A phytosociological study of Signal Hill, Cape Town, utilizing both perennial and ephemeral species

A phytosociological study based on the collection of vegetation and environmental data from 53 randomly stratified sample plots on Signal Hill, Cape Town, was carried out over an area of 124 ha. The survey extended over 12 months to ensure the inclusion of as many plant species as possible, and a list of the vascular plant species was compiled. A total of 81 families, 255 genera and 460 species was identified. The phytosociological method revealed that only one major plant community occurs in the study area and two subcommunities, with a total of five variants correlated mostly with aspect and historic land use, were identified. The perennially and seasonally identifiable species were analysed separately to determine their relative contribution to the phytosociological classification. The two data sets gave similar classifications. A vegetation map as well as a soil map was compiled.


INTRODUCTION
Since the earliest times the flora of the Cape has fascinated travellers, visitors and scientists (Raven-Hart 1967, 1971). The vegetation of Signal Hill has been used for grazing and fuel supplies since prehistoric times, and after European settlement some areas were cultivated and afforested (Joubert 1991). In 1964 the area was proclaimed a nature reserve (Ashton 1985). The natural vegetation has been protected since then.

The vegetation is broadly classified as West Coast
Renosterveld (Moll & Bossi 1984) and falls within the Fynbos Biome (Kruger 1978). The area is a unique West Coast Renosterveld site as it is the only area on Malmes bury shale influenced by coastal fog. No plant community study has previously been made of the area except for a post-fire study (Michell 1922) of the vegetation along the eastern slopes. Werger (1974) states that 'In floristically rich areas ... communities can be clearly characterised floristically on the bases of floristic lists in which only permanently recog nisable species are entered'. He further states that in arid regions perennials are generally better indicators of specific habitat factors because annuals are a relatively unimportant component. Le Roux (no date) found in Namaqualand that species composition and cover of ephemerals vary during the growth period and from year to year as new vegetation associations are formed annually. drops markedly (Fuggle 1981), creating a fire hazard in the area. The prevailing winds in the dry summer are strong southerly to southeasterly winds (Fuggle 1981). Signal Hill is about 350 m high and forms a barrier against the prevailing winds. When the southeaster blows, the northwestern aspect is completely protected, but the northern aspect less so. Likewise, when the cold north wester blows from across the sea, the southeastern aspect is protected, while the northern aspect is once again exposed to the wind. The latter aspect is, therefore, more often exposed to wind. Signal Hill receives a mean of 463.5 mm rain annually. Figure 2 illustrates the data collected during the period 1882 to 1950 (Weather Bureau 1986). Data for the study area and five stations from surrounding areas were collected over a period of about 70 years and appear in Table 1. Topography plays a major role as illustrated by the data collected on different sides of Signal Hill. In all cases the highest precipitation, i.e. 60% of the total, is experienced from May to August with a peak during June, while January and February are the driest months. The mean number of rainy days for Signal Hill is 86. July has the highest number of rainy days i.e. 12, and during December to February there are on average only three rainy days per month.
No temperature data for the study area are available, though data from stations in the vicinity show very clearly that a moderate temperature is experienced with no extremes on a daily or a seasonal basis. The hottest month is February and the coldest month is July.
Fog forms when warm air is blown over the cold Atlantic sea on northwesterly to westerly air drifts (Fuggle 1981), thus the western slopes receive most fog, with a peak during April and May according to data from the Cape Town Harbour.

