Description of a proteoid-restioid stand in Mesic Mountain Fynbos of the south-western Cape and some aspects of its ecology

A description o f the com m unity and its climatic and edaphic environments is given for a stand of Mountain Fynbos vegetation codom inated by Leucadendron xanthoconus and Chondropetalum hookerianum . The paper categorizes aspects o f the study site either according to existing classifications, or by comparison with other fyn­ bos systems. Comparison o f rainfall and tem perature data with those collected at an agricultural research station in the region indicated high variability in the spatial and tem poral pattern o f precipitation, and an air tem perature regime which was influenced by the topography. Analysis o f vegetation data revealed a species richness lower than other fynbos communities, but a species turnover of similar magnitude. A list of flowering plants and ferns found in the stand is appended. The soil of Table Mountain Group origin comprised a colluvial A -E horizon with a well defined stone-line, and residual B and C horizons of shale origin. It had low pH and nutrient status, with a high measured concentration o f aluminium, especially in the B horizon.


INTRODUCTION
Mountain Fynbos is the best preserved vegetation type in the Fynbos Biome of the Cape (Moll & Bossi 1984).The poor, highly leached soils of these upland sites (Kruger 1979) have proved unsuitable for conventional agriculture, and direct commercial utilization is restrict ed almost entirely to silviculture and the wildflower in dustry.As a large-scale international trade, the latter is relatively young, and production techniques are in many instances still experimental (Davis 1984).It is expected that wildflower producers will increasingly favour cultivation over the traditional veld-harvesting method of floricultural production to assist in control ling product quality (Brits et al. 1983).Those parts of the relatively unutilized Mountain Fynbos which con tain the preferred habitats of many of the showy proteaceous species, are seen as the logical locations for this branch of agricultural development.This article is based on observations made during the First phase of a study into the possible effects of physical disturb ance by agricultural tillage on natural Mountain Fynbos.
The primary objective of this paper is to describe the chosen study site in terms of the existing classifica tions and other frames of reference normally used for fynbos systems.Where this is not possible, or is not appropriate, comparison with data from other fynbos studies is attem pted.The rationale for these exercises is two-fold.Firstly, recognition of common sets of at tributes, especially ecologically functional ones, is a necessary basis for formulating the management strate gies required for utilization and conservation of Moun tain Fynbos vegetation.Secondly, where the classifica tions used to describe Mountain Fynbos systems are incomplete, the task of workers motivated to update them, is facilitated by available quantitative data.This paper endeavours also to be a small part of that acces sible repository.

STUDY AREA
The chosen site is on the south edge of the Grabouw Basin, Caledon District, within a region where quartzite, sandstone and thin bands of shale and conglomerate of the upper Table Mountain Group outcrop, as docum ent ed on the 1:125 000 geological map of the area (Govern ment Printer 1966).It lies on a gentle slope of approxi mately 8% , with an aspect of 246° and at an altitude of 375 m.It is 10,5 km from the sea on the landward side of a ridge which rises to a maximum height of approxi mately 500 m.The grid co-ordinates of the site are: 34° 15 ' 38"S and 19° 6 ' 38"E. Until March 1987, the area in which the study site is located was managed by the Directorate of Forestry (Department of Environ ment Affairs) as a mountain water catchment area.It is now under the control of the Department of Nature and Environmental Conservation (Cape Provincial Adminis tration).

Development o f the study site
An experimental plot 50 x 50 m was delineated at the site during 1984.Sample quadrats (2 x 2 m) were delineated at 28 regularly spaced stations, providing a sampling intensity of 4% for the major components of the vegetation.As part of the long-term experimental design the site was cleared by means of a controlled burn in February 1985.

