONICAL SLOPE Undiff. coastal plain (Spearwood Dunes) Sandy beach Estuarine flats and wetlands Limestone Artificial drain \\] Modified terrain Figure 6 Legend to be read in conjunction with Figures 7-11. 35 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. Geographe Bay Sector Regional and Large Scale Geomorphology Geographe Bay is a broad, 100km wide, north-facing embay- ment at the southern end of the Rottnest Shelf. The Quindalup Dune zone is simple and consists, in the central portion of the sector, of a low barrier dune plain which forms a narrow band (average 500m wide), and in the northeast portion, of a shore- line ribbon comprised of parabolic dunes. Along the seaward edge there are beaches, beachridges and low foredunes. Long term Holocene sediment accretion has resulted in progradation of the shoreface along a broad front, generally maintaining the arcuate bay form and developing a beachridge ribbon periph- eral to the bay (Searle 1978, Searle & Semeniuk 1985). The de- velopment of successive beachridges during coastal progradation has impeded the natural drainage from the hinter- land, resulting in the development of elongate fresh to brackish lagoons and inlets in and behind the ridges. In the short term, geologically, the coast has undergone local realignment with de- velopment of small to medium scale accretionary cuspate shore- line and erosional scallops that alternate in time and space along a net progradational shore (Paul & Searle 1978). Medium and small scale geomorphology An undulating plain dominates the Quindalup Dunes in this sector, but there are also multiple sand ridges, each usually less than lm high, but up to 5-6m high, and 50m wide, and corre- sponding linear depressions (swales) and wetlands. There are no blowouts and parabolic dunes in the central portion of the sec- tor, and the Quindalup Dune morphology is essentially similar throughout this portion of the sector. Toward the northeast parabolic dunes and blowouts are present and increase in num- ber as the coast swings to a northerly alignment. Habitats The main habitats within the Quindalup Dunes are undulating plains with low and relict ridges, comprised of crests and slopes and depressions. All habitats are of relatively low relief and consequently depth to the water table is usually less than 3-5m. Humic soils are commonly developed on the aeolian landforms throughout this sector. Leschenault-Preston Sector Regional and Large Scale Geomorphology The Leschenault-Preston Sector extends about 80km in a northerly alignment. The Quindalup Dune zone is characterized by an extensive linear, narrow, barrier dune system (some 20- 30m high and 0.5-2. 0km wide) with its accompanying lagoons (Semeniuk & Meagher 1981b). The barrier is actively retrograding to the east by parabolic dunes and encroaching onto the lagoon environments, and its seaward face is generally undergoing net erosion (Semeniuk 1985). Staggered dune ad- vances in the past several thousand years to the present has re- sulted in an irregular encroachment of the barrier into the barred lagoon. The current coastal landforms of barrier dunes, segmented lagoons and inlets reflect this history. Medium and small scale geomorphology The barrier is dominated by blowouts and eastward migrating attenuated parabolic dunes with their accompanying arms and bowls. These are in various stages of geomorphic degradation and fixing by vegetation. There also are undulating plains (geomorphically degraded dunes), and shore-transverse ridges (= arms of former parabolic dunes). Along the seaward edge of this barrier there are beachridges, foredunes and cliffed dunes. Habitats The main habitats are those associated with parabolic dunes, blowouts and undulating plains. Much of the terrain is of high re- lief, situated well above the water table. Crests and slopes, de- pending on aspect, type of vegetation cover and stage of veg- etation succession, are covered either with minimal humic soil or with moderately developed humic soil. The undulating plains are situated within l-3m of the water table and are underlain by thick humic soil and a calcrete sheet. The dune landforms are of various ages and in various stages of geomorphic degradation, and accordingly support vegetation at different stages of succession. Cape Bouvard-Trigg Island Sector Regional and Large scale geomorphology The Cape Bouvard to Trigg Island Sector extends over 100km in a north to northwesterly alignment. This sector is characterized by complex nearshore bathymetry and discrete cells of Holocene sediment accretion reflecting net, long term, coastal progradation (Searle 1984, Woods & Searle 1983). The Quindalup Dunes mainly form an extensive low cuspate beachridge plain up to 10km wide. They are developed as a beachridge/dune cover to a Holocene sequence of seagrass sedimentary deposits that have extended from the hinterland towards an offshore limestone barrier. The beachridge/dune plain also has extended seaward, linking with emergent rem- nants of an offshore limestone ridge to form tombolos and cus- pate forelands. The resultant Quindalup Dune morphology is well marked by beachridge accretion lines showing successive shorelines. Intermittent erosion, or cessation in progradation, has developed localized blowouts and parabolic dunes which may appear along a specific former shoreline trend. At present, the five major bank and cuspate foreland structures within the sector represent various stages of an evolutionary process from a submarine lobe to fully-emergent cuspate plain stage (Searle 1984, Searle & Semeniuk 1985). Medium and small scale geomorphology Multiple parallel sand ridges, l-3m high and up to 50m wide, and associated depressions dominate the terrain. However, there also are local areas of blowouts and associated parabolic dunes (up to 20-30m high), bowls, wetlands, and residual coni- cal sand hills. The seaward zone contains low foredunes to steep foredunes, beachridges, coppice dunes and locally, cliffed dunes. Habitats The linear crests, slopes and depressions associated with the low parallel sand ridges (or beachridges) are the dominant habi- tats in this sector. These habitats are situated within 3-5m of the watertable. Humic soils are developed over the terrain to a mod- erate extent, and more particularly in depressions between the ridges. The occasional parabolic dunes in this area are habitats of high relief and the crests and slopes of these geomorphic units are situated >5m above the watertable, while flats and bowls have been developed to within lm of the watertable. Whitfords-Lancelin Sector Regional and Large scale geomorphology The Whitfords-Lancelin Sector extends 100km in a north- northwestly alignment. The coast consists largely of eroding rocky shores and pocket beaches interspersed with straight, beached coasts backed by high and perched dunes. Locally, iso- lated large scale dune-topped sandy promontories extend up to 800m seawards (c/ Semeniuk & Searle 1986). These cuspate forelands in the long term are either accretionary or slowly eroding. The Quindalup Dunes are restricted to 1) a thin ribbon 36 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. 37 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Figure 8 Geomorphic features of the Quindalup Dunes typical of Sector 2. Location of area shown in Fig. 1. Note that some units listed as small scale in the legend (Fig. 6) are also evident in the medium scale maps. along the rocky shore, 2) small systems of perched dunes (land- ward advancing parabolic dunes encroaching over the Spearwood Dune terrain) at sites of the large pocket beaches and 3) terrains of the large scale cuspate forelands wherein blowouts and parabolic dunes are dominant, or beachridge plains are present. Medium and small scale geomorphology The strandline is dominated by rocky shores, but pocket beaches have beach ridges, low foredunes and (locally) cliffed dunes. Dunes perched on the limestone hinterland tend to be lin- ear chaots or high relief ridges; they are parallel to the shore and tend to be high relief and well-vegetated. Inland, the perched dunes are either chaots or parabolic dunes. The parabolic dunes are the stabilized forms emanating from large blowouts, and they are uniformly oriented to the prevailing onshore winds. Locally, on the discrete large-scale cuspate forelands there is a complex system of overlapping and detached dunes with beachridges and swales, parabolic dunes, conical residual hills and wetlands. 38 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Figure 9 Geomorphic features of the Quindalup Dunes typical of Sector 3. Location of area shown in Fig. 1. A63378-3 39 Journal of the Royal Society of Western Australia. Vol. 71. Parts 2 & 3. 1989. Figure 10 Geomorphic features of the Quindalup Dunes typical of Sector 4. Location of area shown in Fig. 1. 40 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Habitats Parabolic dunes, chaots and blowouts dominate this sector, and the associated habitats of crests and slopes (with varying de- velopment of humic soils) are situated high above the water table and may overlie limestone with calcrete capstone. The wind-excavated bowl or flat areas are relatively closer to the watertable but are still elevated. In some places these deflation zones expose underlying limestone such that a limestone pave- ment habitat becomes exposed as an inlier within the parabolic dune system. The habitats of the large scale cuspate forelands are slopes, crests and swales of the beachridge plains, and are all situated low in relationship to the watertable. The slopes, crests and flats/swales/bowls associated with parabolic dunes and chaots, are all high relief structures situated >5m above the watertable. Wedge Island-Dongara Sector Regional and Large scale geomorphology The sector from Wedge Island to Dongara extends 180km in an approximately northerly alignment. The coastal geomorphology consists mainly of large erosional scallops into the Pleistocene limestone as well as discrete, accretionary, cus- pate forelands which are net-progradational forms (Woods 1983, Hesp 1984b). The rocky coastline is slowly eroding and contains numerous small pocket beaches, and is interspersed with straight to gently arcuate beached coasts. The dune ter- rains adjacent to the coast are perched dunes and form a discon- tinuous and irregular high relief ridge, or system of chaots. Markedly attenuated mobile dunes (oriented NNE) and old veg- etated parabolic dunes exhibit an alignment between about 010° and 030°. Elongate, roughly shore-parallel wetlands (la- goons or saline marshes) may be developed in the coastal dune terrain. The coastal morphology of this sector reflects the decline in shelter afforded by the offshore bathymetry and the increase in onshore development of NNE migrating dunes. In the northern parts of the sector, the coast is devoid of prominent cuspate forelands, which reflects the lack of offshore bathymetric fea- tures for protection. Medium and small scale geomorphology Markedly attenuated parabolic dunes dominate the terrain. Thus there are successive parabolic dune arms oriented- N to NNE. The bowl areas are vegetated or have developed into wetlands. On the accretionary cuspate forelands there are mul- tiple beachridges 3-5m high and up to 50-100m wide, with as- sociated swales. Interspersed through the beachridge plain there are blowouts and parabolic dune systems aligned along specific former shorelines. These contain parabolic dunes and bowls. The strandline has beaches, beachridges and low foredunes. Habitats Parabolic dunes and blowouts dominate this sector with the concommitant development of the habitats of crests, slopes and bowls. These habitats are situated 5-10m above the watertable with varying development of weakly humic soils. Also present are low beachridge plains with development of crests, slopes and swales situated within 5m of the water table. In local areas there are habitats of extensive limestone-like crust pavements. Discussion The regional variability of geomorphology of the Quindalup Dunes, its use in comparative vegetation studies, and its signifi- cance to conservation of landforms and habitats are discussed below. Regional variability of geomorphology Small scale habitats are determined by the geomorphic pro- cesses that are now operating or have operated in the region. Each sector is identifiable because of its distinct suite of coastal landforms and its location in the climate gradient, and it is axio- matic that habitats for vegetation will have developed as a result of those processes peculiar to a given sector. It is apparent, for instance, that the coastal evolutionary pro- cesses of Sectors 2,3 and 4 are markedly different. Sector 2 contains a retrograding barrier dune system and the entire ter- rain has developed by long term retrogradational dynamics. Sector 3 contains a progradational plain formed by long term net accretion of shoreline sediments to develop a successive series of parallel low beachridges. Sector 4 contains slowly eroding limestone shores and associated pocket beaches, together with the local sedimentary cuspate accumulations in the energy shadow of offshore islands/ reefs, and presents yet another suite of coastal landforms determined by the processes operating in that sector. Clearly, a variety of landforms exist across and along, the ex- tent of the Quindalup Dunes, and that each sector to some de- gree contains a suite of coastal landforms unique to that sector. Thus while there is a common thread in the occurrence of some landforms (eg beachridges, and foredunes), other specific landforms may occur exclusively or mainly only in a given sec- tor. However, even if a given landform may occur in several sec- tors, its dynamic morphology may vary regionally. For instance parabolic dunes appear virtually in every sector, but they are most abundant in Sectors 2, 4 and 5 and are most active only in Sectors 2 and 5. Sector 4 contains a predominance of fixed parabolic forms; Sector 5 contains a proportion of active and relict parabolic dunes, but they are incomparable to those in Sector 2 because they are highly attenuated and are associated with much more extensive bowls. In terms of landform complexity and heterogeneity, the Quindalup Dunes of Sectors 4 and 5 exhibit most variability. There are cusps, perched dunes, parabolics, blowouts, beachridge plains, chaots etc. Sector 1 exhibits least variability probably as a result of the relatively low energy progradational setting, as well as the less intense wind system (resulting in little or no aeolian remobilisation), and higher rainfall (resulting in more marked geomorphic degradation). Sector 3 with its beachridge plain and occasional lines of blowouts and parabolic dunes also exhibits minimal heterogeneity. Sector 4 exhibits moderate landform complexity with its ranges of blowouts, parabolic dunes, plains, shore- transverse ridges etc. Comparative vegetation studies The geomorphic framework presented above provides a ref- erence base for future studies of vegetation of the Quindalup Dunes. This framework should enable future workers to allocate vegetation associations (or vegetation complexes) to more specific geomorphic or habitat settings; the framework should also provide for more realistic comparisons between vegetation complexes of widely dispersed localities, and should enable trends in the variation of species distribution in the region, as re- flecting variation in the species pool, to be determined. The absence/presence of a given species can then be related either to the absence/presence of an appropriate habitat or to exter- nal factors such as climate. 41 Journal of the Royal Society of Western Australia. Vol. 71, Parts 2 & 3, 1989. LARGE SCALE SMALL SCALE Figure 11 Geomorphic features of the Quindalup Dunes typical of Sector 5. Location of area shown in Fig. 1. 42 Journal of the Royal Society of Western Australia. Vol. 71, Parts 2 & 3. 1989. Sector Domi nant Coastal Process Continuity of Coastal Dunes Dominant Land Form (plan) Wedge Island-Dongara Limestone coast retreat and landward dune encroachment alternating with areas of coastal progradation in cusps Discontinuous in discrete accumulations connected by thin ribbons Attenuated parabolic dunes Bowls Shore parallel ridges Whitford-Lancelin Limestone coast retreat and landward dune encroachment alternating with areas of coastal progradation in cusps Discontinuous in discrete accumulations connected by thin ribbons Parabolic dunes Chaots lit Cape Bouvard- Trigg Island Shoreline progradation and development of beachridge plain Continuous extensive cuspate plain Shore parallel ridges Parabolic dunes o Leschenault-Preston Barrier retreat and development of blowouts and parabolic dunes Continuous linear high-relief ridge Parabolic dunes O Blowouts Undulating plain Geographe Bay Shoreline progradation Continuous arcuate low-relief ribbon Undulating plain Shore el ridges Figure 12 Summary of key features of the Quindalup Dunes in each sector. In this figure the term cusp refers to the large scale coastal cusps ( = cuspate forelands). 43 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989 44 Journal of the Royal Society of Western Australia. Vol. 71, Parts 2 & 3, 1989. An example of a vegetation study in a specific geomorphic unit/habitat is provided here to illustrate the suggested manner in which comparative vegetation studies may be undertaken in a regional to subcontinental system. Large scale to medium scale geomorphic units such as parabolic dunes that are com- mon to a majority of coastal sectors were chosen as a basic unit. The vegetated parabolic dunes of Sectors 2, 3, 4 and 5 were studied. The parabolic dunes were subdivided into habitats of crest, slope, toe (= interface between slope and bowl/flat), bowl/flat, and the vegetation was described in terms of structur- ally or numerically dominant species and vegetation structure for each habitat. The results are presented in Fig. 13. These results are only preliminary but serve to illustrate that there are significant changes in structure and/or floristics of the Quindalup Dunes between a similar habitat setting within the same sector and from sector to sector. The conclusion under- scores the need to compare vegetation from similar geomorphic settings and habitats, when assessing the regional significance of flora. Conversely, it is also obvious that if adjoining sectors have markedly different suites of geomorphic units and habitats then the vegetation complexes of those adjoining sectors may be in- comparable, even if there are a number of species in common between the complexes. As a result some sectors may contain unique associations or assemblages of flora. This is not to imply that the components of the flora are rare or endangered but rather that the vegetation associations or complexes may be linked to a specific habitat which is not regionally widespread. This conclusion has implications for allocations of vegetation re- serves and in assessment of regional significance of flora in en- vironmental studies. Adequacy of reserves in the Quindalup Dunes Reserves for purposes of conservation are intended to pre- serve occurrences of rare and endangered flora and fauna, ex- amples of vegetation assemblages, examples of landscape ecol- ogy, areas of scientific interest such as geological features, areas for research and education purposes, or to provide sanctuaries and security for a range of fauna and flora (Frith 1973, Lunney & Recher 1979, Messer & Mosley 1980, McMichael 1980, Ovington 1980, Dept of Conservation & Environment 1983a, b, Anon 1982, Leigh et al 1984). Indeed the various reserves in southwestern Australia have been established for a range of the above reason-s. However, there is inadequate conservation of the variety of geomorphic, habitat and vegetation systems in the Quindalup Dunes. Where reserves are present in the Quindalup Dunes in the Perth metropolitan area there has been a tendency for undue emphasis in preserving the more seaward assemblages at the expense of the more landward assemblages. It is also clear that the regional array of landforms and veg- etation represents a wide spectrum of types and that a few re- serves, as presently distributed, cannot adequately cover and secure sufficient representaion of this variability. It would be preferable to preserve examples of each of the vegetation for- mations, and associated geomorphology and geological features within the Quindalup Dunes. The distribution of existing reserves in the Quindalup Dunes between Geographe Bay and Dongara is shown in Fig. 14. There are a number of other reserves in the Quindalup Dunes but these are for a range of purposes other than conservation of flora and fauna (such as recreation, camping, government re- quirements, explosive reserves). Only five reserves cover areas of Quindalup Dunes, viz Beekeepers-Mt Lesueur Reserve, Nambung National Park, Wanagarran Nature Reserve, the Trigg Island Reserve and Yalgorup National Park. Three of these are located in Sector 5, but of these only one, Nambung National Park, adequately covers a significant area of the Quindalup Dunes, but even here the park does not extend to incorporate the major accretionary beachridge plain cusp in the area. The Wcinagarran Nature Re- serve covers a terrain of perched dunes and an accretionary cusp with blowouts and parabolic dunes. The Beekeepers-Mt Lesueur Reserve covers a portion of perched Quindalup Dunes in Sector 5. The Trigg Island Reserve covers a relatively small part of the perched dune system of Sector 4. Most of the Yalgorup National Park is located on the geomorphic unit, underlain by limestone and wetland deposits, termed by McArthur & Bartle (1980b) the Yoongarillup Plain, and the Quindalup Dunes comprise only some 5 km* of the National Park. The essential features of the Quindalup Dunes that warrant conservation for each coastal sector throughout the southwestern coastal zone are listed in Table 5. 45 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. Table 5 Main natural features within each coastal sector and their conservation status Coastal sector Main natural features particular to a given sector Conservation status 1 Geographe Bay barrier dune and shoreline dune ribbon with plains and parabolic dunes, respectively none of the significant examples of this sector are conserved Leschenault-Preston barrier dune terrain composed predominantly of mobile and fixed parabolic dune systems main significant portion not conserved 2 ; Yalgorup National Park covers a small area of northern part which is not typical of this sector Cape Bouvard- Trigg Island cuspate beachridge plain composed of low relief shore-parallel sand ridges and intervening swales no features of this sector conserved Whitfords-Lancelin perched dunes and accretionary cusps composed of (fretted, crescentic, attenuated) parabolic dunes and chaots no examples of perched dunes and accretionary cusps are conserved Wedge Island-Dongara perched dunes composed of attenuated and fretted parabolic dunes, and accretionary cuspate forelands composed of low-relief shore-parallel sand ridges and intervening swales perched dunes secured in Nambung National Park; examples of cuspate forelands not conserved 1 Conservation status as at January 1987 2 Although there are plans to reserve the Leschenault Peninsula as an example of this sector these plans have yet to be formalized. This list of features illustrates the largely inadequate preser- vation of the variable Quindalup Dune systems. In many areas the major attributes that are specific or typical of a given sector are not reserved: • the shoreline ribbon of the Geographe Bay Sector • the barrier dune of the Leschenault-Preston Sector • the cuspate beachridge plain centred on Becher Point and Rockingham, of the Cape Bouvard-Trigg Island Sector • the perched dunes and accretionary cusps of the Whitford-Lancelin Sector • the beachridge plain cusps of the Wedge Island- Dongara Sector Hence there still is a need for conservation of areas of Quindalup Dunes and this should be based on their landform, scientific interest, representativeness, vegetation and relative lack of disturbance. It should be noted that there are still signifi- cant portions of the Quindalup Dunes that are listed as vacant crown land, land for government purposes or reserves for rec- reation etc., which could be revested or reallocated to become reserves for conservation of flora and fauna. Acknowledgements The manuscript was critically reviewed by Dr D K Glassford, I Leprovost and Dr P J Woods. Their help and discussions are gratefully acknowledged. References Anon 1982 A national conservation strategy for Australia. Dept Home Affairs & En- vironment. Aust Govt Publ Serv. Bates R L & Jackson J A 1980 Glossary of Geology (2nd ed) Arn Geol Inst, Falls Church. Virginia. Beard J S 1976 Vegetation Survey of Western Australia 1:1000000 series, sheet 6: Murchison. Univ W A Press. Beard J S 1981 Vegetation survey of Western Australia 1:1000000 series, sheet 7: Swan. Univ W A Press. Breed C S & Grow T 1979 Morphology and distribution of dunes in sand seas ob- served by remote sensing In: A study of global sand seas (ed E D McKee). U S Geol Surv Prof Paper 1052. 253-302. Bureau of Meteorology 1975 Climatic averages Western Australia. Aust Govt Publ Serv. Canberra. Bureau of Meteorology 1980 Evaporation. Aust Govt Publ Serv. Cooper W S 1967 Coastal dunes of California. Geol Soc Am Mem 104. Cresswell I D & Bridgewater P 1985 Dune vegetation of the Swan Coastal Plain, Western Australia. J R Soc W A 67: 137-148. Davies J L 1980 Geographic variation in coastal development (2nd ed). Longman. Dept of Conservation & Environment 1983a A conservation strategy for Western Australia. Dept Cons & Env Rept 12. Dept of Conservation & Environment 1983b The Darling System - System 6. Dept Cons & Env Rept 13. Dunham R J 1962 Classification of carbonate rocks according to depositional texture In: Classification of carbonate rocks: a symposium (ed W E Ham), Am Assoc Petr Geol Mem 1: 108-121. Frith H J 1973 Wildlife Conservation. Angus & Robertson. Gentilli J 1972 Australian Climatic Patterns. Nelson. Gentilli J & Fairbridge R W 1951 Physiographic diagram of Australia. Geographic Press, Columbia University, NY. Goldsmith V 1985 Coastal dunes In: Coastal Sedimentary Environments (ed R A Davis Jr). Springer-Verlag NY, 303-378. Hack J T 1941 Dunes of the Western Navajo County. Geol Rev 31: 240-263. 46 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Heddle E M 1979 Mapping the vegetation of the Perth Region In: Western Land- scapes (ed J Gentilli), Univ W A Press, 153-182. Heddle E M. Loneragan O W & Havel J J 1980. Vegetation complexes of the Darling System, Western Australia In: Atlas of Natural Resources Darling System Western Australia. Dept Cons & Env, Perth. 37-72. Hesp P A 1984a Foredune formation in southeastern Australia In: Coastal Geomorphology in Australia (ed B G Thom), Academic Press, Sydney, 69-97. Hesp P A 1984b The formation of sand ’’beachridges” and foredunes. Search 15: 289-291. Leigh J. Boden R & Briggs J 1984 Extinct and endangered plants of Australia. MacMillan. Lunney D & Recher H F 1979 National parks: A museum, a garden and an asylum In: A natural Legacy; Ecology in Australia (ed H.F. Recher, D. Lunney & I. Dunn), Pergamon, 184-199. McArthur W M & Bartle G A 1980a Landfords and Soils as a basis for Urban Planning in the Perth Metropolitan North-West Corridor, Western Australia. CSIRO Aust Div Land Resour Manag Ser No 5. McArthur W M & Bartle G A 1980b Soils and Land-Use Planning in the Mandurah- Bunbury Coastal Zone, Western Australia. CSIRO Aust Div Land Resour Manag Ser No 6. McArthur W M & Bettenay E 1960 The development and distribution of the soils of the Swan Coastal Plain, Western Australia. Soil Publ No 16, CSIRO, Melbourne. McKee E D 1979 Introduction to a study of global sand seas In: A study of global sand seas (ed E D McKee), U S Geol Surv Prof Paper 1052: 1-19. McKee E D 1982 Sedimentary structures in dunes of the Namib Desert, South West Africa. Geol Soc Am Spec Paper 188. McMichael D 1980. The Parks. An international perspective. In: The value of national parks to the community (ed J. Messer & G. Mosley), Proc 2nd National Wilderness Conf Univ Sydney Nov 1979. Aust Conserv Found, 35-43. Maingnet M 1984 A classification of dunes based on aeolian dynamics and the sand budget In: Deserts & arid lands (ed Farouk El-Baz), Nijhoff, The Hague, 31-58. Mabbut J A 1977 Desert landforms. ANU Press Canberra. Messer J & Mosley G (eds) 1980 The value of national parks to the community. Proc 2nd National Wilderness Conf Univ Sydney Nov 1979. Aust Conserv Found. Ovington D 1980 The Parks. An international perspective In: The value of national parks to the community (ed J Messer & G Mosley), Proc 2nd National Wil- derness Conf Univ Sydney Nov 1979. Aust Conserv Found, 45-56. Paul M J & Searle D J 1978 Shoreline movement, Geographe Bay Western Australia. 4th Aust Conf Coastal & Ocean Engineering Adelaide Nov 1978, 207-212. Playford P E, Cockbain A E & Low G H 1976 Geology of the Perth Basin, Western Australia. W Aust Geol Surv Bull 124. Ranwell D S 1972 Ecology of salt marshes and sand dunes. Chapman-Hall. London. Searle D J 1978 Sedimentation in Geographe Bay Western Australia. Hons Thesis, Dept Geology, Univ W A. Searle D J 1984 A sedimentation model of the Cape Bouvard to Trigg Island sector of the Rottnest Shelf, Western Australia. Ph D Thesis, Dept Geology, Univ W A. Searle D J & Semeniuk V 1985 The natural sectors of the inner Rottnest Shelf coast adjoining the Swan Coastal Plain. J R Soc W Aust 67: 116-136. Searle D J & Woods P 1986 Detailed documentation of Holocene sea-level record in the Perth region, South Western Australia. Quat Res 26: 299-308. Seddon G 1972 A sense of place. Univ W A Press. Semeniuk V 1983 The Quaternary stratigraphy and geological history of the Australind-Leschenault area. J R Soc W Aust 66: 71-83. Semeniuk V 1985 The age structure of a Holocene barrier dune system and its impli- cation for sealevel history reconstructions in southwestern Australia. Mar- ine Geol 67: 197-212. Semeniuk V 1986 Terminology for geomorphic units and habitats along the tropical coast of Western Australia. J R Soc W Aust 68: 53-79. Semeniuk V & Johnson D P 1982 Recent and Pleistocene beach/dune sequences, Western Australia. Sediment Geol 32: 301-328. Semeniuk V & Meagher T D 1981a Calcrete in Quaternary coastal dunes in Southwestern Australia: A capillary rise phenomenon associated with plants. J Sed. Petr. 51: 47-68. Semeniuk V & Meagher T D 1981b The geomorphology and surface processes of the Australind-Leschenault Inlet coastal area. J R Soc W Aust 64: 33-51. Semeniuk V & Searle D J 1985a Distribution of calcrete in Holocene coastal sand in relationship to climate, southwestern Australia. J Sed Petrol 55: 86-95. Semeniuk V & Searle D J 1985b The Becher Sand, a new stratigraphic unit for Holocene sequences of the Perth Basin. J R Soc W Aust 67: 109-115. Semeniuk V & Searle D J 1986 The Whitfords Cusp - its geomorphology, stratigra- phy and age structure. J R Soc W Aust 68: 29-36. Smith G G 1973 A guide to the coastal flora of south- western Australia. WA Naturalists Handbook No 10, Perth. Smith G G 1985 A guide to the coastal flora of south- western Australia (2nd ed). W A Naturalists Handbook No 10, Perth. Specht R L 1981 Foliage projective cover and standing biomass In: Vegetation classification in Australia (ed A N Gillison & D J Anderson), CSIRO & ANU Press, Canberra. 10-21. Speck N H 1952 The ecology of the metropolitan sector of the Swan Coastal Plain. MSc Thesis, Dept Botany, Univ WA Perth. Thom B G 1965 Late Quarternary coastal morphology of the Port Stevens-Myall Lakes area, NSW. J Proc R Soc NSW 98:23-36. Tinley K L 1985 Coastal dunes of South Africa. S Afr Nat Sci Programmes Rept 109. Woods P J 1983 Selecting a harbour site based on studies of coastal evolution and sedimentology at Jurien, Western Australia. 6th Aust Conf Coastal & Ocean Engineering, Gold Coast, July 1983. Woods P & Searle D J 1983 Radiocarbon Dating and Holocene History of the Becher/Rockinqham Ridqe Plain, West Coast, Western Australia. Search 14: 44-46. A63378-4 47 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989, 49-58 A field guide to the dung beetles (Scarabaeidae:Scarabaeinae and Aphodiinae) common in pastures in South-western Australia T J Ridsdill-Smith 1 , G P Hall 1 & T A Weir 2 ^SIRO Division of Entomology, Private Bag, PO Wembley WA 6014 2 CSIRO Division of Entomology, GPO Box 1700, Canberra ACT 2601 Manuscript received February 1988; accepted April 1988 Abstract A key to 4 native and 7 introduced species of dung beetles (Scarabaeidae:Scarabaeinae and Aphodiinae) common in pastures in south-western Australia is provided, together with notes on their distribution and biology. Scanning electron micrographs are given to assist in the separation of the species and distribution maps are provided. Introduction The species considered here are dung feeding beetles of the family Scarabaeidae (subfamilies Scarabaeinae and Aphodiinae). There are 19 species of Scarabaeinae and 7 species of Aphodiinae which are endemic to the south-western region of Australia, many of which are described by Matthews (1972, 1974), with new records by Ridsdill-Smith et al (1983 and unpubl data). The adults of all these endemic species are trapped in undisturbed vegetation (Ridsdill-Smith et al 1983), during the cool humid period of the year from May to September (Ridsdill-Smith & Hall 1984a). Four of the endemic Scarabaeinae are also trapped in pastures, but only Onthophagus ferox Harold is common (Ridsdill-Smith & Hall 1984b). Six species of cosmopolitan Aphodiinae are present in pastures, of which Aphodius pseudolividus Balthasar is most common (Snowball 1942, Ridsdill-Smith & Hall 1984b). To increase breakdown of cattle dung in pastures throughout Australia, CSIRO has been introducing species of Scarabaeinae from Africa and Europe. Over 500 releases of beetles of 14 species have been made in south-western Australia between 1972 and 1986, of which 9 species are known to be established by 1987. Seven species are common. Since 1978 these introduced species have become dominant members of the dung beetle fauna in pastures in south-western Australia. Because introduced dung beetles are now abundant in pastures, there is considerable interest in which species are present, their biology and seasonal abundance. In this field guide we summarize data we have collected over the past 10 years on the biology and distribution of species common in pastures. We have omitted both introduced and endemic species which are not commonly encountered. The distribution of some of these species may change in the future, and further introduced species may become abundant. A key is provided to separate the 4 native and 7 introduced species which are common in pastures in south-western Australia. Only adults are described and data on occurrence and abundance refer to the adult stage. Identification of species Description of characters Males and females of all species except A. pseudolividis can be distinguished by examining the ventral abdominal segments (Figs IB and C). In the males the segment before the pygidium is constricted in the mid-line, whereas in the female it is of even width. Horn size in horned beetles varies and some specimens may be worn or damaged, and thus a combination of characters should be used for identification. carina frontal horn fore tibia mid tibia elytron mid tarsus hind femur hind tibia terminal spur hind tarsus Figure 1 A Stylized dung beetle showing parts referred to in key. Underneath of abdomen showing method of sexing scarabaeine B males and C females. 49 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Glossary bidentate — with 2 teeth, or teeth-like processes (eg-clypeal margin in Fig 6H) bifid — forked, opening with a medial cleft (eg-horn in Fig 6D) carina — keel-like ridge (Fig 1A) clypeus — the antero-median part of the head (Fig 1A) elytral interval — space between lines (striae) on elytra (Fig 1 A) lamina — a plate-like projection (eg-Figs 6E,F) pubescent — covered with hair (eg- Fig 6C) pygidium — the terminal abdominal segment (Fig IB) rugose — with wrinkles and ridges (eg-clypeus in Fig 3E) Key to species 1 Pronotum green, brown or coppery; elytra brown -2 Pronotum black or with a bronze sheen; elytra black -7 2(1) Small, shining beetles, total length <5mm; pronotum blackish-brown medially with pale margins; elytra totally covering the pygidium; hind tibia with two terminal spurs (Figs 2A ,C)-Aphodius pseudoliuidus Balthasar Larger, less shining beetles, total length >7mm; pronotum variously coloured; elytra not covering the pygidium (Fig 2B); hind tibia with one terminal spur -3 3(2) Total length >12mm; fore tarsi absent; dorsal part of eyes wide (Figs 3E,F); males with 1 or 2 spines on the hind femora (Figs 3G,H) and fore tibiae elongate (Fig 3A); females with a carina on the clypeus (Figs 3E,F) -4 Total length <12mm; fore tarsi present; dorsal part of eyes narrow (Figs 2D-H); males without spines on hind femora, fore tibiae not elongate (Fig 3B); females without a carina on the clypeus (Figs 2F,H,3D) -5 4(3) Male with 1 outwardly facing spine on the posterior edge of the hind femora (Fig 3G); female with a squarish clypeus which has a carina equidistant between the eyes and the anterior edge (Fig 3E) -Onitis alexis Klug Male with 2 inwardly facing spines on the posterior edge of the hind femora (Fig 3H); female with a rounded clypeus which has a carina nearer to the anterior edge than to the eyes (Fig 3F) -Onitis aygulus Fabricius 5(3) Fringe of long hairs on apex of elytra restricted to area near suture; head of male with two carinae (Fig 2G); head of female lacking carinae (Fig 2H); pronotum lacking distinct ‘’speckled” markings -Euoniticellus fulvus (Goeze) Fringe of long hairs on apex of elytra extending along whole apical edge; head of male either with a single horn (Fig 3C) or with 3 carinae (2 in some small males) (Figs 2D,E); head of female with a carina level with front of eyes and another carina at the back of the head (Fig 3D); pronotum with distinct ‘’speckled”markings -6 6(5) Head of male with a single horn (Fig 3C); head of female with clypeus smooth and frontal carina of even height (Fig 3D); pronotal ‘’speckling” appearing light-brown -Euoniticellus intermedius (Reiche) Head of male with 3 carinae (2 in some small males) (Figs 2D,E); head of female with clypeus rugose or heavily punctured, not smooth and frontal carina raised in the centre (Fig 2F); pronotal ‘’speckling” dark-brown to black -Euoniticellus pallipes (Fabricius) 7(1) Male and female with similar head and pronotal armament; head with single horn, entire at apex; pronotum with two forward-projecting horns. Total length >10mm (Figs 5A-D) - Onthophagus ferox Harold Head and pronotal armament different to above; male head with either single horn bifid at apex, a lamina, two horns at the back of the head, or no horns; pronotum with projections or not. Total length clOmm -8 8(7) Elytral intervals with a median row of prominent, shiny, raised lines or beads (Figs 6B,E) -9 Elytral intervals flat, without sculpturing (Figs 4A,E) -10 9(8) Dorsal surface bronzed; pronotum without prominent projections, pubescent; head with some scattered hairs and clypeal margin feebly bidentate (Fig 6G); total length <8mm (Figs6A,B,C) -Onthophagus vermiculatus Frey Entirely black; pronotum with projections in both sexes, not pubescent; head with hairs only along front margin and clypeal margin medially more strongly bidentate (Fig 6H); male with either high bifid horn or two-pointed lamina on head; total length >8mm (Figs6D,E,F) -Onthophagus haagi Harold 10(8) Pronotum without lobes in both sexes; head of male either with a pair of curved horns arising between the eyes and extending back along sides of pronotum or with horns much reduced or replaced by a carina; head of female with a carina between the eyes (Figs 4A,B,C) -Onthophagus taurus (Schreber) Front of pronotum with a prominent median lobe, subquadrate or rounded; head of both male and female without horns, with a carina between the eyes (Figs4D-H) -Onthophagus binodis Thunberg Notes on the species Locations mentioned in the text are shown in Fig 8F. a) Aphodius pseudolividus Balthasar. (Figs 2A,C) A small shining beetle (4-5 mm long) which is elongate and brown. This accidentally-introduced species occurs widely in Australia and in other countries (P Hammond, pers comm). There are two patterns of seasonal abundance in south-western Australia. At sites north of Perth, beetles are common in all seasons and fly during the day. At sites south of Perth they are most common in summer and autumn (December to May) and fly at dusk. Some individuals are found throughout the year at all sites. Distribution: (Fig 7A) Widespread throughout the region, where it is common in pastures, but rare in undisturbed vegetation. b) Euoniticellus fulvus (Goeze). (Figs 2B,G,H) Brownish-yellow beetle (7-1 1mm long) with a plain brown pronotum. An introduced beetle from Europe. The strain released is from France. This species is abundant in summer and autumn. It breeds during summer and flies during the day. Distribution:(Fig 7B) Current records are from Bridgetown and near Bunbury. 50 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. Figure 2 Dung beetles. Elytra covers pygidium in A. A. pseudoliuidus. but not in B E fuluus. C A. pseudoliuidus: D & E E. pallipes males; F E. pallipes female; G E. fuluus male; H E. fuluus female. Scale lines: 1 mm. 51 Journal of the Royal Society of Western Australia. Vol. 71. Parts 2 &. 3. 1989. Figure 3 Dung beetles Foreleg of A Onitis sp. and B Euoniticellus sp.; C E. intermedius male; D E. intermedius female; E Head of O. alexis female; F Head of O aygulus female; G Hind femur of O. alexis male; H Hind femur of O. aygulus male. Scale lines; 1 mm. Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. Figure 4 Dung beetles. A & B O. taurus males; Scale lines: 1 mm. C O. taurus female; D, E & F O. binodis males; G&HO. binodis females. 53 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Figure 5 Dung beetles. A & B O. ferox males; C&DO. ferox females. Scale lines: 1 mm. c) Euoniticellus intermedius (Reiche). (Figs 3C,D) A brownish-yellow speckled beetle (7-10mm long) with a small horn on the head of the male. An introduced species from southern Africa. One of the first species to be released in the region in 1972. Adults are present all year round with peaks of abundance in winter and in summer, and breeds mainly at these times. Beetles fly during the day. E. intermedius is common in pastures near Moora, adjacent to sites of dung beetle studies in undisturbed vegetation (Ridsdill-Smith & Hall 1984a). Out of 19109 scarabaeine and aphodiine dung beetles trapped during 1982-1984 only 8 are E. intermedius. There is no evidence that they are competing with endemic beetles. Distribution:(Fig 7C) Mainly in the hotter parts of the region from Geraldton to Perth and eastward to Bruce Rock. This species is adapted to dry conditions. Adults can breed in very dry, sandy soil (Barkhouse & Ridsdill-Smith 1986). d) Euoniticellus pallipes (Fabricius). (Figs 2D,E,F) A brownish-yellow beetle (6-1 1mm long) with dark-brown to black speckling. An introduced species native to Europe and Asia. The strains released are from Iran and Turkey. Adults are most abundant in summer and autumn, and breed mainly from January to March. Low numbers are also seen in spring. Beetles fly during the day. Distribution: (Fig 7D) Southern half of the region, from Perth to Bruce Rock, but not present along the south coast. e) Onitis alexis Klug. (Figs 3E,G) A robust large beetle (12-19mm long) with greenish pronotum and brown elytra. An introduced species- from southern Africa. The strain released is the cold-adapted strain from summer rainfall regions. This species emerges in November and is abundant for a month; the second generation emerges in March. The species spends the winter as larvae in brood masses in the soil. Beetles fly at dusk. Distribution:(Fig 7E) Mainly in the warmer drier parts of the region from Geraldton to Pinjarra, and Perth to Cunderdin. f) Onitis aygulus Fabricius. (Figs 3F,H) The largest of the dung beetles present in pastures (18-23mm long). A coloured beetle with greenish pronotum and brown elytra. An introduced species from southern Africa. The strain released is a winter-rainfall strain. The biology of this species is very similar to that of Onitis alexis. Beetles fly at dusk. Distribution:(Fig 7F) Mainly in drier areas between Cunderdin and Williams, and from Pinjarra to Hyden. The distribution of this species tends to be more southerly than that of Onitis alexis. 