Flora -historical records
It is impossible to visualise the vegetation of the study area before European settlement and the extent of the disturbances which followed, as the early descriptions only give a sketchy impression. Sparrman (1785) noted that such reports are not always reliable, as these men had spent months at sea, with the result that they tended to over praise the Cape at the sight of greenery after such a long time at sea (Skead 1980).  One of the first descriptions of the study area was made by Van Riebeeck who, on 27 April 1652, 'went along the downs behind the rump of the Lion Mountain where we found between the mountain and the downs the most beautiful land for sowing and for grazing cattle that one can desire ... Crossing the Lion Mountain on the seaward side of the head, found the slopes on the other side dry and stony ...' (Skead 1980). A seaman who visited the seaward slopes during 1685 found the area 'not at all Rocky, but cover'd over with Grass' (Raven-Hart 1971). It appears that grass was already very common at this time as another seaman wrote in 1702 that the '. .. Lion's Rump ... is grown over with luxuriant grass and a few trees ...' (Raven-Hart 1971). A photograph dated ± 1910 (Cape Archives: E 8144) shows grassland along the northern slopes, while another dated 1899 (Cape Archives: Dr. J 80) shows areas densely populated by low scrub, possibly Elytropappus rhinocerotis (renosterbush).
Between 1657 and 1665, some 80 Khoikhoi were living in a kraal on the eastern slopes of Signal Hill. During Kolbe's visit between 1707 and 1713 two large Khoikhoi kraals existed at the foot of the eastern slopes of the Lion's Hill (Fagan 1989). It is possible that their fires maintained the grassland on the slopes, and that they started this firing in an attempt to promote firewood and grazing for their herds, or to stimulate the growth of geophytes (Deacon 1983). Michell (1922) compiled a detailed description of her post-fire study site on the eastern slopes (1919)(1920)(1921). Unfortunately no study was made prior to the fire, with the result that the relative importance of renosterbush and Rhus lucida at this time is not known. She states that the vegetation is 'sclerophyllous bush, the characteristic vegetation of the southwestern region', though the area is 'deficient in several typical southwestern families'. All the species encountered during this period are listed by her, many with notes on their numbers and localities. Of particular interest is the following: The vegetation of the valleys 'show certain marked differences' from that of the open slopes. Michell states that the vegetation in the valleys was not badly burnt owing to the somewhat sheltered position of the watercourse in each valley, and that the valley bottoms were 'covered with Acacia karroo'. Adamson & Salter (1950) also note the presence of this species on Signal Hill. Today only isolated individuals occur. The slopes were dominated by R. lucida after the fire, which 'dotted' the landscape at 'frequent intervals'. The north-facing slopes had a 'more open type of vegeta tion' than the south-facing slopes. The latter had a conspicuously different plant population during the winter months, though the contrast was less during summer.
Clutia pulchella was common at the foot of one of the valleys (similar to relevé 28 of the present study). Michell (1922) also mentions the presence of Noltea africana. Hyparrhenia hirta was common and mentioned several times. Some 20 other grass species are also mentioned. Protea repens, P nitida and Leucadendron argenteum are reported as being scarce. A number of annuals was associated with cattle manure. Michell (1922) concludes by stating that the fire favoured the renosterbush which was 'far in advance of any others' and, with the exception of the northern slopes of the valleys, evenly distributed all over the area. The vegetation type established after the fire is termed 'Rhenosterveld' and considered an 'artificial one'. Mem bers of the Proteaceae, Rutaceae and Ericaceae were 'only occasionally seen'. She states that 'especially in the case of the Proteaceae, bush fires have been largely instrumen tal in eradicating large numbers of species from these slopes'.
A few other descriptions from the early part of this century give an idea of what the vegetation was like at the time. According to these descriptions Proteaceae were common on the slopes of Signal Hill. Luckhoff (1951) states: 'Most of the original Signal Hill flora has been destroyed. The older generation still speaks of fields of proteas that once grew on the hill. Today only a few plants survive. For the rest we find pines, gums, taaibos and abundant grass-the latter always reliable evidence of repeated burning'. It appears that Protea repens was once extremely common (Jackson 1977) on the lower slopes of the study area, whereas large tracks of P lepidocarpodendron were described by Marloth during the early part of this century (Luckhoff 1951). ' Almost the whole of Signal Hill used to be covered with Proteaceae, mainly P repens, P lepidocarpodendron, P. nitida, Leucospermum conocarpodendron and Leucadendron argenteum' (Luckhoff 1951). Today only isolated individuals of some of these species have survived. Adamson (1929) noted that renosterbush is 'well developed on the slopes of Signal Hill' and that the community is 'relatively pure and slow changing' on the western slopes. The area carried no appreciable forests (Luckhoff 1951).