Climatic data
A weather station was set up on the cleared area and a data-logging device (MC Systems, Cape Town) instal led.This monitored a set of environmental parameters, including precipitation and air temperature.Regional long-term precipitation data were obtained from records of the Weather Bureau (1985), and from a statistical report issued by the Soil and Irrigation Research Insti tute (Agrometeorological Division 1983) for the follow ing stations respectively: Highlands Forest Station (34° 17'S; 19° 6'E; 426 m) over the period 1938-1984; and the experimental farm of the Fruit and Fruit Tech nology Research Institute in Elgin (34° 8'S; 19° 2'E; 305 m) over the period 1 9 6 3 -1 9 8 3 .As an estimate of the long-term mean air tem perature at the study site, long-term data from the Elgin Station (Agrometeorolo gical Division 1983) were adjusted by the differences recorded for this same parameter at the two stations during the period July 1 9 8 5 -Ju n e 1986 (see Results).
For periods when the data-logging equipment was non-functional (a total time of approximately six weeks during the sample period of July 1985-D ecem ber 1986), precipitation data recorded at the Highlands Forest Sta tion, 2,5 km to the south-west, have been used.Means o f the monthly totals at these two stations during 1986 agreed to within 1,3%.Temperature data were not aug mented in this way.

Vegetation
Mature vegetation was sampled at the 28 stations mentioned above.A 1 x 1 m subquadrat was used for close inspection of the less conspicuous species.A list of all identified species of ferns and flowering plants observed on random scans of the plot and its immediate surroundings (a total area of ± 0,65 ha) is given in the Appendix.Further species recognized as distinct are not recorded because the material found was such that it could not be identified.
A single set of nested quadrats (after Whittaker et al. 1979) was marked out in veld adjacent to the study plot for the construction of a species-area curve to per mit comparison of the site with data from other studies.The number of different species was measured in qua drats of: 1 m 2 (10 replicates); 10 m 2 (2 replicates); 100 m 2 and 1 000 m2 (no replication).
Age of the stand was estimated by counting the num ber of nodes on the largest individuals of the dominant shrub species, Leucadendron xanthoconus (Kuntze) K. Schum., and cross-checking against aerial photographic records of the Department of Surveys and Mapping.

Soil
Description of the soil profile was provided by three shallow soil pits (approximately 0,8 m deep), and a single deeper one (1,8 m).For analysis of the physical and chemical characteristics of the soil, samples were taken from the A horizon, the top of the B horizon, and a single sample from saprolitic parent material at 2 m.Each of these samples was air-dried and sieved to 2 mm.Nutrient analyses were performed by the regional Soil Analytic Laboratory of the Department of Agricul ture and Water Supply (Winter Rainfall Region) at Elsenburg using methods described by Jackson (1958), Hesse (1971), the Fertilizer Society of South Africa (1974), andMoore &Chapman (1986).Bulk density and field capacity were determined on undisturbed soil cores, and texture on 2 mm sieved samples.
As possible factors influencing pedogenesis at the site, incidental observations of plant or animal inter actions with the soil were noted; the most apparent of these was the presence of a number of termite mounds.

Climate
The climatic diagram (after Muller 1982), derived for the study site from the adjusted data of Highlands Forest Station and Elgin Experimental Farm, is given in Figure 1. Figure 2 shows the total monthly rainfall measured at the study site (with adjustments for missing datasee Methods), together with concurrent and long-term data from other sites.Figure 3 demonstrates that on a weekly basis during the sample period the rainfall was very unevenly distributed between the Highlands study site and Elgin, although total precipitation received during 1986 at each station was similar (Highlands, 1110 mm and Elgin, 1 090 mm For reasons dictated by the completeness and reliabil ity o f the data, air temperature regimes are given for the period July 1985 to June 1986 (see Figure 4).The most noticeable differences between these two locations are the consistently warmer mean temperatures during the spring and summer months, and the year-round colder minima at the Elgin Station.A 10-day period o f missing logged data at the study site during February may cause the reported extreme values to be inaccurate for that month.The mean species richness in the set of twenty-eight 1 m2 quadrats was calculated to be 7,7 species per m2 (1,7 S.D.), whereas the 1 m2 quadrats of the nested set afforded a slightly higher value of 8,6 species per m2 (1,7 S.D.).The overall mean for these two sets is 7,9 species per m2 .The larger quadrats of the nested set contained 19 (mean of 2); 37; and 56 species in 10 m2, 100 m2 , and 1 000 m2 respectively.A linear regression between the number of species (S), and the log10 o f the quadrat area (LogA) gave the following relationship: S = 6,12 + 16,02 LogA (r2 = 0,984)