54 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Figure 6 Dung beetles. A & B O. vermiculatus males; C O.vermiculatus female; D & E O. haagi males; F O.haagi female; G Head of O. vermiculatus male; H Head of O. haagi male. Scale lines: 1 mm. 55 Journal of the Royal Society of Western Australia. Vol. 71. Parts 2 & 3. 1989. Figure 7 A - F Distribution maps for Aphodius. Euoniticellus and Onitis spp. Full circles represent establishment and open circles represent releases where the species has not yet been recovered. 56 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Figure 8 A - E Distribution maps for Onthophagus spp. Full circles represent establishment and open releases where the species has not yet been recovered. F Locations mentioned in text. circles represent 57 Journal of the Royal Society of Western Australia. Vol. 71. Parts 2 & 3, 1989. g) Onthophagus binodis Thunberg. (Figs 4D-H) A matt black beetle (7-12mm long) with a lobe on the pronotum. An introduced beetle from southern Africa. Nearly all releases are of the winter rainfall strain. This species is very abundant in summer and autumn, and in lower numbers in winter. Mainly breed in spring but some breeding also occurs during the summer. Flies during the day. Out of 19109 scarabaeine and aphodiine dung beetles trapped in undisturbed vegetation over two years at six sites (Ridsdill-Smith & Hall 1984a) there are 12 0. binodis. There is no evidence that they are competing with endemic beetles. Distribution:(Fig 8A) Mainly in the higher rainfall areas along the coast from Moora to Esperance. Does not persist in drier areas. More abundant at sites with some summer moisture or irrigation. Adults cannot not breed in dry, sandy soil (Barkhouse & Ridsdill-Smith 1986). h) Onthophagus ferox Harold. (Fig 5) A large robust shining black beetle (12-20mm long) with one horn on its head and two on the pronotum. This is a native species. Adults are active during the cool humid period of the year (Ridsdill-Smith & Hall 1984a, b). To the north of Perth this is from May to September and to the south from May to December. It is most abundant in May and June and breeds in the spring. Adults fly at night and are commonly found at lights. It is rarely seen during the summer, although a few individuals are found at this time along the south coast, and some emerge after summer thunderstorms at Cunderdin. Distribution:(Fig 8B) Widespread throughout the south-west region. There are two unconfirmed records from Alice Springs. It is also found in undisturbed vegetation, where it is more common at jarrah forest and heath sites than in karri forest (Ridsdill-Smith et al 1983). i) Onthophagus haagi Harold. (Figs 6D,E,F,H) A black, shining beetle (8-10mm long) with a single horn or a lamina with two points on the head. This is a native species. Adults are active during the cool humid period from May to September. It is most abundant in May, June, August and September and breeds in the spring. Adults fly during the day. Adults are rarely found during the summer, except along the south coast. Distribution: (Fig 8C) Mainly in the higher rainfall areas along the coast from Perth to Bremer Bay. There is one unconfirmed record from Norseman. It is present also in undisturbed vegetation, where it occurs in jarrah forest and heath. It appears to be more common near swamps. j) Onthophagus taurus (Schreber). (Figs 4A,B,C) A shining black beetle (7-10mm long) with two long curving horns on the head of the male. This is an introduced species, and the strains released are from Greece, Spain, Italy and Turkey. It is most abundant in the summer and breeds in the spring. Adults fly during the day. Beetles are more abundant at sites with summer moisture or irrigation and breeding continues at these sites during the summer. Distribution: (Fig 8D) Moora to Margaret River, and from Pinjarra to Williams. k) Onthophagus vermiculatus Frey. (Figs 6A,B,C,G) A small black beetle (5-8mm long) with bronze reflections. This is a native species. Adults are present mainly during the cool humid period from March till November. Distribution:(Fig 8E) Perth to Albany in undisturbed vegetation including karri forest, jarrah forest and heath. It appears in pastures near the south coast at Albany. Acknowledgements We thank Alec Mahon, John Matthiessen, Lynne Hayles and officers of the WA Department of Agriculture for help in releasing beetles, and members of the CSIRO Dung Beetle Groups for collecting and rearing the beetles. The Elec- tron Microscopy Unit, CSIRO Division of Entomology, Canberra took the scanning electron micrographs and prepared the plates. References Barkhouse J & Ridsdill-Smith T J 1986 Effect of soil moisture on brood ball production by Onthophagus binodis Thunberg and Euoniticellus intermedius (Reiche) (Coleoptera: Scarabaeinae). J Aust ent Soc 25: 75-78. Matthews E G 1972 A revision of the Scarabaeine dung beetles of Australia. I. Tribe Onthophagini. Aust J Zool Suppl Ser No 9 Matthews E G 1974 A revision of the Scarabaeine dung beetles of Australia. II. Tribe Scarabaeini. Aust J Zool Suppl Ser No24. Ridsdill-Smith T J, Weir T A & Peck S B 1983 Dung beetles (Scarabaeidae: Scarabaeinae and Aphodiinae) active in forest habitats in southwestern Australia during winter. J Aust ent Soc 22: 307-309. Ridsdill-Smith T J & Hall G P 1984a Seasonal patterns of adult dung beetle activity in south-western Australia. In: Medecos IV Proc 4th Int Conf Mediterranean Ecosystems, (ed Dell B). Univ W Aust, Nedlands, 139-140. Ridsdill-Smith T J & Hall G P 1984b Beetles and mites attracted to fresh cattle dung in southwestern Australian pastures. CSIRO Aust Div Entomol Rep No 34: 1-29. Snowball G J 1942 A consideration of the insect populations associated with cow dung at Crawley, WA. J R Soc West Aust 28: 219-244. 58 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989, 59-67 The early evolution of the plant life of South-western Australia* J S Beard 6 Fraser Road, Applecross, WA 6153 Abstract A general review is given of the evolution of plant life in Australia, with special reference to south-western Australia where the evidence permits, from the beginning of life on Earth to the end of the Cretaceous, at which time recognizably modern floras had come into existence. Western Australia comprises the whole span of life as it has fossil stromatolites of the earliest known, 3 500 million years old, and living colonies of stromatolites at the present day. Only very primitive forms of life existed until the Cambrian, 570 my ago; land plants first appeared in the Upper Silurian (400 my) and became well established in the Devonian (395-345 my). An accurate impression of the plant life in Western Australia in the Lower Permian (270 my) can be formed due to the abundant remains in coal deposits when the Glossopteris flora of seed ferns was dominant. Development of higher forms of life continued into the Mesozoic Era (225-65 my) with rising temperatures and climates at first arid and then humid tropical. Cycads and cycad-like plants, and gymnosperms ancestral to modern southern conifers — kauri pines, podocarps and Araucarians — became dominant but were superseded by the flowering plants (Angiosperms) during the Cretaceous (135-65 my). Following a discussion of the origin of the Angiosperms, I suggest that colonization of Earth by plants was not immediate but gradual. At the outset, relatively primitive plants could grow only in wet places but gradually more difficult habitats could be colonized. It is argued that even so it was only with the coming of the Angiosperms with their greater evolutionary plasticity and ability to adapt to adverse habitats, that the entire surface of the earth could be colonized for the first time and a complete plant cover established. Introduction Much has been written and much discussion has taken place about the evolution of the plant life of SW Australia, encouraged by the numerous special characteristics which the present-day flora possesses, but this interest has been devoted largely to the later stages subsequent to the appearance of flowering plants during which the present-day flora came into being. This paper attempts to fill in the picture by an account of the early times during which plant life on Earth first appeared, adapted, and gradually colonized the earth’s surface. The history of the vegetation of the Australian region as a whole has been addressed in two recent books by Smith (1982) and White (1986) . The first comprises a collection of five papers prepared for the International Botanical Congress in Sydney. The second book gives the whole history of the evolution of Australian vegetation in a well illustrated book. The story told is not without its difficulties and inconsistencies, due to the uncertain evidence and conflicting theories. The formation of Earth is generally accepted to have taken place between 4 500 and 5 000 million years ago. Just how or when life on earth originated is not known, but fossil evidence of very primitive forms of life has been detected in Presidential address 1987, delivered 22 July 1987. rocks as old as 3 500 my. It is a strange fact that for the next 3 000 million years after that — an extremely long period — these forms of life underwent only a very slow development, and it was not until the geological period known as the Cambrian which began around 570 my ago that the evolution of life really got under way. This horizon was originally chosen for the starting point of the Cambrian period and also for that of the whole Phanerozoic Eon (which means the period of visible life) because it is the point at which readily identifiable fossils appear. It was originally thought that earlier rocks were devoid of evidence of life on earth. With subsequent work it is known that humble forms were in fact there and slowly evolving. None the less it appears that at the opening of the Cambrian period a critical threshold was crossed, beyond which abundant life began to be possible. Earth in Precambrian times was very different from today; the atmosphere was of very different composition, some say it consisted largely of methane and ammonia (Echlin 1966), others say water vapour, carbon dioxide, nitrogen and various sulphurous gases (Cloud 1968) . They agree that it contained little or no free oxygen. Oxygen is vital to the functioning of life, and it is interesting that the supply of oxygen on which life depends had to be built up gradually by the action of life itself. The process of photosynthesis by which plants support their life processes, drawing energy from sunlight to combine carbon dioxide and water into sugars and higher compounds, involves liberating an excess of oxygen from these molecules. 59 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. For a very long time most of the oxygen produced in this way would have been absorbed again in the oxidation of iron and other surface minerals, a stage that may have lasted 2 000 my, but later the photosynthesis of primitive organisms gradually increased the free oxygen content of the atmosphere. Prominent among the early forms of life were the Cyanobacteria or Blue-green “Algae” whose colonies trapped sand and silt and form those curious structures known as stromatolites which we can still see today at Shark Bay (Fig. 1) . Stromatolites have been detected in the early fossil record, and Western Australia can claim the oldest known deposit, at the locality “North Poie” in the Pilbara, 3 500 my old. By the beginning of the Cambrian it is supposed that the oxygen content of the atmosphere may have reached a critical level of perhaps 1% (whereas it is 20% today), permitting the evolution of life to progress more rapidly. In the words of Plumstead (1969): “When Phanerozoic time commenced 600 million years ago, the world stage was set for the great biological drama about to be enacted on its surface.” The Palaeozoic Era At first, the developing forms of life which become abundant in the fossil record are found virtually only in marine deposits, and another 150 million years had to pass, until the oxygen content of the atmosphere had increased still further to perhaps 10% , before indisputably terrestrial vascular plants appear. One reason for this may have been that the early atmosphere, with its lack of oxygen, possessed little or no ozone layer to buffer harmful incoming radiation, so that early forms of life could only have survived under a protective cover of water at least 5m deep (Plumstead 1969). It used to be tacitly assumed (eg Plumstead 1969) that as aquatic plants came first in the fossil record, land plants evolved from them by adaptation to sub-aerial conditions. Plumstead suggested that land plants evolved gradually at the margins of the sea or other large bodies of water, adapting the ability to survive short periods and even seasons of low water level, until complete adaptation occurred. Even then for a long time early land plants must have had to grow with their roots in water, as their tissues were not sufficiently evolved for efficient uptake and conductance of water, and they could not tolerate desiccation (Plumstead 1969) . More recently this has been questioned by Stebbins & Hill (1980) and others who suggest that large land plants evolved from unicellular soil algae independently of multicellular aquatic plants. If the new suggestion is true, the evolutionary process may have commenced in the Cambrian or earlier, and in fact some fossils found in the Middle Cambrian of Queensland have been claimed to represent land plants (Fleming & Rigby 1972). The occurrence of spores in the fossil record appears to document colonization of the land by non-vascular plants as early as the Ordovician (Gray 1985) , but it is not until the Late Silurian about 410 my ago that we find megafossils definitely interpretable as land plants. Whether they evolved from aquatic plants or independently, terrestrial plants can be distinguished by the morphological characters needed to fit them for sub- aerial life, and generally if fossil plants possess cuticle, stomata or lignified vascular tissue they are assumed to be terrestrial. During the Devonian period, 395 to 345 my, there was a rapid radiation of vascular plants, and it is possible to identify lycopods, ferns, pteridosperms (fern-like plants which were large woody trees and probably seed-plant ancestors), and Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. possibly even the earliest seed-plants themselves. It has been suggested that this was an “experimental” phase of evolution, a time when evolution ran riot as innovators flourished in the largely unoccupied ecospace (Runnegar 1982) . Except for the Angiosperms which did not appear in identifiable numbers until the Cretaceous, all the extant phyla of plants appeared together very early and were established if not during the Devonian, at least by the Carboniferous. All the same there is a definite progression in time of dominant plant forms, the simpler and more primitive being the more abundant in the early stages, and progressively more advanced forms becoming dominant later. The flora consists predominantly of relatively primitive forms such as lycopods in the Devonian, of ferns and seed- ferns in the Carboniferous and Permian, of gymnosperms by the Jurassic. The earliest period in Western Australia for which adequate data show what the landscape and its vegetation really looked like is the Permian, 280 to 225 my ago, for two reasons: there is an exceptionally large amount of fossil material, and the geological events of that time have left some traces still recognizable in the landscape. At the outset of the Permian, Australia still formed part of a great continental landmass, Gondwana, from which the separate continents of today would later one by one split off, and it occupied a polar position. To find the pole of those days one must resort to palaeomagnetic data which are confusing for Gondwana. However, a position for the Permo-Carboniferous South Pole was given by McElhinny & Embleton (1974) where readings from South America, Africa and Australia agree within a reasonable framework of confidence in placing it in the vicinity of Tasmania. Subsequent to the Permian the pole gradually moved away from Australia to its present position. During the Sakmarian stage of the Early Permian, in keeping with the continent’s polar position, the continental shield area of Western Australia was extensively glaciated. The sedimentary basins which surround it — the Perth, Carnarvon, Canning and Officer Basins — contain sediments dated to that time which are of glacial and fluvio-glacial origin. The widespread occurrence of these sediments indicates that the higher land was covered by a continental ice-sheet similar to that of Antarctica today, and the traces left indicate too that the country was extensively abraded by the ice. To form a picture of the landscape after the withdrawal of the ice-sheets one can refer to Quebec and Labrador and Finland, where there is country of comparable geological structure based upon ancient crystalline rocks, which has been quite recently glaciated. Such country is of comparatively low relief, scoured into a structure of boss and hollow, and with the hollows now occupied by strings of long narrow lakes showing the direction of ice movement. The nature of the plant cover in the early Permian of WA is shown to us by the fossiliferous coal beds which were laid down after the withdrawal of the ice, not only in the well known Collie Basin where the coal is mined commercially today, but in the Perth and Canning Basins also, in the eastern States and in other parts of Gondwana. The local plant macrofossils have been described by White (1961) and Rigby (1966) and demonstrate the presence of the “Glossopteris flora” which was widespread in the Southern Hemisphere at this period. An artist’s reconstruction of it was published as a frontispiece to Plumstead (1969). While this illustration is from a South African source, Africa and Australia formed part of the same continental land mass in those days and shared the same flora. The immense coal deposits laid down during the Permian are being exploited today both here at Collie and in the eastern States. The stereotype of coal swamps being steamy tropical swamp- jungles, as was the case in the northern hemisphere during the Carboniferous, is not applicable to southern Permian coals. These coal deposits are the product of cold, swampy bogs in which Horsetails grew in immense profusion like rushes. Ferns and seed-ferns, and probably mosses as in modern peat, as well as herbaceous lycopods like Selaginella formed a low, dense, swamp vegetation. Trees and shrubs of the Glossopterids, with special aeration tissues in their roots suited to the boggy conditions, grew in the swamps and in adjacent areas with high water tables (White 1986) . It is impossible to reconstruct accurately the vegetation of any past period because relative abundance of species, dominance and structural features of vegetation cannot be known. One can only depict some typical plants whose form is known from macrofossil remains. The Glossopteris flora has been so named because that form-genus is the commonest element. Seven “species” have been identified in Western Australia and constitute 25% of all known macrofossil taxa of that period. Glossopteris is a name originally attached to fossilized leaves thought (at first) to represent a fern . Further study has shown that these plants are actually a diverse group ranking taxonomically more at the level of an Order according to White (1986). Glossopterids as they should preferably be called were trees (Fig. 2) with large tongue-shaped leaves up to 40 cm long which were deciduous in the cold winters, so that great mats of them have been preserved where they fell and were stratified in the swamp. The wood had annual rings and secondary thickening, and resembled modern Araucaria. Roots had a segmented internal structure which probably had an aerating function. Most interesting of all, the leaves bore various kinds of fruiting bodies which bore seeds, and may well have been the ancestors of later groups of plants such as the southern conifers, the eye ads and even the angiosperms. Figure 2 A Glossopteris tree with details of leaves, inflorescences, stem and roots (from White 1986) . 61 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. The Lower Permian macrofossil record shows us vegetation of low-lying and swampy areas. What then was the vegetation of the much greater areas of upland? Did it exist at all? At that epoch most plants were still of a low grade of organization, eg mosses, hepatics, ferns, seedferns, whose modern descendants mostly inhabit wet places. Could they have colonized dry stony uplands in a cold post-glacial period? The answer is to be sought in high latitudes of the Antarctic today where we find communities of bryophytes, lichens and ferns. A number of different cryptogam tundra communities have been described from the maritime Antarctic by Gimingham & Lewis Smith (1970) composed in order of abundance of the following classes of organisms: lichens, mosses, algae, hepatics, fungi, bacteria, and a few flowering plants. Considerable areas of low-lying ground may become free of snow in summer, a large number of habitats are available for cryptogam colonization, and water is locally plentiful during the short growing season. It is likely that these conditions prevailed also on our uplands in Western Australia during the immediately post-glacial Permian, and that the land was therefore occupied by similar cryptogam communities, while the Glossopteris forest occupied sheltered and low-lying places. A similar juxtaposition prevails today in Tierra del Fuego, with subantarctic tundra communities on higher ground , side by side with dense rain forests of Nothofagus on protected sites. The point here is that the cryptogam vegetation is able to tolerate cold but not drought. It must have sufficient moisture during the growing season. The Mesozoic Era Following the Permian the ensuing era of the Mesozoic spanned 160 my and is divided into the Triassic, Jurassic and Cretaceous periods. The fragmentation of Gondwana took place then (Owen 1983) . The portion which is now Australia began to separate on the west from Greater India at the end of the Jurassic (Heirtzler et al 1973) and from Antarctica during the Cretaceous (Owen 1983). Australia moved steadily towards the equator and experienced a warming of climate, much greater than could be expected merely from the equatorial movement. In the Jurassic and Cretaceous the global picture is of very widespread tropical conditions associated with a progressive warming of the entire earth. The Jurassic period is found to have been already warmer than the present, and by the end of the Cretaceous temperatures had risen to a maximum which according to Hughes (1976) was probably the hottest experienced at any time since land plants first appeared. Subsequently during Cainozoic time global temperatures have been progressively falling away from this maximum until glacial phenomena have once more appeared (Kemp 1978) and the earth has regressed to conditions approaching those of the Permian period. Evidence for this is based primarily on the past temperature of the sea which is calculable for example by the well-known oxygen isotope determinations from belemnites which give apparently high values generally for Cretaceous seas and much higher temperatures than now in high latitudes (Lowenstam 1964). Hughes (1976) also cited the widespread carbonate sedimentation of the late Cretaceous which he attributed to high ocean temperatures, and pointed to the geographical distribution of certain types of pollen in the fossil record, eg Classopollis, as indicating that the equatorial belt was at least 80° of latitude wide (40° on each side of the equator) . Under such conditions warmer temperatures would have prevailed in high latitudes as well with absence of glaciation and of polar ice caps. At a number of localities in both hemispheres which were situated at high latitudes in the Cretaceous, often between 70° and 80° from the equator, well within the polar regions, the fossil record shows luxuriant forests of warm-temperate type, (eg West, Dawson & Hutchinson 1977 for the northern hemisphere, Douglas & Williams 1982 for the southern) and it is difficult to account for this unless the climate of the earth was radically different in late Cretaceous and early Tertiary time. Various causes can be suggested to account for the situation, the most plausible being either variation in the amount of energy received in the earth’s atmosphere, or variation in the tilt of the earth’s axis relative to the plane of the ecliptic. The principal difference between these two mechanisms would lie in the effect at high latitudes. With the obliquity as it is today and a higher solar output, it would be much warmer than now but there would still be seasonal variation in day length and continuous high illumination at high latitudes during the summer months only. With null obliquity there would be no seasonality, no summer or winter, no variation in day length but a low level of illumination, due to the low angle of the sun, during daylight hours throughout the year. Evidence from the late Cretaceous-early Tertiary vegetation of the Canadian high arctic (West, Dawson & Hutchinson 1977) supports the radiation hypothesis. The fauna, which included large land tortoises and alligators, indicates a frost-free climate, whereas the flora was deciduous and petrified wood shows well-marked annual rings indicative of seasonal growth. Furthermore there is an unusual leaf gigantism, inferred from modern studies to be a response to continuous photoperiod (Choi, Traverse & Hickey 1980) . These facts appear to disprove any variation in obliquity. Australian evidence from an early Cretaceous flora of Victoria which lay at more than 70° south at that time (Douglas & Williams 1982) shows a similar warm-temperate fauna and flora, although in this case leaf gigantism has not been reported. Axelrod (1984) showed that it would be quite possible for mesothermal plants and animals to tolerate the light conditions and seasonality prevailing at high latitudes at the present day provided much more warmth were available. Throughout the Mesozoic, therefore, climate was progressively changing, and at the same time the evolution of the Western Australian landscape continued, evolving from the glaciated landscape of the early Permian to a base-levelled landscape which we may call the Gondwana surface and which was perfected by the close of the Cretaceous. King (1972) recognized a series of erosional surfaces in Natal resulting from episodic continental uplift, the oldest of which is the Gondwana surface and is held to be a surviving portion of the surface of Gondwana as it was before rifting and separation. In South Africa the Gondwana surface has been uplifted to 3000m above sea level and most of it has been destroyed by erosion. In Western Australia it has been uplifted less than 300m and is visible as the Yilgarn Plateau, in the interior behind the Meckering Line, where it has remained substantially unmodified since at least the Eocene (Mulcahy 1967). The fossil record shows us something of the evolution of plant life in response to all these changes. Whether Western Australia remained sufficiently humid in the Triassic for conjectural tundra vegetation of the early Permian to survive, is not known. The likelihood is that it became extinct with the warming and drying of the climate. We cannot be certain whether anything replaced it, because the macrofossil record is probably not showing us the upland flora, while with pollen material we cannot be certain what it represents nor where it came from. White (1986:97) wrote confidently that by late Permian times the drier hillsides and places away from permanent water were habitats for early conifers, Ginkgos and cycad-ancestors, but there is little evidence for this assertion. Microfossils are evidence for the existence of this flora but not for where it grew. An ecologist must examine the likely capabilities of the contemporary plants before making a guess 62 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. at their habitat and in this case it seems that they may well have been confined to river plains and other damp places upstream, and did not necessarily grow on hillsides. Retallack (1977) dealt with the Triassic vegetation of eastern Australia. There is much more fossil material than in the West and Retallack adopted a new approach, treating the eleven different fossil plant assemblages which he found as plant associations in the modern sense, giving them names in Braun- Blanquet terminology, eg Dicroidietum odontopteroideum, and inferred habitats. Much of this may appear frankly far- fetched and fanciful. Seed-ferns such as Dicroidium (Fig. 3) Figure 3 Reconstruction of Dicroidium, leaf and inflorescences (White 1986). were dominant in most of the eleven associations, but one was a coniferous forest dominated by Voltziopsis a podocarpaceous conifer resembling modem Dacrydium (not to be confused with Dicroidium) . Retallack took the view like so many other palaeo- botanists that because terrestrial flora existed it necessarily occupied all habitats and formed a universal plant cover. I find this view hard to support. Retallack confidently drew several landscape diagrams showing the distribution of his hypothetical associations, making them extend to the tops of the hills, and asserting equally confidently that they grew up to 200 km inland from the coast of that time (Fig. 4) . White (1986) says of this period that the conifers “grew on dry hillsides, remote from water”, but how does she know? Careful reading of Retallack’s paper shows no evidence that any association grew on a site where water was not plentiful, and I regard the existence of an upland flora as not proven. During the ensuing Jurassic period in Western Australia the flora comprised as before lycopods, horsetails, ferns and seed- ferns, but dominance was assumed by gymnosperms. Many of these were cycad-like plants such as Pentoxylon, (Fig. 5) and others were coniferous trees belonging to those groups of southern conifers which still survive today, the Podocarps, Araucarias and Kauri Pines (Agathis), as well as Ginkgo of which a single species survives in China. Here for the first time we begin to find a vegetation which is recognizable in modern terms. A beautifully preserved Jurassic flora is found in the Talbragar Fish Beds of northern New South Wales which were laid down in an ancient lake and show us a forest of kauri pine interspersed with podocarps, and with Pentoxylon in the understorey. This is particularly interesting because relict Agathis forests on the Atherton Tabeland have Podocarps growing among the Kauri Pines and a tree-like cycad, Lepidozamia hopei, is present in the understorey. Thus a modern assemblage of plants exists with the same basic composition (though at a more evolved stage) . White (1986) claimed from this that “we know exactly what the Jurassic vegetation looked like”. At least, we know what some Jurassic vegetation looked like. The Talbragar flora is likely to have grown under cool, constantly humid conditions as its modern counterpart does and cannot be assumed to have formed a universal cover extending over less favourable habitats. This gymnosperm-dominated flora persisted through to the middle Cretaceous after which it was superseded gradually by a flora of angiosperms (that is, of flowering plants) of modern type and the earlier forms became extinct. This radical biological change was accompanied apparently by an increase in 'the number and variety of insects and birds, and their co-adaptation as pollinators with the developing angiospermous plants, by the radiation of mammals and the extinction of dinosaurs, not with precise contemporaneity but within the 70-million year timespan of the Cretaceous period. This period therefore witnessed a biological revolution as profound and important as that of the Cambrian when higher forms of life “suddenly” radiated in the sea, and transcending that of the Devonian when the first land plants evolved. The appearance and early development of the angiosperms occurred from the Barremian to the Cenomanian stages of the Cretaceous and is usually described as “sudden”, but actually spanned about 20 million years. The flowering plants or angiosperms possess a whole range of well-marked characters readily recognizable even in fossil form which distinguish them from their precursors. In addition to their entirely new and distinctive reproductive system, they possess large reticulate-veined leaves, wood with vessels, and distinctive pollen. The earliest of this pollen, in an early and simple form, appears in the Barremian and radiates into more common and complex types by the Cenomanian (Wolfe, Doyle & Page 1975, Fig. 6) . The earliest known fruits and seeds are Barremian, while leaves become generally recognizable in the Albian. The earliest known angiosperm remains are found in western Gondwana, and we have no records within Australia until the Albian . The origin of these angiosperms is unknown. They appeared relatively suddenly and apparently already fully developed, leading to the postulate that angiosperms underwent a long previous period of development which failed to register in the fossil record. The Russian botanist Takhtajan discussed the problem in the English version of his book (1969). Accepting that there is no factual basis for attempts to derive the angiosperms from ferns or other lower forms without a transitional gymnospermous stage, he proceeded to examine all the known 63 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3. 1989. Figure 4 Reconstruction of Triassic vegetation cover at a site in present north-western New South Wales, after Retallack (1977). Communities shown are Phoenicopsetum levee scrub (dashed shading and conical pattern), Dicroidietum flood plain forest (rounded trees and undulating pattern) and Johnstonietum mallee-like woodland (serrated pattern on hillsides). This reconstruction implies total vegetation cover of the land area. groups of Jurassic gymnosperms, looking for indications as to which might be candidates for angiosperm ancestry. In a sense this approach was unproductive, since it showed that all of them in one way or another, mainly in wood anatomy and the reproductive organs, possessed characters already too advanced along their own evolutionary line, so that they could not have evolved into angiosperm characters. Takhtajan eventually concluded that the angiosperms and their nearest Mesozoic gymnosperm relatives, the Bennettitales, probably had a common seed-fern ancestor as far back as the Carboniferous, although no trace of such an ancestor had been found. Stebbins (1976) wrote that “all we can say at present is that towards the end of the Palaeozoic era the evolutionary line leading to the angiosperms entered a dark tunnel of ignorance . . . and remained there until the angiosperms emerged, fully differentiated, in the early part of the Cretaceous period.” In the 1960s when Edna Plumstead in South Africa was engaged in her pioneer work on the Glossopterids, she suggested that these may have been among the missing angiosperm ancestors. Takhtajan’s conclusion would support this but the idea found little favour among most botanists, who were wedded to the concept that the Ranales, with Magnolia- like flowers, were the sole ancestral type. Melville of Kew supported Plumstead from the beginning (Melville 1970) and recently (1983) came out strongly in favour, citing morphological evidence which he contended indicates that Glossopterids and Angiosperms “belong to one lineage which extends back in time for 300 million years”. Recently White (1986) has given added support to Glossopterid ancestry. While of course the angiosperms must have had ancestors, we cannot satisfactorily identify them in the present state of our knowledge. It seems useful however to discuss further the why and the where of their origin, as it seems to me that certain points have been overlooked. Figure 5 Pentoxylon, a cycad-like plant of the Jurassic. The name is based on the division of the trunk into five stelae which have secondary thickening unlike modern cycads (White 1986). 64 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3. 1989. Stage o to -p a) M V & a u cO ti o Cenomanian Albian Upper Middle Lower Aptian Barremian Figure 6 Evolution and radiation of Angiosperm pollen forms during the Cretaceous (after Wolfe, Doyle & Page 1975). Authors dealing with pre-Cretaceous origins usually follow the “upland origin” hypothesis, whereby angiosperms evolved in rolling hilly terrain and the lower slopes of mountains in the ancient humid tropics. Because these sites were far from the depositional basins, the opportunity for material to become preserved in fossil form would be extremely low except as pollen. Various authors too have speculated as to what the earliest angiosperms were like and what habitats they occupied. On the basis of fossil evidence, the earliest angiosperms may have been small woody evergreen plants with simple, entire, pinnately veined leaves, their flowers pollinated by insects and their seeds distributed by animals (Takhtajan 1969). They may have originated in seasonally arid climates, migrating to more mesic regions later (Hickey & Doyle 1977) initially as riparian weeds or colonists of talus slopes, streambanks and other unstable habitats (Stebbins 1976). Hughes (1973) suggested that mangrove environments may have been critical to their dispersal and early succession as they moved to upland sites and disturbed areas along edges of stream valleys and coasts. These ideas run counter to the traditional concept that early angiosperms were mesophytic tropical trees with pinnate leaves and clusters of large arillate follicles as their fruiting mechanism. However these new ideas are attractive. Angiosperm colonization of the rain forest would have been secondary, and come later, perhaps at the stage when we begin to detect them in the fossil record. All of the new hypotheses have in common the concept of initial evolution to fill a vacant niche or niches which would comprise habitats too adverse in one way or another for the more simply organized earlier phyla of plants to tolerate . It seems to me that the logical conclusion of this approach is that it was only with the coming of the angiosperms that something approaching a universal plant cover of the continents was achieved. Even today it is not universal for plants do not grow where the ground is frozen for all or most of the year, or in deserts where too little moisture is available. Before the Cretaceous the range of unoccupiable habitats must have been much greater. Hughes (1976) is the sole author I have come across who has not assumed that the earth essentially was completely colonized by early land plants in the Palaeozoic and Mesozoic. Hughes observed that if the time from the Devonian period to the present day has recorded progressive degrees of colonization of the land by plants, it follows that many possible habitats now filled would not have been filled in early Cretaceous time. 65 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. Support is given to this from a table given by Hughes showing the number of vascular plant taxa existing at various selected moments of geological time. The great diversity of form in the modern flora is expressed in the huge total of 300 000 living species, the great majority being angiosperms. The table shows however that the ferns and gymnosperms have also increased in number since the Cretaceous, having undergone a secondary adaptation to Cainozoic conditions mostly in forms different from those of the Mesozoic. The lower levels of the table give the number of taxa known from the fossil record and it is astonishing to find only 500 taxa of plants in the entire world at the end of the Carboniferous, and 2 500 in the Jurassic. Of course the record is distorted by the selectivity of fossilization, but if the numbers were multiplied by a factor of 10 they would still be astonishingly small. I am convinced by this that the colonization of the land must have been a long and slow process, not completed before the coming of the angiosperms. I envisage the first small delicate land plants of the Silurian, confined to permanently wet places, perhaps the edges of lakes and ponds. Then in the Devonian and Carboniferous plants increase in size and occupy the deltaic habitats including coal swamps which the fossil record shows us. In the Permian in the southern hemisphere extensive coal swamps were formed and perhaps upland cryptogam communities in a sub- Antarctic climate. In the Triassic and Jurassic as the gymnosperms came to prominence they would first colonize river flood plains and lowlying ground. In the Triassic it is not uncommon to find petrified wood representing rafted tree trunks but these can only have come from the river- bank vegetation ripped away by floods. Trees growing on hills do not get washed down rivers, so that rafted logs are not evidence for upland forests. By the Jurassic, however, gymnosperm forests may have established themselves on favourable upland sites under humid climates, but there would still remain the unfavourable sites, the shallow rocky soils, the steep hillsides, and of course the less humid, hot and dry climates, all of which offered vacant ecospace for the later radiation of angiosperms. It seems very unlikely that a mere 500 taxa of plants — or even 5 000 — known world-wide in the Carboniferous provided a complete cover for the uplands of the earth. 