Soil map
A soil map was compiled for Signal Hill. Soil profiles were studied throughout the area. Soil samples were taken at four different localities and analysed by the Faculty of Agriculture at the University of Stellenbosch. The pH was measured in KC1. Resistance is expressed in ohms, using a standard USDA soil cup. The localities and soil profiles are marked A, B, C and D (Figure 3). The results of the soil analyses and the exposition as regards the symbols in Tables 3 & 4, are given in Table 2, as well as dominant soil families, a brief description of the soils and a topographical description of the sample plots.

Phytosociological study
The field work for the vegetation survey was started in December 1988 and completed in December 1989. Relevés were compiled from 53 stratified random plots (Werger  Table 2  1974). These were permanently marked with a steel dropper, 1.3 m high. A 700 mm long white PVC conduit tubing with a diameter of 20 mm was fixed onto the dropper (Van Blerk 1990), rendering the plots easily detectable from a distance, especially in mid-high, mid-dense shrubland (Campbell etal. 1981). A numbered metal tag was fixed to the top of the dropper to ensure the location of the plots even after a veld fire.
Stratification was done on a topographical basis. Six such units were distinguished within the study area. The number of plots was determined on an area basis within the three major units, i .e. the three aspects with open slopes and ravines (Figure 4). Cultivated and built-up areas were excluded from the survey. Randomly stratified sample plots were sited in stands of vegetation which appeared floristically, structurally and environmentally as homogeneous as possible (Campbell & Moll 1977). The sample intensity  Table 3

A+
A 0 + + + ++ ♦ BB +1 .+ 00    ( (Table  3). The plots were visited on a regular basis for 12 months in order to record as many species as possible and a separate table of these seasonally identifiable species was made (Table 4). (Hill 1979) was used to derive a first approximation of the vege tation types. A suite of computer programs used by Boucher & Shepherd (1988) was also used and proved most useful in this respect. The results of this classification were refined by the Braun-Blanquet classification technique as described by Mueller-Dombois & Ellenberg (1974), using a word processor package to shuffle relevés and species in order to obtain a final classification. All plots sampled are retained in the phytosociological tables (Tables 3 & 4).

Two-way indicator species analysis TWINSPAN
The perennial species were treated separately from the ephemeral species to determine the contribution of each to the vegetation. Perennial species are considered to be those which are perennially identifiable (Werger 1974) and ephemeral species those which are not perennially identifiable. The latter includes both therophytes and cryptophytes. The results from the perennial species are listed separately from those of the ephemeral species in order to determine the relationship between the two groups. Species with a single occurrence were listed below  . (1985, 1987), or recent revisions, or by consulting specialists in particular taxa. Outdated taxon names which could not be updated were excluded from the list. Species mentioned by Adamson & Salter (1950) as occurring on Signal Hill were checked with the Guthrie Collection (Bolus Herbarium). Only those species which occur as actual specimens in the herbarium and which were collected on Signal Hill were added to the list. Species encountered incidentally during herbarium work, for which Signal Hill is given as the locality, were also included.  Table  2 for the exposition of the symbols on the soil map.

Soil
The phytosociological classification, as well as the floristic relationship between the different subcommuni ties and variants, is given in Tables 3 & 4

. A list of the vascular plant species recorded appears in the Appendix.
In this survey the taxonomic entities Mohria caffrorum and Cheilanthes contracta were merged. Introduced woody species such as Pinus pinaster; P. radiata, Acacia cyclops, A. saligna and Eucalyptus spp., which occur as scattered individuals, were excluded from Table 3.