Vegetation
The species list (see Appendix) contains the names of all taxa recorded at the site both before the experimental burn in 1985, and for two subsequent seasons.  in the size of their largest dimension from less than 10 mm, to more than 300 mm, and were usually heavily ferruginized.The B horizon was composed exclusively of the shale-derived material, showed weak structure, and tended to be gleycutanic.The deeper soil pit which was dug outside o f the study plot and adjacent to an area with outcropping sandstone, revealed in the sub soil horizon a layer of pre-weathered sandstone approxi mately 1,2 m thick, bounded above and below by shalederived material.This band dipped at an angle of ap proximately 45°, and it is thought that the C horizon throughout the study plot was effectively within the upper shale stratum.With regard to the classification system developed by MacVicar et al. (1977), the soil could be placed in the Kroonstad Form (Mkambati or Avoca Series), although where the B horizon displayed more prismatic structure and darker cutans, association with the Estcourt Form was stronger (Uitvlugt or Estcourt Series).Identification o f the soil series was equi vocal on account of the variability of the clay content of the E horizon, a diagnostic feature of both forms.

Soil
Results of the physical and chemical analyses per formed on samples taken from the site are summarized in Table 1.In terms of the textural classification includ ed by MacVicar et al. (1977), soil of the A horizon lies on the border between loamy sand and sandy loam.Field capacity of the top layer of soil, expressed as gravimetric water content, was measured as 22.5% (3,57 S.D.).
The following features which might influence profile development were observed at the site: 1, surface soil movement under the influence of winter runoff; 2, waterlogging of the colluvial stratum , but not the B horizon during winter; 3, the presence of termite colonies (Amitermes sylvestris) whose mounds were present with a mean density of 120 per ha, ard a irean height of 350 mm; 4, the occurrence of earthworms (infrequently observed); 5, occasional mole ot mole rat activity; and 6, the penetration of roots into the dense B horizon.This latter phenomenon was limited to structural faults and was noted as occurring to the maximum investigated depth of 1,8 m.These roots probably belonged to Leucadendron xanthoconus in dividuals, the only species whose roots were positively identified as penetrating into the B horizon.Fungal hyphae were also observed in old root channels in this horizon.

Gimate
The climate diagram (Figure 1) based on long-term data depicts a typical humid mediterranean-type with winter half-year rainfall (May to October) exceeding 65% of the 928 mm annual total, and a distinct winter with at least one mean monthly temperature less than 15°C (Aschmann 1973).As described by Fuggle & Ash ton (1979), the climates of the Fynbos Biome form a 'spatially diverse mosaic' on account of its mountainous topography.Comparison of the observed climatic para meters at the study site and at the Elgin Experimental Farm illustrates this diversity, with precipitation patterns in the region being especially non-uniform (Table 2).The heterogeneity of the rainfall may also play a critical role in the fire ecology of these seasonally flam mable areas.The single highest rainfall event at the Highlands Forestry Station during the sample period occurred in February, the height of the fire season.Such temporal and spatial patchiness of rainfall acting over the millennia of fynbos evolution could have contributed significantly to a patchy fire history, and hence to the heterogeneous mosaic of the present-day vegetation.
Regional patterns of ambient air temperature appear to be more predictable than those of precipitation (Figure 3).The warmer spring and summer mean air temperatures, and the colder year-round minima of the Elgin Experi mental Farm relative to the Highlands study site, can probably be explained by the location of the former station.Elgin almost certainly experiences more restrict ed air movement during the windier spring and summer period than the study site (partly supported by un published data from this study); and perennial nocturnal drainage of cold air (Barry & Chorley 1982) from the large mountains o f the H ottentots Holland and Franschhoek to the north.See Davis (1987) for a brief discus sion of wind at the study site.