2 500 taxa or even 25 000 in the Jurassic can only represent a partial cover. The ecological amplitude required of those early taxa would have been far too great. Diverse habitats must have existed then as now, and required diverse adaptations to populate them. A feature of the angiosperms, and the very basis of their success, is their immense evolutionary plasticity. They evidently radiated quickly in the Cretaceous into a multitude of new and specialized habitats. Ecological diversity is the basis for the great number of angiosperm species in the world today. This plasticity far exceeds that of the ferns and gymnosperms as we can see from their relative numbers at the present time. Pre-angiosperm floras evolved slowly and lacked capacity to adapt to adverse habits. This thesis is compatible with the conclusions of others that early angiosperms were small- leaved shrubs or small woody trees originating in dry climates. If “under adverse conditions” is substituted for “in dry climates”, these are just the type of plants we should expect to be capable of first colonizing the previously empty spaces of the earth. Further supporting evidence for this comes from the global geological change to predominantly carbonate sedimentation which occurred in the second half of the Cretaceous. Whereas previously sediments had been mostly of erosional origin — sandstones, siltstones and claystones — now they become mostly depositional, i e limestones, which accumulated from the settling out of calcium carbonate in the skeletons of marine organisms. Hughes (1976) took account of this phenomenon but attributed it to the high temperature of the sea during his “Radmax” period. While this may certainly account for the high quality of the limestone in chalk deposits of that time, we must remember that during the subsequent Cainozoic era carbonate sedimentation has continued to predominate down to our own day even though global sea temperatures have been falling steadily. Other causes have been suggested such as the development of world oceans in the Cretaceous as Pangaea broke up, with their thermohaline circulations and concomitant evolution of calcareous plankton. Such no doubt were contributory but carbonate sedimentation may have been favoured by a reduction in sediment load, resulting in clearer water off shore, and this was the result of establishment for the first time of a general protective plant cover over the uplands. I suggest that by the close of the Mesozoic a general plant cover was more or less established for the first time, by an already ecologically diverse angiosperm flora which had radiated rapidly into a variety of habitats. This flora occupied not only the lowlands, mingling with the Jurassic gymnosperms, but a variety of upland habitats as well. By the close of the Cretaceous, 65 my ago, there was a fairly general angiospermous vegetation diversified in form from scrub to rain forest, growing in a warm, non-seasonal, humid climate. Since the Western Shield of Australia lay above sea level during the Cretaceous, there is little fossil evidence of the vegetation and we have to rely heavily on data from eastern States sites, supported by local fossil evidence of slightly later date from the early Tertiary. These sources show us a flora whose modern counterparts occur in rain forests of the highlands of New Guinea and New Caledonia, with related types in New Zealand and Tasmania. We can recognize something similar to today’s forests of southern beech (Nothofagus) and of southern conifers such as Dacrydium or mixtures of these, and with other living genera of sub-tropical trees including numerous Proteaceae. Table 1 Estimated number of seed-plant and pteridophyte species taxa existing at the selected instants of time (after Hughes 1976) Selected instant Time (my) Gymno- sperm Pterido- phyte Angio- sperm Approximate total Recent 0 640 10 000 286 000 300 000 End Cretaceous 65 500 2 000 20 000 22 500 Beginning Cretaceous 135 1 500 1 500 0 3 000 Mid Jurassic (end-Bajocian) 170 1 500 1 000 0 2 500 Late Carboniferous (end-Westphalian) 300 200 300 0 500 66 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Acknowledgements I thank Prof B E Balme, Dr A E Cockbain, Dr K J McNamara and Dr G Kendrick for critically reading the manuscript and for their helpful suggestions r and Ms M E White for permitting me to use four illustrations from her book (White 1986). References Axelrod D I 1970 Mesozoic palaeogeography and early Angiosperm history. Bot Rev 36:277-319. Axelrod D 1 1984 An interpretation of Cretaceous and Tertiary biota in polar regions. Palaeogeogr, Palaeoclimatol, Palaeoecol 45:105-147. Choi D K Traverse A & Hickey L J 1980 Preliminary report on the paleo-palynology of the Eureka Sound formation, Ellesmere and Axel Heiberg Islands, Canadian Arctic. Amer Assoc Strat Paly, Progr and Abstracts 7. Keystones, Colorado. Cloud P E 1968 Atmospheric and hydrospheric evolution on the primitive earth. Science 160:729-736. Douglas J G & Williams G E 1982 Southern polar forests: the Early Cretaceous floras of Victoria and their paleoclimatic significance. Palaeogeogr, Palaeoclimatol, Palaeoecol 39:171-185. Echlin P 1966 Origins of photosynthesis. Sci Journ 2:42-47. Fleming P J G & Rigby J F 1972 Possible land plants from the Middle Cambrian, Queensland. Nature 238:266. Gimingham C H & Lewis-Smith R I 1970 Bryophyte and lichen communities in the maritime Antarctic. In Antarctic Ecology (ed M W Holdgate) vol 2: 752-785. Academic Press, London. Gray Jane 1985 The microfossil record of early land plants. Phil Trans Roy Soc Lond B309: 167-195. Heirtzler J R et al 1973 Age of the floor of the eastern Indian Ocean. Science 180:952-4. Hickey L J & Doyle J A 1977 Early Cretaceous fossil evidence for Angiosperm evolution. Bot Rev 43:3-104. Hughes N F 1973 Mesozoic and Tertiary distributions and problems of landplant evolution. In Organisms and continents through time (ed N F Hughes). Special Paper in Paleontology 12:189-98. Hughes N F 1976 Palaeobiology of Angiosperm origins — problems of Mesozoic seed-plant evolution. Cambridge Univ Press. Kemp E M 1978 Tertiary climatic evolution and vegetation in the SE Indian Ocean region. Palaeogeog/Palaeoclim/Palaeoecol 24:169-208. King L 1972 The Natal Monocline. Univ Natal, Durban. Lange R T 1982 Australian Tertiary vegetation. In: A History of Australasian Vegetation (ed J M B Smith), 44-89, McGraw-Hill, Sydney. Lowenstam H A 1964 Palaeotemperatures of the Permian and Cretaceous periods. In: Problems in Palaeoclimatology (ed A E M Nairn), Interscience New York. McElhinny M W & Embleton B J J 1974 Australian palaeo-magnetism and the Phanerozoic plate tectonics of Eastern Gondwanaland. Tectonophysics 22:1-29. Melville R 1970 Links between the Glossopteridae and the Angiosperms. Proc 2nd Gondwana Symposium CSIR South Africa, 585-588. Melville R 1983 Remoration: an overlooked process in Angiosperm evolution. Kew Bull 37:613-632. Mulcahy M J 1967 Landscapes, laterites and soils in southwestern Australia. In Landform Studies in Australia and New Guinea (eds J N Jennings & J A Mabbutt), 211-230, ANU Press Canberra. Owen H G 1983 Atlas of Continental Displacement 200 my to Present. Cambridge Univ Press. Plumstead E P 1966 The story of South Africa’s coal. Optima 16:186-202. Plumstead E P 1969 Three thousand million years of plant life in Africa J Geol Soc S Afr Annexure to Vol 72:1-72. Retallack G J 1977 Reconstructing Triassic vegetation of eastern Australasia; a new approach to the biostratigraphy of Gondwanaland. Alcheringa 1:247-277. Rigby J F 1966 The Lower Gondwana floras of the Perth and Collie Basins, W A. Palaeontographica 1218 B: 113-152. Runnegar B 1982 The Cambrian explosion: animals or fossils? J Geol Soc Aust 29:395-411. Smith J M B 1982 (ed) A History of Australasian Vegetation. McGraw-Hill, Sydney. Stebbins G L 1976 Seeds, seedlings and the origin of Angiosperms. In Origin and Early Evolution of Angiosperms (ed C B Beck), 300-311, Columbia Univ Press. Stebbins G L & Hill G J C 1980 Did multicellular plants invade the land? Amer Nat 165:342-353. Takhtajan A 1969 Flowering Plants: Origin and Dispersal. Oliver & Boyd Ltd, Edinburgh. West R Dawson M & Hutchinson J 1977 Fossils from the Paleogene Eureka Sound Formation, N W T Canada: occurrence, climate and paleogeographic implications. In: Palaentology and Plate Tectonics, (ed R M West), Milwaukee Mus Spec Publ Biol Geol 2:77-93. White M E 1961 Plant fossils from the Canning Basin W A. In: Geology of the Canning Basin (ed J J Veevers & A T Wells), Bur Min Res Bull 60:291-320. White M E 1986 The Greening of Gondwana. Reed Books Sydney. Wolfe J A Doyle J A & Page V M 1975 Paleobotany. In: The Bases of Angiosperm Phylogeny (ed JW Walker), Ann Miss Bot Gdn 62:801-824. 67 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989, 69-75 Spatial variation in fish communities in two South-western Australian river systems B J Pusey, A W Storey 1 , P M Davies & D H D Edward Aquatic Research Laboratory, Department of Zoology, The University of Western Australia, Nedlands, WA 6009 Manuscript received March 1988; accepted June 1988 Abstract The fish fauna of two river systems, the Canning and North Dandalup catchments, were sampled every 3 months over an 18 month period. Similar numbers of species and individuals were recorded from both systems. The majority of headwater streams were temporary, predominantly colonized by the native minnow, Galaxias occidentals . Physical obstruction to the seasonal migration of this species by both natural and man- made barriers was evident. At lowland sites an exotic, Gambusia affinis, was dominant in the Canning River, and was the second most abundant species in the North Dandalup River. The distribution of G. affinis is discussed in relation to streamflow regulation, disturbance and habitat requirements. Introduction The freshwater piscifauna of South-western Australia is considered a distinct element of the Australian fish fauna and has been described as depauperate, with a high degree of endemism (Whitley 1947, Allen 1982, Merrick & Schmida 1984). The fauna is represented by eight families (one monotypic and endemic) containing 12 genera and 17 species. Nine species and six genera, four of which are monotypic, are endemic to the State. Within the state, regional patterns of endemism are also apparent for some species (Christensen 1982). A number of exotic species are present, most notably the mosquito fish, Gambusia affinis (Bird & Girard). Despite the low number of species, little is known of their biology, apart from the above gross patterns of distribution. In addition, little is known of the impact of introduced species. G. affinis is widespread throughout the South-west of Australia (Allen 1982) and much of the Australian continent (Merrick & Schmida 1984). This species has been implicated in the elimination of native species from many systems (Myers 1975, Mees 1977, Sarti & Allen 1978, Arthington et al 1983), yet little is known of the interactions between this and native species in South-western Australia. This paper describes patterns in the distribution of the fish fauna of two river systems, the Canning and the North Dandalup and formed part of an extensive biological monitoring programme for environmental impact assessment and water quality using fish and macroinvertebrates. 1 Correspondence Study areas The location of sampling sites within each river system is illustrated in Figure 1. The headwaters of both rivers are situated in forested regions (Eucalyptus marginata and E. calophylla) of the Darling Scarp, the western edge of the Great Plateau of Western Australia (Jutson 1950). In the Canning catchment these streams are intermittent while in the North Dandalup they are more permanent. This is related to the higher annual rainfall and the presence of swamps on the headwaters of the latter system. The climate of the area is mediterranean (Seddon 1972) with predictable patterns of rainfall and stream discharge (Bunn et al 1986) . Both river systems are regulated, resulting in reduced summer flow at the majority of lowland sites. On the Canning River a large dam (built c 1932) is situated 2km upstream of site LC1. Below this dam a number of short tributaries, one of which is Stinton Creek, arise from adjoining sub-catchments. On the North Dandalup River a pipehead dam, which overflows each winter, is situated upstream of site ND5. In contrast this river receives little additional input from sub- catchments. Many of the streams of both catchments are impounded by V-notch weirs for gauging discharge. A riparian release valve, periodically opened to augment low summer flow, is situated immediately upstream of site LC6. The headwater streams of both rivers are enclosed in a thick canopy of riparian vegetation, which is also a feature of some of the Lower Canning River sites (LC 1-4) . The lower reaches of the Canning system flow through urban areas, while downstream sites on the North Dandalup are situated in rural areas where riparian vegetation has been reduced by stock grazing. 69 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. Figure 1 The location of study sites in the Canning and North Dandalup Rivers. k m 1 0 70 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. Methods Sampling regime Quarterly sampling was implemented from March 1985 to September 1986. A total of 30 sites was routinely sampled, with 11 sites in the North Dandalup and 19 in the Canning catchment. Sites were initially selected by the presence of riffle zones for macro invertebrate sampling; however fish samples included both riffle and pool habitats. At each site fish communities were sampled by seine and hand netting. The 5m wide purse type seine net, with a 9mm stretched mesh size, was placed across the stream and fish were driven downstream into the net over a 50- 100m reach. Submerged vegetation, logs and large rocks were swept with standard FBA handnets (1.00mm mesh size). Sampling time was standardized to 2 person -hours at each site. All fish taken were identified, enumerated and released. Species richness (S) , taken as total number of species in each sample, was determined for each site on each sampling occasion. Data analyses One-way ANOVAs were used to test the significance of between-site changes in the above parameters. Prior to analyses, Cochran’s C and Bartlett’s Box tests were used to measure homogeneity of variances (Zar 1974) . Square root or logarithmic transformations were used if variances were heteroscedastic . Results Composition of the fish fauna Canning Catchment Eight species of fish were sampled (Figure 2). The most commonly taken fish was the exotic, Gamhusia affinis, comprising 48.3% of the total. The pygmy perch, Edelia vittata Castelnau was the next most abundant species (23.5%), followed by the western minnow, Galaxias occidentals Ogilby (16.9%). Two other species; silverside, Atherinosoma wallacei Prince, Ivantsoff & Potter and goby, Pseudogobius olorum (Sauvage) contributed 7.2 and 2.9% of the total respectively. The remainder (1.2%) was composed of the nightfish, Bostockia porosa Castelnau, the cobbler, Tandanus bostocki Whitley and the goby Favonigobius suppositus (Sauvage) . A total of 2 593 fish was caught in the Canning River, 80.2% of which were collected from lower river sites, with the remainder spread evenly between the headwater streams. No fish were collected at the headwater sites SCI, SC2, CD1A & CD4. North Dandalup Catchment Eight species of fish, totalling 2 182 individuals were sampled from the North Dandalup Catchment (Figure 3) . This catchment contained one additional species, the exotic rainbow trout Salmo gairdneri Richardson. The goby, F. suppositus, present in the Canning Catchment, was not taken. The most common fish, comprising 71.5% of the total number caught, was G. occidentals. G. affinis comprised 19.2% of the total and was the second most abundant species. E. vittata and A. wallacei made up 6.7% and 1.6% of the total respectively. The remaining four species; B. porosa, T. bostocki, P. olorum and S. gairdneri, comprised 1% of the total abundance. Spatial variation in community structure Canning Catchment Between-site differences in species richness (Figure 2) were significant (ANOVA F = 2.2062, P < 0.05, df 8,45). Sites LC1, LC5 & LC6 had a lower species richness than all other lowland sites (Duncan’s Multiple Range test (DMR), P < 0.05). G. affinis, the dominant species in the lower Canning system, demonstrated a significant between-site difference in relative abundance (ANOVA F = 6.3905, P < 0.001, df 8,45), with sites LC6 & LC7A significantly higher than all other lowland sites, with the exception of site LC2 (DMR, P < 0.05). E. vittata, widely distributed throughout the lower Canning system (Figure 2), demonstrated significant between-site differences in relative abundance (ANOVA, F = 6.369, P < 0.001, df 8,45) . This species had a significantly higher relative abundance at sites LC1, LC3, LC4 and LC6A than all other lower sites (DMR, P < 0.05). E. vittata was not taken from site LC6. G. occidentals, the third most abundant species in the Canning system, was widely distributed throughout the catchment, with a higher relative abundance in the headwater sites. The remaining five species were mainly restricted to lower sites, with low relative abundances. A. wallacei was an exception, comprising approximately 40% of the total catch taken from site LC4. North Dandalup River Between-site differences in species richness were significant (Figure 3) (ANOVA, F = 6.5184, P < 0.05, df 10,66). Sites ND4, ND6, ND8, ND9, ND10 & ND11 had a higher species richness than sites ND1, ND2, ND3, ND5 & ND7 (DMR, P < 0.05). G. occidentals, the most abundant species, was widespread throughout the North Dandalup catchment (Figure 3) and dominated the headwater sites. G. affinis and E. vittata, the second and third most dominant species respectively, were common but restricted to the lowland sites, downstream of the dam. P. olorum, A. wallacei and T. bostocki were also restricted to the lowland sites but were neither widespread nor abundant (Figure 3). The remaining two species, B. porosa and S. gairdneri, were uncommon and taken only from headwater sites. Discussion The majority of the native species of fish recorded in freshwater rivers within 160km of Perth (Allen 1982) were collected during this study. Those not collected are either migratory or estuarine species that occasionally penetrate freshwater. Of the native species collected, only one, Favonigobius suppositus, was not recorded from both rivers, being absent from the North Dandalup and rare in the Canning system. The distribution of species in the Canning and North Dandalup rivers fits that predicted by Horwitz (1978) for temporally variable rivers. Both systems demonstrated a low overall species richness, increasing slowly downstream and within the common species there was no downstream replacement, only additions. The greatest within-system differences in species richness occurred between headwater streams and the lower reaches of both rivers. This may be a reflexion of the effects of physical barriers to fish movement. Several sites exhibited a community composition different from that expected. Species richness at site LC6 was markedly reduced from site LC5, 100 m upstream. This site is downstream of a riparian release valve from which treated water (chlorinated to lppm) is periodically released to augment reduced summer flow. It is unlikely that the change in community structure was solely due to elevated chlorine levels. Increased siltation and a disrupted food chain (macroinvertebrate community structure) may also be important factors. The effects do not appear to be long-lived 71 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. STINTON rDcci/ SPECIES DATE i 2 3 SITE CANNING DAM 2 3 4 5 6 1A LOWER CANNING 2 3 4 5 6 7 6 A 7 A Galaxias occidentalis (16.89) Edelia vittata (23.52) Gambusia affinis (48.32) Bostockia porosa (0.92) T andanus bostocki (0.15) Pseudogobius olorum (2.93) Atherinosoma wallacei (7.21) Favonigobius suppositus (0.08) MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 MARCH 1985 JUNE 1 985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 Figure 2 Spatial variation in the distribution of fish species in the Canning Catchment, March 1985 to September 1986 (Symbols indicate sample size: •, 1-5; #, 6-20; O , 21-50; , > 50; +, site dry); Values in parentheses rep- resent percentage of total abundance of each species over the sampling period. 72 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. SPECIES Galaxias occidentalis (71.5) DATE MARCH JUNE SEPT. DEC. MARCH JUNE SEPT. (6.7) Gambusia affinis (19.2) MARCH JUNE SEPT. DEC. MARCH JUNE SEPT. 1985 1985 1985 1985 1986 1986 1986 Edelia vittata MARCH 1985 JUNE 1985 1985 1985 SEPT DEC. MARCH 1986 JUNE 1986 SEPT. 1986 1985 1985 1985 1985 1986 1986 1986 SITE NORTH DANDALUP 4 5 6 7 8 9 10 11 if i? r • • • • i • • Bostockia porosa (< 1 . 0 ) MARCH 1985 JUNE 1985 SEPT. DEC. MARCH 1986 JUNE 1986 SEPT. 1986 1985 1985 Tandanus bostocki (<1.0) MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 Pseudogobius MARCH 1985 olorum (<1.0) 1985 1985 1985 JUNE SEPT DEC. MARCH 1986 JUNE 1986 SEPT. 1986 Atherinosoma wallacei (1.6) MARCH 1985 JUNE 1985 SEPT. 1985 DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 Salmo gairdneri MARCH 1985 JUNE 1985 SEPT. 1985 (<1 .0) DEC. 1985 MARCH 1986 JUNE 1986 SEPT. 1986 Figure 3 Spatial variation in the distribution of fish species in the North Dandalup Catchment, March 1985 to September 1986 (Symbols indicate sample size: •, 1-5; • , 6-20; • , 21-50; ® , > 50; + , site dry); Values in parentheses represent species over the sampling period. percentage of total abundance of each 73 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. with the river recovering by site LC6A, the next site downstream. Site LC7A, the most downstream Canning River site, also demonstrated a reduced species richness and diversity. This part of the river is both wide and deep reducing the effectiveness of the sampling method. More intensive sampling in the deeper sections of the river at this site may 'detect additional species. Conversely, species richness at site CD5 of the Canning Dam catchment was higher than expected, with this site comparable to lower river sites. This stream, unlike the other catchment streams, did not dry up totally, but remained as large isolated pools throughout the summer. These pools may act as refugia and hold residual populations of E. vittata, B. porosa and G. affinis which may then rapidly recolonize site CD5 and to a lesser extent site CD6. These three species were not collected from any other catchment streams. This may be a reflexion of their inability to colonize newly inundated streams demonstrating, like many small temperate streams, that these are not highly favourable environments for fish colonization (Moyle & Vondracek 1985). E. vittata and B. porosa are known to colonize rapidly floodplains of the south-west of Western Australia (Pusey unpubl) . Spawning of these and a number of other sympatric species on the floodplains occurs in spring with the young feeding predominantly on planktonic Crustacea. Temporary catchment streams, because of the paucity of planktonic fauna, may represent areas unfavourable for the development of young and this may restrict the distribution of these species. Horwitz (1978) reported planktivores to be uncommon in the headwaters of mid-western American rivers, presumably due to the unavailability of a suitable food source. The temporary headwater streams of both river systems were recolonized rapidly by G. occidentals soon after flow resumed. Little is known of the life history of this species in the streams of Western Australia, but it is thought that G. occidentals performs an upstream migration to spawn in tributaries and headwater streams (EPA 1987). During this study large aggregations of G. occidentals were frequently sampled at the base of V-notch weirs and on one occasion fish were observed, fully emersed, mid- way up the concrete spill- way below a weir face. This supports the view of an intended upstream movement. Spawning in headwater streams is also supported by the collection of gravid females at these sites. The same process may be occurring in the lower river because larvae and fry have been regularly sampled from drainage channels and flooded areas of the Swan Coastal Plain in winter (Edward unpubl). Migration up drainage channels and lowland tributaries may be an adaptation of populations of G. occidentals which are isolated downstream of physical barriers eg waterfalls, dams and V-notch weirs. The action of such structures as barriers to fish movement would explain the absence of fish from two of the headwater streams, sites CD1A & CD4, which were both upstream of V-notch weirs. Fish were also absent from sites SCI & SC2 which were upstream of a steep waterfall. This natural feature may be acting as a physical barrier to fish movement since aggregations of G. occidentals were present at the base of the waterfall, at site SC3. The construction of dams and V-notches is likely to have had a significant impact on the seasonal movements of G. occidentals, affecting its reproductive biology and the recruitment of juveniles. Temporary headwater streams may be recolonized by residual populations of G. occidentals within the reservoirs. Adult fish may actively retreat to the reservoir as stream flow decreases and it is also likely that if larvae are pelagic they will be swept downstream into the reservoir. Both river systems were shown to have the same species richness. In the North Dandalup catchment the native species G. occidentals was dominant while in the Canning system the introduced species G. affinis was most frequently encountered. G. affinis was first introduced to the fresh waters around Perth in 1936 (Mees 1977) to control mosquitoes. The species now dominates many streams and lakes in the Perth area and is widespread throughout South-western Australia. Sarti & Allen (1978), in a survey of the wetlands of the northern Swan Coastal Plain, found that G. affinis was the most abundant species present in lentic habitats. Where native species were present in such habitats they were in low densities and usually confined to inlet streams. Native species were abundant in lotic habitats only (Moore River; ten native and one exotic species, Ellen Brook; seven native and one exotic species). It is not known when G. affinis was first introduced to the North Dandalup River but as it is capable of rapid increases in population size it is unlikely that the difference in abundance between rivers is due to differing periods of residency. It is inferred from this study that between -system differences in the stream environment influence the population size of G. affinis. The Canning system is regulated by a substantial dam which rarely overflows. As a result, the lower reaches of this river are wide, deep and relatively slow flowing, especially so in summer. This habitat is comparable to the lentic environment to which G. affinis seems particularly suited. On the North Dandalup system the pipehead dam above site ND5 overflows each winter making this river more prone to spates. Also, the lower reaches of the North Dandalup system are shallower and faster flowing possibly reducing suitability for colonization by G. affinis. The role of disturbance in structuring fish communities may be important. Meffe (1984) found that G. affinis populations introduced to a Sonoran Desert stream incurred great losses during flashfloods. The Sonoran topminnow, Poeciliopsis occidentals, endemic to the region, did not suffer such losses as a result of behavioural responses. Meffe (op cit) argues that fish which evolve in habitats with frequent perturbations exhibit behavioural responses that minimize the impact of disturbance. It seems likely that this is also the case in the Canning and North Dandalup systems. The high level of endemicity of the south-western fish fauna suggests that the evolution of the fauna has occurred in situ allowing sufficient time for the adaptation of behavioural responses to the seasonal regimes of stream discharge. Meffe (1984) suggests that abiotic disturbance culminating in the almost complete removal of G. affinis allows the coexistence of this species and P. occidentals in tributaries of the Santa Cruz River. A number of studies listed show replacement of P. occidentals by G. affinis and all occurred in lentic environments. Such a response, similar to the storage effect (Warner & Chesson 1985) in which coexistence is mediated by fluctuations in recruitment, may be involved in the coexistence of native species and G. affinis in streams and rivers of South-western Australia. Acknowledgements The authors thank all past and present members of the Aquatic Research Laboratory, who were involved in data collection. Special thanks go to Mrs P Gazey and Mrs M Jones for data processing and the preparation of figures. This work was funded by the Water Authority of Western Australia as part of their wider water quality management studies of Darling Range catchments. 74 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. References Allen G R 1982 A field guide to inland fishes of Western Australia. Western Australian Museum, Perth. Arthington A H, Milton D A & McKay R J 1983 Effects ^>f urban development and habitat alterations on the distribution and abundance of native and exotic freshwater fish in the Brisbane region, Queensland. Aust J Ecol 8:87-101. Bunn S E. Edward D H & Loneragan N R 1986 Spatial and temporal variation in the macroinvertebrate fauna of streams of the northern jarrah forest, Western Australia. Freshwater Biol 16:67-92. Christensen P 1982 The distribution of Lepidogalaxias salamandroides and other small freshwater fishes in the lower south-west of Western Australia. J R Soc W Aust 65: 131-141. EPA 1987 The effects of gauging station control structures on the native fish migration in freshwater streams of south west Australia. Environmental Protection Authority, Perth, Western Australia. Bull 282. Horwitz R J 1978 Temporal variability patterns and the distribution patterns of stream fishes. Ecol Monogr 48:1173-1185. Jutson J L 1950 The physiography of Western Australia, 3rd edn. Bull Geol Survey W Aust 95. Mees G F 1977 The status of Gambusia affinis (Baird & Girard) in south-western Australia. Rec West Aust Mus 6:27-31. Meffe G K 1984 Effects of abiotic disturbance on co-existence of predator-prey species. Ecology 65:1525-1534. Merrick J R & Schmida G E 1984 Australian freshwater fishes: biology and management. J R Merrick, School of Biological Sciences, Macquarie University NSW. Moyle P B & Vondracek B 1985 Persistence and structure of the fish assemblage in a small California stream. Ecology 66:1-13. Myers G S 1975 Gambusia, the fish destroyer. Aust Zool 13:102. Sarti N L & Allen G R 1978 The freshwater fishes of the Northern Swan Coastal Plain. In: Faunal studies of the Northern Swan Coastal Plain: A consideration of past and future changes. Report published by Western Australian Museum for the Department of Conservation and Environment pp 204-220. Seddon G 1972 Sense of place. Univ W Aust Press, Nedlands. Warner R R & Chesson P L 1985 Coexistence mediated by recruitment fluctuations: a field guide to the storage effect. Am Nat 125: 769-787. Whitley G P 1947 The fluvifaunalae of Australia with particular reference to freshwater fishes in Western Australia. West Aust Nat 1:49-53. Zar J H 1974 Biostatistical analysis. Prentice Hall, New Jersey. 75 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989, 77-81 A simple model to forecast wheat yield in Western Australia D J Stephens 1 , T J Lyons 1 & M H Lamond 2 Environmental Science, School of Biological and Environmental Sciences, Murdoch University, Murdoch, WA 6150 2 Austweather Pty Ltd, 13 Buntine Rd, Wembley Downs, WA 6019 Manuscript received October 1987; accepted June 1988 Abstract Stress, defined as the lack of sufficient water to maintain maximum growth rates and yields, has long been recognized as a dominant factor in Australian wheat yield. Thus, by relating stress to crop yield and incorporating seasonal forecasts, a simple predictive crop model can be developed, the data requirements of which are commensurate with available meteorological seasonal forecasts. Such a model has been applied to the Merredin district of Western Australia and its potential is illustrated through the use of yield hindcasts for the 1984-1986 seasons. Introduction The growth and development of crops from sowing to harvest is influenced by a number of climatic and soil factors which interact in a very complex way. In recent years research has gone into the development of models to predict and simulate plant growth. Baier (1979) classified these crop- weather models into three broad categories: (i) crop growth simulation, (ii) crop- weather analysis, assessing crop response to weather and climate, and (iii) empirical-statistical models, where several weather variables are related to yields. In Western Australian numerous studies have used the latter approach (Gentilli 1946, 1959; Fitzpatrick 1970; Hill & Goodchild 1981; Wigley & Tu Qipu 1983) to quantify the sensitivity of wheat yields to climate. They are particularly useful for zoning and mapping areas in terms of their suitability for growing crops and estimating yield potential. Nevertheless, statistical methods are limited for crop yield forecasting as they incorporate complex non-linear interactions between independent variables and there is no evidence to suggest normality of errors (Matis et al 1985). Baier (1977) and Hill & Goodchild (1981) noted interactions between weather and technology as well as weather variables themselves. Also, Hill & Goodchild (1981) found that long term historical data bases are likely to include unquantifiable historical events that have statistically intractable effects on yields. A crop growth simulation model, adapted from CERES- Wheat (Ritchie & Otter 1985), has been developed for conditions at the Western Australian Department of Agriculture’s Merredin Research Station (118.17°E, 31.29°S) in the eastern wheat belt of Western Australia (Perry 1986, pers comm) . This model relies on a fallow-cereal crop water balance and computes soil water flow, crop growth and phenological development (McMahon 1983). Yield predictions can normally only be made at the end of the season and are sensitive to daily changes in temperature, rainfall and radiation. Duchon (1986) used such a model, CERES-Maize, to predict yield, using a combination of current weather, and sequences of past weather for the time between prediction and harvest. Application of this approach to Merredin yielded a low correlation between predicted and observed yields for past trials. This is a reflexion of the present inability to simulate accurately the plant-environment system on a daily basis, when significant biological events happen in much shorter time intervals, beyond the resolution of available standard meteorological data. However, it is neither essential nor practical to model at a level greater than that required for useful predictions and the model sophistication should be commensurate with routinely available input data, such as meteorological forecasts. Simple models can have a powerful predictive value when one or two major factors dominate the performance system (Ritchie 1983) . Stress, defined as the lack of sufficient water to maintain maximum growth rates and yields (Mederski 1983) has long been recognized as dominating the performance of Australian wheat (Nix & Fitzpatrick 1969). Accordingly, a simple crop- weather analysis model that relates stress to crop yield (Frere & Popov 1979) was adapted to the Merredin Research Station using results from a ten year direct drilling wheat trial (1977-86) (Jarvis et al 1986) . Incorporation of meteorological seasonal forecasts into the model, meant that yield predictions were possible and this procedure is illustrated with hindcasts based on the seasonal forecasts issued for 1984-86. 77 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3. 1989. Model Description The crop-weather analysis model we used was originally designed by Frere & Popov (1979) . It was designed to provide developing countries in semi-arid conditions with a simple technique for monitoring crop conditions, thereby allowing the preparation of quantitative yield assessments. As such it requires a minimum amount of actual data and calculations. Central to the method is the determination of the crop water balance which shows directly whether the crop is experiencing stress or not. An accumulated stress factor (stress index) is determined from the water balance and as the season progresses better reflects the ability of the crop to produce yields. Crop assessments are based on the past relationship between final stress indices and actual yields. The basis of the model is a cumulative water balance which is summed for 1 0 day intervals over the whole growing season . This is done by adding the difference between precipitation received and water lost by evapotranspiration, to the existing stored moisture which is first estimated at the sowing date. A potential evapotranspiration (PET) is defined as the maximum quantity of water which may be evaporated by a uniform cover of dense short grass when the water supply to the soil is not limited (Penman 1948) . When the available water supply can satisfy the PET rate, maximum growth is assumed to occur; but when it does not, stress is implied so that growth rates and final yields suffer. The stress factor gives a direct measure of the expected reduction in yield. The only input data needed throughout the growing season is the actual precipitation (Pa) received in 10 day periods (decades), ie from days 1-10, 11-20, and from day 21 until the end of the month for each month. The last decades of some months have 11 days to ensure continuity of the monthly notation and the use of standard meteorological information. Rainfall is rounded to the nearest millimetre to eliminate small showers (< 0.6 mm) which are considered to have little significance, being evaporated rapidly in most conditions. Runoff is not accounted for since no measurements of this are routinely available. The average daily PET for a given month was calculated from the Penman formula (Penman 1948) using climatological records of mean monthly temperature, relative humidity, pressure, sunshine duration and wind speed (Frere & Popov 1979). Estimation of total radiation can be found from direct observations or measurements of sunshine duration (eg Frere & Popov 1979; Edwards & Lyons 1982), whereas the remaining input data are estimated from standard climatological observations. Mean monthly wind speed was estimated from the Merredin wind rose. Monthly values were found by multiplying the average of the mean 0900 and 1500 wind speeds with a ratio, determined from the ratio maps of Hutchinson et al (1984). In the absence of significant errors in the 0900 and 1500 records the estimates obtained are expected to be within 10% of the actual values (Hutchinson et al 1984) . Frere & Popov (1979) found that the Penman formula under-estimated evapotranspiration in dry environments by not accounting for dry air advection. To rectify this, they modified the Penman coefficient affecting the wind speed whenever the average minimum temperature was above 5 C and the difference between monthly average maximum and minimum temperatures was more than 12 C. For larger temperature differences, a larger value of the coefficient was used. Consequently, for Merredin the coefficient was modified for the warmer months of September through to April, utilizing values given by Frere & Popov (1979). Monthly values of PET were obtained by multiplying the mean daily PET by the number of days for each respective month. These values were then divided into decadal periods by a simple mathematical procedure based on the curve fitted to the monthly values X c , for c = 1 to 12. By representing the three decades for each month by Y lt Y 2 and Y 3 , their individual values are determined by the equations; Y 2 = (Xe/3) (1) Y, = (X c /3) - (X c - Xc-d/9 (2) Y 3 = (Xc/3) + (X c+1 - X c )/9 (3) such that X c = Y, + Y 2 + Y 3 (4) The values for the three decades (Y x , Y 2 , Y 3 ) were adjusted so that the monthly total X c is preserved. If the total of the three decades is less than X c , 1mm is added to the decade Y 3 or Y 1? depending on whether the slope between X c and X c +i, increases or decreases from the previous monthly interval (X c _i to X c ). Alternatively if the total is greater than X c , 1mm is subtracted from Y 3 or Y lf depending on whether the slope decreases or increases from the previous interval. When a point of inflexion occurs at a particular month, Y 2 is given the largest (or smallest) decadal value depending on whether the curve is convex or concave. The water requirements of the crop are found by multiplying the decadal PET (Er) with the respective crop coefficient (KcJ for that period: WR = ErK cr (5) Such an equation allows for the fact that cultivated crops pass through several stages from emergence until maturity. Over this period the plant cover varies in comparison to the reference short grass used in defining PET and this variation is expressed as the crop coefficient, being the ratio of maximum actual evapotranspiration over PET. The total water requirements of a potential highest yielding crop are based on conditions experienced at the trial site during 1974, when yields reached maximum expected values of 3 tonne/ha (Jarvis 1987, pers comm). For that year, the sum of stored soil moisture at sowing, estimated from the water balance subroutine of CERES- Wheat (McMahon 1983), and rainfall during the growing season came to 280mm. Therefore it was assumed, that if this amount of moisture was available and evenly distributed to a crop, no stress would be incurred and maximum yields close to 3 tonne/ha would result. The most regularly used crop coefficients of Doorenbos & Pruitt (1977) were slightly increased so that when they were multiplied with the decadal PET values the total water requirement for the season came to 280mm. At the same time the distribution of the decadal water requirements was related to the observations of French & Schultz (1984) which showed that 70% of the total water use occurred by anthesis. At Merredin, anthesis usually occurs in mid-September and hence the crop coefficients were adjusted to ensure that 70% of the total water requirement was between sowing and anthesis. The relative proportions of the total water requirement were approximated as 0.3 for the period (c 2 months) between sowing and tillering, 0.4 for the period (c 2 months) between tillering and anthesis, 0.2 for the following month up to soft dough, and 0.1 for the remaining time to maturity. The current readily available moisture, RS i? is found from: RSi = RSi-i + (Pa - WR) (6) where the difference between actual precipitation and crop water requirements is added to the existing stored water, RSj-i. RS indicates the amount of usefully stored or readily available water in the soil, and this is commonly referred to as the water reserve between the field capacity and the 78 Journal of the Royal Society of Western Australia, Vol. 71. Parts 2 & 3, 1989. permanent wilting point. The amount readily available depends on the depth of the soil exploited by the roots and the physio- chemical characteristics of the soil (Frere & Popov 1979) . For the heavy, calcic red brown earth at Merredin, the maximum amount of stored moisture held in the 1.2 m depth rooting zone is 180 mm (Perry 1986 pers comm). Originally designed for parts of Africa and Asia which experience a brief wet-season, this procedure assumes that there is no stored soil moisture before the first opening rain. Such an assumption cannot be applied to Western Australian conditions, where moisture is almost always carried through from either autumn or summer. Thus the validated water balance subroutine from CERES-Wheat (McMahon 1983), was used to estimate the stored moisture accumulated before the sowing date. The model was run with daily values of rainfall, radiation and temperature from the first of January through to the beginning of the 10 day period in which sowing occurred. Values of stored water determined varied from 0mm in 1985 to 64mm in 1984. When the final stress factor was plotted against yield for the 10 years, two of the years, 1979 and 1984, heavily over- estimated the stress in comparison to the observed yield. However, the end of both 1978 and 1983 were very wet, suggesting that moisture could have been carried through from these years and this is not accounted for in initializing the soil moisture balance from January. By using the final soil moisture profile for the previous year to initialize the profile for the following year, very different soil moisture estimates were obtained for 1979 and 1984. For these two years the values increased from 53mm to 96mm and 64mm to 106mm respectively. Negligible increases of 1 to 4mm were found for the other years. If RS is greater than the total possible stored water, in this case 180mm, a water surplus is registered and stress is assumed to occur. Likewise, if insufficient moisture is available to maintain maximum growth rates stress is also inferred. This is the case when RS becomes negative — a water deficit. As such, RS only relates what the available moisture content would be if a crop was to be growing at its potential rate not limited by stress. Because Merredin has a very high water holding capacity and a dry climate, only deficits were observed. Stress is represented by a culminative stress index (I) and is calculated (for deficits, D) as: 1 = 1-1 + (D/WRT) (7) where WRT represents the total water requirement for the growing season of the potential highest yielding crop, and Ij_i is the previous value of I. Expressed as a percentage, this factor indicates the extent to which the water requirements of a crop have been satisfied in a cumulative way up to that point in its development. At the beginning of the season, the index is given a value of 100. It will remain at that value for successive decades until a surplus or deficit appears. If a deficit of say 28mm appears, the quotient between 28 and 280 (the total water requirement) is 0.1. This corresponds to 10% of the water requirements not satisfied, so the index drops from 100 to 90. The calculation is continued through to maturity where the final stress index reflects the cumulative stress endured by the crop throughout the season. As such, it is usually closely linked with yield unless other harmful factors such as pests, diseases or frost have had an over-riding influence. Results and Discussion The accumulated stress-yield relationship for the 10 year direct drilling trial (Fig. 1) has a least squares regression of: Y = 37.12 I - 1083.58 (8) where Y is the expected yield in tonne/ha and I the accumulated stress over the growing season. The Stress Index Figure 1 Comparison between yield and computed stress indices for the Merredin Research Station, accompanying correlation coefficient is 0.94 (r 2 = 0.88) and is significant at P = 0.001. This illustrates that moisture stress is a significant yield determining variable for heavy soils in the Merredin district. The lower than expected yield in 1978 may have resulted from surface detention and evaporation of water due to poor soil structure at the start of the trial and the high rainfall (410 mm) in 1978 (Jarvis et al 1986). Equation (8) formed the basis of the predictive mode of the model. Yield predictions made after the growing season but based on information available before the season are known as hindcasts. These were made by using decile rainfall corresponding to the seasonal forecasts issued by Austweather prior to the corresponding season. Such forecasts, based on large scale ocean-atmosphere indicators, endeavour to predict whether rainfall will be in one of three possible categories: (i) below normal (signifying the lower 30% of climate data), (ii) near normal (signifying the middle 40% of recorded values), or (iii) above normal (signifying the upper 30% of recordings). Consistent with a similar decision-making model (Brown et al 1986), the three forecast categories were represented by the deciles 1.5, 5 (median), and 8.5 of the distribution of growing season precipitation. For certain situations, intermediate 2 event forecasts were made by Austweather and these were represented by decile 3 (for normal to below normal forecasts), and decile 7 (for normal to above normal forecasts), as these values are positioned at the boundaries between the two categories used in each respective 2 event forecast. The hindcasts were first made for the beginning of the 10 day period in which sowing occurred and the initial soil moisture was estimated from CERES-Wheat. In place of actual rainfall, however, the decadal decile rainfall was used as an indicator of the distribution of the rainfall over the forecast period. Thus the level of sophistication of the climate data fits that of the seasonal forecasts. The monthly values were separated into 10 day intervals by the same mathematical procedure outlined in equations (1-4). Such a process was carried out for early winter (April- June), late winter (July- September) and early summer (October-December). Estimated stress was used to predict yield via equation (8). An updated yield prediction followed at the end of June. Actual rainfall figures for June replaced the decile rainfall figures and another estimate was made. This updating continued until the end of the growing season when the final yield estimate was made. Initial and updated predictions for 1984-1986 are shown in Fig. 2. 