Phytosociological study
The analysis resulted in the identification of one major plant community and two subcommunities, one of which has three and the other two variants. Each of these vegetation types is related to a particular set of environ mental conditions, of which aspect and historic land use differences were the most significant. There appears to be no clear correlation between soil forms and vegetation types. These community types have been mapped in Figure 5.  Depending on local site factors, the shrub canopy closure varies from sparse to mid-dense; similarly the grass understorey varies from very sparse to open (Campbell et al. 1981). There are very few trees, but the multi stemmed shrub, Rhus lucida, is abundant and more or less evenly distributed.
Three strata are generally recognised, namely a canopy of mid-high sparse to mid-dense shrubs; a sparse to open grassy/low shrub stratum; and a very sparse seasonal herb layer. Rhus lucida, Anthospermum spathulatum and Otholobium hirtum are the most constant species present in the canopy, having a height of 1-2 m and a cover of 10-60%. Viscum capense was often recorded as growing on R. lucida. The second stratum is dominated by the woody shrubs Chrysocoma coma-aurea, Helichrysum patulum, Salvia africana-caerulea, Rhus laevigata, Hermannia hyssopifolia and Senecio pterophorus, and the grass Merxmuellera stricta with a height of 0.25-1.0 m and a cover of 10-25%.

The herbaceous layer with a cover of generally < 5 % consists of the forbs Senecio hastatus and Orobanche ramosa, and the fern Cheilanthes hastata. Geophytes are a feature on the northern aspects in particular, and autumn to spring ephemerals are prominent in areas of sparse vege tation. Characteristic ephemeral species are the forbs Pelargonium elongatum, P. lobatum\ the geophytes Oxalis glabra, O. purpurea, O. pes-caprae, O. bifida, Cyanella hyacinthoides, Moraea bellendenii and the introduced grass Brizja maxima. Lichens and mosses were recorded, but omitted from the phytosociological table. It appears that lichens are generally less common on the east and southeast-facing slopes than in the rest of the area.
Plant litter cover varies between 5% and 95%, and consists mainly of leaves of R. lucida and H. patulum, grasses and twigs. The total canopy cover of the vege tation varies from 35-140%, and an average of 31 perennial and 22 ephemeral species was recorded per relevé.

Very few species were found growing under the Rhus bushes. Those which did occur include Senecio hastatus, Stachys aethiopica, Scabiosa columbaria, Annesorrhiza capensis, Oxalis spp., annual orchids and fern clusters under the larger bushes. The climber, Helichrysum patulum, with a height of about 0.7 m, often scrambles into other shrubs such as R. lucida and R. glauca, and when thus supported can reach a height of about 1.2 m. Other scramblers in R. lucida are Helichrysum cymosum and Chrysocoma coma-aurea, though to a lesser extent.
This community is divided into two subcommunities. Table 3). The diagnostic ephemeral species for the latter subcommunity are less well defined and are listed in Species Groups D and E (Table 4).

Hyparrhenia hirta-Rhus glauca Subcommunity
This subcommunity of 35 relevés is the most extensive and occupies approximately 80 ha (64% of the study area). It occurs mostly on the north and northwest-feeing seaward side, though two relevés, Nos 39 and 40, face east-northeast. This is the more diverse subcommunity, probably related to the diversity of microhabitats available because of the occurrence of a variety of soil types, and as a result of disturbance history including fire. The soils are generally shallow, usually gravelly to cobbly, with the topsoil overlying shale rock and having 6-40% clay in the topsoil. Rocky outcrops are rare to common.
The high constancy (100%) and relatively high canopy cover, ranging from <1% to 40%, of H. hirta and relatively constant presence (80%) and locally high canopy cover, up to 45%, of R. glauca, are features of this sub community. The rest of the differential species in this subcommunity have a relatively low canopy cover. The high constancy and canopy cover of H. hirta, and the low cover of woody species is an indication of the generally poor condition of this vegetation, attributed to frequent fires, especially in variants 1.1.1 and 1.1.2.

Species which differentiate this subcommunity are listed in Species Group C (Tables 3 & 4). An average of 32 perennial and 26 ephemeral species was recorded per relevé.
In this open grassy/low shrubland three strata are evident. The mid-high sparse shrub stratum, with a canopy cover of 10-20%, is 1.  Most of this area was burnt during early 1986, less than three years prior to the survey.