Vegetation
The criteria established by Taylor (1978) for the de finition of fynbos are amply satisfied by vegetation at the study site, and the species which characterize it.all have distributions restricted to the Fynbos Biome as delineated by Moll & Bossi (1984).Species richness at the site was lower than the much quoted fynbos figure of 121 flowering plant species within an area of 100 m 2 (Taylor 1972).Unfortunately this figure has been quoted in the literature as a bench mark of species richness in fynbos (e.g.Bond 1983;Jarman 1982;and Taylor 1978), when, in fact, it is given in a semi-popular article in which descriptive de tails are omitted.Better documented figures are pre sented by Bond (1983), who reports a maximum figure of 104 species in an area of 1 000 m 2 in a Jonkershoek stand of Protea nitida (waboomveld).the extrapolation of which on the log-scale would agree well with the total of 126 species recorded at the Highlands study site in an area of approximately 0,65 ha.In the same paper, Bond presents a synoptic species-log area curve for fyn bos vegetation in the southern Cape mountains.For the formulation S = b + dlog10A, where S is the number of species in an area A, he found b = 16,4 and d = 15,8.These constants of the linear equation represent 'point diversity' and species turnover (or community patchiness) respectively (Bond 1983).Highlands data indicate a sig nificantly lower point diversity (t-test; p <0.001), but a similar patchiness for the community at the study site.They are more similar to those obtained by Whittaker et al. (1979) for mallee vegetation in New South Wales, Australia (b = 5,3 and d = 15,3).Based on a sample area of 100 m 2 , Cowling (1983) reported species-richness of 26,5 for Mountain Fynbos in the south-eastern Cape.This is lower than the Highlands figure, while on the other hand the mean of his 'point diversity' (sensu Bond 1983) for fynbos shrubland sites was twice that of the study site.The measures of diversity discussed above lend a valuable perspective to the description of the According to the description of post-fire succession in fynbos by Kruger & Bigalke (1984), as summarized by Rutherford & Westfall (1986), the 12-year post-fire stand of the study site was in an early stage of m aturity, a phase during which the codominance of phanerophytes, chamaephytes, and hemicryptophytes is best de veloped.However, the abundance of restioid shoot tissue and the consequent build-up of a dense mat of litter may effectively advance the maturation process in the site community by causing premature reduction in species richness.
Subjectively, the mature vegetation of the study site was best described as a Leucadendron xanthoconus stand, with an understorey dominated by Chondropetalum hookerianum and Erica cristata.These species have distributions as follow's: L. xanthoconus and C. hooke rianum occur from the Cape Peninsula eastward as far as Bredasdorp (Vogts 1982)and Riversdale (Linder 1985) respectively, while E. cristata is restricted to the area between Sir Lowry's Pass and the Klein River Mountains (Baker & Oliver 1967).Grobler (1964), Boucher (1972Boucher ( , 1978)), Kruger (1974), andDurand (1981) have all con ducted vegetation surveys within a 15 km radius of the study site, and although they cite some species con spicuous in the Highlands vegetation, none of their community descriptions characterize it.Inspection of Boucher's (1978) data for the occurrence of the above three species in his study area between Cape Hangklip and the Palmiet River revealed a pattern (Figure 5), which suggests that the convergence of all three at Highlands may be a characteristic feature of the site.In two of the three instances where this occurred in Boucher's study, soil was of the duplex Estcourt Form.(It is possible that E. cristata and C hookerianum form a commensalistic association, in which physical support of the trailing ericoid by the erect restioid may be an element, a phe nomenon observed in the mature vegetation at the study site.) Comparison of the Highlands species list (see Appen dix) with those of Boucher (1978) and Kruger (1974) Highlands study site, but as yet the body of available information is insufficient for this parameter to be used as an accurate classifier.confirmed the small degree of overlap at the species level.Of the Highlands species, approximately one third of the total number was contained in each o f the other lists, while less than 20% were common to all three.Kruger & Taylor (1980) have previously demonstrated that a 60% difference in species composition exists between Cape Hangklip and Jakkalsrivier.
The above discussion suggests that while many phytosociological elements of the region are represented at the study site, regional patchiness might easily make mani festation of a previously recognized community unlikely.In an attem pt to improve upon the phytosociological approach to classification of Mountain Fynbos vegeta tion, Campbell (1985Campbell ( , 1986) ) invested considerable effort in constructing a structural classification with a priori rules for classifying communities.He pointed out (Camp bell 1985), using stands dominated by Leucadendron gandogeri as an example, that some assemblages of plant life will necessarily defy classification by that particular system.Interpretation of the Highlands vegetation ac cording to the key of structural features affords it a simi larly equivocal position.Careful consideration of the Highlands vegetation may offer additional information to resolve that particular shortcoming of the structural classification.