79 Yield (tonne/hectare) Month Figure 2 Yield predictions made at various stages during the growing seasons for 1984-1986 based on seasonal forecasts, compared with the 10 year average yield and the actual yield . 80 Journal of the Royal Society of Western Australia, Vol. 71, Parts 2 & 3, 1989. In 1986 the yield prediction gradually rose and levelled out near 2 tonne/ha. The rise in predicted yield was due to the actual rainfall being slightly wetter than the decile rainfall given by the seasonal forecasts. A similar result occurred for 1985 except that the predicted and actual yields were all about a tonne/ha lower. In 1984 the reverse happened at the beginning of the season. An initial estimate of 2.26 tonne/ha declined to 1.78 tonne/ha at the end of June. This sudden change in yield estimate was essentially caused by the rigidness of the seasonal forecasts, individually restricted to a three month time-span. The outlook was for average to above average rainfall for the period April to June and below average rainfall from July through to September. What eventuated was an average to above average April and May, followed by below average rainfall for June, July and August. A rigid three month forecast could not account for such variation and thus the possibility of monthly forecasts is being considered. Ultimately the value of the predicted yield is dependent on the stress/yield relationship and the accuracy of the seasonal forecasts. The relative success of the predictive model over the 1984-86 seasons is a direct function of the success of the input seasonal forecasts. Nevertheless, this simple model is commensurate with the available meteorological data and is able to express seasonal forecasts directly in terms of expected yield. The value of such predictions is not in the actual yield predicted, but rather as a comparative measure of how a particular year is expected to compare to previous years. At this stage, the model has only been applied to one soil type at one location. For other regions and soil types, different crop responses are expected but could easily be accounted for through modification of the maximum stored water and water required for maximum yield. Conclusions A crop-weather analysis model similar to Frere & Popov (1979) was developed and applied to the Merredin Research Station. By including soil moisture estimates at past sowing dates from CERES- Wheat, a correlation of 0.94 between yield and final stress was obtained. This strong relationship between stress and yield is in agreement with the observation of Nix & Fitzpatrick (1969) that lack of sufficient water to maintain maximum growth rates and yields is a dominant influence on Australian wheat. Such a relationship can be used to provide a direct link between expected seasonal weather conditions and yield as the data input required is of a similar sophistication to available seasonal forecasts. Hindcasts for the 1984-86 seasons illustrate the potential of the method as a basis for a dynamic decision making model. Acknowledgements This project has benefited from the assistance, advice and encouragement of Mr Mike Perry, Miss Geraldine Pasqual, Mr David Tennant, Mr Ron Jarvis, Mr Reice Pitt (Western Australian Department of Agriculture) , Dr Bill Scott (Murdoch University) Mr Bob Southern (formerly Austweather Pty Ltd), and Mr Yuri Kuuse (Commonwealth Bureau of Meteorology). Throughout it, one of the authors (DJS) was in receipt of a Neville Stanley Studentship provided by the Western Australian State Government through its West Australian Technology and Development Authority. All of this assistance is gratefully acknowledged. References Baier W 1977 Crop-weather models and their use in yield assessments. WMO Tech Note 151, World Meteorological Organization, Geneva, 1-48. Baier W 1979 Note on the terminology of crop-weather models. Agric Meteor 20: 137-145. Brown B G, Katz R W & Murphy A H 1986 On the economic value of seasonal- precipitation forecasts: the fallowing/planting problem. Bull Amer Meteor Soc 67: 833-841. Doorenbos J & Pruitt W O 1977 Crop water requirements. Irrigation and Drainage Paper 24, FAO, Rome, 1-144. Duchon C E 1986 Corn yield prediction using climatology. J Climate Appl Meteor 25: 581-590. Edwards P R & Lyons T J 1982 Estimating global solar irradiance for Western Australia, Part II. Arch Met Geoph Biokl B30: 371-382. Fitzpatrick E A 1970 The expectancy of deficient winter rainfall and the potential for severe drought in the southwest of Western Australia. Misc Publ 10/1, Agronomy Dept, Univ W Aust. French R J & Schultz J E 1984 Water use efficiency of wheat in a Mediterranean- type environment. 1. The relation between yield, water use and climate. Aust J Agric Res 35. 743-764. Frere M & Popov G F 1979 Agrometeorological crop monitoring and forecasting. Plant Production and Protection Paper 17, FAO, Rome, 1-66. Gentilli J 1946 Mimeographed tables and maps of rainfall and climate in Western Australia and the rainfall — wheat relationship. Publ Geography Dept, Univ W Aust, 1-37. Gentilli J 1959 Weather and Climate in Western Australia. W Aust Govt Tourist & Publicity Bureau, W Aust Govt Printer, Perth. Hill J & Goodchild N A 1981 Analysing environments for plant breeding purposes as exemplified by multivariate analyses of long term wheat yields. Theor Appl Genet 59: 317-325. Hutchinson M F, Kalma J D & Johnson M E 1984 Monthly estimates of wind speed and wind run for Australia. J Climatol 4: 311-324. Jarvis R J, Hamblin A P & Delroy N D 1986 Continuous cereal cropping with alternative tillage systems in Western Australia. W Aust Dept Aqric Bull 71 1-37. Matis J H, Saito T, Grant W E, Iwig W C & Ritchie J T 1985 A Markov chain approach to crop yield forecasting. Agric Syst 18: 171-187. McMahon T A (Ed) 1983 A general wheat crop model for Australia. Agric Eng Rep 67/83, Univ Melbourne, 1-49. Mederski J H 1983 Effects of water and temperature stress on soybean plant growth and yield in humid, temperate climates. In: Crop Reactions to Water and Temperature Stresses in Humid, Temperate Climates (ed C D Raper & P J Kramer), Westview Press, Boulder, Colorado, 35-48. Nix H A & Fitzpatrick E A 1969 An index of crop water stress related to wheat and grain sorghum yields. Agric Meteor 6: 321-337. Penman H L 1948 Natural evaporation from open water, bare soil and grass. Proc R Soc London A 193: 120-145. Ritchie J T 1983 Reduction of stress related to soil water deficit. In: Crop Reactions to Water and Temperature Stresses in Humid, Temperate Climates (ed C D Raper & P J Kramer). Westview Press, Boulder, Colorado, 329-340. Ritchie J T & Otter S 1985 Description and performance of Ceres-Wheat: a user oriented wheat yield model. In: ARS Wheat Yield Project (ed W D Willis), US Dept Agric, Agric Res Ser, ARS-38, 159-175. Wigley T M L & Tu Qipu 1983 Crop- climate modeling using spatial patterns of yield and climate. Part 1: Background and an example from Australia. J Climate Appl Meteor 22: 1831-1841. 81 INSTRUCTIONS TO AUTHORS The Journal publishers (after refereeing) • papers dealing with original research done in Western Australia into any branch of the natural sciences; • papers concerning some biological or geological aspect of Western Australia; • authoritative overviews of any subject in the natural sciences, integrating research already largely published in the more specialized national or international journals, and interpreting such studies with the general membership of the So- ciety in mind; • analyses of controversial issues of great scientific moment in Western australia. Prospective authors of papers in the last two categories should consult the Hon. Editor for further advice. Contributions should be sent to The Honorary Editor, Society of Western Australia, Western Australian Museum, Francis Street, Perth, Western Australia, 6000. Publication in the Society’s Journal is available to all categories of mem- bers and to non-members residing outside Western Australia. Where all authors of a paper live in Western Australia at least one author must be a member of the society. Papers by non-members living outside the State must be communicated through an Ordinary or an Honorary Member. Submission of a paper is taken to mean that the results have not been published or are not being considered for publication elsewhere. Free reprints are not provided. Reprints may be ordered at cost, provided that orders are submitted with the return galley proofs. Authors are solely responsible for the accuracy of all information in their papers, and for any opinion they express. Manuscripts. The original and two copies must be submitted. They should be typed on opaque white paper with double- spacing throughout and a 3 cm margin on the lef-hand side. All pages should be numbered consecutively, including those carrying tables and captions to illustrations, which appear after the text. Illustrations, both line drawings and photographs, are to be numbered as figures in a common sequence, and each must be referred to in the text. In composite figures, made up of several photographs or diagrams, each of these should be designated by a letter (e.g. Figure 2B). To avoid risk of damage to original figures, authors may retain these until after the paper is accepted. The copies of the figures accompanying the manu- script must be of good quality. Authors are advised to use the most recent issue of the Journal as a guide to the general format of their papers. Words to be placed in italics should be underlined. To facilitate editing, papers must be accompanied by a table of contents, on a separate sheet, showing the status of all headings. References must be set out as follows: Paper Jackson A 1931 The Oligochaeta of South-Western Australia. J R Soc W Aust 17:17-136. Twigg L, Majer J D & Kotula R 1983 The influence of fluoroacetate producing plants upon seed selection by seed harvesting ants. Mulga Res Centre W Aust Inst Technol, Bentley, Ann Rep 6:75-80. Book Jacobs M R 1955 Growth Habits of the Eucalypts. For Timb Bur, Canberra. Chapter in book Dell J 1983. The Importance of the Darling Scarp to Fauna. In: Scarp Symposium (ed J D Majer) W Aust Inst Technol, Bentley, 17-27. The Title should begin with a keyword. The Abstract should not be an expanded title, but should include the main substance of the paper in a condensed form. The metric system (SI units) must be used. Taxonomic papers must follow the appropriate international Code of Nomenclature, and geological papers must adhere to the International Stratigraphic Guide. Spelling should follow the Concise Oxford Dictionary. Authors should maintain a proper balance between length and substance, and papers longer than 10 000 words would need to be exceptional importance to be considered for publication. Authors will be charged page costs (currently $30 per page) if papers exceed 8 printed pages. Short papers (2-4 printed pages) are particularly sought as these often ensure full use of the 32 pages available in each part. Illustrations. These should be prepared to fit single or double column widths. Illustrations must include all necessary letter- ing, and be suitable for direct photographic reduction. No lettering should be smaller than 1 mm on reduction. To avoid unecessary handling of the original illustrations, which are best prepared between 1.5 and 2 times the required size, authors are advised to supply extra prints already reduced. Additional printing costs, such as those for folding maps or colour blocks, will be charged to authors. Supplementary Publications. Extensive sets of data, such as large tables or long appendices, may be classed as Sup- plementary Publications and not printed with the paper. Supplementary Publications will be lodged with the Society’s Library (C/- Western Australian Museum, Perth, WA 6000) and with the National Library of Australia (Manuscript Section, Parkes Place, Barton, ACT 2600) and photocopies may be obtained from either institution upon payment of a fee. JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA CONTENTS VOLUME 71 PARTS 2 & 3 1989 Page The Quindalup Dunes: the regional system, physical framework and veg- etation habitats V Semeniuk, I D Cresswell & P A S Wurm 23 A field guide to the dung beetles (Scarabaeidae: Scarabaeinae and Aphodiinae) common in pastures of South-western Australia T J Ridsdill-Smith, G P Hall & T A Weir 49 The early evolution of the plant life of South-western Australia J S Beard 59 Spatial variation in fish communities in two South-western Australian river systems B J Pusey, A W Storey, P M Davies & D H D Edward 69 A simple model to forecast wheat yield in Western Australia D J Stephens, T J Lyons & M H Lamond 77 Edited by I Abbott Registered by Australia Post — Publication No. WBG 0351 No claim for non-receipt of the Journal will be entertained unless it is received within 12 months after publication of Part 4 of each Volume The Royal Society of Western Australia, Western Australian Museum, Perth Circulation of this Journal exceeds 600 copies. Nearly 100 of these are distributed to institutions and societies elsewhere in Aus- tralia. A further 200 copies circulate to more than 40 countries. The Society also has over 350 personal members, most of whom are scientists working in Western Australia. The Journal is indexed and abstracted internationally. A63378/8/88 — 700 — L/4233 GARRY L. DUFFIELD, Government Printer, Western Australia VOLUME 71 (PARTS 2 & 3) JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA Volume 71, Part 4, 1989 Banksia WOODLANDS SYMPOSIUM ISSN 0035-922X (illje iSoyal §ociety of Western Australia To promote and foster science in Western Australia and counteract the effects of specialization PATRON Her Majesty the Queen VICE-PATRON His Excellency Professor Gordon Reid Governor of Western Australia COUNCIL 1988-1989 President J S Pate Ph D, D Sc, FAA, FRS Vice-Presidents M Candy M Sc, FRAS B Dell B Sc (Hons), Ph D Past President J T Tippett B Sc, Ph D Joint Hon Secretaries K W Dixon B Sc (Hons), Ph D L Thomas M Sc Hon Treasurer J Dodd B A, M Sc, Ph D Hon Librarian M A Triffitt B A, ALAA Hon Editor 1 Abbott B Sc (Hons), Ph D Hon Journal Manager J Backhouse B Sc (Hons), M Sc, Ph D Members J S Beard M A, B Sc, D Phil W A Cowling B Agric Sc (Hons), Ph D S J Hallam M A, FAHA L E Koch M Sc, Ph D K McNamara B Sc (Hons), Ph D J Majer B Sc, DIC, Cert Ed, Ph D V Semeniuk B Sc (Hons), Ph D Cover illustrations by Mrs Joan Bacon. Reproduced from originals with kind permission of artist. (Clockwise from top of page): Dryandra niuea. D. sessilis. Banksia prionotes. B. attenuata. Journal of the Royal Society of Western Australia, 71 (4), 1989, 83-84 Preface J S Pate, FAA, FRS President 1988/89 Royal Society of Western Australia The concept of holding a Symposium specifically on Banksia woodlands was the brainchild of the late Dr Jenny Arnold, to whom this volume is dedicated. In 1987 she approached the Royal Society of Western Australia as a possible host society for the Symposium, the Council of the Society agreed wholeheartedly, and, with Jenny as convener, the program which you now see recorded in this volume was generated. It was always Jenny’s belief, now shared by many of us, that ecosystems close to the Metropolitan area are being so rapidly depleted and degraded that some form of widescale publicity was immediately required to evaluate their current state, and suggest how future management might be implemented within the inevitable context of multi-purpose usage by a range of pub- lic bodies and interests. Having focussed previously on wetland habitats and to a certain extent on the virtually extinct tuart for- est ecosystem, the Banksia woodlands of the Swan Coastal Plain seemed a highly appropriate topic for reasoned scientific evalu- ation and debate. This is essentially what took place at our Symposium, and I commend this volume enthusiastically to you as a focus of cur- rent knowledge, and, hopefully, a basis on which future manage- ment strategies might be considered. JENNIFER MARY ARNOLD B Sc, B A, M Sc, Ph D— An Appreciation Jenny was born and spent her childhood on a farm which her parents pioneered near Waiki in the outer wheat belt of Western Australia, and although from primary school she had to board away from home she always retained an understanding of farm- ing and a sympathy for farm life. It must have been there that her love of nature and concern for conservation was born. She went to Northam High School and then came to the Uni- versity of Western Australia where she studied Biology and completed a double major in Zoology and Botany in 1957. Shortly after graduating she went to work in the Climatology Section of the CSIRO Division of Land Research and Regional Survey in Canberra. Then in 1961 she went to the University of Queensland as Senior Demonstrator in the Zoology Department before return- ing to a similar position at the University of Western Australia in 1964. There two of her many talents were displayed: her skill in writing and her deep concern for others. Her doctoral thesis on the biology of the Native Cat was a masterful piece of writing which received the highest praise from the examiners, one of whom wrote: “I have only high praise for this exceptionally fine thesis. Her perseverence and ingenuity has resulted in an impress- ive assembly of new information, which is well reported, and well related to the general subject....” Jenny was a scientist, but her interests were much wider. She had a love of good literature and when she joined the Depart- ment of Conservation and Environment in 1977 she returned to the University to study English Literature and graduated again with a Bachelor of Arts Degree. Her feeling for the language never allowed her to accept any- thing but the best in her writing. She was a perfectionist with her own work and drove herself hard to achieve a high standard, one that put a great strain on her at a time during her long illness when she might well have relaxed. But she was a kind and con- structive critic of the work of others who sought her help. Jenny’s main job at UWA was in running first year laboratory classes and there she gave of herself in helping the many students. In the words of her Professor she was “almost obsessionally conscientious.” She was perceptive of student needs, she was interested in them personally and in what they were doing, and gave advice with sympathy and understanding, but without any pretence that she knew best. The local branch of ANZAAS and the Royal Society of West- ern Australia especially owe her much for her tireless efforts in organizing their activities - which she was doing almost to the end - and in so doing endeared herself to her colleagues. One of her last responsibilities was to the Fitzgerald River National Parks Association and it was wonderful to see the high esteem in which she was held and the affection the members felt for her. A67232- 1 83 Journal of the Royal Society of Western Australia 71 (4), 1989 Another of Jenny’s talents was as an artist, with a love of art, and only recently while she was fighting the cancer from which she died she took a course in drawing at the Fremantle Arts Centre. Her PhD Thesis is illustrated with her lifelike drawing of the animal she studied. She also had a love of good music and after she returned from Queensland she sang with the UWA Choral Society choir. While she was with the Environmental Protection Authority Jenny worked for a year (1985-86) at the WA Water Authority where she was greatly respected and appreciated. There she compiled an inventory of the wetlands of the Swan Coastal Plain, a mammoth job. This she did with her usual meticulous concern for accuracy and her distrust of the political demand for decisions before there was the relevant information on which to base judgements. Those of us who visited Jenny in hospital in the last weeks of her life will remember with gratitude how through all her pain and distress she rallied to show her interest and involvement in our interests and activities, and to display her wonderful sense of humour and recall the fun she had shared with us. It was a joy to be with her. Ernest P Hodgkin 84 Journal of the Royal Society of Western Australia, 71 (4), 1989, 85-86 Definition and location of the Banksia woodlands J S Beard 6 Fraser Road Applecross WA 6153 A recent book (Pate & Beard 1984) entitled Kwongan; the Plant Life of the Sandplain dealt with the vegetation of sandplains in the Extra-Dry Mediterranean Bioclimatic Zone (Pate & Beard 1984) on which kwongan is the characteristic plant formation. Banksia woodland (technically Banksia low woodland) is the characteristic formation of sandplains in the more humid Dry Mediterranean Bioclimatic Zone where the length of the dry season averages 5-6 months as against the 7-8 months of the Extra-Dry Zone. Banksia woodland may be said to represent the plant life of the sandplain within this bioclimatic zone, which corresponds to the Darling Botanical District. Deep well-drained sandy soils form the characteristic habitat for Banksia woodland, and these occur principally within the Drummond subdistrict, which includes the Swan Coastal Plain and the Dandaragan Plateau. Area calculations from the Vegetation Survey of WA (Beard & Sprenger 1984) estimate that Banksia low woodland orig- inally covered 6 229 km 2 of which 61% is now alienated land where the vegetation is presumed cleared. Banksia woodland with scattered emergent eucalypts is estimated to have covered a further 680 km 2 , all of which is now on alienated land. The amount of remnant vegetation on alienated land is not known. The woodland is dominated by small trees of about 6-8m among which Banksias are the most numerous. On the Swan Coastal Plain these are B. attenuata, B. menziesii , less commonly B. ilicifolia, together with Eucalyptus todtiana and Nuytsia floribunda, and in the southern part Casuarina fraseriana. Further inland, in the northeastern occurrence on the Dandaragan Plateau, B. menziesii is replaced by B. burdettii and B. prionotes, while Actinostrobus arenarius and Xylomelum angustifolium also join the assemblage. Leaves of the Banksias are 10-15 cm long, rough, chartaceous, serrate; though large, they are relatively few so that the crown is thin, and branching of the trees is somewhat open. Bark is very thick, dark, scaly. The formation resembles the oak ( Quercus ) woodlands of other mediterranean regions, in particular the cork oak (Q. suber) woodlands of the mediterranean itself. Individually, Banksias re- semble Curatella americana of savannas in tropical America, which significantly is named chaparro, Spanish for cork oak. Some convergence in evolution is perhaps impressed by the poor siliceous sands in these cases. The Banksia woodlands have a well developed sclerophyll shrub understory for which Speck (1952) listed 28 spp of tall shrubs, 130 of low shrubs, and 182 spp of ground-layer plants including many Restionaceae, Cyperaceae and other her- baceous monocots, for the woodlands on the Swan Coastal Plain. Speck indicated for each of these components whether it occurred on limestone ( i e on the Spearwood Dune System), on Bassendean sand or in a so-called “Moist Phase” which de- scribes swampy areas which may be present throughout. These are the three most readily recognizable substrate types within the Banksia woodlands of the Plain. Prior to Speck’s detailed study, the Banksia woodlands had received scant attention in the literature. Diels (1906) did not recognize this formation precisely. The vegetation of the Bassendean sand was treated as “Mixed woodlands of the Coastal Plain” with emphasis on Eucalyptus marginata and Casuarina fraseriana. Ffowever Diels named the component Banksia spp (B. attenuata, B. menziesii and B. ilicifolia) as ”sev- eral Banksia species which as smaller trees or tall-growing shrubs characterise many places in these woodlands of the plain” (p 228). Gardner (1944) made no mention at all of these woodlands but as his treatment of the vegetation of the State was much briefer than Diels’, it was necessary to omit describing many of the less extensive communities. Subsequent to Speck’s work, Havel (1968) published a study of the understory communities of the Banksia woodlands found in State Forest No. 65, employing mathematical analysis which distinguished 11 understory groupings associated with particu- lar site conditions. 5 of these occurred on the Spearwood Dune System, 3 on the Bassendean System, 1 on the transition be- tween the two, and 2 on swampy sites which might occur throughout the area, thus confirming Speck’s earlier habitat divisions. Subsequently again, Speck’s work was summarized by Seddon (1972) together with a vegetation map scale 1:500 000 of the Swan Coastal Plain from the Moore to the Murray River, distinguishing 11 units. In 1979 as part of the System 6 study, Heddle published a vegetation map scale 1:400 000 covering the same part of the Swan Coastal Plain and extending further inland. F G Smith mapped vegetation in detail at 1:250 000 on the Busselton-Augusta sheet in 1973 and the Collie sheet in 1974. My own mapping at this scale of the Moora/Hill River, Perth and Pinjarra sheets was published in 1979. A composite map from these sources showing the extent of the Banksia woodlands is shown in Fig 1. The mapping shows that Banksia woodland, while generally considered typical of the Swan Coastal Plain, is by no means confined to it. A substantial area, perhaps one third of the whole, occurs on the Dandaragan Plateau. There are even some minor occurrences east of the Darling Range. All these occur- rences are within the Darling Botanical District. Banksia wood- land occurs throughout the length of the Swan Coastal Plain, a distance of some 400 km north to south. Rainfall and tempera- ture as well as soil and topography vary throughout this range, so that the distribution of the woodlands varies from point to point. Banksia woodland reaches an optimum on the Plain between Muchea and Cataby, declining in both the northerly and southerly directions. In this central node it extends west onto the Spearwood Dune System and even reaches the sea in several places. It forms almost the total vegetation of the Plain here and extends widely over the Dandaragan Plateau immediately to the east. North of the central node the Banksia woodland is af- fected by declining rainfall so that it becomes restricted to the deeper sands. It is replaced by kwongan on the coastal lime- stone, and on the Dandaragan Plateau also except for a large extension from Moora to Watheroo. On the Coastal Plain the Banksia woodland finally pinches out near Jurien. The decline of the Banksia woodland is a gradual process and in many places there is a continuum from the woodland into Banksia-dominated kwongan which is much lower and more open than the wood- land proper. In mapping in such cases one has to exercise a sub- jective decision and to decide somewhat arbitrarily where a boundary should be drawn. This applies particularly to the northern limit of inland Banksia woodland about the middle of the Watheroo National Park. In a southerly direction the Banksia woodland is affected by increasing rainfall, the appearance of heavier soils, and of fre- quently more swampy conditions. South of Yanchep, tuart and tuart with jarrah form taller eucalypt woodlands on the coastal limestone. The Banksias persist as an understory. On the inland side of the Swan Coastal Plain south of Gingin there are heavier A67232-2 85 Journal of the Royal Society of Western Australia 71 (4), 1989 Figure 1 Distribution and extent of Banksia woodlands in southwestern Australia. soils under jarrah-marri woodland and a swampy belt which has (or used to have before clearing) a mosaic of eucalypt woodland, Banksia woodland, paperbarks and Casuarina obesa. South of Wanneroo the Banksia woodland, which has already acquired Casuarina fraseriana as a co-dominant since north of Yanchep, is further joined by scattered larger emergent jarrah trees, a structural type separately mapped. This occurs also at the southern extremity of the Dandaragan Plateau. At Wellard opposite Rockingham the last of the extensive stretches of Banksia woodland comes to an end, pinched out by the heavier soils at the foot of the escarpment and the increasing wetness of the Plain. From here southward the Banksia woodland forms a component of a tripartite mosaic with eucalypt and paperbark woodlands, or occurs as occasional patches within predominant eucalypt woodland. This situation is modified only in the ex- treme south where jarrah-marri forest (rather than woodland) is believed to have been the principal vegetation of the Coastal Plain. There are still some specific patches of Banksia on former coastal dunes, including one patch outside the Coastal Plain. Along the south coast from Cape Leeuwin east to Albany a different bioclimatic zone is entered, the Mesomediterranean where the dry season is shortened to 3-4 months. Banksias con- tinue of course to occur but mainly as components of other veg- etation. Banksia woodland as such may occur on limited areas of well-drained sand forming an intermediate belt between jarrah forest and swamps, or on low sandy ridges rising from swamps (F G Smith 1972). Small patches of typical Banksia low wood- land have been recorded in the Narrikup area (Beard 1979d). Apart from these instances sandy habitats are otherwise occu- pied. On south coast dunes Agonis flexuosa occurring in forms from mallee to well-grown trees is the dominant species, rather than Banksias. Outwash plains of white sand carry Casuarina fraseriana low forest in pure stands or mixed with Eucalyptus staeri if there is poor drainage. A novel Banksia community in this region however consists of B. uerticillata forming low forest under swampy conditions. Further to the east there is a transition to kwongan as there is in the northern sandplains with Banksias continuing as a con- spicuous element on deeper sandy soils as far east as Israelite Bay. References Beard J S 1979 Vegetation Survey of Western Australia 1:250 000 Series. Vegmap Publications, Applecross. 1979a Moora & Hill River; 1979b Perth; 1979c Pinjarra; 1979d Albany & Mt. Barker. Beard J S & Sprenger B S 1984 Geographical Data from the Vegetation Survey of Western Australia. Veg Surv WA Occ Paper 2. Vegmap Publications, Applecross. Diels L 1906 Die Pflanzenwelt von West-Australien sudlich des Wendekreises. Vegn Erde 7, Leipzig. Gardner C A 1944 The vegetation of Western Australia with special reference to cli- mate and soils. J R Soc W Aust 28:11-87. Havel J J 1968 The potential of the northern Swan Coastal Plain for Pinus pinaster Ait. plantations. Bull For Dept. W Aust 76. Heddle E M 1979 Mapping the vegetation of the Perth region. In: Western Land- scapes (ed J Gentilli) Univ WA Press. Pate J S & Beard J S (eds) 1984 Kwongan; the Plant Life of the Sandplain. Univ W Aust Press, Nedlands. Seddon G 1972 Sense of Place. Univ W Aust Press, Nedlands. Smith F G 1972-4 Vegetation Survey of Western Australia 1:250 000 series. Dept Agriculture, Perth. 1972 Pemberton & Irwin Inlet; 1973 Busselton & Augusta; 1974 Collie. Speck N H 1952 Plant ecology of the metropolitan sector of the Swan Coastal Plain. MSc Thesis, Univ W Aust. 86 Journal of the Royal Society of Western Australia, 71 (4), 1989, 87-88 Bassendean and Spearwood Dunes: their geomorphology, stratigraphy and soils as a basis for habitats of Banksia woodlands V Semeniuk 1 & D K Glassford 2 l 2l Glenmere Road, Warwick WA 6024 2 33 Rockett Way, Bull Creek WA 6155 Introduction Banksia woodlands and allied plant associations inhabit the extensive sandy soils of the Bassendean and Spearwood dune systems of the Swan Coastal Plain of SW Australia. These dune systems also have formed the basis for identifying vegetation as- sociations in relationship to landscape (Heddle et al 1980, Cresswell & Bridgewater 1985). However, to date, there has not been a systematic description of the dunes in terms of their landforms, stratigraphy, soils, inter-relationships and age struc- ture specifically for the purposes of landscape ecology, ie for de- lineation of vegetation habitats (cf Semeniuk et al 1989). This paper reviews the physical features of the Bassendean and Spearwood dunes as habitats for Banksia woodlands, describes some of the smaller scale variability of the dune systems as a basis for identifying potential habitats for vegetation, and suggests guidelines for identifying habitats. Description of geomorphology, stratigraphy and soils The Swan Coastal Plain is subdivided into 5 regional scale geomorphic units which are in narrow belts oriented north-south (McArthur & Bettenay 1960; Fig. 1). These units are: Ridge Hill Shelf - a landform underlain by Pleistocene laterite and sand; Pinjarra Plain - Pleistocene to Holocene fluvial landforms and sediments; Bassendean Dunes - undulating hills and degraded Pleistocene aeolian landforms underlain by yellow and white quartz sand; Spearwood Dunes - shallow to deep yellow quartz sand overlying shore-parallel ridges and depressions of Pleistocene aeolianite limestone (locally, there are ridges of yel- low sand with little or no underlying limestone); and Quindalup Dunes Holocene calcareous coastal dunes. The Bassendean and Spearwood dunes dominate the longitudinally central part of the Swan Coastal Plain. Bassendean Dunes exhibit high relief (up to 80m above AHD) in the northern Swan Coastal Plain, and low relief in the south. Locally in both settings linear ridges and mounds of sand have relative relief of 20-40m. In contrast, the Spearwood Dunes have a relief of c50-80m both in the northern and southern Swan Coastal Plain. Geomorphologically, the Bassendean and Spearwood dunes can be subivided into drylands and wetlands. Drylands can in- clude dunes and interdunes. Dunes can be progressively subdivided into high dunes and low dunes, then into dune crests, dune flanks, and then into north flanks and south flanks. Simi- larly, interdunes can be systematically divided into thick sand sheets and thin sand sheets, then into sand sheets with a deep water table and sand sheets with a shallow water table. Alternatively, the dunes may be divided on geometry into simple linear dunes, composite linear dunes, sinuous linear dunes, star dunes, dome dunes, parabolic dunes, barchan dunes, and lunette dunes (Fig.l). Wetlands can be classified ac- cording to CASemeniuk (1987). Stratigraphically, Bassendean Sand is equivalent to sands of the Bassendean Dunes, and Tamala Limestone is equivalent to the limestone portion of the Spearwood Dunes. The Tamala Limestone and Bassendean Sand interfinger along unconformity interfaces, the Bassendean Dunes may contain isolated lenses of limestone, and the yellow sands of the Bassendean and Spearwood dunes can be traced into each other (Fig.l). To date, the bulk of the yellow quartz sand Figure 1 Schematic diagram showing distribution and relation- ships of the geomorphic units across the Swan Coastal Plain, the relationships between the underlying stratigraphic units, and de- tails of the medium to small scale geomorphology of the Bassendean and Spearwood dunes as a basis for habitats of the Banksia woodlands. For details see Semeniuk & Glassford (1988) and Glassford & Semeniuk (1989). 87 Journal of the Royal Society of Western Australia 71 (4), 1989 overlying the Tamala Limestone has not been formally recog- nised as a separate formation, except in local areas (Glassford & Semeniuk 1989). Lithologically, yellow sands of the Bassendean and Spearwood dunes consist of quartz, felspar, heavy minerals and kaolin. A goethite-stained coating of silt-clay sized kaolin and quartz on the grains imparts the yellow colouration (Glassford & Semeniuk 1989). Information on the soils is difficult to review because the soil units have been inconsistently and imprecisely described, and inconsistently mapped; eg compare the definition of soil associ- ations, series and phases (Bates & Jackson 1987; and dis- cussion in Conacher & Dalrymple 1977), with the use of these terms in the local literature, and similarly, compare the various descriptions in map legends and texts of soil units which appear to largely overlap or be co-incident. Also, compare the different designations of landform-soil units near Mandurah (McArthur & Bettenay 1960, Churchward & McArthur 1980, McArthur & Bartle 1980), and the inconsistent location of the boundary of the Bassendean and Spearwood dunes in various maps. The definitions of the soil units do not differentiate between a stra- tigraphy resulting from complex and multiple pedogenic alter- ation, and the more simple, shallow products of pedogenesis superimposed on a complex primary stratigraphy, particularly where landforms are degraded, ie tending towards planation. Furthermore, primary sedimentary features and secondary alteration features frequently are not identified or separated. Further discussion of these aspects of landform-soil units of the Swan Coastal Plain are presented in Semeniuk (1989). Soil and landform-soil units (Churchward & McArthur 1980) are the features most used by phyto-sociologists to correlate vegetation with physical setting, and so we summarise infor- mation on the soil units, without implication that we accept these subdivisions. Soils of the Bassendean Dunes are subdivided into Bassendean and Southern River soil associations. Bassendean soils occur on dunes and interdune sand sheets, and comprise deep grey sands with humic or ferruginous layers. Southern River soils are similar to Bassendean soils, but differ in the oc- currence of sandy clay, clay and swamp deposits. Soils of the Spearwood Dunes are subdivided into Karrakatta and Cottesloe soil associations. Karrakatta soils comprise deep yellow sands over limestone, and are divided into yellow and grey soil phases. Cottesloe soils consist of exposed limestone, or shallow brown sands over limestone. The deeper stratigraphy underlying the Bassendean and Spearwood dunes, rather than the actual surficial soils, has been the basis for subdividing soil units. Separation of soil types on the Spearwood Dunes has a genetic basis, ie yellow sand is as- sumed to be a residual soil derived by leaching from limestone, and depth of yellow sand is used as an index of soil develop- ment. If yellow sand overlying limestone has not formed by leaching in situ, the actual soils within the Spearwood dunes are humic and bioturbated surface alterations of a parent yellow sand of variable thickness (Glassford & Semeniuk 1989). How- ever, even if it is assumed that yellow sands formed by in situ leaching, then these sands, which are mostly Pleistocene, are relict. Therefore, the parent material for present day soils is not limestone, but a relict, Pleistocene yellow sand. Accordingly, where limestone is covered by yellow sand, the soils of the Spearwood Dunes should be classed as humic and bioturbated quartz sand on a parent yellow sand. Where limestone is nearly exposed at the surface, the soils of the Spearwood Dunes should be classed as humic sands and rendzinas on limestone. Similarly, soils of the Bassendean Dunes have been separated on the basis of the stratigraphy underlying the dunes and interdunes, rather than on the actual surficial soils. Vegetation habitats The Banksia woodlands inhabit a dry, nutrient-poor, quartz sand terrain. In detail, other physico-chemical features may be habitat/vegetation determinants, particularly at the small scale (cf Semeniuk et al 1989). The following aspects can determine environmental conditions, and hence variations in habitats and understorey assemblages in the Banksia woodlands: location within dune type or interdune; depth to water table; aspect; or- ganic soil development; kaolin content; Fe-mineral content; colour; thickness of bleached soil; moisture content in the vadose zone; and subsurface stratigraphic or pedogenic fea- tures, particularly for their influence on hydrology, and reten- tion of water and nutrients. Some vegetation studies already have identified habitats, to various levels of detail, in terms of some of these edaphic features. Havel (1968) correlated veg- etation types to some detailed edaphic information such as as- pect, soil depth, soil moisture etc. Heddle et al (1980) recog- nized landform and soil as determinants of vegetation floristics and structure, but did not proceed beyond correlating broad as- sociations to the large scale geomorphic and soil systems of McArthur and Bettenay (1960) and Churchward & McArthur (1980). Cresswell and Bridgewater (1985) related vegetation associations within the Bassendean and Spearwood dunes to lo- cation on dune crests (tops), slopes, swales, an approach which effectively identifies location of habitat within dune terrain. However, we consider that to adequately address the variability of floristics in the Banksia woodlands of the Bassendean and Spearwood dunes, that phyto-sociologic studies should be undertaken in conjunction with a determination of the physical features listed above. It should be apparent from the above, that the geomorphology, stratigraphy, and soils of the Bassendean and Spearwood dunes, in our opinion, have not been documented in sufficient detail, using modern systematic approaches, at the full range of large to fine scales. However, the development of an adequate descriptive framework of these physical features of the dune terrains should be an essential first stage requirement both for purposes of landscape ecological studies and for the identification of resource variability within Banksia woodland systems. References Bates R L & Jackson J A 1987 Glossary of Geology. Am Geol Ins, Virg. Churchward H M & McArthur W M 1980 Landforms and soils of the Darling System, Western Australia In: Atlas of Natural Resources Darling System Western Aus- tralia. Univ W A Press. 