The very sparse herb/dwarf-shrub stratum (< 0.25 m) with a cover of <5% is represented by the woody Hermannia althaeifolia, Helichrysum cylindriflorum, H. asperum, Aspalathus acuminata, Pelargonium myrrhifolium and Leysera gnaphalodes. A fairly common climber in R. lucida is Pelargonium candicans. The most common succulents are Lampranthus emarginatus, Ruschia rubricaulis, Orbea variegata and Cynanchum zeyheri. The forbs include Indigofera incana, Cineraria geifolia, Euphorbia crispa, Aizoon sarmentosum and Commelirui africarui; and the ephemerals as listed in Species Group C (
These two variants can be treated as a single unit as they are closely related. The difference between the two variants is characterised by the two species, Emgrostis curvula and Erepsia anceps present in variant 1.1.1, but absent from variant 1.1.2. The relationship between these two variants is shown by the relative absence of species listed in Species Group B (Tables 3 & 4). These variants are also discerned from variant 1.1.3 by the absence of these species. Between 15% and 55% of the surface is bare. Litter cover is between 5% (relevé 7, a fire having been experienced a year previously) and 60% (relevé 6, on the open slope). Total canopy cover during summer varies between 45 % and 65 %. During winter cover increases by about 10% and is associated with the appearance of ephemeral species. The height of the vegetation does not exceed 1.5 m.

The total number of plant species recorded per relevé varies between 40 and 62, with about 32 perennial and 24 ephemeral species per relevé. Species Groups I and J (Table 3) list those perennial species which occur throughout the study area, though generally not in these two variants. Species listed in Species Group I do not occur in the valleys on the southeast-facing slopes either. Conspicuous species which are absent from these two variants are the woody shrubs Elytropappus rhinocerotis, Pelargonium cucullatum, Lobostemon argenteus, Passerina filiformis, Cliffortia ruscifolia, Gnidia laxa, Montinia caryophyllacea, Athanasia trifurcata, Diospyros glabra; and the herb Scabiosa columbaria.
Species Group J (Table 4) H. hirta and R. glauca with a cover of 25% and 6.5% respectively. Other species in this stratum with a cover of some 40% as well as those in the very sparse herb/dwarf-shrub stratum are similar to those listed in the general discussion on this subcommunity and to a lesser extent those listed in Species Groups A and B (Table 4) It appears that frequent fires played an important role in determining vegetation structure and floristic compo sition. Evidence from historical photographs indicates that parts of this vegetation have been grassveld for at least nine decades. The high cover of Hyparrhenia hirta seems to be a result of these frequent fires. The frequency of fires has, however, decreased in recent years.
There appears to be no difference floristically between the open slopes and dry ravines. An attempt was conse quently made to determine whether a difference on a structural basis existed between these two habitats. The canopy cover and height of the woody R. lucida, R. glauca and Otholobium hirtum were studied. It was found that R. glauca attains the highest canopy cover value and height in the ravines; i.e. in relevés 1, 2 and 5. A mean canopy cover of 18% and a height of 1.1 m was recorded in the ravines, versus a mean canopy cover of 2% and a height of 0.75 m on the open slopes. The other two species did not show this tendency. It was also noted during field work that O. hirtum forms extensive thickets in the valley bottoms, though no data were collected in these areas. A mean canopy cover of < 3 % for this species in these two variants was recorded, though local high values were recorded; i.e. 15% in relevé 2 (a dry ravine) and 10% in relevé 8 (with moderately deep soil on the open slope).
The cause of the similarity in species composition between the dry ravines and open slopes may be the extremely shallow soil in the former as bedrock attains its highest cover in the ravines. None of these ravines is wet for a long period, and runoff is probably rapid.
It appears that these two variants are at present in a relatively stable subclimax stage induced abiotically by excessive firing over a relatively long period.  Table 4). The total canopy cover for summer varies between 35% (relevé 17) and 70% (relevé 19).
The relatively high canopy cover of Elytropappus rhinocerotis in relevé 33 suggests frequent fires in the past (Michell 1922). The extreme rockiness of this reíevé may be attributed to surface erosion under such conditions. Brownlie (1982) found a positive correlation between E. rhinocerotis and eroded areas on clay soil. Many large (4 m2) bare patches are conspicuous in relevés 20 and 32 which occur on the same southwest-facing slope. In these relevés, 30-35% of the surface is bare. The vegetation is about 20 years old. Very large specimens of both old and dead E. rhinocerotis individuals were recorded.
Litter cover varies between 30% in a high lying rocky relevé with several succulent species (relevé 12) to about 75-85% in ravines at lower altitudes (relevés 13, 24, 25 and 26). The highest litter values appear to be correlated with wet areas and not with the age of the vegetation. B (ikbles 3 & 4), although these species also occur to a lesser extent in variant 1.1.2 of this subcommunity.