25-33. Conacher A J & Dalrymple J B 1977 The nine unit landsurface model: an approach to pedogeomorphic research. Geoderma 18: 1-154. Cresswell 1 D & Bridgewater P 1985 Dune vegetation of the Swan Coastal Plain, Western Australia. J R Soc W Aust 67: 137-148. Glassford D K & Semeniuk V 1989 Stratification and disconformities in yellow sands of the Bassendean and Spearwood dunes. Swan Coastal Plain, southwestern Australia J R Soc W Aust (in press). Havel J J 1968 The potential of the northern Swan Coastal Plain for Pinus pinaster Ait. plantations. For Dept W Aust Bull 76. Heddle E M, Loneragan O W & Havel J J 1980. Vegetation complexes of the Darling System, Western Australia In: Atlas of Natural Resources Darling System West- ern Australia. U W A Press, 37-72. McArthur W M & Bartle G A 1980 Soils and land-use planning in the Mandurah- Bunbury coastal zone, Western Australia. CSIRO Aust Div Land Resour Manag Ser No 6. McArthur W M & Bettenay E 1960 The development and distribution of soils of the Swan Coastal Plain. Western Australia. CSIRO Soil Publ No 16. Semeniuk C A 1987 Wetlands of the Darling System - a geomorphic approach to habitat classification. J R Soc W Aust 69: 95-111. Semeniuk V 1989 An assessment of the geomorphology and soils of the Yoongarillup Plain, southwestern Australia. J R Soc W Aust (in press). Semeniuk V & Glassford D K 1988 Significance of aeolian limestone lenses in quartz sand formations: an interdigitation of coastal and continental facies, Perth Basin, southwestern Australia. Sediment Geol 57: 199-209. Semeniuk V. Cresswell 1 D & Wurm PAS 1989 The Quindalup Dunes: the regional system, physical framework and vegetation habitats. J R Soc W Aust 7 1 : 23-47. 88 Journal of the Royal Society of Western Australia, 71 (4), 1989, 89-90 Floristics of the Banksia woodlands J Dodd 1 & E A Griffin 2 J Western Australian Department of Agriculture, Baron-Hay Court, South Perth WA 6151 2 47 McMillan Street, Victoria Park WA 6100 Floristic studies are concerned with the botanical composition of vegetation. While the Banksia woodlands of the Swan Coastal Plain have been described in detail in recent vegetation surveys (eg Beard, this symposium, Heddle et al 1980), there is little published information on their floristic composition. The num- ber of plant species in Banksia woodlands is relatively large and approaches that of kwongan (sclerophyllous shrublands), but is substantially lower than the Jarrah forest (Table 1). However, the Banksia woodland and kwongan values refer to individual vegetation types, whereas the Jarrah forest total covers a var- iety of vegetation types experiencing a range of topographic, edaphic and climatic factors. The limited data available suggest that species richness (species per unit area) of Banksia woodlands is less than in most heathlands (George et al 1979, Griffin et al 1983) but more than in some forest and woodland types such as Wandoo (Griffin & Hopkins, unpublished data from Mt Lesueur) and York gum/Wandoo (Lamont 1984). Milewski & Davidge (1981) recorded a cumulative total of 77 shrub species after sampling 52 consecutive 2 m 2 quadrats, while Dodd (unpublished) measured a mean richness of 28 shrub species and 3 tree species in 83 400 m 2 stands of Banksia woodland. The species richness is also very variable (16-53 species per stand; Dodd, unpublished), reflecting variation in edaphic, climatic and geographic factors (Havel 1968). Broadly speaking, the Banksia woodlands are floristically rep- resentative of the State’s south-western flora, since their domi- nant families and genera (measured by number of species) are also the dominant taxa throughout the south west. Although a large number of families is represented in Banksia woodlands, most species belong to only a few. The families of woody plants with the greatest number of species are the Proteaceae, Myrtaceae, Papilionaceae and, to a lesser extent, Epacridaceae (Table 1). Amongst non-woody plants, the most important famil- ies are the Orchidaceae, Cyperaceae, Haemodoraceae, Anthericaceae (part of Liliaceae sensu lato) and Asteraceae. These families, except the Orchidaceae, are also prevalent in kwongan. Indeed, the similarities are such in some areas that Banksia woodlands could be considered as kwongan with a Banksia canopy (but see Beard & Pate 1984). As in kwongan, some genera are often represented by several species within a single stand (eg Banksia, Calytrix, Conostylis, Dauiesia, Hakea, Hibbertia, Petrophile and Schoenus). Other genera well rep- resented throughout Banksia woodlands are Acacia, Conospermum, Eremaea, Jacksonia, Leucopogon and Melaleuca. The dominant canopy species are Banksia attenuata and B. menziesii, with Eucalyptus todtiana and Nuytsia floribunda occur- ring less frequently. In some wetter stands, B. ilicifolia is present. In southern areas of the Swan Coastal Plain, E. calophylla, E. marginata and Allocasuarina fraseriana become increasingly im- portant and, eventually, dominate (Beard, this symposium) while B. menziesii is absent. Banksia prionotes may be present in some areas and is the dominant tree in woodlands on the Spearwood dunes near Jurien. The understorey shows much greater variation than the canopy. Species found commonly on both Bassendean and Spearwood dunes are Bossiaea eriocarpa, Eremaea pauciflora, Gompholobium tomentosum , Hibbertia hypericoides, Lyginia barbata, Petrophile linearis and Xanthorrhoea preissii (Dodd, unpublished). Some species occur frequently only on one dune system eg Calytrix flauescens, Conostephium pendulum, Hibbertia subuaginata, Leucopogon conostephioides, Patersonia occidentalis and Scholtzia inuolucrata on Bassendean dunes and Mesomelaena stygia, Petrophile macrostachya and Leptospermum spinescens on Spearwood dunes. None of these understorey or canopy species is exclusive to Banksia woodlands, however, and all can be found in other vegetation types, especially kwongan on sand. Many of the characteristic species also occur in the understorey of those coastal plain woodlands south of Perth, in which banksias form a secondary canopy beneath E. calophylla, E. marginata and A. fraseriana (Griffin, unpublished). Banksia woodlands lack floristic uniformity and, instead, consist of a number of different floristic types. Very few species are consist- ently found throughout the range of these woodlands. Only 13% of understorey species from 45 Bassendean dune sites and 1 1% from 31 Spearwood sites were found in more than 50% of stands surveyed by Dodd (unpublished). Havel’s (1968) study of the vegetation of part of the northern Swan Coastal Plain de- fined seven types of Banksia woodland which reflected differ- ences in topography and soil depth, moisture characteristics and degree of leaching. The two main factors that determined floristic composition, namely the degree of soil leaching and the moisture availability of the site (Havel 1968), have been found to apply to Banksia woodlands throughout the coastal plain (Dodd, unpublished). Conclusions Despite their simple structure and seemingly uniform appear- ance, Banksia woodlands are floristically rich and taxonomically diverse. Floristically, they appear to have close affinities to the kwongan of regions north of the Swan Coastal Plain. The wood- land understorey exhibits a high degree of variability indicating responses by the component species to a range of environmen- tal variables, of which edaphic factors are the most important. At the same time, the canopy shows little variation in compo- sition. Hence, on the basis of their understorey composition, Banksia woodlands can be divided into a number of floristic types (mostly undefined as yet) in terms of topography, soil type and moisture status and geographic location. The degree of flor- istic variation found in Banksia woodlands has significant impli- cations for conservation, since adequate conservation requires that the range of variation should be represented in reserves. 89 Journal of the Royal Society of Western Australia 71 (4), 1989 Table 1 Floristic composition of Banksia woodlands and adjacent vegetation types Families Genera Species Dominant families of woody plants Reference Banksia woodlands Jandakot 31 - 122 Myrt. (13/11)* Papil. (8/ 7) Prot. (7/ 6) Epac. (7/ 6) 1 Perth region 45 122 236 Prot. (29/12) Myrt. (19/ 8) Papil. (17/ 7) Epac. (11/ 5) 2 Perth region 57 - 377 + Prot. (33/ 9) Myrt. (33/ 9) Papil. (32/ 9) Epac. (26/ 7) 3 Swan Coastal Plain 59 78 187 Prot. (33/18) Myrt. (33/18) Papil. (27/14) Epac. (22/12) 4 Brookton Kwongan 24 63 98 Prot. (23/24) Myrt. (18/18) Epac. (5/ 5) Papil. (5/ 5) 5 Mt Lesueur 43 131 287 Prot. (46/16) Myrt. (33/12) Papil. (28/10) Mimos. (12/ 4) 6 Badgingarra 41 112 238 Prot. (51/21) Myrt. (30/13) Papil. (25/11) Epac. (13/ 6) 7 Eneabba 50 162 429 Prot. (71/17) Myrt. (55/13) Papil. (27/ 6) Epac. (19/ 4) 8 Eneabba 38 125 317 Prot. (61/19) Myrt. (37/12) Papil. (28/ 9) Epac. (15/ 5) 9 Tutanning Other - 315 Prot. (45/14) Myrt. (30/10) Papil. (23/ 7) Epac. (11/ 3) 10 York gum/Wandoo 36 . 85 Mimos. (4/ 5) Papil. (4/ 5) Prot. (3/4) Myrt. (3/4) 11 woodland Coastal heath 66 192 413 Myrt. (56/14) Prot. (42/10) Papil. (21/ 5) Mimos. (18/ 4) 12 Jarrah forest 95 784 Prot. (70/ 9) Papil. (68/ 9) Myrt. (63/ 8) Mimos. (37/ 5) 13 References: 1 Milewski & Davidge 1981 2 Speck 1952 3 Marchant et al 1987 4 Dodd, unpublished 5 Beard & Hnatiuk 1981 6 Griffin & Hopkins 1985 7 van der Moezel et al 1987 8 Hopkins & Hnatiuk 1981 9 Griffin et al 1983 10 Brown & Hopkins 1983 11 Lamont 1984 12 Wills et al 1989 13 Bell & Heddle 1988 ‘First value = number of species; second value = percentage of total species + Habitat descriptions suggest occurrence in Banksia woodlands Endemic and rare species have not been assessed fully for Banksia woodlands. A number of rare and endangered species are discussed by Hopper & Burbidge (this symposium). Until the regional variation of Banksia woodlands has been documented fully, the adequacy of existing reserves for en- compassing the variation remains unknown. References Beard J S & Hnatiuk R J 1981 A large remnant of yellow-sand kwongan near Brookton, Western Australia. W A Herb Res Notes 5: 1-4. Beard J S & Pate J S 1984 Foreword. In: Kwongan, Plant Life of the Sandplain (eds J S Pate & J S Beard). Univ W Aust Press, Nedlands. Bell D T & Heddle E M 1988 Floristic, morphologic and vegetational diversity. In: The Jarrah Forest (eds B Dell et al). Kluwer Academic Publishers, Dordrecht, 53-66. Brown J M & Hopkins A J M 1983 The kwongan (sclerophyllous shrublands) of Tutanning Nature Reserve, Western Australia. Aust J Ecol 8: 63-73. George A S, Hopkins A J M & Marchant N G 1979 The heathlands of Western Aus- tralia. In: Heathlands and Related Shrublands of the World, A. Descriptive Studies (ed R L Specht). Elsevier, Amsterdam, 211-230. Griffin E A, Hopkins A J M & Hnatiuk R J 1983 Regional variation in mediterranean type shrublands near Eneabba, south-western Australia. Vegetatio 52: 103-127. Griffin E A & Hopkins A J M 1985 The flora and vegetation of Mt Lesueur, Western Australia. J R Soc W Aust 67: 45-57. Havel J J 1968 The potential of the northern Swan Coastal Plain for Pinus pinaster Ait. plantations. For Dept W Aust Bull 76. Heddle E M, Loneragan O W & Havel J J 1980 Vegetation complexes of the Darling System, Western Australia. In: Atlas of Natural Resources, Darling System, Western Australia. Dept Cons Env, Perth. Hopkins A J M & Hnatiuk R J 1981 An ecological survey of the kwongan south of Eneabba. Western Australia. W A Wildl Res Bull 9. Lamont B B (ed) 1984 Natural resources of the Avondale flora and fauna reserve. W A Inst Tech School Biol Bull 6. Marchant N G, Wheeler J R, Rye B L. Bennett E M, Lander N S & Macfarlane T D 1987 Flora of the Perth Region. W A Herbarium, Dept Agric, Perth. Milewski A V & Davidge C 1981 The physical environment, floristics and phenology of a Banksia woodland near Perth, Western Australia. W A Herb Res Notes 5: 29-48. Speck N H 1952 The ecology of the metropolitan sector of the Swan Coastal Plain. M Sc thesis, Univ of W Aust. van der Moezel P G, Loneragan W A & Bell D T 1987 Northern sandplain kwongan: regeneration following fire, juvenile period and flowering phenology. J R Soc W Aust 69: 123-132. Wills R, Lyons M N & Bell D T 1989 The European honeybee in Western Australian kwongan - foraging preferences and some implications for management. Proc Ecol Soc Aust 15 (in press). 90 Journal of the Royal Society of Western Australia, 71 (4), 1989, 91-92 Water relations of Banksia woodlands J Dodd 1 & E M Heddle 2 ’Western Australian Department of Agriculture, Baron-Hay Court, South Perth WA 6151 2 E M Mattiske & Associates, PO Box 437, Kalamunda WA 6076 Introduction The interaction of the climate, soils and geology of the Swan Coastal Plain has an important bearing on the water relations of Banksia woodlands. The coastal plain experiences a dry Mediterranean climate (Beard 1981) with 5-6 dry months each year and receives 86% of annual rainfall between May and Oc- tober. The deep, leached sands that support Banksia woodlands have an extremely low water holding capacity and, conse- quently, virtually no water is available from the top few metres of soil during the summer months. Paradoxically, the presence of groundwater, usually at several metres depth, provides a po- tentially unlimited water supply for those deep rooted plants capable of reaching it. Water use by Banksia woodlands Water use by Banksia woodlands and other native vegetation is estimated to return 70-90% of average annual rainfall to the atmosphere through evapotranspiration and therefore has a sig- nificant impact on the amount of water available to recharge the groundwater body. Net recharge of groundwater, estimated to vary from 9% to >30% of annual rainfall depending on location, is affected by plant water use since a) plants remove water from the soil profile, preventing that water - which arrived as rain - from reaching the groundwater body; and b) some plants with deep tap roots make direct use of groundwater. Net recharge is greatest at upland sites where the water table is many metres below ground surface, and virtually inaccessible to even the deepest root systems. Rooting depth and root system morphology determine which source of water is utilized by plants. Excavation of the root sys- tems of shrub and tree species from Banksia woodlands demon- strated a diversity of rooting types and a range of rooting depths (Grieve 1956, Dodd et al 1984). At Gnangara, shallow root sys- tems that penetrated up to 1 m were the commonest, being found in 25 of the 43 shrub species examined, including Acacia pulchella, Calytrix fraseri, Eriostemon spicatus, Bossiaea eriocarpa, Hibbertia aurea, H. helianthemoides and several species of Leucopogon. Medium-depth root systems that pen- etrated 1-2 m were found in eight of the species, including Gompholobium tomentosum, Scholtzia involucrata, Adenanthos cygnorum and Conostephium pendulum. Wood- land plants with shallow or medium depth root systems would depend on profile-stored water. Deep tap roots, potentially capable of reaching ground water, were found in 13 species, in- cluding Banksia attenuata. B. menziesii and B. ilicifolia. Dauiesia triflora, Jacksonia densiflora, PetrophUe linearis, Calytrix flavescens, Melaleuca scabra. M. seriata and Stirlingia latifolia (Dodd et al 1984). This diversity of rooting depths and morphologies allows virtually complete occupation of the soil profile and leads to exhaustion of soil moisture reserves by late summer. The moisture requirements of a large number of Banksia woodland species have been defined by Havel (1968). The trees Melaleuca preissiana. Eucalyptus marginata. Banksia littoralis and B. ilicifolia occur in moist sites near swamps, in depressions and on lower dune slopes. Banksia attenuata and B. menziesii charac- terize the drier upper slopes and dune crests, while E. todtiana and Nuytsia floribunda show no distinct preference. Havel (1968) also defined a number of understorey species as indi- cators of particular combinations of site moisture character- istics, soil type and topography. The moisture requirements of the indicator species are reflected in the site conditions which they characterise. Responses to annual fluctuations in water supply The various seasonal patterns of transpiration and water stress which have been measured in a range of Banksia wood- land trees and shrubs closely reflect plant rooting depths and, consequently, the nature and longevity of the water supply (Grieve 1956, Dodd et al 1984). At a site with a water table at 6-7 m depth, high transpiration rates and relatively low levels of water stress were measured throughout summer in the deep rooted Banksia attenuata, B. menziesii, Stirlingia latifolia, Dauiesia triflora and PetrophUe linearis, indicating continuous ac- cess to groundwater. Eremaea pauciflora and Jacksonia densiflora also showed these features until the abrupt onset of severe water stress in mid summer, which was due possibly to their roots losing contact with the falling water table. Deep rooted plants are not obligate users of groundwater, however, since some of the species exhibited reduced transpiration with relatively high levels of water stress during summer at other sites which lacked accessible groundwater (Grieve 1956, Grieve & Hellmuth 1970). Shallow rooted species generally showed severe water stress during summer, the early onset and severity reflecting the shallowness of the root system and the exhaustion of moisture reserves in the soils above the groundwater body (Dodd et al 1984). One consequence of these differences in water relations is that maximum transpiration occurs during spring and early summer in the understorey, but during summer in the canopy. Responses of vegetation to changes in water availability Havel (1968) demonstrated that soil moisture conditions and the degree of soil leaching are the main determinants of the com- position of the vegetation of the Swan Coastal Plain. This study involved detailed measurements of species composition and 91 Journal of the Royal Society of Western Australia 71 (4), 1989 plant cover at a large number of Banksia woodland sites on the northern Swan Coastal Plain in the mid-1960s, before the start of groundwater extraction and, therefore, constitutes a major base-line survey for subsequent ecological investigations. Havel’s sites included nine transects which ran from swamp to dune crest, covering the range of topographical situations characteristic of the coastal plain. Four of these transects have been re-surveyed, first in 1976 (Heddle 1980) and at 1-3 year intervals subsequently, while several additional transects have been established to permit monitoring over a wider range of sites, including some close to groundwater pumping bores. Over the study period, since 1966, the Swan Coastal Plain has experi- enced a drought with 15 years receiving below average rainfall. The results reveal certain trends in soil moisture conditions and in the vigour and composition of Banksia woodlands. There has been a reduction in soil moisture contents and a lowering of the water table on a regional scale throughout the coastal plain. At the same time, soils of the upper slopes and crests of dunes have remained relatively dry. This drying of coastal plain soils has been largely due to a reduction in annual rainfall. Land-use changes and groundwater pumping have also had an impact at some locations. Changes in vegetation include: • a shift to drier types of vegetation with reductions in some tree and understorey species which tolerate wetter soils, eg Eucalyp- tus marginata, Banksia littoralis, Hypocalymma angusti folium and Regelia ciliata. This has been accompanied by replacement of older and larger trees with seedlings since 1966, as in B. littoralis and B. ilicifolia. Some trees, such as the paperbarks (Melaleuca spp), have responded to drought by producing additional stems. In contrast, at sites where soil moisture conditions have re- mained relatively stable, the composition and vigour of the can- opy have been maintained. • many species which tolerate drier soils or which are not site- specific in occurrence have maintained or increased their abun- dance. These include Banksia attenuata, B. menziesii, Gompholobium tomentosum. Hibbertia subuaginata and Leucopogon conostephioides. • certain short-lived understorey species have declined, prob- ably because of the lack of suitable conditions for germination and establishment. Many of these changes had been predicted in earlier studies by Aplin (1976), Havel (1975) and Heddle (1980). The major cause of these changes in the Banksia woodlands has been the long-term drought which has caused a lowering of water table levels and has affected vegetation on a regional scale. The ex- traction of groundwater has added to these effects in sections of the Gnangara Mound. Responses to ground water extraction Many of the vegetational changes likely to be associated with groundwater pumping have already been observed as re- sponses to long-term, regional drought. The major differences would relate to the rapidity, extent and permanence of ground water draw-down, which would affect the banksias and a num- ber of the understorey species. The majority of species are inde- pendent of groundwater, however, and should be little affected by pumping. Even where adult trees are killed, replacement from seed and from suppressed seedlings will occur, leading eventually to restoration of woodland. The resulting vegetation would probably have fewer and smaller trees and would re- semble undisturbed woodland on dry, upland sites. In order to predict the effects of groundwater extraction and water table lowering in greater detail, further information is needed on the extent of groundwater use by the plants of the Swan Coastal Plain. Conclusions Plant species of the Banksia woodlands exhibit a variety of physiological responses to changes in water availability. Amongst many understorey species, the intensity of water stress is inversely related to rooting depth. This relationship does not hold for the canopy banksias and certain deep rooted shrubs which utilize groundwater at some sites and which are in- dependent of soil-stored moisture. Knowledge of the water re- lations of plants from Banksia woodlands provides insights into the adaptations of plants in a Mediterranean-type environment and has helped explain the results of long-term monitoring, which in turn can be used for predicting the changes likely to occur in vegetation of the Swan Coastal Plain as a result of con- tinuing droughts and/or increased groundwater extraction. Water use by existing vegetation is a major component of the water balance of the Swan Coastal Plain and is, therefore, of di- rect relevance to planning the development and management of the plain’s groundwater resources. References Aplin TEH 1976 Consequences of variations of the water table level : vegetation and flora. In: Groundwater Resources of the Swan Coastal Plain (ed B A Carbon). CSIRO Div Land Resources Management, 126-139. Beard J S 1981 Vegetation Survey of Western Australia, Sheet 7, Swan. Map and Explanatory Notes. Univ W Aust Press, Nedlands. Dodd J, Heddle E M, Pate J S & Dixon K W 1984 Rooting patterns of sandplain plants and their functional significance. In: Kwongan, Plant Life of the Sandplain (eds J S Pate & J S Beard). Univ W Aust Press, Nedlands, 146-177. Grieve B J 1956 Studies in the water relations of plants: 1. Transpiration of Western Australian (Swan Plain) sclerophylls. J R Soc W Aust 40: 15-30. Grieve B J & Hellmuth E O 1970 Eco-physiology of Western Australian plants. Oecol Plant 5: 33-67. Havel J J 1968 The potential of the northern Swan Coastal Plain for Pinus pinaster Ait. plantations. For Dept W Aust Bull 76. Havel J J 1975 The effects of water supply for the city Perth, Western Australia, on other forms of land use. Landscape Planning 2: 75-132. Heddle E M 1980 Effects of changes in soil moisture on the native vegetation of the northern Swan Coastal Plain, Western Australia. For Dept W Aust Bull 92. 92 Journal of the Royal Society of Western Australia, 71 (4), 1989, 93-94 Fire in the Banksia woodlands of the Swan Coastal Plain A J M Hopkins 1 & E A Griffin 2 department of Conservation and Land Management, PO Box 51, Wanneroo WA 6065 2 47 MacMillan Street, Victoria Park WA 6100 Introduction As is the case for most remnants of native vegetation in south western Australia, fire is an important management consider- ation for the Banksia woodlands of the Swan Coastal Plain. The proximity of these woodlands to the major urban and semi-rural developments within the State ensures that issues of fire protec- tion have considerable prominence. Yet despite the ease of ac- cess for study purposes and the importance of effective fire management of these woodlands, relatively little is known of their fire ecology. In this paper we review briefly relevant infor- mation on this aspect and seek to provide guidelines for future management and research. Historical burning Judging by the records of observations compiled by Hallam (1979), there were concentrations of aboriginal people around the estuaries of the coastal plain from Moore River to Albany and perhaps further afield. These people did use fire: charcoal associated with artifacts in the Upper Swan archaeological de- posit gave dates of almost 40 000 years (Pearce & Barbetti 1981). The establishment of the Swan colony in 1829 would almost certainly have led to changes in frequency, intensity, seasonality, and spatial distribution of fires (regime, sensu Gill 1975, Hopkins 1985a). The nature of these changes can only be speculated although it is probable that useful information exists in various historical accounts. Some fire records have been kept by the Wanneroo District Office of the Department of Conser- vation and Land Management over the past 30 years; these show that there have been some fuel reduction burns, mainly in spring, and some wildfires, mainly in autumn and recurring at in- tervals of about 6-8 years. Fire fuels Burrows & McCaw (in press) have constructed fuel accumu- lation curves on the basis of extensive sampling of the Banksia woodlands in the Wanneroo area. After c 6 years the total avail- able fuel stabilized at between 6 and 8 tonnes ha' 1 oven dry weight. These levels of fire fuels seem surprisingly low es- pecially when compared with total above-ground biomass of Banksia ornata woodland at Keith, South Australia (460 mm annual rainfall) (Jones et al 1969). The curves, together with in- formation on fire behaviour in these fuel types, suggest that re- peated fuel reduction burning on a broad acre basis would be of limited practical value (Burrows & McCaw in press). Effects of fire on plants To provide this overview on the effects of fire regimes on the Banksia woodlands, we have focussed on selected aspects of the biology of the component species. This has enabled us to piece together the results of studies of fire on Eucalyptus-Banksia- Allocasuarina woodland at King‘s Park (Baird 1977), on Banksia woodlands at Mooliabeenee east of Gingin (R J Hobbs unpubl data) as well as studies on other aspects of Banksia woodlands (eg Dodd et al 1984). The plant communities that now make up the Banksia woodlands of the Swan Coastal Plain contain very few long-lived perennial plant species that regenerate only from seed following 100% crown scorch (Table 1). Only 6 of the 13 species identified are in the most vulnerable category (Hopkins 1985b), being fire sensitive and having seed storage on the plant in bradyspores. This feature, together with the general observation that signifi- cant areas of Banksia woodlands apparently in good condition still exist, indicates that the present-day plant communities com- prising these woodlands must be tolerant of a wide range of fire regimes. Generalizations about the impact of recurrent fire on the Banksia woodland communities could be developed by collect- ing data on the time it takes for species in Table 1 from germi- nation of seed to production of sufficient, viable seed to permit population replacement in the event of a further fire. In the ab- sence of detailed data, a rough guide of 2.5 to 3 times the time from germination to first flowering can be applied as a minimum between-fire interval if local extinctions are to be avoided (c/Gill & McMahon 1986). There are no data from these Banksia woodlands that indicate best season of burn for conservation but the study of Banksia burdetti at Watheroo National Park by Lamont & Barker (1988) may be indicative. That study shows best seed release, germi- nation and establishment after a hot fire in late summer/ autumn. Season of burn also has some bearing on fire intensity. Bur- rows (1985) showed that the extent of death of stems of Banksia grandis in the jarrah forest was a direct function of fire intensity: hotter fires kill more stems. A similar effect could be expected for Banksia spp on the coastal plain. A conspicuous, complicating factor in the process of develop- ing management guidelines is the likely invasion of burnt areas by weeds. As Keighery (this volume) points out, weeds are com- monly associated with disturbed sites including many with a his- tory of recurrent fire (see also Baird 1977, Bridgewater & A67232-3 93 Journal of the Royal Society of Western Australia 71 (4), 1989 Backshall 1981). Not only does weed invasion lead to loss of nature conservation values, it also can lead to a vicious spiral of degeneration of the vegetation through recurrent burning be- cause the presence of weeds alters the characteristics of the fire fuel bed, engendering an increase in flammability. Fire and animals The other important interaction associated with fire is be- tween plants and animals. Whelan & Main (1979) showed how grasshoppers can modify regenerating vegetation by grazing selectively on seedlings and presumably on other types of shoots. The impact of herbivores is particularly acute after small and/or patchy burns. Bamford (1986) also looked at effects of fire on invertebrates but more in terms of their role as a food resource for the ver- tebrates. Bamford’s study area was east of Gingin but sup- ported Banksia woodlands similar to those of the Swan Coastal Plain. He found that a spring fire reduced invertebrate numbers more than an autumn fire and suggested that the impact would be greatest on diurnal, terrestrial invertebrates and that this would have particular repercussions for the reptiles. This was not obvious in the trapping results; indeed the vertebrate fauna generally appeared to have coped quite well with the historical fires of the study area and the experimental fires. However, Bamford (1986) did observe that there had been some extinctions (three species of birds and perhaps some mammals) from the Swan Coastal Plain north of Perth which may be a consequence of the frequent, intense and extensive fires associ- ated with European settlement (see also Bamford & Dunlop 1984). The interaction between fires and weeds is also relevant when considering fauna. How & Dell (this volume) observe a decline in open area feeding reptiles with the invasion of the open areas by weeds. As noted above, weed invasion can be promoted by recurrent burning. Concluding remarks Because of the extensive clearing and disturbance of the Banksia woodlands of the Swan Coastal Plain these woodlands are now at a point where effective conservation and manage- ment is critical. Despite their proximity to Perth, these Banksia woodlands have been neglected scientifically; this applies as much to the issue of fire - its effects and its use in management - as to the many other important aspects of their biology. There is a real need to redress this situation. In respect of fire alone, the present-day plant and animal com- munities of the Banksia woodlands appear to be relatively ro- bust. They contain few plant species that we could describe as vulnerable. The fauna has also been shown to tolerate a certain regime of fire. We have, of course, no real insight into the extent of and reasons for disappearances of species from the Swan Coastal Plain in historical times. This apparent robustness is no justification for continuation of laissez-faire management; further species losses and degeneration are likely consequences of such an approach. It would be a relatively simple matter to develop fire manage- ment guidelines for these Banksia woodlands starting with the gathering of data on rates of regeneration of vulnerable plant species provided that we are prepared to accept the rule of thumb suggested here as the basis for those guidelines. How- ever, such an approach does not take into account the import- ant interaction that we have identified here - that is the interac- tion between fire (or any other form of disturbance) and weed invasion. Fire has the potential to promote weed invasion which in turn leads to increases in flammability of the vegetation and the loss of nature conservation values. This fire-weed interac- tion is probably the most important issue to be taken into ac- count in the development of any fire management strategies in the future. References Baird A M 1977 Regeneration after fire in King's Park, Perth, Western Australia. J R Soc W Aust 60: 1-22. Bamford M J 1986 The dynamics of small vertebrates in relation to fire in Banksia woodland near Perth, Western Australia. Ph D thesis, Murdoch Univ. Bamford M J & Dunlop J N 1984 The ecology of small mammals in patches of Banksia woodlands with particular reference to fire. In: The management of small bush areas in the Perth metropolitan region (ed S A Moore). Dept Fish Wildl, Perth, 54-7. Bridgewater P B & Backshall D 1981 Dynamics of some Western Australian ligneous formations with special reference to the invasion of exotic species. Vegetatio 46: 141-8. Burrows N D 1985 Reducing the abundance of Banksia grandis in the jarrah forest by the use of controlled fire. Aust For 48: 63-70. Burrows N D & McCaw W L (1988) Fire studies in Banksia low woodlands in Western Australia. I. Fuel characteristics. Unpubl rept. Dodd J, Heddle E M, Pate S & Dixon K W 1984 Rooting patterns of sandplain plants and their functional significance. In: Kwogan. Plant life of the sandplain (ed J S Pate & J S Beard). Univ W Aust Press, Nedlands, 146-77. Gill A M 1975 Fire and the Australian flora: a review. Aust For 38: 4-25. Gill A M & McMahon A 1986 A post-fire chronosequence of cone, follicle and seed production in Banksia ornata. Aust J Bot 34: 425-33. Hallam S 1979 Fire and Hearth. A Study of Aboriginal Usage and European Usurp- ation in South-western Australia. Aust Inst Aboriginal Studies, Canberra. Hopkins A J M 1985a Planning the use of fire on conservation lands in south-western Australia. In: Fire Ecology and Management in Western Australian Ecosystems (ed JR Ford). WAIT Environmental Studies Group Rep 14, W Aust Inst Technol, Perth 203-8. Hopkins A J M 1985b Fire in the woodlands and associated formations of the semi- arid region of south-western Australia. In: Fire Ecology and Management in Western Australian Ecosystems (ed J R Ford). WAIT Environmental Studies Group Rep 14, W Aust Inst Technol, Perth, 83-90. Jones R. Groves R H & Specht R L 1969 Growth of heath vegetation. III. Growth curves for heaths in southern Australia : A reassessment. Aust J Bot 17: 309-14. Lamont B B & Barker M J 1988 Seed bank dynamics of a serotinous, fire-sensitive Banksia species. Aust J Bot 36: 193-203. Pearce R H & Barbetti M 1981 A 38 000 year old site at Upper Swan, Western Aus- tralia. Archaeol Oceania 16: 173-8. Whelan R J & Main A R 1979 Insect grazing and post-fire plant succession in south- west Australian woodland. Aust J Ecol 4: 387-98. Table 1 Long-lived perennial plant species which occur in Banksia woodland communities on the Swan Coastal Plain (from the list compiled by Griffin & Dodd for this symposium) and which are killed by fire causing 100% canopy scorch and which regenerate only from seed. Species with seed storage on plant Species with seed storage in soil Banksia prionotes Adenanthos cygnorum Dryandra sessilis Astroloma xerophyllum Hakea trifurcata Leucopogon striatus Hakea obliqua Leucopogon cordatum Beaufortia elegans Lysinema ciliatum Beau/ortia squarrosa Astroloma heterophylla Acacia pulchella 94 Journal of the Royal Society of Western Australia, 71 (4), 1989, 95-96 Terrestrial invertebrate fauna J D Majer School of Biology, Curtin University of Technology, Kent Street, Bentley WA 6102 Introduction The interactions between certain plant and invertebrate species in Banksia woodlands have been reviewed by Byron Lamont in this Symposium. This paper reviews what little infor- mation is known about the invertebrate communities of this plant formation. It is arranged in the following way. First, some thoughts on the species composition of invertebrate communi- ties in Banksia woodlands are presented. This is then followed by a review of the effects of disturbance on certain components of the invertebrate fauna. Next, the few remaining invertebrate community studies which have been carried out in this veg- etation formation are reviewed and, finally, a prognosis for the conservation of Banksia woodland invertebrates is presented. Invertebrate communities in Banksia woodland In simplistic terms, the Banksia woodlands consist of small trees of about 6-8 m under which there is a well developed sclerophyll shrub understorey. This formation tends to be bounded on the western edge by tuart (Eucalyptus gomphocephala) open-forest and to the east by jarrah (E. marginata)-mani (E. calophylla) open forest. Both of these for- mations have a well developed tree stratum of at least 25 m height and, like the woodlands, the latter of the two formations has a well developed shrub understorey. The Banksia woodland therefore differs from the adjoining formations in lacking a taller Eucalyptus tree stratum, and from the tuart open-forest in pos- sessing a well developed shrub layer. The majority of the woodlands occur on the older Bassendean dune system which, in comparison with the adjoining soils, has soil of poorer nutrient status. Taking these factors into account, it is likely that the in- vertebrate community of the woodlands may differ from that of the adjoining plant formations. There are few published accounts of the composition of indi- vidual invertebrate faunas in Banksia woodland and nearby plant formations. Abbott (1982) surveyed earthworm distri- bution in the Perth metropolitan area. He found seven native species, of which only two were confined to undisturbed habitat, principally woodland or swampland. The other species were also found in habitats which had been modified by humans. In re- viewing the literature on earthworm distribution on the part of the Darling plateau close to Perth, Abbott (1982) found seven additional species which were absent from the metropolitan coastal plain area. It therefore appears that the coastal plain earthworm fauna differs substantially from that of the Darling plateau and this may be associated with differences in climate, soil type or other factors. Rossbach & Majer (1983) surveyed the composition of the ant fauna in two coast-to-Darling Range transects which ran through the various plant formations. They found that while many species occurred in a range of plant formations, some were confined to one type such as the Banksia woodland. An or- dination of the various sites in terms of their ant species compo- sition indicated that the jarr ah- Banksia woodland sites had a characteristic ant fauna, which was allied to that of the tuart open-forest and the Banksia- sheoak (Allocasuarina fraseriana)- prickly bark (E. todtiana) woodland. The ant fauna of the wood- land differed from that of the coastal scrub and the Darling pla- teau open-forest. On the basis of this rather limited information on invertebrate species composition in various plant formations, it seems likely that the overall invertebrate fauna of Banksia woodland is to some extent distinctive, although many species would probably be shared with adjacent plant formations. Although not peculiar to the Banksia woodlands, the inver- tebrate fauna of this region also exhibits a strong seasonality. Koch & Majer (1980) and Majer & Koch (1982) compared the seasonality of surface-active invertebrates at Reabold Hill in tuart-jarrah woodland, where some Banksia spp. were present, with that of jarrah open-forest at Dwellingup and Manjimup. Their data indicated that the various functional groups exhib- ited seasonal patterns which differed between the woodland and forests in their time or length of activity. This is well illustrated by the variation in duration of activity of slaters (Isopoda), a group which is involved in the decomposition of litter. The dur- ation of activity increased progressively from Reabold Hill to Dwellingup to Manjimup, being restricted to the wetter months at Reabold Hill and Dwellingup, but active throughout the year at the southern-most site. This trend appears to be related to the duration of rainfall, which is least at the Perth site. Effects of disturbance John Beard has already pointed out that some 61% of Banksia woodland is now alienated land where the original veg- etation has largely been cleared. A number of studies have looked at the effect of habitat modification on selected inver- tebrate groups in what was formerly Banksia woodland. Springett (1976a) looked at the species richness and popu- lation density of soil microarthropods in three Pinus pinaster stands at Gnangara and compared this with densities in the native woodland of the area. Whilst the species richness of selec- ted microarthropod taxa was generally halved by replacing the woodland with pines, the density of microarthropods in the pines was in the same range as that of the woodland. Accom- panying decomposition studies indicated that the less species- rich soil fauna of the pine plots was unable to decompose pine or sclerophyll litter as fast as the fauna of the woodland. Another fate of Banksia woodland is urbanization. Majer & Brown (1986) surveyed the ant fauna in 33 Perth gardens, most of which were situated in former woodlands. They found that 95 Journal of the Royal Society of Western Australia 71 (4), 1989 ant species richness was significantly reduced in gardens when compared to native vegetation. By quantifying the makeup and management practices of each garden and correlating these parameters with the ant fauna, they found that the variety of the ant fauna was enhanced by the length of time the garden had been established, by increasing the size of the garden and by providing a thick leaf litter and ground cover. Gardens where pesticides were used, where tall shrubs were dense or where management practices such as watering were intense, had a depauperate ant fauna. In view of the fact that the variety of ants tends to reflect that of other invertebrate groups, these con- clusions might also apply to other components of the inver- tebrate fauna. The remaining Banksia woodlands are subject to a number of pressures. One is the invasion of woodland by veldt grass ( Ehrharta calycina). Barendse et al. (1981) surveyed the spider fauna of King’s Park and found that areas colonized by veldt grass harboured a spider fauna of low abundance and species richness. Presumably this is either because the veldt grass occu- pies the feeding space which is normally used by spiders or be- cause it harbours less invertebrate prey items than the native vegetation. A widespread influence on Banksia woodlands is burning so, not surprisingly, a number of studies have been performed on this phenomenon. Springett (1971, 1976b) looked at the effect of fire on soil fauna in woodland at Gnangara which had been replaced by pines. Discussion of this study is outside the scope of this paper. Around the same period Bornemissza (1969) described post-fire changes in the soil fauna within the woodlands of King’s Park. Although this was an extensive long-term study, only an abstract of the results was ever published. It concludes that changes in numbers and species composition could be detected up to 5 years after a fire. Whelan, Langedyk & Ashby (1980) and Bamford (1986) looked at the effects of burning on surface-active invertebrates. Pitfall trapping was employed in both of these studies. The for- mer study, which was performed near Jandakot, found that in- vertebrate catch increased in the immediate post-fire period. Although this was in part an artifact of the sampling method, it does highlight the ability of a large component of the inver- tebrate fauna to survive a fire. Concurrent hand collections per- formed by Whelan et al. (op cit ) revealed that many animals sur- vived the fire by congregating in the crowns of Macrozamia riedlei and Xanthorrhoea preissii or under fallen logs. Bamford’s (1986) study was performed at Mooliabeenee and looked at both short-term (< 1 year) and long-term (1-22 years) impacts of fire. His monthly data collected within a year of burn- ing indicated that spring burning had a greater impact on the ‘ant’ and ‘other invertebrate’ categories than did autumn burn- ing. In the long-term, ant numbers declined to lower levels after reaching a peak one year after burning, while ‘other inver- tebrates’ progressively increased with time after fire as the litter layer built up. A more restricted program of sampling the understorey foliage invertebrates produced higher numbers 6 years after burning than those obtained after 23 years. This trend could be associated with the stimulation, and later the sen- escence, of vegetation after fire. The relationship between invertebrates and post-fire plant succession was examined in more detail by Whelan & Main (1979). This work was carried out in small and large area burns near Jandakot. Whilst grasshopper invasion was rapid in the smaller burnt areas, grasshoppers were absent from the larger areas for 1-2 years after fire. This absence of grazing grass- hoppers allowed escape of seedlings from predation, while in the smaller areas the susceptibility of seedlings to grazing was influenced by the relative palatability of species. Other studies Davidge’s (1979) study of the herpetological community of a Banksia woodland near Jandakot included an analysis of the diets of the various frogs and lizards. Analysis of gut contents re- vealed that with the exception of the frog Myobatrachus gouldii, a termite specialist, all species were opportunistic feeders which fed on a wide range of prey items including vertebrates, plant material and sixteen categories of invertebrates. The only other published community study from Banksia woodland which I am aware of is Bornemissza’s (1957) analysis of arthropod succession in carrion. This study, which was car- ried out in King’s Park, identified five different stages of carcass decomposition and these were correlated with the animal com- munities present in the decomposing tissue. Bornemissza also looked at the reinvasion of the carrion zone by soil arthropods and found that this was not complete after one year. Conservation of Banksia woodland invertebrates This review of the literature on terrestrial invertebrates has indicated how community composition can be altered by urbanization, veldt grass invasion, the frequency and season of burning, as well as by clearing of the original woodland for some new land use. Therefore, if the conservation of invertebrates in Banksia woodlands is to be catered for, we need to consider if, when, and how frequently to burn such areas and we also need more information on the impact of weed invasion on the terres- trial invertebrate fauna. The fragmentation of the remaining woodlands by agriculture, roads, urbanization and other land uses is also reason for concern. However, unlike vertebrates, the conser- vation of relict areas of only a few hectares in size can be ad- equate to preserve almost all the invertebrate species of the community (Key 1978). This, of course, assumes that the relict areas are managed in an appropriate way to maintain environ- mental quality. References Abbott I 1982 The distribution of earthworms in the Perth Metropolitan area. Rec WA Museum 10: 11-34. Bamford M J 1986 The dynamics of small vertebrates in relation to fire in Banksia woodland near Perth, Western Australia. PhD Thesis, Murdoch Univ. Barendse W J, Boulton A J, Collins L M, Craven L, Pusey B J, Sorokin L M & Ward B H R 1981 Spiders in King’s Park - an appraisal of management. Honours Thesis, Univ W Aust. Bornemissza G F 1957 An analysis of arthropod succession in carrion and the effect of its decomposition on the soil fauna. Aust J Zool 5:1-12. Bornemissza G F 1969 The reinvasion of burnt woodland areas by insects and mites. Proc Ecol Soc Aust 4:138. Davidge C 1979 The herpetological community of a Banksia woodland near Perth, Western Australia. PhD Thesis, Murdoch Univ. Key K H L 1978 The conservation status of Australia’s insect fauna. Aust National Parks & Wildlife Service Occ Paper 1: 24pp. Koch L E & Majer J D 1980 A phenological investigation of various invertebrates in forest and woodland areas in the south-west of Western Australia. J R Soc W Aust 63:21-28. Majer J D & Brown K R 1986 The effects of urbanization on the ant fauna of the Swan Coastal Plain near Perth, Western Australia. J R Soc W Aust 69:13-17. Majer J D & Koch L E 1982 Seasonal activity of hexapods in woodland and forest leaf-litter in the south-west of Western Australia. J R Soc W Aust 65:37-45. Rossbach M H & Majer J D 1983 A preliminary survey of the ant fauna of the Darling Plateau and Swan Coastal Plain near Perth, Western Australia. J R Soc W Aust 66:85-90. Springett J A 1971 The effects of fire on litter decomposition and on soil fauna in a Pinus pinaster plantation. Proc 4th Colloq Zool Committee Int Soc Soil Sci: 529-535. Springett J A 1976a The effect of planting Pinus pinaster Ait. on populations of soil microarthropods and on litter decomposition at Gnangara, Western Australia. Aust J Ecol 1:83-87. Springett J A 1976b The effect of prescribed burning on the soil fauna and on litter decomposition in Western Australian forests. Aust J Ecol 1:77-82. Whelan R J Langedyk W & Pashby A S 1980 The effects of wildfire on arthropod populations in jarrah - Banksia woodland. W Aust Nat 14:214-220. Whelan R J & Main A R 1979 Insect grazing and post-fire plant succession in south- west Australian woodland. Aust J Ecol 4:387-398. 96 Journal of the Royal Society of Western Australia, 71 (4), 1989, 97-98 Vertebrate fauna of Banksia woodlands R A How & J Dell Western Australian Museum, Francis St, Perth WA 6000 Banksia woodlands with their floristic richness, extensive flowering regimes, juxtaposition to other vegetation formations and extensive distribution on deep near-coastal sands are im- portant for vertebrates. However, no vertebrate species is unique to banksia woodlands. Amphibians Nine species may occur in Banksia woodlands (Davidge 1979, Murray 1980, Bamford 1986, A H Burbidge pers comm, How & Dell unpubl). Most require water for larval development, consequently their entire life cycle cannot occur in Banksia woodlands; they are captured throughout the year after rain, but activity peaks in spring (How & Dell unpubl). The two arboreal hylid species (Litoria moorei and L. adelaidensis) are seldom found far from water and thus occur as transients and then only peripherally in Banksia woodlands. Species of the leptodactylid genera Heleioporus. Limnody- nastes, Pseudophryne and Ranidella also depend on water, but have the ability to burrow while H. eyrei and L. dorsalis have been caught in Banksia woodlands several kilometres from the nearest water. The Turtle Frog Myobatrachus gouldii has direct develop- ment, ie no tadpole stage is involved (Watson & Saunders 1959, Robertsl981), and is independent of water for its development making it the only species capable of surviving entirely in Banksia woodlands. Reptiles Studies of major geographic regions (Kitchener et al 1980, McKenzie et al 1987) have found that few, if any, reptile species have disappeared since European settlement. The goannas Varanus gouldii. V. rosenbergi and V. tristis have become less nu- merous in Banksia woodlands due to habitat fragmentation and changing resource availability and there are no recent records of the Carpet Python Morelia spilota. Two arboreal geckos, Diplodactylus spinigerus and Phyllodactylus marmoratus are widespread and abundant, while the terrestrial, D. polyophthalmus and D. ornatus are in- frequently recorded (Davidge 1979, Murray 1980, Bamford 1986). Agamids whose distributions encompass the extent of Banksia woodlands ( Pogona minor and Tympanocryptis adelaidensis) are recorded on most sites, however, Banksia woodlands occur outside the main geographical distribution of most geckos and agamids. Skinks are the richest family with 10 genera. The arboreal Cryptoblepharus plagiocephalus is widespread and common as are the terrestrial Ctenotus fallens . C. lesueurii, Lerista elegans , Menetia greyi. Morethia obscura. M. lineoocellata . Tiliqua rugosa and the fossorial Lerista praepedita. Several species in Banksia woodlands are near the limits of their range; these include Ctenotus impar, C. schomburgkii. Egernia multiscutata, Lerista christinae. and the rare L. lineata. It is probable that Leiolopisma trilineatum. Egernia napoleonis and Omolepida branchialis only occupy those woodlands adjacent to denser and moister veg- etation types. Hemiergis quadrilineata and Lerista lineopunctulata, appear to be more common in woodlands occu- pying the coastal Spearwood Dune systems. The blind snake Ramphotyphlops australis occurs in many Banksia woodlands and elapid snakes are recorded in most Banksia woodlands that have been sampled for more than a year. Bamford (1986) recorded Demansia reticulata . Notechis curtus. Rhinoplocephalus gouldii. Vermicella calonotus and V. bertholdi at Mooliabeenie while at Bold Park (How & Dell unpubl) V. bertholdi. V. calonotus. V. bimaculata. V. fasciolata occur sympatrically with Pseudonaja affinis. The diversity of Vermicella can be explained by the abundance of fossorial and epigaic lizards which constitute their principal food source. The composition of the reptile assemblage reflects the sandy substrates of Banksia woodlands. Genera that are fossorial [eg Lerista and Vermicella) are well represented, while those that use burrows (eg Diplodactylus and Egernia) are poorly rep- resented. Litter inhabiting genera ( Hemiergis . Morethia. Menetia) occur in most habitats, although this is correlated with time since fire (Bamford 1986). In some isolated patches of Banksia woodlands extensive weed invasion (Keighery this publ) covers the ground between shrubs. This has severely impinged on reptiles which forage in the open between bushes eg Tympanocryptis adelaidensis and Ctenotus lesueurii. Dense rooting patterns may also inhibit the movement of near-surface fossorial species. Birds Bird studies have focussed on the importance of Banksia species to the maintenance and structuring of both nectar and insect feeding groups and entire assemblages. Bamford’s (1986) detailed study on Banksia woodlands of dif- ferent ages after fire provides the most detailed account of the annual composition of the avifauna. Bamford recorded 86 bird species within his study area at Mooliabeenie; 17 were princi- pally birds of Banksia woodlands, 48 occurred in both woodland and adjacent farmland and 21 occurred principally in cleared farmland. Of the 86 species, 19 were migratory or present for only part of the year, 24 were rare or uncommon vagrants, 35 showed marked seasonal variation and 8 slight variation in num- bers. The strong seasonality in bird numbers of these B. menziesii/B. attenuata woodlands resulted principally from mi- gratory species, particularly insectivores, moving in towards the end of spring to make use of increased invertebrate availability. Honeyeaters showed a bimodal peak in numbers of individ- uals (Bamford 1986) that corresponded to the peak flowering of B. attenuata (Dec. -Jan.) and B. menziesii (June-July). Insects are taken by all honeyeater species (Tullis et al 1982, Collins 1985) and form an important protein source, while nectar is the princi- pal energy source. Tullis et al (1982) indicated the importance of Banksia woodlands to honeyeaters in winter, but their study only covered the period April-July. Newland & Wooller (1985) compared nectar feeding and insectivorous honeyeaters and other insectivores in Banksia littoralis woodland and adjacent B. menziesii/B. attenuata woodland. They present important data on the flowering phenologies of Banksia and understory species in explaining contrasting richness and abundance in honeyeaters over the year. B. littoralis had more species and in- dividuals during the winter, when peak nectar occurred; adjac- ent woodlands had more constant numbers throughout the year resulting from overlapping flowering times of the dominant species. Large and moderate sized honeyeaters predominated in B. littoralis woodland while smaller species dominated in other Banksia woodlands. Migratory insectivores, eg Bronze Cuckoos, Bee-eaters, move into Banksia woodlands to take advantage of seasonal peaks in arthropods, while resident insectivores, eg thornbills,’ 97 Journal of the Royal Society of Western Australia 71 (4), 1989 silvereyes, retain relatively constant numbers throughout the year with fluctuations in numbers attributed to recruitment of the young (Newland & Wooller 1985). Most resident species breed during winter-early spring so that food resources are maximized for nesting and breeding when en- ergetic costs are greatest. Banksias themselves infrequently develop hollows conse- quently, parrots and cockatoos that are dependent on tree hol- lows for nesting rely on adjacent eucalypt woodlands, or use hol- lows in the few emergent eucalypts in Banksia woodlands. Fire in Banksia woodlands advantages those species which prefer to forage in open vegetation (Bamford 1986). These are generally colonizing species that can take advantage of modified environments. Small resident species, especially insectivores, are disadvantaged by a loss of the structurally diverse and dense understory species after fire. These birds are generally those that need special conservation measures since they are susceptible to local extinctions after major environmental modifications. Storr & Johnstone (1988) presented a list of birds of the Swan Coastal Plain and commented on status changes since European settlement. Unlike birds of other associations, such as Tuart for- ests and woodlands fringing lakes, Banksia woodland birds have not declined to the same extent. Mammals Range reductions of mammals since European settlement have been attributed to several causes, principally habitat alter- ation, changed fire frequency and predation. (Kitchener et al 1978, How et al 1987). Consequently, the fauna of Banksia woodlands has probably changed substantially since European settlement. The most abundant and widespread mammal in Banksia woodlands is the introduced Mouse Mus domesticus. Other species that have established feral populations since European settlement and which also occur in Banksia woodlands are the Black Rat Rattus rattus, Red Fox Vulpes vulpes, Ferret Mustela putorius, Cat Felis catus, and European Rabbit Oryctolagus cuniculus (Kitchener et al 1978). The Dingo Canis familiaris, Tammar Macropus eugenii , Quokka Setonix brachyurus and Woylie Bettongia penicillata may all have used Banksia woodlands but are unlikely to maintain viable populations in these habitats. Of the small native semi-arboreal mammals, the Honey Possum Tarsipes rostratus occupies many Banksia woodlands, while the Western Pigmy Possum Cercartetus concinnus may also occur in woodlands that have remained unburnt for a long period. Small native terrestrial mammals are never abundant in this habitat, although 3 species of dunnart, Sminthopsis dolichura . S.griseouenter, S. granulipes and the Ashy-grey Mouse, Pseudomys albocinereus, have been recorded in the more northern woodlands (Murray 1980, Bamford 1986). Little is known of the bats occupying Banksia woodlands as no systematic surveys have been done. However, the Lesser Long- eared Bat Nyctophilus geoffroyi frequently roosts under the bark of dead Banksia trees (Kitchener et al 1978). The Echidna Tachyglossus aculeatus probably feeds on termites within and adjacent to these woodlands. Similarly, the marsupial carnivore Chuditch Dasyurus geoffroii may have foraged in Banksia woodlands and adjacent areas. The Brushtail Possum Trichosurus uulpecula occurs in Banksias only where there are emergent eucalypt trees that provide the necessary hollows for dens; there are no records of this possum feeding on Banksia flowers, fruit or leaves. The Quenda Isoodon obesulus, occurs in Banksia attenuata/B. menziesii and swamp Banksia (B. littoralis) woodlands where these surround ephemeral swamps and lakes. The effect of fire on small mammals in Banksia woodlands has been documented by Bamford (1986). Fire coupled with habitat fragmentation, has greatly reduced the range and abundance of most species. Recent work in three isolated Banksia woodlands within the metropolitan area show that Mus domesticus occurs in all three and Isoodon in only one (How & Dell unpubl) despite the apparent suitability of these woodlands for other species. Conclusions The vertebrate fauna occupying Banksia woodland consists of species that generally have distributions focussed on the south- west of the state with its Mediterranean type climate. The great majority of vertebrates breed in late autumn to early summer and the young become independent when food is most abun- dant. Several species of reptile are more characteristic of semi- arid and arid parts of the state, and frequent more northern Banksia woodlands where the higher temperature and sandy soils are determining factors. Similarly some arid-zone bird species are moving into areas where Banksia woodlands occur adjacent to cleared land. Strong patterns of seasonal abundance are apparent in many species and these result from different causal mechanisms. For highly mobile bird species, numbers fluctuate in response to changing food resources such as nectar or insects. Reptile abun- dance reflects changes in temperature, while seasonality in am- phibians is associated with changing temperature and moisture regimes. Mammals show only slight seasonal patterns that gen- erally reflect recruitment into the population after breeding. Unlike many eucalypts, Banksia trees do not readily form hol- lows in limbs or trunks, consequently, species such as parrots and Brushtail Possums, which are obligate hollow users, seldom occur exclusively in Banksia woodlands. Land clearing and fire have both affected the composition of vertebrate species occu- pying Banksia woodlands. The frequency and time since burning can be shown to have a pronounced effect on species compo- sition; species dependent on vegetation that has been unburnt for long periods have been greatly reduced since European settlement. References Bamford M J 1986 The dynamics of small vertebrates in relation to fire in Banksia woodland near Perth. Western Australia. PhD thesis, Murdoch University. Collins B G 1985 Energetics of foraging and resource selection by honeyeaters in for- est and woodland habitats of Western Australia. NZ J Zool 12: 577-587. Davidge C 1979 The herpetofaunal community of a Banksia woodland near Perth, Western Australia. PhD thesis, Murdoch University. How R A Dell J & Humphreys W F 1987 The ground vertebrate fauna of coastal areas between Busselton and Albany. Western Australia. Rec West Aust Mus 13: 553-574. Kitchener D J Chapman A Dell J Muir B G & Palmer M 1980 Lizard assemblages and reserve size and structure in the Western Australian wheatbelt - some impli- cations for conservation. Biol Conserv 17: 25-62. Kitchener D J Chapman A & Barron G 1978 Mammals of the northern Swan Coastal Plain. In: Faunal studies of the northern Swan Coastal Plain, (ed R A How) W.A. Museum, Perth, 54-92. McKenzie N L & Robinson A C 1987 A biological survey of the Nullarbor Region - South and Western Australia in 1984. Dept of Environment & Planning, South Australia. Murray P J 1980 The small vertebrate community at Badgingarra, Western Aus- tralia. Honours thesis, Murdoch University. Newland C E & Wooller R D 1985 Seasonal changes in a honeyeater assemblage in Banksia woodland near Perth, Western Australia. NZ J Ecol 12: 631-36. Roberts J D 1981 Terrestrial breeding in the Australian leptodactylid frog Myobatrachus gouldii (Gray). Aust Wildl Res 8: 451-462. Storr G M & Johnstone R E 1988 Birds of the Swan Coastal Plain. Rec W Aust Mus Suppl 28. Tullis K J Calver M C & Wooller R D 1982 The invertebrate diets of small birds in Banksia woodland near Perth, W.A., during winter. Aust Wild Res 9: 303-09. Watson J A L & Saunders L M 1959 Observations on the reproductive system of the female Myobatrachus gouldii (Gray). W Aust Nat 7: 1-6. 98 Journal of the Royal Society of Western Australia, 71 (4), 1989, 99-100 Biotic and abiotic interactions in Bcmksia woodland Byron Lamont School of Biology, Curtin University of Technology, Perth WA 6001 Introduction The aim of this synopsis is to explore interrelationships be- tween representative plant species in banksia woodland and their environment. The two species selected are Banksia menziesii and Adenanthos cygnorum. Both are characteristic species of the Swan Coastal Plain, but extend further north on the coastal sands and, to a lesser extent, south-east in sandy pockets on the plateau (Nelson 1978, Taylor & Hopper 1988). Banksia menziesii This banksia flowers in winter, when few other species are available for nectar-dependent animals, especially honeyeaters (Newland & Wooller 1985). Florets open in response to foraging by honeyeaters (Ramsay 1988). These birds probably play a major role in pollen transfer as this species is self-incompatible (Scott 1980). There is minor consumption of pollen by staphylinid beetles (Ramsay 1988), while specialist moth and weevil larvae may feed on the flower heads (Scott 1982). Insect damage affects fruit set markedly. Cockatoos may feed on these larvae, giving some biological control (Cowling, Lamont & Pierce 1987, Lamont & van Leeuwen 1988). Less than 1% of florets produce follicles, for various reasons (Fig. 1). A beetle larva feeds only on the seed of B. menziesii (Scott 1982). This banksia stores little seed in its canopy, although the proportion increases northwards, corresponding to an increase in likelihood of fires reaching the canopy to release the seed (Cowling & Lamont 1985). Otherwise, mature seed is released in late autumn. The seed and young seedlings are highly nutritious (S Holman & B Lamont unpubl) and are eaten by granivores and herbivores (Fig. 1; Whelan & Main 1979, Cowling & Lamont 1987). Seedling establishment is negligible in the absence of a post-fire seed bed, and rare even in its presence (Cowling & Lamont 1987). This is an evolutionary tradeoff with the ability of the parent plants to resprout after fire. The first summer drought plays a major role in seedling death, but there are other causes as well (Fig. 1). Adult trees harbour many small animals, providing food and shelter, although there are no details in the literature (Fig. 1). Whole branches sometimes die and the loosened bark and dead wood are sources of food and shelter for additional small pri- mary consumers, carnivores and parasitoids. B. menziesii is one of the few species susceptible to a minor disease which causes a coralloid distortion of the branches. The symptoms are con- sistent with invasion by a mycoplasm carried by cicadelloid leaf- hoppers, but the phenomenon has received no study. The leaves drop after 4-5 years. The litter serves as food and shelter for decomposers but the rate of decomposition is ex- tremely slow. Proteoid roots from the surface laterals grow up into the fresh litter. These hairy rootlets probably short-circuit the nutrient absorption path by adhering directly to the litter particles and promoting nutrient release (Lamont 1986). The proteoid roots mat the surface soil and increase its stability for geophytes and other small plants, but make inter-fire seedling establishment less likely. The sticknest ant, Iridomyrmex coni- fer, uses the litter for nest-building and is a major ‘robber’ of banksia nectar - it is believed to site its nest near nectar-rich trees (R P MacMillan, pers comm). 1 Flower heads : Nectar and pollen for birds, insects, marsupials. Food for larvae -> cockatoos vl. Jtf/.lf, Inadequate pollen, insufficient nutrients, canopy fire in spring, larval borers C ^>X t/ u/ Seed release: /r ~ \autumn or fire no canopy fire, granivory by cockatoos Dead branches ,. ^ Food and protection 4 Adult tree : Nesting sites, food Litter : Food and habitat " l , for springtaSls\ mites, drought, (frequent) fire, herbivory, trampling y/} 3 Seed and seedlings : Food for insects, rodents, marsupials, birds Figure 1 Interactions between Banksia menziesii and other components of banksia woodland on the Swan Coastal Plain. Line through arrow refers to inhibition of the process. 99 Journal of the Royal Society of Western Australia 71 (4), 1989 Adenanthos cygnorum The woolly bush has many unusual features which give it a special place in the ecology of the region. It is a colonizer of soil- and fire-disturbed sites, growing as a thicket. This species is the tallest shrub and yet one of the few to be killed by fire. Unlike B. menziesii, the woolly bush flowers from spring to autumn, and has a major role in maintaining resident nectar-feeding birds throughout the year (Newland & Wooller 1985). The single flower is immersed in a terminal whorl of leaflets whose overlap- ping hairs, together with the constricted neck of the perianth, prevent access to nectar-robbing ants (Fig. 2). About 1-3% of the flowers are converted to fruits whose fate may follow three paths: (a) The green fruit is eaten by parrots on the plant, (b) The bracts surrounding the fruit open out pushing the leaflets into a cup from which the fruit drops to the ground (Fig. 2). Here it is eaten by granivorous birds and probably rodents, (c) The fruit usually remains in the cup where it is visited by ants. Most ants remove the fruits to their nests where they consume the basal elaiosome (Lamont & Grey 1984). The fruits remain dormant until the soil is disturbed or a fire occurs. Presumably in response to a change in the temperature regime (Brits 1987), most then germinate during the next winter from an average depth of 35 mm. As with banksias, drought probably greatly influences eventual recruitment into the new stand, but this has not been studied directly. Apart from the cotyledons, the foliage is fibrous and probably not very attract- ive to herbivores. However, there is a specialist phytophagous moth, Xylorycta sp., which webs together the terminal leaves. It is preyed upon by a parasitoid wasp, Campoletis sp. Seven of 10 xyloryctid pupae we hatched yielded these ichneumonid wasps (Grey & Lamont unpubl). The first leaves of the new season’s branchlets in mature shrubs bear extrafloral nectaries on their tips (Fig. 2). Seventeen species of ant, as well as the Campoletis sp. and other nectar- seeking insects, visit these glands. The nectaries are a reliable, albeit small, source of sugar for the predatory ants and wasps throughout the year. The location of the elaiosome-bearing fruits and xyloryctid larvae respectively are secondary and ir- regular events for these insects, but vital in maintaining the fit- ness of this species. A. cygnorum tends to collapse as it senesces and is readily invaded by termites which eventually consume the woody stems after death, before or after fire. Discussion There are major obstacles to the completion of one phase and commencement of the next in the life cycle of these two indi- cator species of banksia woodland. None of the threats to this fine balance as imposed by nature in the past are of the same order as currently caused by wholesale clearing for urban and rural expansion. For the remnants that escape the bulldozer, new obstacles remain, such as fertilizer and herbicide appli- cation or drift, weed invasion, Phytophthora dieback disease, trampling by hooves and shoes, frequent fire and unpre- cedented changes in soil water levels. As outlined here, the de- mise of these two species alone will take with them many depen- dent animals, to varying degrees. Although the honeyeaters readily switch to new nectar sources, some of the plant-insect re- lationships outlined here appear to be obligate. From a biologi- cal point of view, there is much to commend the conservation of this threatened ecosystem. 1 Flowers : Nectar for birds, insects Plant death Extrafloral inadequate pollen, insufficient nutrients, drought, fire 4 Seedlings : Food for marsupials? Food for birds, rodents Figure 2 Interactions between Adenanthos cygnorum and other components of banksia woodland. References Brits G J 1987 Germination depth vs. temperature requirements in naturally dis- persed seeds of Leucospermum cordifolium and L. cuneiforme (Proteaceae). S Afr J Bot 53: 119-124. Cowling RM & Lamont B B 1985 Variation in serotiny of three Banksia species along a climatic gradient. Aust J Ecol 10: 345-350. Cowling R M & Lamont B B 1987 Post-fire recruitment of four co-occurring Banksia species. J Appl Ecol 24: 645-658. Cowling R M, Lamont B B & Pierce S M 1987 Seed bank dynamics of four co- occurring Banksia species. J Ecol 75: 289-302. Lamont B 1986 The significance of proteoid roots in proteas. Acta Horticulturae 185: 163-170. Lamont B & Grey J 1984 Ants, extrafloral nectaries and elaiosomes on a pioneer species. Proc 4th Int Conf Mediterranean Ecosystems, Botany Dept, Univ W Aust, Nedlands, 89-90. Lamont B B & van Leeuwen S J 1988 Seed production and mortality in a rare Banksia species. J Appl Ecol 25: 551-559. Nelson E C 1978 A taxonomic revision of the genus Adenanthos (Proteaceae). Brunonia 1: 303-406. Newland C E & Wooller R D 1985 Seasonal changes in a honeyeater assemblage in Banksia woodland near Perth, Western Australia. N Z J Zool 12: 631-636. Ramsay M W 1988 Floret opening in Banksia menziesii R.Br.; the importance of nec- tarivorous birds. Aust J Bot 36: 225-232. Scott J K 1980 Estimation of the outcrossing rate for Banksia attenuata R.Br. and Banksia menziesii R.Br. (Proteaceae). Aust J Bot 28: 53-59. Scott J K 1982 The impact of destructive insects on reproduction in six species of Banksia L.f. (Proteaceae). Aust J Zool 30: 901-921. Taylor A & Hopper S 1988 The Banksia Atlas. Aust Govt Pub Serv, Canberra. 257 PP- Whelan R J & Main A R 1979 Insect grazing and post-fire plant succession in south- west Australian woodland. Aust J Ecol 4: 387-398. 100 Journal of the Royal Society of Western Australia, 71 (4), 1989, 101-102 Management of groundwater resources for protection of native vegetation Jeff Kite & Ken Webster Water Authority of Western Australia, 629 Newcastle Street, Leederville WA 6007. Introduction The Perth region, which has a population well over one mil- lion people, is underlain by significant fresh and brackish groundwater resources. These resources play a very significant role in the lifestyle of the people who live in the area. The in- creasing demand for these groundwater resources among com- peting users is requiring more effective planning and manage- ment strategies. Water Authority activities have considerable potential to im- pact on wetlands and Banksia woodlands, as direct impacts from the provision of water services such as the construction of water and sewer mains and the less obvious impacts from the management and development of the groundwater resources. The Resource Underlying the Swan Coastal Plain to depths of 14 km are large groundwater resources. These occur as the unconfined water-table aquifers which constitute the major Gnangara and Jandakot groundwater mound systems and as confined groundwater. Recharge to these systems is directly from rain- fall. Some of the confined groundwater is artesian to the extent that pressure heads are above the land surface. Whilst these groundwater sources are relatively large, at least in the local context, the increasing demands are such that care- ful management is required to ensure equitable use and to pro- tect water quality. Unlike many other commodities, manage- ment of Perth’s groundwater resources is undertaken in accordance with the concept of sustainable yield as defined under the State Conservation Strategy (Department of Conser- vation and Environment 1987). The concept of sustainable yield applies particularly to fresh resources and requires that abstraction is maintained within limits that can be sustained in perpetuity without adverse effects on the beneficial uses of the resource. Current Water Use Total water consumption within the Perth region (between Gingin and Rockingham) is close to 400 million m 3 annually of which approximately half is supplied by the Water Authority and the remainder is extracted by private users (Webster 1989). In 1985, 262 million m 3 were obtained from groundwater sources of which 206 million m 3 were extracted privately and the remainder was pumped by the Water Auth- ority from public water supply areas. The importance of the shallow unconfined aquifer can be seen by the fact that of the 262 million m 3 of groundwater used some 223 million m 3 /annum were from the shallow aquifer. This water is used for a range of purposes, namely private domestic irrigation, local authority and institutional parks and gardens, Water Authority public supplies, horticulture and in- dustrial and commercial purposes. Other important users of the shallow unconfined groundwater in the Perth metropolitan region are the wetlands and the areas of lowland vegetation including Banksia woodlands. Future Water Use The Source Development Plan for future development of water supply sources for Perth’s public water supplies identifies sources for the next 25 years of Perth’s development (Water Authority of WA 1987). Within this plan, groundwater sources continue to play a prominent role, with groundwater being expected to contribute in similar proportion to the current one- third of the total public supply. By the year 2000, based on current trends, it is estimated that the total water consumption in the Perth metropolitan region could be close to 550 million m 3 /annum of which 350 million m 3 /annum could come from groundwater resources. The value of these resources is considerable. If the fresh shal- low groundwater resources in the Perth region were valued en- tirely in terms of their potential for public water supply, their net present worth would be in excess of $1000 million. Their value in terms of other potential uses would be less in some instances, and more in others. The value in environmental terms is difficult to estimate but relates at very least to the high value of the qual- ity of life that we enjoy in Perth. The importance of groundwater to existing future public water supplies lies in its proximity to centres of demand. For ex- ample, the Gnangara Mound is adjacent to the North West Cor- ridor, and its attribute of large storages compared to replenish- ment rates allows water to be borrowed from these storages in drier periods. There are, however, some quite contrary aspects to these positive attributes. The proximity of groundwater to the urban demand centres is seeing pressure for the urban development to move onto the primary groundwater source areas with the con- sequential threats of degradation of these sources. The need to maintain groundwater levels within limits necessary to support environmental requirements is requiring the Water Authority to review its use of unconfined groundwater as a drought protec- tion strategy and to review its policy of firm licensed allocations to private users. These conflicting requirements of the groundwater resource increase the need for management of not only the water resources but of the land uses in the surrounding area. 101 Journal of the Royal Society of Western Australia 71 (4), 1989 The main pressure regarding groundwater quantity relates to the increasing competition from the users of these resources. Groundwater Allocation Strategy Deciding how to manage groundwater resources to meet in- creasing competition is not easy. In broad terms, the competing uses are public water supplies, private use, and the environ- ment. It is thus a matter of determining how a finite resource can be shared equitably between the users. In the past, the management of water resources of the State has been undertaken largely through the proclamation of water resource management areas ie Catchment Areas, Water Re- serves, Groundwater Areas and Public Water Supply Areas. This provides the Water Authority with statutory mechanisms to ensure that activities do not degrade the quantity and quality of water and to allocate the resource. To a large extent, the proclamation of water resource man- agement areas has also allowed for protection of the environ- ment. A good example of this is the Gnangara Water Reserve. This area of 835 km 2 on the Gnangara Mound between Gnangara Road and Gingin Brook was proclaimed in 1973. Although a considerable proportion of this area is pine plan- tation, the largest part is either State Forest or vacant crown land which is mainly vegetated with Banksia woodland. Large areas of the Gnangara Mound Banksia woodland are now pro- posed for more secure vesting in the Department of Conser- vation and Land Management’s latest regional management plan -eg Carabin, Wabling and Yeal Nature Reserves (CALM 1987). Without the proclamation which was originally aimed at protecting the water resource, some of these areas would prob- ably have been subdivided for agricultural purposes. Because of increasing competition for a limited resource, management of water resources is now significantly more com- plicated. Satisfying this emerging competition and resolving the conflicts are the fundamental objectives of the water allocation process. This process will become increasingly complex as com- munity water demands grow. For water allocation strategies to be successful, they will need to become part of a wider strategic plan. The traditional ap- proach of allocating water to specific individual users for specific purposes needs to become part of a process of developing re- gional land and water resource strategy plans. These strategy plans should include a hierarchical process that will ensure adequate protection of the water resource for the intended uses. The highest level within this hierarchical pro- cess should be allocation of priority beneficial uses to the water resources of a region or basin. Such an allocation defines the pri- mary desirable use or uses for a specific water resource in the long term. These priority beneficial uses fall into two broad categories of: • environmental and recreational uses; and • consumer uses (public and private water supply). The next level of allocation should specify the bulk allocations for particular uses within the identified priority beneficial use. An example of bulk allocation within the consumer uses cat- egory is between public and private water supplies where the usage by each is broadly constrained within specified annual ab- straction volumes. The lowest level of allocation that has been actively used in resource management is the individual user allocation or licenced allocation. It is proposed that these water allocation principles will be embodied into a Environmental Protection Policy under the En- vironmental Protection Act, for the Gnangara Mound. This will offer greater protection to both the water resource and the environment. Protection of Lowland Vegetation The Water Authority is very aware of its responsibility to manage groundwater levels to protect the environment. The Minister for Environment has set a number of conditions on the next phase in the development and management of the Gnangara Mound groundwater resources. These conditions are founded on ensuring adequate conservation and management of wetlands. In protecting wetlands, they also protect lowland vegetation surrounding wetlands. Future extensions of wellfields on the Gnangara Mound are aimed at keeping wells in upland areas where the vegetation is not dependent on the water table for its water supply. The Water Authority now monitors a significant number of vegetation transects on both the Gnangara and Jandakot Mounds. Monitoring commenced on some of the 14 sites in the Gnangara Mound as early as 1966 when the Forests Depart- ment established four transects. The monitoring has so far shown that climatic factors have the major effects on the Banksia woodlands with many areas being drought affected in recent years and succession occurring from water loving species to more drought resistant species. (Mattiske & Associates 1988). Certainly there have been some dramatic effects on local areas in the vicinity of wells which must be attributed to pumping. These are now being carefully monitored. It appears the main reason for Banksia deaths was the sudden drop in water table level occuring soon after the commissioning of wells. The Water Authority is looking at means of avoiding this such as developing the full yield for new wells over more than one year to give the trees time to adapt to lower water tables. An ad- ditional new transect has also been established radiating out from one of the new Pinjar bores to monitor the effect of the pumping strategy. Five vegetation transects have recently been established on the Jandakot Mound to provide baseline monitoring as part of the Environmental Impact Assessment for the Jandakot Groundwater Scheme Stage 2. Developing environmental cri- teria for valued areas of the environment including Banksia woodlands in Jandakot area is a key component of the early stages of this environmental assessment. References Department of Conservation & Environment 1987 A State Conservation Strategy for Western Australia. Bull 270. DCE, Perth. Department of Conservation & Land Management 1987 Northern Forest Region - Regional Management Plan 1987 - 1997. CALM, Perth. Mattiske E M & Associates 1988 Monitoring the effects of groundwater extraction on native vegetation on the northern Swan Coastal Plain. Prepared for Water Auth- ority of Western Australia. E M Mattiske & Associates, Perth. Water Authority of Western Australia 1987 Planning Future Sources for Perth’s Water Supply. Water Resources Planning Branch Rep No WP63. Water Auth- ority of WA, Perth. Webster K C (1989) Groundwater Management on the Swan Coastal Plain - An Overview. In: Conference Proceedings Swan Coastal Plain Groundwater Man- agement. (ed G Lowe) W Aust Water Resources Council, Perth, October 1988. WAWRC, Perth. 