This variant is characterised by the prominence of species listed in Species Group
Three strata are evident in this sparse grassy/low to midhigh open shrubland. The mid-high sparse shrub stratum, about 1.3 m tall and with a canopy cover of about 20%, is dominated by R. lucida, which has a canopy cover of 11% in this variant versus 6% in the former two variants. Other species in the mid-high shrub stratum (1-2 m tall) are Anthospermum spathulatum, Otholobium hirtum, Chrysanthemoides monilifera, Pelargonium cucullatum, Cliffortia ruscifolia and Athanasia trifurcata. H. hirta (0.85-0.90 m) with a canopy cover which ranges from < 1-30%, and a mean of 10%, is generally conspicuous, and the stunted Themeda triandra (< l-2 0 % ), Cymbopogon marginatus (< l-6 % ) and Tribolium uniolae are locally prominent.
Low-growing succulents are found scattered in this variant. These are often found among rocks and on rocky outcrops at higher altitudes in particular. Soil is generally shallow (< 2 5 0 mm)
As in the former variant, the ravines appear to be no different floristically from the open slopes. Structural differences occur in that both Rhus glauca and Hyparrhenia hirta are taller in some ravines (relevés 24 and 25), though they do not have a higher canopy cover. The same applies to H. hirta in relevé 35, though not to R. glauca (Figure 8). Species listed in Species Groups G, I and J (Table 3) Conspicuous in this variant is the much drier appearance of the north-north-west-facing slopes where H. hirta is prominent, versus the southwest-facing slopes where shrubs are dominant, giving the latter a green lush appearance. A vertical firebreak on a west-facing slope is maintained by using bush-cutters (Figure 9). The shrubs are thus prevented from reaching any considerable height. The vegetation in this firebreak is consequently artificially converted to grassveld dominated by H. hirta. The occur rence of this grass firebreak and the high cover of H. hirta on the north-north-west-facing slopes indicate the strong relationship among the three variants in this subcom munity.
Two of the relevés on the east-facing slopes appear to belong to this variant rather than to subcommunity 1.2. This may be due to the fact that these releves occur along an east-north-east-facing slope. Variant 1.1.3 may be con sidered as transitional between the two subcommunities.   vegetation is lower than that in the rest of the subcom munity.

Species listed in Species Group I (Tables 3 & 4) indicate
Two variants are recognised in this subcommunity; i.e. variant 1.

Cliffortia polygonifolia-Rhus tomentosa Subcommunity
This subcommunity of 18 relevés occupies some 45 ha (36% of the study area) and occurs on the mesic eastern and southeastern inland slopes and valleys. The slope varies between 17° and 34°. The soils are generally shallow to deep, gravelly or non-gravelly with 15-40% clay in the topsoil. The deepest soils occur in the valley bottoms. This vegetation has been protected from fire for about 16 years. Prior to 1972, extensive pine plantations existed on the southeastern slopes, but it appears that the veld along these slopes has recovered to a large extent, except for the area in which relevés 29 and 51 are situated. This latter area has been subjected to afforestation for a much longer period than the rest of the area. Floristically this part has also recovered and cannot be distinguished from the rest of the slope, though the general height of the The total number of species per relevé varies between 26 and 61, and for perennials between 19 and 45. An average of 29 perennial and 12 ephemeral species per relevé was recorded. The number of ephemerals was exceptionally low in some relevés and varied between 4 and 24 species. Between 5% and 40% of the surface is bare in summer. Litter cover varies between 45% and 95%, while total canopy cover varies between 50% and 80%.

The high constancy and canopy cover values of many species such as
The east-facing slopes (relevés 36, 37, 38, 41 and 42) are significantly disturbed and occur on a multiple fire break which is shifted periodically. The vegetation cover in this firebreak is not lower than in the rest of the variant. The species richness in these relevés, as well as in relevés 29 and 51, is generally lower than in the rest of the area. In the disturbed area occupied by relevés 29 and 51, Helichrysum patulum attains its highest canopy cover in the study area, namely up to 25% with a mean of 16%, giving this short veld a silver-grey colour.
Some of the former eucalypt firebreaks along these slopes are at present reduced to the occasional tree stump on which coppices frequently appear. Isolated pine trees, mainly Pinus pinaster, occur on the slopes. Dead, burnt pine tree stumps and relicts from dry-packed stone walls are also encountered.

Variant 1.2.2
This variant occurs in five narrow valleys along the southeast-feeing slope. These valleys occur on moderately to highly organic-rich, well-drained soils which are deep (> 1.0 m), usually reddish in colour, apedal, with 20-40% clay and 1-5% organic carbon in the topsoil. This is the highest organic carbon content found in the study area. Rocks are mostly of a smaller size (2 0 -4 0 mm), though larger rocks as well as bedrock were recorded. The slope varies between 23° and 30°.
The total canopy cover varies between 50% (relevé 44) and 140% (relevé 48) in summer. During summer between 5 % and 15 % of the surface is bare. In winter the canopy cover increases, mainly due to the growth of the fern Cheilanthes capensis (especially in relevé 44) and an added component in the form of ephemeral species as listed in Species Groups E, H, J and K (Table 4). The litter cover is generally high (90%).  Table 5 shows an increase in cover by R. lucida from 6% on the north-facing slopes to 11% on the northwest-facing slopes. H. hirta shows a decrease from 25% to 10% in these areas. The prominence of H. hirta on the north-north-west-facing slopes in variant 1.1.3 as well as in a firebreak maintained by bush-cutting shows the relationship among the three variants of this subcommunity.

Four strata are recognised in this mid-high mid-dense shrubland. The low tree layer with an open canopy cover of 15% is represented by
Rhus lucida shows an increase from 11% in variant 1.1.3 to 21% on the wetter open slopes and to about 13% in the valleys of subcommunity 1.2. A large number of other mid-high shrubs was also recorded in these valleys. R. lucida showed an increase in height from the drier to the wetter parts. H. hirta is almost completely absent from Mid-high shrubs (1-2 m) showed an increase from the drier to the wetter parts throughout the five variants. Low shrubs (0.25-1.0 m) also showed this tendency, though a decrease was recorded in the wet valleys on the southeastfacing slopes. Variant 1.1.3 is a transitional area between the xeric north-facing parts and wetter east and southeast-facing parts. This is indicated by the presence of certain shrub and forb species and the high local cover values for H. hirta. Relevés 39 and 40 occur on an east-north-east-facing slope, which means that these slopes receive more solar radiation than the other slopes in subcommunity 1.2. Its relationship with variant 1.1.3 (subcommunity 1.1) shows that the east-north-east-facing slopes of subcommunity 1.2 can with inadequate management be converted into the more xeric variant 1.1.3. 1.1.1 and 1.1.2). Very few young E. rhinocerotis individuals were recorded. These occurred mainly along the edge of natural vegetation adjacent to the eucalypt firebreaks.

Subcommunity 1.1 indicated that Elytropappus rhino cerotis can increase with an increase in fire frequency and/or erosion (relevé 33). With even higher fire frequencies grassland may become dominant (Variants
During winter an increase in total canopy cover is experienced which is caused by ferns in the wetter valleys and forbs, especially ephemeral species, along the drier northern slopes. A mean of 26 ephemeral species per relevé was recorded for subcommunity 1.1, versus a mean of 15 ephemeral species per relevé for subcommunity 1.2.
The relative stability of the vegetation was indicated by examination of aerial and historical photographs. There appears to have been little permanent change in the vegetation structure during the last century.