102 Journal of the Royal Society of Western Australia, 71 (4), 1989, 103-104 Urban development M D Poole State Planning Commission, 22 St George’s Terrace, Perth WA 6000 There is little doubt that urban growth and consolidation will result in deterioration of many of the natural plant communities along the Swan Coastal plain, even though the people respon- sible for directing that urban growth may do their utmost to pro- tect these plant communities (or at least representative samples) through the various legislative powers that are available. To illustrate the methods of controlling urban expansion, some history of regional planning needs to be outlined. In 1955, the State Government commissioned consultant town planner Gordon Stephenson to prepare, in conjunction with the Town Planning Commissioner, Alistair Hepburn, a re- gional plan for Perth and Fremantle. This was the state’s first at- tempt at regional planning. It did little for the Banksia woodland areas, other than identify them as suitable for “development as woodlands and forests”. , The Stephenson Hepburn Plan, formed the basis of the Metropolitan Region Scheme, which was gazetted in 1963. As a statutory Scheme, it was the first in Australia, and is still the basis for urban development and expansion within the Perth area. It zones land for industrial, rural and urban uses and re- serves land for a variety of purposes, including parks and recreation. As a policy guide to identify future expansion needs for Perth, the Corridor Plan was adopted in 1970. This, although some- what controversial, set down the basis for a series of urban corri- dors. The rural wedges between the corridors will experience significant pressures from growth that will affect the remaining areas of Banksia woodlands. During the late 1970s a Conservation-through-Reserves Committee was established by the Department of Conservation and Environment in an attempt to identify those areas of West- ern Australia that needed to be protected from development. The Darling System, System 6, covered the Perth area. Rec- ommendations made in the System 6 study have been endorsed by the State government, and will eventually be recognised in the Metropolitan Region Scheme. Several areas contain Banksia woodlands, but generally the wetlands received greater atten- tion by this committee. The future expansion of the region is now being re-examined as a result of a formal review of the Corridor Plan. This review is examining the various choices available for the region’s future growth. These are: 1 . Continued corridor growth as outlined in the Corridor Plan. 2. Peripheral Expansion. 3. Easterly Expansion. 4. Coastal Expansion. In order to make these decisions, there must be some aware- ness of the nature of the expansion and the constraints. The fol- lowing gives a broad basis for understanding the growth rates applicable to the Perth region. • The region’s population will continue to grow at a rate of be- tween 1.5 and 2% or some 30 000 people per year. This means an increase in excess of 800 000 people by the year 2021 . • Residential development is the largest consumer of land, re- quiring some 12 000 lots per annum or 1 500 ha. • There will be a need for up to 50 000 extra ha of land by the year 2021. • 97% of the metropolitan population lives in the urban zone. • There is expected to be a decline in the average number of persons per household from 2.84 in 1986 to 2.51 in 2001 and 2.38 in 2021. • The North west corridor is experiencing the highest rate of lot consumption comprising nearly 25% of the metropolitan total. • Increased population and urban expansion will place further pressure on the finely balanced and limited capacity of the metropolitan environment. The team undertaking the corridor review is using a con- straints mapping system to identify that land most suitable for urban development. The various constraints to development were identified and mapped. These include environmental con- straints such as: • The geomorphology of the coastal plain (V Semeniuk & D K Glassford - This publication). • The water resources. The groundwater resources in particu- lar are a constraint to further urban development, and a bonus to the further protection of the Banksia woodlands (J Kite & K Webster - This publication & J S Beard - This publication). • The natural vegetation (J S Beard - This publication). There is some natural vegetation remaining on the coastal plain, but even this is often contaminated with weed species, and dieback disease (G J Keighery - This publication, S D Hopper & A H Burbidge - This publication). • Significant landscape. These landscapes have been identified and plotted, and are considered of sufficient importance to be protected where possible. It is expected that the Metropolitan Region Scheme will be the primary vehicle for this protection. The above constraints and others were used in the prep- aration of a comprehensive constraints map and, as a result of this, areas for urban expansion were identified. The planning process does not stop there, as it involves public participation through submissions from individuals and groups that have studied the published concepts. These submissions will influence the final decisions on the future patterns of urban development. There are two choices available for future urban expansion in Perth. Either, the environmental and planning recommen- dations are adhered to and development is restricted to the already established urban corridors, and therefore protect the wetlands and the Banksia woodlands from urban development, OR peripheral urban expansion is allowed, which will result in eventual destruction of the wetlands and woodlands. Obviously every effort will be made to follow the first option, and to use the Metropolitan Region Scheme to reserve the valu- able environmental areas between the urban corridors. The Pro- posed Metropolitan Parks system, outlined in Figure 1, is the preferred means of protecting these areas. One wonders, how- ever, just how much protection will be forthcoming, given the 103 Journal of the Royal Society of Western Australia 71 (4), 1989 pressures that will result from adjoining residential commercial industrial and agricultural land uses, and from the following: • Groundwater use, (J Kite - This publication). • Disease (B L Shearer and T C Hill - This publication). • Weeds (G J Keighery - This publication). • Groundwater pollution (R Humphries - This symposium). In the report ‘Planning for the Future of the Perth Metropoli- tan Region’ (State Planning Commission 1987) it is commented that reservation and acquisition of is one of the successes of the Corridor Plan. The annual cost of maintaining such areas is now the major problem. There is a dearth of reliable information about recreation use patterns in and around Perth. Available evidence suggests that parts of the coastal foreshores, the Swan and Canning Rivers and Rottnest Island, are under the most intensive pressure. They are also the areas that are most attractive to tourists. However, increases in population, affluence and leisure time are likely to cause growth in demand for other regional-scale open space (as distinct from local open space) and for rec- reational facilities in general. Cultural influences, including greater awareness of the natural environment and health-based life-styles, are likely to result in greater political support for such facilities. The key issues for the future are the need for acquisition of more open space the better co-ordination of the planning and development of the regional open space system, and the intro- duction of a metropolitan system of management and adminis- tration. Funds will be needed for acquisition, development and maintenance of the metropolitan park system. Some revenue may be obtained from user charges in future. In broader terms, the Perth coastal plain is a delicately bal- anced environment, far less able to withstand the impact of long- term urbanisation than the sites of most other Australian cities. This fact has been recognised in the formulation of the Corridor Plan and by Stephenson and Hepburn in their 1955 Plan on which the Metropolitan Region Scheme was based. It is through the recognition of environmental constraints that the sandy coastal plain can be separated into patterns of urban and non- urban land, and that a continuously sprawling metropolitan area may be avoided in the long term. Although the ocean beaches, riverine environments and coastal wetlands system are under the greatest pressure from both use and development, the Banksia woodlands are by no means immune to this pressure. Their protection should con- tinue to be a high priority in a revised strategy. Adequate water resources are fundamental to the future de- velopment of the region and to maintain Perth’s pleasant life- style and attractive environment. Land-use has a significant im- pact on both the quantity and quality of water resources. The groundwater recharge areas are under the most immediate threat and their continued protection should be given a high pri- ority by the State Planning Commission, Water Authority and other responsible agencies. It is important that water quality be maintained and per capita consumption managed by encourag- ing land-use and development with low water demands, ensur- ing that groundwater is used efficiently, increasing recharge by surface drainage, and by containing pollution. There will also be a need to acquire more public land in the rural wedges to con- serve environmentally important areas (such as the Banksia woodlands), to add to the metropolitan open space system, and to assist in the management of available water resources. The establishment of a continuous regional open space chain has considerable potential for conserving the flora and fauna en- vironments of the region. The most important are the northern chains of wetlands in Wanneroo, similar habitats in the Beeliar and Serpentine wetlands and the native forests, and Banskia woodlands some of which still remain intact. Figure 1 Proposed Metropolitan Parks Source: State Planning Commission With careful management and restricted access it may be possible to retain some areas of Banksia woodlands in original condition. But the balance will, at best, be altered, as Kings Park has been altered to cope with the increasing demands of an urban population that has more leisure time for use in passive and active recreational. References Stephenson G & Hepburn J A 1955 Plan for the Metropolitan Region Perth and Fremantle. Government Printing Office, Perth, W. Aust. Environmental Protection Authority 1983 Conservation Reserves for Western Aus- tralia. The Darling System, System 6 1983. Dep Conservation & Environment, Perth, W. Aust. State Planning Commission 1987 Planning for the Future of the Perth Metropolitan Region State Planning Commission of Western Australian , Perth, W. Aust. Metropolitan Region Planning Authority 1970 Corridor Plan for Perth. Town Plan- ning Department, Perth, W. Aust. 104 Journal of the Royal Society of Western Australia, 71 (4), 1989, 105-106 Impact of horticulture on Banksia woodlands I R McPharlin & B A Stynes Division of Horticulture, Western Australian Department of Agriculture, South Perth WA 6151 Introduction Horticultural production in Western Australia was valued at $200M for the year ending 30/6/87 of which 44% ($90M) was vegetables, 35% ($70M) fruit and 20% ($40M) nurseries and cut flowers. The sands (Spearwood, Karrakatta and Bassendean) are important soils for the production of vegetables, flowers and to a lesser extent fruits such as citrus, avocadoes and straw- berries. The alluvial soils of the Pinjarra plain such as the neutral red earths (Belhus) in the Guildford/Swan area are preferred for the production of grapes. About 40% of Western Australia’s vegetables (2 726 ha) are grown on these sands in the Perth region (Moore to Serpentine River) of the coastal plain. The area planted to fruit on the sands (including some grapes) is much less significant being currently c 4.0% of total State plantings (301 ha). It has increased sub- stantially in recent years with a big increase in plantings of avocadoes (45% of WA) and citrus (especially oranges) near Gingin (20% of WA). There are about 360 ha of nurseries and flower crops (both sheltered and field plantings) representing about 36% of the State total. Most of the vegetables grown in the region are exported. They set the quality standard in Asia and distinguish the Swan Coastal Plain as one of the most important of vegetable pro- duction regions in the southern hemisphere. In this paper we summarize the impact of horticulture on the coastal plain, make some projections for the future, and discuss what is currently being done to reduce problems and minimize any in the future. Current Management and Impact Total area The total area of land in the Perth region of the coastal plain (Moore to Serpentine Rivers) is 380 000 ha or about 34% of the total area of the coastal plain (1.125 Mha)(Allen 1981). Nearly 80% of the total area (c 300 000 ha) of the plain in the Perth re- gion is sands (Bassendean, Karrakatta, Spearwood). The area of sands under horticultural crops in the Perth region is around 3 400 ha or just over 1.0% of the total area. This is small com- pared with the area of other activities such as urbanization and pine forests which currently cover c 55 000 and 22 000 ha re- spectively. The impact of horticulture on other components of the system (soils, water) is now considered. Soil Physical properties The soils of the Banksia woodlands have little natural struc- ture. The negative impact of horticulture on the structure of the Bassendean, Karrakatta and Spearwood sands has therefore been negligible. In fact the structure of these sands has almost certainly been improved by extensive additions of organic amendments such as poultry manure (10-100 m 3 ha' 1 ). Clearing land usually exposes the soil to erosion of some type (wind, water). The Bassendean (Jandakot, Joel, Gavin) sands have developed on wind blown material thus they are particu- larly vulnerable to wind erosion following the removal of veg- etative cover. Sprinklers are used to stabilize sandy soils during windy periods to prevent sand blasting of tender seedlings or rendering leafy vegetables unmarketable through direct wind damage or sand. There is some use of artificial windbreaks such as Paraweb® for protecting vegetable crops on the coastal plain but little use of trees. Chemical properties The sands of the coastal plain are highly leached and there- fore very infertile in their natural state with the Bassendean and Karrakatta sands being more infertile than the Spearwood sands. Almost all the macronutrients (N, P, K, S, Mg, Ca) and micronutrients (Cu, Zn, Mn, B, Mo, Fe) regarded as essential for plant growth must be added as inorganic or organic fertilizers for the production of vegetables, fruits and exotic flowers on the coastal sands. In addition these sands have low capacity to adsorb cations and anions. As these nutrients readily leach, it is not possible to build up a nutrient bank in these sands to any extent and regular applications of fertilizers for crop production are required. The sands differ in their capacity to adsorb P with the Bassendean sands having the lowest adsorption capacity (< 2 ppm P) and the Spearwood sands the highest (10-15 ppm P) (Ozanne & Shaw 1967). Nutrients not used by the crop are prone to leach- ing from the soil into ground and surface waters. This problem is exacerbated by over-irrigation. Bassendean and Karrakatta sands have been extensively leached of CaC0 3 and have low pH (^ 5. 0-6. 5) in their natural states. These soils require liming prior to fruit and vegetable production as only a limited number of crops are productive at low pH (potato, sweet potato, watermelon, rhubarb). The Spearwood sands are not as leached and are normally neutral to alkaline in reaction in their virgin state. Regular liming of Bassendean and Karrakatta sands is necessary to maintain pH when used for vegetable production. As the coastal sands have low capacity to hold either water or nutrients, salt build up is not a problem since it is readily leached. Water Quantity The annual consumption of groundwater from the superficial aquifer was 223xl0 6 m 3 pa in 1985/86 of which 38xl0 6 m 3 (17%) was for irrigated agriculture. The quantity used for horti- culture is c 32xl0 6 m 3 or c 15% of total ground water consump- tion. Thus the impact of horticulture on total water supplies is much greater than on total land area (c 1%) in the Perth region. The future of horticulture on the Swan coastal plain depends on an adequate supply of good quality (< 500-750 mg salts L' 1 ) 105 Journal of the Royal Society of Western Australia 71 (4), 1989 ground water. Currently about 66% of the total annual renew- able recharge of groundwater in the superficial aquifers in the Perth region is consumed (337 x 10 6 m 3 pa). Much of the re- maining 33% in the superficial aquifers would also be suitable for irrigation of most crops. Quality Groundwater sources in the Jandakot and Wanneroo licence areas are used for blending with dam water for drinking. Ferti- lizers leached from horticultural properties on sands may lead to increased levels of nutrients in ground water especially when ir- rigation is in excess of soil water holding capacity. Nutrients of most concern are nitrate, sulphate and salt. Nitrate and sulphate High levels of nitrate in drinking water have led to health problems such as methaemoglobinaemia in infants. Thus upper limits (10 mg L' 1 , NOo N) are put of the level of nitrate con- sidered safe in water destined for human consumption. Nitrate is very mobile in soils and is leached even in heavily textured soils. There is a positive correlation between urbanization, horti- cultural activities and nitrate levels in groundwater in the super- ficial aquifers on the Swan Coastal Plain (Cargeeg et al 1987). Nitrate in groundwater is higher (> 10 mg L’ 1 N0 3 , N) under suburbs with a high incidence of septic tanks [eg Applecross). Vegetable production has been implicated in high levels of ni- trate found in some bores (1-29 mg L’ 1 N0 3 , N) in the Gwelup groundwater area. Nitrate levels in water outside the urban areas is well below potable limits. Sulphate (S0 4 ) concen- trations in excess of 400 mg L' 1 are considered unsafe for drinking. High concentrations of sulphate (> 200 mg L’ 1 ) have been recorded in groundwater near the coast between Wood- man Point and Kwinana. This is probably associated with indus- trialization rather than any other activities. Fertilizers applied to vegetables may have resulted in elevated sulphate levels in the groundwater in some bores in the Gwelup area. Nevertheless these were well below potable limits of 400 mg L -1 . Salt High salt in water (^ 500-750 mg L' 1 ) severely limits the range of crops than can be grown. Very few vegetable crops (Asparagus, Silver beet) can tolerate high salinity. The best quality water (< 150 mg salt L’ 1 ) occurs at the crest of the Jandakot and Gnangara groundwater mounds. Salinity in- creases (> 500 mg L’ 1 ) from the crest of the mount to the coast. By far the biggest risk to the quality of groundwater in the Perth region of the coastal plain is the intrusion of the salt water wedge in coastal areas due to overpumping from the large number domestic (88 000) and agricultural bores. Surface Waters Oligotrophic surface water systems such as estuaries, rivers and lakes are characterized by low levels of P [ie usually < 0.1 mg L' 1 ). These systems are vulnerable to eutrophication if P inputs increase substantially. P L' 1 leached from soils where fer- tilizers have been applied to agricultural (pastures) and horticul- tural (vegetables) crops in the catchments of the Harvey, Murray and Serpentine Rivers have been implicated in the eutrophication of the Peel Inlet and Harvey Estuary. Whilst most (85%) of the P originates from fertilizers applied to agricul- tural land about 5% originates from vegetable land and 10% from intensive animal (piggeries, sheep holding yards) indus- tries. Given the high rates of P applied to horticultural crops such as vegetables (30-120 kg P ha’ 1 crop’ 1 ) cf pastures (8.5 kg P ha^yr’ 1 ; Kinhill Engineers 1988) any large scale expansion of horticultural activities on sands with very low P absorption ca- pacity (eg Bassendean sands) could see this quantity increase substantially. At present horticultural crops are grown on 536 ha respresenting 0.26% of the coastal plain portion of the catchment area. Only a small percentage of wetlands of the Banksia woodlands are close to their pristine condition (Halse 1988). Almost all the remaining wetlands are nutrient enriched ie P concentration of water in excess of 0.04 mg L’ 1 (Chalmers & Davis 1988). Horticulture along with intensive animal enterprises, industry and urbanization has contributed to this nutrient enrichment and associated algae pollution. Future Impact and Management Horticultural production in Western Australia is predicted to at least double by the year 2011. The greatest impact will be on the water resources (both quantity and quality) of Banksia woodland. Water quantity Demand for groundwater will intensify considerably above the current 32 x 10 m 3 pa with the predicted expansion in horti- culture. Horticulture will most likely be forced to access more unconfined groundwater either further north (Lancelin) or south (Scott River). Confined water from the Leederville and Yarragadee formations may increasingly be used for horticul- ture although supplies from this source are much smaller than the unconfined sources. Growers will be pressured to improve their efficiency of water use through crop selection and irri- gation management. Improved irrigation management will also aid in the reduction of nutrient leaching. Movement of horticul- ture expecially vegetables greater distances from the metropoli- tan area will increase costs (freight) and reduce quality (es- pecially leafy vegetables). Fertilizer Management and Water Quality Increased horticultural production will increase the pressure on groundwater and surface water quality. Improved fertilizer management to reduce nutrient pollution of ground and surface waters of the coastal plain is a major chal- lenge facing horticulture currently. Fertilizers have not tradition- ally been a major production cost (c 5-15% of direct costs of pro- duction) in horticultural crops and growers have tended to oversupply fertilizers since the financial penalties of under ferti- lization, especially with vegetables, are severe. Growers will be pressured to increase their efficiency of fertilizer management as water pollution presents a major cost to the community. Man- agement strategies likely to be employed include preservation of better quality soils (Spearwood rather than Bassendean sands) for vegetable production, more regular fertilizer appli- cations in smaller quantities (compared with less frequent and larger applications) to more closely match nutrient supply to crop demand and the use of soil and plant testing to monitor fer- tilizer programmes. The use of amendments high in Fe and Al oxides and hydroxides such as ‘red mud’ (Barrow 1982) and lateritic loams to increase the P absorption capacity of the Bassendean and Karrakatta sands may also be included in a strategy to minimize the impact of horticulture of the ground and surface waters of the coastal plain. Improved irrigation management (as mentioned above) will be an integral part of this programme. References Allen A D 1981 Ground water resources of the Swan Coastal Plain, near Perth, Western Australia. In: Groundwater Resources of the Swan Coastal Plain (ed B R Whelan). CSIRO, 29-80. Barrow N J 1982 Possibility of using caustic residue from bauxite for improving the chemical and physical properties of sandy soils. Aust J Agric Res 33: 275-85. Cargeeg G C. Boughton G N Townley L R, Smith G R, Appleyard S J & Smith R A 1987 Perth Urban Water Balance Study. Vol 1 — Findings WAWA. Chambers J M & Davis J A 1988 How Wetlands Work. In: Proc Swan Coastal Plain Groundwater Management Conference (ed G Lowe), WA Water Resources Council, 97-103. Halse S A 1988 Wetlands — Past and Present. In: Proc Swan Coastal Plain Manage- ment Conference (ed G Lowe). WA Water Resources Council, 105-112. Kinhill Engineers 1987 Peel inlet and Harvey estuary management strategy. ERMP — Stage 2. WADA, Dept Marine & Harbours. Ozanne P G & Shaw T C 1967 Phosphate sorption by soils as a measure of the phos- phate requirement for pasture growth. Aust J Agric Res 18: 601-611. 106 Journal of the Royal Society of Western Australia, 71 (4), 1989, 107-108 Forestry and Banksia woodlands on the Swan Coastal Plain E R Hopkins Department of Conservation and Land Management, Hackett Drive, Crawley WA 6009 Introduction From earliest settlement days the forests and woodlands of the Swan Coastal Plain were heavily exploited for their wood re- sources. The target species was jarrah used for fuel and pit sawn and later milled for heavy structural timber. Tuart was revered for its structural qualities and since early days has been man- aged for special timbers for flooring and waggon construction. Tuart management and preservation was a major concern of early forest management and State forests 1 and 2 were estab- lished for this purpose. Banksia was used as firewood but was of secondary interest. Early Management Apart from tuart management the first organized forestry at- tention saw the Coastal Plain as an opportunity to grow pines and reduce the heavy costs to the Colony of importing softwoods. For this purpose the Spearwood and Bassendean Dune systems offered the following advantages: 1 The apparently otherwise “worthless nature” of much of the woodlands for alternative economic purposes (ie for agriculture, urbanization). 2 The close proximity to the main centres of development (wood markets). 3 Relatively gentle topography and cheap clearing costs for plantation establishment. 4 The example of afforestation with Pinus pinaster in a cli- mate and on sands considered to be similar in the Gascony region of France during the period 1781 to 1864. The first initiative to convert areas of the Swan Coastal Plain woodlands to pine is recorded in an annual report submitted in 1987 by Ednie-Brown, Conservator of Forests. He noted that an endeavour should be made to favour plantations of some of the softwoods of commerce and thus make the Colony independent of outside supplies. Trial plantings were made that year on coastal sand dunes near Bunbury. They failed. Other sites were tried and early reports on the growth of pines at Hamel and Ludlow were enthusiastic. Pinus radiata was favoured because of its great success in South Australia. The early promise with coastal pine planting was not maintained and in 1916 most of the failed P. radiata was felled and replaced by P. pinaster. In 1916, Sir David Hutchins, a prominent British forester em- ployed to report on the forestry situation in Western Australia observed that “one problem was to fill up the sparse stocking of the jarrah forest with the stocking of timber that the climate fits it to carry and so improve the yield and straighten up the too often crooked jarrah” (Hutchins 1916). He noted that pines nat- urally suggest themselves and favoured P. pinaster of the Portu- guese provenance for the coastal sands. Plantation Development Three centres of interest were concentrated on for pine plant- ing on the coastal sands — Ludlow, Myalup and Gnangara. Ludlow operations were largely in conjunction with managing the tuart reserve. The few stands that remain are the best and were concentrated on the Tuart sands of the Spearwood Dunes types. Most of the Ludlow plantings were, however, on the poorer non-tuart sites on the Coolilup sands. These had a high banksia component in the original flora and have now been largely removed in a mineral sands operation. The Gnangara area was first considered as a pine planting proposition by C E Lane Poole in 1917. He was greatly influ- enced by the similarity of the sandy areas available to those of the Landes in Gascony. Some 3 100 ha were subdivided into compartments and a railway was surveyed from Bayswater to market the logs and firewood removed in clearing the sites. A number of trial plots were cleared and planted with P. pinaster raised in nurseries or direct sown. Plantings at Myalup and Gnangara concentrated on sands of the Spearwood (Myalup) and Bassendean Dunes types. Prob- lems with nutrition, establishment, genetics and spacing (Kessell 1927; Kessell & Stoate 1938, Perry 1939, Stoate 1939, 1946, Hopkins 1960a, b Perry & Hopkins 1967, Havel 1968, Butcher & Havel 1976; Butcher 1977a, b) needed to be overcome. Up to 1988 c 28 000 ha of pine plantation had been estab- lished on sands of the Swan Coastal Plain. Of this area c 10 500 ha and 5 730 ha are on grey and transitional sands of the Bassendean System and 11 000 ha are on yellow sands of the Spearwood System. Extension has now ceased due largely to the changing values placed on the land types by a developing urban population. Current Plantation Values Ludlow is now managed completely for conservation values of the native forests. Ludlow plantations will only remain until they can be converted to best suit the new national park role of the area. 107 Journal of the Royal Society of Western Australia 71 (4), 1989 Myalup-McLarty plantations have provided a small but stra- tegic source of pines which can be logged in winter when logging is difficult in the major P.radiata plantations of the Blackwood Valley. It is probable much of the P. pinaster will be converted to P.radiata to provide a valuable economic resource. For Gnangara, Pinjar and Yanchep the prime value of the area is now recognized as water production from the Gnangara mound (Butcher 1979a, b). A fibre board plant is projected for 1989 to provide for commercial thinning of the plantations and from approximately 1990, the saw log yield will increase to sup- port a significant milling industry. Whether these northern pines will be replanted for a second rotation to favour water manage- ment and wood production is yet to be determined. Recreation in these coastal plantations is an increasing ben- efit to population development. Impact on Banksia Woodland Havel (1968) has described the site types considered for pine planting in State forest 65 on the northern Swan Coastal Plain. On the Bassendean Sands and Transition Areas the Banksia dominated woodlands have proved unsuitable for commercial pine planting due to both low fertility and poor water relations. Generally only the sites carrying jarrah and marri have been suitable. On Spearwood Dunes, favourable sites containing tuart and or jarrah cease north of Yanchep and most areas planted are on banksia woodland types with deep yellow sands. In the absence of moisture associated species such as jarrah, tuart, B.grandis and B.ilicifolia, pine planting is of questionable value without supporting heavy thinning costs. Unplanted areas of the original forestry reserves are now managed as portion of the major conservation reserves to re- main on the Swan Coastal Plain. Apart from the plantation areas, conservation is now the highest value to be managed for. This is a challenging requirement as the increasing population, adjacent urbanization and high value for recreation renders management of fire and public access a major problem. Of 60 573 ha of State forest north of Perth, 23 050 ha have been converted to pine, 10 000 ha have been singled out for conservation reserves and some 27 500 ha of Banksia wood- land remain for general conservation and recreation purposes. For the 9 000 ha of State forest on the Swan Coastal Plain south of Perth, 4 000 ha have been cleared for pine plantation, 1 400 ha have been selected for conservation reserves and 3 600 ha are managed for general conservation and protection purposes. References Butcher T B 1977a Gains from the Pinus pinaster Ait. improvement programme in Western Australia. In: Proceedings, Third World Consultation on Forest Tree Breeding, Canberra. Butcher T B 1977b Impact of moisture relationships on the management of Pinus pi- naster Alt. plantations in Western Australia. For Ecol Manage 1:97-107. Butcher T B 1979a Management of Pinus pinaster plantations on the Swan Coastal Plain for timber and water yield. Aust Water Res Council Tech Paper 42. Butcher T B 1979b Growing sawlogs on the Gnangara Mound. Hydrology and Water Resources Sympo, Perth, 97-98. Havel J J 1968 The potential of the northern Swan Coastal Plain for Pinus pinaster Ait. plantations. W Aust Forests Dept Bull 76. Hopkins E R 1960a The fertilizer factor in Pinus pinaster Ait. plantations on sandy soils of the Swan Coastal Plain, Western Australia. W Aust Forests Dept Bull 68. Hopkins E R 1960b Variation in the growth rate and quality of Pinus pinaster Ait. in Western Australia. W Aust Forests Dept Bull 67. Hutchins D E 1916 A discussion of Australian forestry with special references to for- estry in Western Australia. W Aust Forests Dept Bull 5. Kessell S L 1927 Soil Organisms. The dependence of certain pine species on a bio- logical soil factor. Empire For J 6:70-74. Kessell S L & Stoate T N 1938 Pine Nutrition. An account of investigations and ex- periments in connection with the growth of exotic conifers in Western Australia plantations. W Aust Forests Dept Bull 50. Perry D H 1939 The effect of superphosphate on Pinus pinaster. Aust For 4:1 2- 14. Perry D H & Hopkins E R 1967 Importation of breeding material of Pinus pinaster Ait. from Portugal. W Aust Forests Dept Bull 75. Stoate T N 1939 Pine establishment. W Aust Forests Dept Bull 53. Stoate T N 1946 Pine establishment. W Aust Forests Dept Bull 53A. 108 Journal of the Royal Society of Western Australia, 71 (4), 1989, 109-110 Mineral resources and mining of the Spearwood and Bassendean Dune Systems J R Gozzard & M J Mouritz Geological Survey of Western Australia, and Mining Engineering Division, Department of Mines of Western Australia, 100 Plain Street, East Perth WA 6004 Introduction The development of Perth is very much dependent upon an assured access to reliable supplies of industrial minerals (sand, limestone, clay, gravel and hardrock) which are essential for road and building construction. The Banksia-dominated woodlands of the Spearwood and Bassendean Dune Systems contain all the limestone resources and some of the more significant sand resources of the metro- politan region (Fig. 1). Deposits of sand and limestone have always been thought to be abundant and freely available in the metropolitan area. These materials do occur extensively throughout the region, but the occurrence of economic deposits is limited, and proven re- sources are generally restricted to isolated pockets in specific geological units (Metropolitan Region Planning Authority 1984). Sand resources Geology Although several geological units in the Perth area contain sandy strata, most sand supplies come from the Bassendean Sand and Tamala Limestone which form the Bassendean and Spearwood Dune Systems respectively (Fig. 1). The main areas of extraction are Wanneroo, Gnangara, Beechboro, Ffenley Brook, Jandakot, Spearwood and Baldivis (Biggs 1979). The Bassendean Sand is typically yellow at depth beneath a surface cover of pale to dark grey humic sand. The sand com- prises fine - to medium - grained, general subrounded quartz with occasional feldspar and heavy minerals. It is moderately well sorted with a low silt and clay content. In the Spearwood Dune System a residual sand formed as a product of weathering of the underlying Tamala Limestone. It is humic-grey at the surface and yellow at depth, becoming orange close to the parent limestone bedrock. The sand comprises fine- to medium-grained, subangular to subrounded quartz with rare feldspar and heavy minerals, and is moderately well sorted with a small, but significant, clay and silt content. In general, it is coarser than the Bassendean Sand but the difference is slight and cannot be detected in all samples. Uses The major use of sand in Perth is for land fill - freeway con- struction, bridges, housing pads and rubbish disposal by sani- tary land fill all require sand filling. Sand is also used as bedding for pipes to prevent damage, especially in the hills area where soil movement may cause problems. Standard specifications re- quire that the sand be free of vegetable matter. Construction sands - those used in concrete, brick work and plaster - need to meet a set of standards relevant to their end use. These standards set permissible percentages of various grain sizes, the rate of water absorption, particle shape, and the amount of contained impurities. In Perth the most valuable sands are those with a high silica content, such as those found in the Gnangara and Jandakot areas. These are exported to Japan for glass manufacture and for some types of moulding, and to the Philippines for cement manufacture. Figure 1 Sand and Limestone resources of the Spearwood and Bassendean Dune Systems 109 Journal of the Royal Society of Western Australia 71 (4), 1989 Mining and treatment Because large areas of Perth are built on sand, there is a gen- erally held belief that supplies of sand are cheap and unlimited. However, as urbanization advances, potential resources are sterilized, existing pits are forced to close and relocate, and transport costs increase. In 1986 there were 49 actively worked sand pits in the metro- politan area. These were worked by a total of 32 operators, many of whom operated pits on a full time basis. The remaining pits were worked on an ad hoc basis according to the operators’ needs. It is partly due to the simplicity of pit operation that the number of operations is high and ex-pit prices are very low. Sand is the least expensive of all raw materials to extract, and the simplest to exploit. A typical operation involves a dozer, a front-end loader and a screening plant. Once the vegetation is cleared, the overburden is stripped and stockpiled for future rehabilitation. In the case of sands not requiring treatment, the sand can be loaded directly from the pit face to the truck. Construction and other specialized sands usually require either dry screening or washing to remove or- ganic matter and oversize material. Limestone resources Geology The Tamala Limestone contains all of the limestone re- sources of the Perth region (Gozzard 1987). It occurs as a series of ridges parallel to the coast, and most is dunal in origin although marine beds are also present. The limestone typically ranges between 50% and 90% calcium carbonate (CaC0 3 ). The magnesium content is normally low, ranging from 0.5% to 1.5% MgC0 3 , but in exceptional cases it may be as high as 3% MgC0 3 . Silica (Si0 2 ), in the form of quartz grains, is the only sig- nificant contaminant and usually exceeds 12% of the rock. Aver- age concentrations of minor constituents are: 1.1% AL0 3 , 1.1% Fe 2 0 3 , 0.5% K 2 0, 1.14% Na 2 0, and 0.013% Cl. The in situ moisture content is normally about 5%. The higher grade material is only found in isolated pockets within two areas. One is between Spearwood and Tamworth Hill, and the other is to the north west of Wanneroo (Fig. 1). Uses There are three industries that require high-grade limestone and cannot function with any substantial proportion of lower grade calcium feed. These are cement manufacture, lime pro- duction, and iron and steel smelting. The main use for high- grade material is in cement manufacture, which requires lime- stone with a CaC0 3 content of at least 80%. Currently the two cement companies (Swan Portland Cement and Cockburn Cement) operating in the metropolitan area use this rock to produce at least 600 000 tonnes of cement annually. The main uses of low- and medium-grade limestone include soft material for building, hard caprock for groynes and breakwaters, and rubble for road construction. Mining and treatment In 1986 a total of 24 operators had licences issued by local authorities to extract limestone within the metropolitan area. These operators include three local authorities and the Main Roads Department. Six of the operators had interests in three or more sites. In addition there are approximately 100 mining tenements for limestone extraction within the metropolitan area. The quarrying of limestone generally proceeds in stages, the first of which is the removal of overburden. This material, con- sisting principally of uncemented silica sand and variable quan- tities of caprock derived from limestone pinnacles, is normally used for the restoration of worked out areas. Initially the over- burden is stockpiled, but, as the quarry is developed, all over- burden is transferred directly into the worked-out areas to effect progressive restoration, thus avoiding double handling. Follow- ing this, weathered limestone is removed to expose the usable material. Whenever possible, higher and lower grade materials are blended in order to extend the life of a quarry but this practice is too expensive when the quarry is yielding only low- and medium-grade material. Low-grade limestone can be mixed with binders such as bitumen, lime, clay, fly ash, and Portland cement, and used in road construction. Environmental Aspects Approval to operate sand or limestone pits depends on the proposed use of the material and the tenure of the land. These factors will determine whether the operations are approved under the Mining Act or under the Extractive Industries By-laws of Local Government. If the location is in an area that is environmentally sensitive approval may also be required from the Environmental Protection Authority. In all cases the proponent is required to prepare a mine plan, operational guidelines, final landform and a rehabilitation pro- gram. The final landform and method of rehabilitation depends very much on the final use envisioned for the site. In some cases pits have been used for landfill waste disposal sites and in other cases suburban development has occurred over worked-out pits. In recent months an old limestone pit north of Wanneroo, which was in operation before the creation of the Neerabup National Park, has been re-contoured and re- habilitated to encourage forest regeneration to a standard ac- ceptable for return of the area to National Park status. The environmental aspects of these types of extractive indus- tries are presently under review by a Government Committee into Conservation and Rehabilitation in the Mining Industry. This committee will formulate recommendations aimed at en- suring that basic raw materials supplies are always available while ensuring that final landform objectives are met. Acknowledgements This paper is published with the permission of the Director of the Geological Survey and the Director of the Mining Engineering Division. References Biggs E R 1979 Sand in the Perth Metropolitan Area. W Aust Geol Survey Record 1979/6, 58p. Gozzard J R 1987 Limesand and limestone resources between Lancelin and Bunbury, Western Australia. W Aust Geol Survey Record 1987/5, 36p. Metropolitan Region Planning Authority 1984 Availability of Basic Raw Materials, Perth Metropolitan Region, An information base to complement MRPA Policy: Perth, Western Australia. 110 Journal of the Royal Society of Western Australia, 71 (4), 1989, 111-112 Banksia woodland weeds G J Keighery Western Australian Wildlife Research Centre, Department of Conservation and Land Management, PO Box 51, Wanneroo WA 6065 Introduction Floristic studies of Banksia woodlands usually mention intro- duced species (Milewski & Davidge 1981, Bell et al 1979, Foulds 1988), but no comprehensive survey has been under- taken of these taxa. This paper reports a survey of the natural- ized flora of 100 sites distributed between Mandurah and Moore River. Results 120 species were recorded as naturalized in Banksia woodlands within this region (Table 1). Most are Eurasian (chiefly Mediterranean) or South African in origin (Fig. la); However, a small but growing group originates in Eastern Aus- tralia and the Americas. This may have considerable impli- cations for future composition of this area’s weed flora. In life form the majority of weeds are annual or bulbous (chiefly the South African species) herbs, and the trees and shrubs are Aus- tralian species (Fig. lb). Half of the naturalized taxa were recorded at one (41 taxa) or two sites (19 taxa). Another 37 were located at fewer than 10 sites. Only 5 species (Ehrharta calycina, Ehrharta longiflora, Lagurus ovatus, Romulea rosea, Hypochaeris glabra and Ursinia anthemoides ) were recorded at more than 30 sites. Naturalized taxa occurred in a variety of disturbed areas or micro habitats within each site. Some were restricted to natural openings (Pelargonium capitatum) or moss swards (mainly Caryophyllaceae), edges of the remnants, tracks (43 taxa con- fined to track edges) or in litter under trees. Only 7 taxa were found to be abundant throughout remnants ( Avena barbata, Ehrharta calycina, Ehrharta longiflora, Romulea rosea, Gladi- olus caryophyllaceus, Pelargonium capitatum and Homeria flaccida). The major avenue of introduction of weeds was rubbish dumping and soil transportation. Spread within a site was pri- marily via too many tracks being created in each remnant. Other disturbance factors were present and past grazing, clear- ance, frequent fires and tree felling. From this survey the major weeds of Banksia woodlands are Ehrharta calycina, Avena barbata (chiefly in Banksia prionotes woodlands), Gladiolus caryophyllaceus, Pelargonium capitatum and Homeria flaccida (the last two mainly in Spearwood dune woodlands). Two special microhabitats are under threat; moss swards (Spearwood dunes invaded by small annuals) and deep litter under trees (invaded by Myrsiphyllum species, Freesia leichtlinii and Fumaria species). Management of Banksia woodlands should aim to lower dis- turbance and prevent further introductions occurring. References Bell D T, Loneragan W A and Dodd J 1979 Preliminary vegetation survey of Starr Swamp and vicinity; Western Australia. W Aust Herbarium Res Notes 2: 1 - 21 . Foulds W 1988 Ecology of Pinnaroo Valley Memorial Park. Western Australia: floristics and nutrient status. Kingia 1: 27-48. Green J S 1985 Census of the Vascular Plants of Western Australia. Dept Agriculture, South Perth. Milewski A V & Davidge C 1981 The physical environment, floristics and phenology of a Banksia woodland near Perth, Western Australia. W Aust Herbarium Res Notes 5: 29-48. Herbs (Tubers) Figure 1 Area of origin and life form of Banksia woodland weeds. Table 1 Species recorded as naturalized in Swan Coastal Plain Banksia woodlands. Figures in parentheses indicate number of times each species was recorded. .Gymnosperms Pinaceae Pinus pinaster Ait. (1) Angiosperms Monocotyledons Poaceae Aira cupiana Guss. (22) Aira caryophyllea L. (3) List is arranged in systematic order, after Green (1985). Avena barbata Link. (21) Avena fatua L. (6) Briza maxima L. (11) Briza minor L. (9) Bromus diandrus Roth. (17) Bromus hordeaceus L. (2) Bromus madritensis L. (1) Cynodon dactylon (L.) Pers. (2) 111 Journal of the Royal Society of Western Australia 71 (4), 1989 Ehrharta calycina Sm. (72) Ehrharta longi flora Sm. (74) Eragrostis curvula (Schrad.) Nees (2) Hordeum leporinum Link. (14) Hyparrhenia hirta (L.) Stapf. (1) Lagurus ovatus L. (37) Lolium rigidum Gaud. (17) Pentaschistis thunbergii Stapf. (4) Stenotraphum secundatum (Walter) Kunze. (1) Trachynia distachya (L.) Link (1) Vulpia bromoides (L.) Gray (7) Vulpia myorus (L.) C. Gmelin. (4) Asparagaceae Myrsiphyllum asparagoides (L.) Willd. (2) Myrsiphyllum declinatum (L.) Oberm. (1) Agavaceae Agave americana L. (1) Asphodelaceae Trachyandra divaricata (Jacq.) Kunth. (2) Hyacinthaceae Albuca canadensis (L.) F.M. Leighton (1) Lachenalia reflexa Thunb. (1) Lachenalia orchioides L. (1) Cyanella hyacinthoides L. (1) Iridaceae Babiana disticha Ker. Gawler (1) Babiana stricta (Ait.) Ker. Gawler (1) Chasmanthe floribunda (Salisb.) N.E. Br. (2) Ferrari a crisp a Bur man (1) Freesia aff. leichtlinii Klatt (4) Gladiolus angustus L. (4) Gladiolus caryophyllaceus (N. Burm.) Poir. (17) Gladiolus undulatus L. (1) Hesperantha falcata (L.f.) Ker. Gawler (1) Homeria flaccida Sw. (4) Ixia polystachya L. (2) Romulea rosea (L.) Ecklon (41) Romulea flava (Lam.) De Vos (7) Spar axis bulbifera (L.) Ker. Gawler (4) Watsonia aletroides (Burm. f.) Ker. Gawler (1) Orchidaceae Monadenia bracteata (Sw.) T. Durand et Schinz. (2) Dicotyledons Polygonaceae Rumex acetosella L. (1) Emex australis Steinh. (4) Phytolaccaceae Phytolacca octandra L. (1) Aizoaceae Carpobrotus edulis (L.) L. Bolus (5) Caryophyllaceae Aren aria serpyllifolia L. (1) Cerastium glomeratum Thuill. (3) Minuartia hybrida (Vill.) Schischkin (2) Petrohagia velutina (Guss) P. Ball et Heyw. (16) Polycarpon tetraphyllum (L.) L. (1) Sagina apetala Ard. (4) Silene gallica L. var. gallica (9) Silene gallica uar. quinqueuulnero (L.) Koch (1) Silene nocturna L. (1) Spergula arvensis L. (1) Stellaria media (L.) Villars (2) Fumariaceae Fumaria capreolata L. (3) Fumaria muralis Sond. ex Koch. (2) Brassicaceae Brassica juncea (L.) Czernj. (1) Brassica tournefortii Gouan. (7) Diplotaxis muralis (L.) DC. (1) Heliophila pusilla L.f. (11) Crassulaceae Crassula thunbergiana Schultes (1) Resedaceae Reseda alba L. (1) Fabaceae Lupinus angustissimus L. (4) Lupinus cosentinii Guss. (3) Medicago polymorpha L. (17) Ornithopus pinnatus (Mill.) Druce (1) Tri folium angustifolium L. (1) Trifolium arvense L. (3) Trifolium campestre Schreber (4) Trifolium dubium Sibth. (7) Tri folium glomeratum L. (9) Vicia hirsuta (L.) Gray (1) Vicia sativa L. (7) Geraniaceae Erodium botrys (Cav.) Bertol (2) Erodium cicutarium (L.) L’Her. (2) Erodium moschatum (L.) L’Her. (4) Geranium molle L. (3) Pelargonium capitatum (L.) L’Her. (12) Oxalidaceae Oxalis pres-caprae L. (4) Oxalis purpurea L. (2) Zygophyllaceae Tribulus ter rest is L. (1) Euphorbiaceae Euphorbia peplus L. (2) Malvaceae Malva parviflora L. (1) Sterculiaceae Brachychiton populneus (Schott) R.Br. (2) Myrtaceae Agonis flexuosa (Spreng.) Schau. (1) Eucalyptus citriodora Hook. (1) Eucalyptus maculata Hook. (1) Leptospermum laevigatum (Gaertn.) F. Muell. (1) Primulaceae Anagallis arvensis L. var. caerulea Gonan. (5) Anagallis arvensis L. var. arvensis (1) Gentianaceae Centaurium erythraea Rafn. (4) Lamiaceae Stachys arvensis (L.) L. (3) Solanaceae Solanum nigrum L. (2) Scrophulariaceae Dischisma arenarium E. Mey. (3) Dischisma capitatum (Thunb.) Choisy (2) Parentucellia latifolia (L.) Caruel (7) Kickxia spuria (L.) Dumort. (1) Orobanchaceae Orobanche minor Sm. (1) Rubiaceae Galium murale (L.) All. (1) Campanulaceae Wahlenbergia capensis (L.) A. DC. (29) Asteraceae Arctotheca calendula (L.) Levyns (7) Conyza bonariensis (L.) Cronq. (5) Cotula bipinnata Thunb. (1) Hypochaeris glabra L. (89) Hedynopis rhagioloides (1) Lactuca serriola L. (2) Osteospermum clandestinum (Less) Norlindh (14) Pseudognaphalium luteoalbum (L.) Burtt et Hillard (16) Sonchus oleraceus L. (5) Urospermum picroides (L.) Scop, ex F.W. Schmidt (6) Ursinia anthemoides (L.) Poir (94) Velleroeophyton dealbatum (Thunb.) Hilliard et Burt (4) 112 Journal of the Royal Society of Western Australia, 71 (4), 1989, 113-114 Diseases of Banksia woodlands on the Bassendean and Spearwood Dune Systems B L Shearer 1 & T C Hill 2 1 Dwellingup & 2 Como Research Centres, Department of Conservation and Land Management, PO Box 104, Como WA 6152 Current knowledge Diseases of Banksia woodlands have been a neglected area of plant pathology. Of the c 250 000 publications on plant diseases abstracted in the Review of Plant Pathology since 1922, only about 30 refer to diseases of Banksia. Only 6% of these 30 publi- cations refer to diseases of Banksia in woodlands compared with 55% for forest and 39% for Banksia species used in floriculture. Observations on the impact and spread of Phytophthora cinnamomi by Podger (1972) and Havel (1979) are the only pub- lished account of disease on Banksia woodlands of the Bassendean Dune system. Nevertheless, despite the lack of published information, disease is an important factor affecting the ecology of Banksia communities. Phytophthora species have been the most frequent cause of disease of Banksia (73% of the 30 publications) followed by wood rots (12%), leaf spots (9%) and Cylindrocladium scoparium (6%). The leaf spot Asterina systema-solare (Shivas in press), wood rots caused by Armillaria luteobubalina (Shearer & Tippett 1988), Ganoderma , Polyporus, Poria and Stereum (Hilton collection, WA Herbarium) and the canker pathogen Botryosphaeria ribis (Shivas in press) have been recorded on Banksia species occurring on the Bassendean Dunes. However the most destructive impact on the Banksia community of the Bassendean Dune system is disease caused by Phytophthora species, especially P. cinnamomi (Podger 1968, 1972). Phytophthora cinnamomi is distributed widely in Banksia woodlands of the coastal plain killing most of the overstorey and shrub layers in affected areas (Podger 1972). Incidence of dis- ease is greatest south of Perth, decreasing north of Wanneroo (Havel 1979). The Moore River National Park is the most north- erly known occurrence of P. cinnamomi on the coastal plain. Geographically restricted and susceptible B. laricina is being killed in affected areas in this park. The incidence of P. cinnamomi on the Spearwood Dunes is much less than on the Bassendean Dunes (Podger 1968), even though plant species are susceptible and disturbance from human activity is high (Havel 1979). Phytophthora cinnamomi is an introduced soil-borne fungus belonging to the Oomycetes or “water moulds”. As the name “water mould” suggests, the life cycle of P. cinnamomi depends on moist conditions that favour survival, sporulation and disper- sal of the fungus, and host infection. Warm, moist conditions and interactions with soil microflora favour vegetative production of sporangia and thick walled chlamydospores from mycelial strands in the soil or host tissue. Interaction of mycelium of dif- ferent mating types may produce thick-walled sexual oospores. However reproduction in soil is mainly by the asexual sporangium-zoospore cycle which produces large numbers of in- fectious spores when conditions are favourable. Sporangia release motile zoospores in free water. Zoospores can swim over short distances in water, but are mainly dispersed over large distances in flowing water or in infected moist soil moved by human activity. Zoospores in moist soil are chemotactically attracted to root surfaces where they germinate to produce germ tubes that penetrate roots. Infection by P. cinnamomi is probably favoured by the thin bark and prolifer- ation of rootlets associated with the specialized proteoid roots of the Banksia species occurring on the Bassendean Dunes. The fungus actively grows through root systems or is passively dis- persed in infected roots transported in soil. Root to root contact facilitates mycelial growth between root systems and initiation of new infections. The pathogen infects at least 1000 species of known hosts from taxonomically diverse families (Zentmyer 1980). The fam- ilies Epacridaceae, Myrtaceae and Proteaceae, important components of Banksia woodlands, contain many susceptible species. Figure 1 Destruction of Banksia woodland following infection by Phytophthora cinnamomi (light grey) compared with rem- nants of healthy woodland (dark grey) on a gently sloping Bassendean Dune between two swamp systems at Gnangara on October 1964. The infected area was healthy woodland prior to 1953. Scale 1:12 500. 113 Journal of the Royal Society of Western Australia 71 (4), 1989 Current research Following Podger’s (1968) initial monitoring of sites from Ravenswood to Moore River, little research has been done on the occurrence of P. cinnamomi in Banksia woodlands on the coastal plain. Investigations are now in progress to determine the factors influencing disease development, impact and methods of control of P. cinnamomi and other Phytophthora species in Banksia woodlands. Disease Development The spread of P. cinnamomi in 132 ha of Bassendean Dunes at Gnangara has been mapped from aerial photographs. The area includes 97 ha of Banksia woodland and 35 ha of ephem- eral swamp. Four small patches of dead vegetation totalling 0.15 ha occurred in 1942 alongside tracks that radiated from nearby strawberry farms. The area of infected woodland and swamp had increased to 55 ha by 1959, because of expansion of original infections, new infections from a nearby farm and contamination of the swamp system with associated destruction of low lying Banksia woodland (Fig. 1). These disease fronts ex- panded at 1.0 m yr 1 and 67 ha was infected by 1974, increas- ing to 82 ha (63%) in 1988. Slope or depth to water table did not appear to influence the rate of spread. However the rate of spread at Gnangara was slower than a mean downslope spread of 8 m yr 1 observed by Podger (1968) on a gently sloping dune of Gavin sand in the Bassendean system near North Dandalup. The pattern of disease development at Gnangara reflects the ability of P. cinnamomi to exploit various mechanisms of spread. Disturbance associated with market gardening, roads, tracks and off-road driving have resulted in the dispersal of P. cinnamomi in infected soil, and is responsible for the widespread distribution of the pathogen throughout the coastal plain. Active and passive dispersal of zoospores in free water contributes to spread within an area and results in the contamination of swampy areas. Zoospore dispersal in coarse-textured sands may also be assisted by movement of the water table or by lat- eral drainage from perched layers of saturated soil over clay or iron hardpan. We have recovered the fungus from groundwater at 3 and 5 m below the soil surface in affected Banksia woodland on Gavin sand near Hamel and south of Busselton. Growth of P. cinnamomi in roots of susceptible hosts ensures continued spread through summer, even though activity of the fungus in dry soil ceases. For example, P. cinnamomi can grow up to 1 cm day' 1 in roots of susceptible B. grandis in summer when tempera- tures are optimal for fungal growth (Shearer et al 1987). The destructiveness and persistence of P. cinnamomi in Banksia woodlands is partly determined by the ability of the pathogen to survive the dry soil environment over summer and resume activity when moist conditions return. The fungus sur- vived throughout the year in soil sampled from a depth of a metre in an affected Banksia woodland on Gavin sand south of Busselton. Recovery rates at depth from this site were often higher than those obtained from a high impact site in the jarrah forest. Infected host tissue provides a buffered environment for P. cinnamomi survival during dry conditions. For example, the fungus survived summer in 65% of colonized pine plugs buried at 30 cm in an affected area at Gnangara, even though soil moisture at this depth decreased to 0.6% in February. Impact Phytophthora cinnamomi infection destroys the structure and diversity of Banksia woodland. The dominant overstorey of B. attenuata . B. ilicifolia and B. menziesii is killed and only scattered Eucalyptus todtiana and Nuytsia floribunda remain in affected areas. Many understorey shrub species are similarly affected. Species richness in 64 m 2 quadrats decreased from 56 species in healthy woodland to 41 species in an affected area. Biomass can be reduced by up to 90% following infection (Fig. 1). Despite the impact of P. cinnamomi on Banksia woodland, information is lacking on the long term structural and floristic changes in affec- ted areas. Other Phytophthora species Phytophthora citricola, P. cryptogea (Aj), P. megasperma var. megasperma and P. megasperma var. sojae have been isolated from dying vegetation on Bassendean Dunes north of the Moore River. Many of the affected areas were low-lying and seasonally inundated or received off-road drainage. The susceptibility of native vegetation to these Phytophthora species needs to be de- termined before their relative significance to the health of native plant communities can be accurately evaluated (Shearer et al 1988). Control Eradication of P. cinnamomi from spot infections by Ridomil and fumigation with formaldehyde is being assessed in Jandakot sands of the Bassendean Dunes at Gnangara with promising re- sults. In addition, the systemic fungicide phosphorous acid has arrested lesion extension in B. grandis. Evaluation of these con- trol methods is continuing. Conclusions Systematic surveys of Banksia woodlands in southwestern Australia are needed to address the lack of information on dis- eases of Banksia. Phytophthora cinnamomi and other Phytophthora species are major factors affecting the ecology and management of the di- verse, but susceptible, Banksia communities on leached sands. Information is lacking on the specific requirements for pathogen survival, sporulation and spread as well as host infection and susceptibility in sandy soils. Such information is essential for the development of hazard and risk systems to minimize introduc- tion and spread of Phytophthora species. Knowledge of the diversity of Banksia woodlands, similar to the site-vegetation classification of Havel (1979), is needed in the development and application of hazard and risk systems. Long term effects of Phytophthora spp on diversity clearly needs to be quantified. An understanding of the low incidence of P. cinnamomi on Spearwood Dunes may provide clues for the control of the dis- ease. Control strategies must be developed and applied to pre- vent spread and intensification of disease favoured by disturb- ance caused by increasing urbanization and sand mining. References Havel J J 1979 Vegetation: Natural factors and human activity. In: Western Land- scapes (ed J Gentilli). Univ W Aust 122-152. Podger F D 1968 Aetiology of jarrah dieback. A disease of dry sclerophyll Eucalyptus marginata Sm. forests in Western Australia. M Sc thesis, Univ Melbourne. Podger F D 1972 Phytophthora cinnamomi. a cause of lethal disease in indigenous plant communities in Western Australia. Phytopathology 62: 972-981. Shearer B L & Tippett J T 1988 Distribution and impact of Armillaria luteobubalina in the Eucalyptus marqinata forest of south-western Australia. Aust J Bot 36: 433-445. Shearer B L, Shea S R & Deegan P M 1987 Temperature-growth relationships of Phytophthora cinnamomi in the secondary phloem of Banksia grandis and Euca- lyptus marginata. Phytopathology 77: 661-665. Shearer B L, Michaelsen B J & Somerford P J 1988 Effects of isolate and time of in- oculation on invasion of secondary phloem of Eucalyptus spp. and Banksia grandis by Phytophthora spp. Plant Disease 72: 121-126. Shivas R G (in press) Fungal and bacterial diseases in Western Australia. J R Soc W A. Zentmyer G A 1980 Phytophthora cinnamomi and the diseases it causes. Monograph 10. American Phytopathological Soc, St Paul. Journal of the Royal Society of Western Australia, 71 (4), 1989, 115-116 Conservation status of Banksia woodlands on the Swan Coastal Plain Stephen D Hopper & Allan H Burbidge Western Australian Wildlife Research Centre, Department of Conservation and Land Management, PO Box 51, Wanneroo WA 6065 Introduction Banksia woodlands are such a familiar sight to urban Perth dwellers that few would consider these communities to be of serious conservation concern. However, many areas of wood- land have been cleared of native vegetation recently, and an as- sessment of the conservation status of Banksia woodlands is, therefore, both appropriate and long overdue. The Banksia woodlands chosen for study were those confined to the Bassendean and Spearwood dune systems of the Swan Coastal Plain from Lancelin southwards, and in mapped veg- etation complexes in which banksias were described as being dominant by Heddle et al (1980). Using satellite imagery at a scale of 1:250 000, the area of Banksia woodland communities on the Swan Coastal Plain extant in 1986 was determined and compared with original areas (see Burbidge & Rolfe, in prep., for a detailed account). Clearing of Banksia woodlands By 1986 an estimated 55% of the 281 000 ha of Banksia woodland complexes between Lancelin and Capel on the Bassendean and Spearwood dune systems had been cleared en- tirely of native vegetation (Table 1). The destruction of the seven complexes varied from 90% for the small area of Bootine Complex (north of Gingin) to only 2% of the equally small Karrakatta Complex-North-Transition Vegetation Complex (NE of Lake Pinjar). The two largest complexes, Bassendean — North and Bassendean — Central and South, differed substantially (37% us 85% respectively cleared). Essentially, little remains of the once extensive Banksia woodlands from Perth southwards to Busselton. Only in the Moore River National Park — Yeal Swamp — Melaleuca Park areas north of Perth are large tracts of Banksia woodlands still to be found. Conservation on reserves Only 7% of the original 281 000 ha of Banksia woodlands in- vestigated was on conservation reserves (Table 1, and Fig. 1 of J.S. Beard, this volume), and those reserves were uneven in geographical distribution and among vegetation complexes. Some of these deficiencies will be addressed by proposed ad- ditions to the conservation estate in CALM’s Northern Forest and Central Forest Region Management Plans 1987-1997. However, in these and other areas the conservation of remnant Banksia woodlands currently lies very much in the hands of owners of private property or Crown lands set aside for pur- poses other than flora and fauna conservation. Rare and threatened species Although poorly studied, Banksia woodlands appear to con- tain few rare localized endemic species. However, the extensive clearing of these woodlands has resulted in some plants and ver- tebrate animals declining in numbers to the point where they are now considered highly vulnerable or endangered. Two declared endangered reptiles occur in the Banksia woodlands of the Swan Coastal Plain. The Carpet Python ( Morelia spilota) is widespread in Australia but scarce through- out its range. The Black-striped Snake ( Vermicella calonotus) is almost confined to the Swan Coastal Plain, from Lancelin to Mandurah and most commonly in the deep white sands of the Bassendean and Spearwood dunes. A third reptile, the Lined Skink ( Lerista lineata), is not declared endangered but is of lim- ited occurence, being found only on the Swan Coastal Plain from Perth to Yalgorup. Varanid lizards have also declined markedly in Banksia woodlands (How & Dell, this publ.). Many species of birds have declined in numbers on the Swan Coastal Plain (Storr & Johnstone 1988) although no bird species is restricted to the area and the Banksia woodlands do not consti- tute a major part of the breeding habitat of any declared en- dangered bird species. However, the Banksia woodlands do pro- vide an important feeding resource for non-breeding flocks of Carnaby’s Black-Cockatoo [Calyptorhynchus funereus latirostris) which has declined markedly due to clearing of native veg- etation in the wheatbelt (Saunders et al 1987). Amongst the mammals, it is possible that only a few species of small mammal maintain viable populations in Banksia woodlands (How & Dell, this publ). For example, the Western Quoll ( Dasyurus geofjroii) is possibly locally extinct and in Banksia woodlands the Numbat ( Myrmecobius fasciatus) is now restricted to a small, highly vulnerable population in the Can- ning Vale area. Little is known about the conservation status of Banksia woodland invertebrates (Majer, this volume). The King Spider Orchid ( Caladenia huegelii sens str) and two hammer orchids ( Drakaea jeanensis and D. micrantha Hopper ined) are the only plants currently declared as rare flora that occur mainly in Banksia woodlands on the Swan Coastal Plain. However, ephemeral wetlands dotted through the Banksia woodlands have three additional declared rare plants — Purdie’s Donkey Orchid ( Diuris purdiei). Stalked Water Ribbon (Aponogeton hexatepalus) and Minute Pygmy Sundew ( Drosera occidentalis) . 115 Journal of the Royal Society of Western Australia 71 (4), 1989 The area of Banksia woodland vegetation complexes the total CALM estate and in conservation reserves. Table 1 still extant on the Swan Coastal Plain in 1986, and their representation in Woodland Complex* Original % extantf % in CALM % area (ha) in 1986 estate on reserves 37 Bootine 3911 9.8 1.0 1.0 43 Bassendean-North 78 261 63.0 32.8 11.1 44 Bassendean-Central 86 123 14.5 8.8 1.6 & South 45 Bassendean-North- 20 845 86.5 44.9 39.3 Transition 47 Karrakatta-North 43 868 41.0 42.6 0 48 Karrakatta-North- 5 282 97.7 10.7 0 Transition 51 Cottesloe-North 43 062 55.6 34.5 4.5 Totals 281 353 45.2 27.3 7.2 * as defined and mapped by Heddle et al (1980) t these figures represent very conservative estimates of the amount of habitat destruction in the Banksia woodlands. Much of the remaining woodlands exists only in small patches and in many of these the understorey is highly modified by disturbance agents such as grazing. While still locally common, several other plant species en- demic or nearly so to Banksia woodlands need to be monitored in the future, eg Banksia laricina, Eremaea pupurea, and Caladenia speciosa Hopper ined. The woodlands also contain a number of outlying populations well removed from their main areas of occurrence, eg southern populations of Winter Bells ( Blancoa canescens) and Conostylis latens in Canning Vale. Management for conservation Other papers in this publication allude to major management concerns facing Banksia woodlands — fire, dieback disease, herbivory by rabbits, fertilizer drift, groundwater extraction and weed invasion to name but a few of the most important. To this list might also be added various recreational activities (trail bike riding, off road vehicle use) and commercial pursuits (wildflower picking, bee keeping, grazing by stock) that require management. While these issues deserve our concerted efforts, perhaps of greatest and immediate concern is the ongoing pace of clearing of Banksia woodlands for agricultural, urban and industrial de- velopment projects. While some land uses such as mining are obliged to go through an environmental impact assessment pro- cess before they can clear native vegetation, this does not apply under current legislation to others such as residential develop- ments. Any program to rectify this situation must include a con- certed effort at public education concerning environmental issues. The management of remnants of Banksia woodlands set aside for conservation poses a number of problems, as does the man- agement of remnants of native vegetation generally (Saunders et al 1987). It seems likely that inappropriate fire regimes can promote invasion of perennial weeds such as veldt grass, and lead to gradual degradation of conservation values (Hopkins, this publ). We know little of the population ecology of Banksia woodland species, and require this sort of understanding to plan future management strategies (eg Lamont, this publ). Future Directions New initiatives are needed to improve on the conservation es- tate in both Crown and private ownership. Those proposed and already reviewed in the public arena (System 6, CALM Re- gional Management Plans) need to be implemented as soon as possible if the land involved is not to suffer degradation through other inappropriate land uses. Indeed, biologists will have to work harder at communicating the value of conserving Banksia woodlands if we are not to witness the rapid attrition of these communities over the next few decades. Few rare and threatened species are known to be endemic to these Banksia woodlands, but there are some among plants. These require research and the development of appropriate management strategies. The possible local extinction of preda- tory medium sized mammals like Quolls is, however, an early warning that many common Banksia woodland species will be- come threatened unless we research, plan for and actively man- age the remnants extant today. References Heddle E M, Loneragan O W & Havel J J 1980 Vegetation complexes of the Darling System. Western Australia. In: Atlas of Natural Resources, Darling Sys- tem, Western Australia. Dept Conservation & Environment. Perth. 37 - 72. Saunders D W, Arnold G W, Burbidge A A & Hopkins A J M (eds) 1987 Nature Con- servation: The Role of Remnants of Native Vegetation. Surrey Beatty & Sons, Sydney. Storr G M & Johnstone R E 1988 Birds of the Swan Coastal Plain and adjacent seas and islands. Rec W Aust Museum Suppl 28. Journal of the Royal Society of Western Australia, 71 (4), 1989, 117-118 Banksia woodlands: Summary and conclusions Andrew A Burbidge Department of Conservation and Land Management, Western Australian Wildlife Research Centre, PO Box 51, Wanneroo WA 6065 Introduction The Banksia woodlands that have been the subject of these series of papers are surprisingly poorly studied and docu- mented. Perhaps this is due, at least partly, to their location close to Perth — they are too familiar to many scientists who have preferred to work in less familiar and perhaps more “exotic” surroundings. However, it is their closeness to Perth that makes the Banksia woodlands an ideal place for scientific research since they are ideally located for low cost studies and have been and are subjected to various land-use pressures that are causing many changes. Environment and Conservation The Banksia woodlands of the Swan Coastal Plain grow on deep Quaternary sands with a very low nutrient level. They are subjected to a typically Mediterranean climate of cool, mild win- ters and hot, dry summers. They have a rich flora and fauna, with much variation over 4° of latitude and between the differ- ent soil types. However, there are relatively few endangered species. Because of their proximity to Perth, Banksia woodlands are being destroyed at a rapid rate. While some conservation re- serves protect samples of Banksia woodland, not all the types of woodland are protected at present, nor are all the reserves large enough to be viable. Additionally, the reserves are threatened by a variety of disturbers. Resources The resources of the Banksia woodlands can be divided into three categories: Land Resources The land is used for urban and industrial purposes, food grow- ing, horticulture and pine plantations, and recreation. Urban de- velopment and its associated land uses are leading to the de- struction of increasing areas of Banksia woodland. Natural Non-renewable Resources The chief demands for non-renewable resources are for basic raw materials used in construction, brick making, road building, etc. and for minerals. The demand for minerals is low and re- stricted mainly to silica and limestone. The amount of Banksia woodland destroyed by demand for basic raw materials is re- lated largely to the growth of Perth. Natural Renewable Resources The most important of these are groundwater and the biologi- cal resources — the indigenous plants and animals that make up a genetic storehouse for the future. Demand for groundwater is related to the growth of Perth. Disturbers of Natural Banksia Woodlands Those Banksia woodlands that are not cleared for urban or other use are subject to many disturbers. Among the most im- portant are: Fire Mediterranean climatic areas are typically affected by high in- tensity summer wildfires and the Banksia woodlands are no ex- ception. Aboriginal firing would have occurred for many thou- sands of years but there are few data on their extent, frequency or timing. With recent urbanization and increases in human population fires are likely to have increased in frequency at least. Disease Phytophthora cinnamomi, an introduced fungus that destroys roots, is now a major disease of native and exotic plant com- munities of the Banksia woodlands. Other Phytophthora spp. occur also. Weeds Numerous environmental weeds are now established in Banksia woodland communities. They are competing with and probably eliminating some native species. Feral Animals A variety of introduced animals has become feral in Banksia woodlands. This includes the rabbit, house mouse, black and brown rats, cats, foxes, dogs and, near Perth, even polecats (or ferrets). Their effects on plant and animal communities are only beginning to be understood. Overgrazing and selective grazing are destroying components of the flora, and predation has elim- inated or is eliminating some native animals. Pollution Urban and industrial pollution has so far had the greatest ef- fect on wetlands, via increased nutrients in the water table, rivers, estuaries and lakes. Groundwater Extraction Groundwater extraction can lower the water table and lead to the death of some plants and a reduction in the size of wetlands or the length of time that they contain water. Recreation Human use can lead to degradation of the land and its plant and animal communities. Damage can come from off-road ve- hicles, rubbish-dumping and walking paths. Synergism A major problem of understanding the effects and relative im- portance of the various disturbers is that most of them are inter- related and, indeed, their effects may be synergistic. For ex- ample, frequent fires combined with exotic grazing and 117 Journal of the Royal Society of Western Australia 71 (4), 1989 browsing animals will lead to the degradation of the vegetation much faster than would be expected from the simple addition of the effects of the two disturbers measured in isolation, and the invasion of weeds is much faster in the presence of soil disturb- ance or frequent fire. Future Management Strategies If Banksia woodlands are to be used for the long-term benefit of the people of Western Australia it is clear that strategies will have to be developed and applied to prevent their total destruc- tion or near destruction plus degradation of the remnants. Of particular importance are: 1 Further documentation of biological resources and environ- mental dynamics. 2 Increased research into and management of • plant disease, especially Phytophthora cinnamomi • fire regimes • groundwater extraction • recreation • the “greenhouse effect” 3 Continued land-use planning, including a refinement of the system of nature conservation reserves. 4 A continuing debate in the community about the acceptable limits to the growth of Perth. 5 Education. Without improved education about Banksia woodlands, most of the above points are unattainable. INSTRUCTIONS TO AUTHORS The Journal publishes (after refereeing) • papers dealing with original research done in Western Australia into any branch of the natural sciences; • papers concerning some biological or geological aspect of Western Australia; • authoritative overviews of any subject in the natural sciences, integrating research already largely published in the more specialized national or international journals, and interpreting such studies with the general membership of the Society in mind; • analyses of controversial issues of great scientific moment in Western Australia. Prospective authors of papers in the last two categories should consult the Hon Editor for further advice. Contributions should be sent to The Honorary Editor, Royal Society of Western Australia, Western Australian Museum, Francis Street, Perth, Western Australia, 6000. Publication in the Society’s Journal is available to all categories of members and to non-members residing outside Western Australia. Where all authors of a paper live in Western Australia at least one author must be a member of the Society. Papers by non-members living outside the State must be communicated through an Ordinary or an Honorary Member. Submission of a paper is taken to mean that the results have not been published or are not being considered for publication elsewhere. Free reprints are not provided. Reprints may be ordered at cost, provided that orders are submitted with the return page proofs. Authors are solely responsible for the accuracy of all information in their papers, and for any opinion they express. Manuscripts. The original and two copies must be submitted. They should be typed on opaque white paper with double- spacing throughout and a 3 cm margin on the lef-hand side. All pages should be numbered consecutively, including those carrying tables and captions to illustrations, which appear after the text. Illustrations, both line drawings and photographs, are to be numbered as figures in a common sequence, and each must be referred to in the text. In composite figures, made up of several photographs or diagrams, each of these should be designated by a letter (eg Figure 2B). To avoid risk of damage to orig- inal figures, authors may retain these until after the paper is accepted. The copies of the figures accompanying the manuscript must be of good quality. Authors are advised to use the most recent issue of the Journal as a guide to the general format of their papers. Words to be placed in italics should be underlined. To facilitate editing, papers must be accompanied by a table of contents, on a separate sheet, showing the status of all headings. Following acceptance and editing of the manuscript, authors with access to word processors/computors should amend their manuscript and forward to the Hon Editor their disk (noting the computer and word-processing program used) and one copy of the paper. References must be set out as follows: Paper Jackson A 1931 The Oligochaeta of South-Western Australia. J R Soc W Aust 17:17-136. Twigg L, Majer J D & Kotula R 1983 The influence of fluoroacetate producing plants upon seed selection by seed harvesting ants. Mulga Res Centre W Aust Inst Technol, Bentley, Ann Rep 6:75-80. Book Jacobs M R 1955 Growth Habits of the Eucalypts. For Timb Bur, Canberra. Chapter in book Dell J 1983. The Importance of the Darling Scarp to Fauna. In: Scarp Symposium (ed J D Majer) W Aust Inst Technol, Bentley, 17-27. The Title should begin with a keyword. The Abstract should not be an expanded title, but should include the main substance of the paper in a condensed form. The metric system (SI units) must be used. Taxonomic papers must follow the appropriate international Code of Nomenclature, and geological papers must adhere to the International Stratigraphic Guide. Spelling should follow the Concise Oxford Dictionary. Authors should maintain a proper balance between length and substance, and papers longer than 10 000 words would need to be of exceptional importance to be considered for publication. Authors will be charged page costs (currently $30 per page) if papers exceed 8 printed pages. Short papers (2-4 printed pages) are particularly sought as these often ensure full use of the 32 pages available in each part. Illustrations. These should be prepared to fit single or double column widths. Illustrations must include all necessary letter- ing, and be suitable for direct photographic reduction. No lettering should be smaller than 1 mm on reduction. To avoid un- necessary handling of the original illustrations, which are best prepared between 1.5 and 2 times the required size, authors are advised to supply extra prints already reduced. Additional printing costs, such as those for folding maps or colour blocks, will be charged to authors. Supplementary Publications. Extensive sets of data, such as large tables or long appendices, may be classed as Sup- plementary Publications and not printed with the paper. Supplementary Publications will be lodged with the Society’s Library (C/- Western Australian Museum, Perth, WA 6000) and with the National Library of Australia (Manuscript Section, Parkes Place, Barton, ACT 2600) and photocopies may be obtained from either institution upon payment of a fee. JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA CONTENTS VOLUME 71 PART 4 1989 BANKSIA woodlands symposium Page Preface J S Pate 83 Jennifer Mary Arnold, BSc, BA, MSc, PhD — An appreciation E P Hodgkin 83 Definition and location of the Banksia woodlands J S Beard 85 Bassendean and Spearwood dunes: their geomorphology, stratigraphy and soils as a basis for habitats of Banksia woodlands V Semeniuk & D K Glassford 87 Floristics of the Banksia woodlands J Dodd & E A Griffin 89 Water relations of Banksia woodlands J Dodd & E M Heddle 91 Fire in the Banksia woodlands of the Swan Coastal Plain A J M Hopkins & E A Griffin 93 Terrestrial invertebrate fauna J D Majer 95 Vertebrate fauna of Banksia woodlands R A How & J Dell 97 Biotic and abiotic interactions in Banksia woodland B Lamont 99 Management of groundwater resources for protection of native vegetation J Kite & K Webster 101 Urban development M D Poole 103 Impact of horticulture on Banksia woodlands I R McPharlin & B A Stynes 105 Forestry and Banksia woodlands on the Swan Coastal Plain E R Hopkins 107 Mineral resources and mining of the Spearwood and Bassendean dune systems J R Gozzard & M J Mouritz 109 Banksia woodland weeds G J Keighery 111 Diseases of Banksia woodlands on the Bassendean and Spearwood dune systems B L Shearer & T C Hill 113 Conservation status of Banksia woodlands on the Swan Coastal Plain S D Hopper & A H Burbidge 115 Banksia woodlands: Summary and conclusions A A Burbidge 117 Edited by I Abbott Registered by Australia Post — Publication No. WBG 0351 No claim for non-receipt of the Journal will be entertained unless it is received within 12 months after publication of Part 4 of each Volume The Royal Society of Western Australia, Western Australian Museum, Perth Circulation of this Journal exceeds 600 copies. Nearly 100 of these are distributed to institutions and societies elsewhere in Australia. A further 200 copies circulate to more than 40 countries. The Society also has over 350 personal members, most of whom are scientists working in Western Australia. The Journal is indexed and abstracted internationally. A67232/3/89- 1M-AL/753 GARRY L. DUFFIELD, Government Printer, Western Australia VOLUME 71 (PART 4)