Soil
As with the composition of plant communities, soil is a characteristically variable component of the Fynbos Biome (Moll & Jarman 1984).Boucher (1978) counted eight soil forms (14 series) in his study area of 115 km 2, but some of his classified mountain plant communities included up to six o f these.Estcourt, one o f the forms identified at the Highlands site, occurred at 15% of his mountain releves as the Soldaatskraal Series, while Kroonstad was not listed at all.Kruger (1974), noted six forms within the 1,58 km 2 Jakkalsrivier catchment, none of which was in common with the Highlands site.Campbell (1983), in his extensive survey of montane environments in the Fynbos Biome also encountered none of the forms identified at the study site, although the Highlands data are consistent with the generalized gradients which summarize his work.Considering the shale-derived component of the Highlands soil, his warning against equating non-quartzitic origin in Moun tain Fynbos soil with nutrient-richness is borne out.
In the broad context of fynbos soils, topsoil at the study site is typical in that it is acid, leached, and nutrient poor (Kruger 1979).Being duplex in nature, however, the dense B horizon acts as an impediment to the vertical loss of many of the soil constituents that might normally be removed from the system during the podzolization process, although throughflow (Trudgill 1977) may account for loss via seeps.(The working definition of nutrient-poorness supplied by Campbell (1983) is easily met for both A and B horizons.) Apart from some intensive studies on lowland sys tems with narrowly defined objectives (e.g.Low 1983;Mitchell et al. 1984;Stock 1985;Witkowski & Mit chell 1987), published data which describe the nutrient status and cycling processes in fynbos soils are limited.Information on nutrients in mountain systems is sporadic in the literature, and usually incidental to broader eco logical studies.Comparison of the Highlands data with those describing other Mountain Fynbos sites (Low 1983), indicated that total N in the Highlands topsoil was greater than at these other sites by factors of be tween 1,1 and 3,2.The measured available P was com parable to the values of between 2,5 and 4,5 ^g.g'1 re ported by Read & Mitchell (1983) for coastal fynbos.The C.E.C. measured at the Highlands site fell into the wide range of values measured by Kruger (1974) for soils at Jakkalsrivier (0,5 to 44,0 m e/100 g), while it was appropriately lower (for an oligotrophic soil) than the approximate mean of 14 m e /100 g given by Tucker (1983) for a range of non-carbonate soils in Australia and the USA.
Accumulation of clay particles at the top of the B horizon clearly increased the measured C.E.C. at this level (Table 1), but parallel concentration of aluminium may outweigh the advantage of this to plants by reduc ing the availability of phosphorus under the inherently acid conditions (White 1979).The ability of Scottish heathland plants to survive on soils with high Al content is demonstrated in a study cited by Woolhouse (1981) where concentrations of 0,17% (18,9 m e/100 g) are re ported for the Bj horizon.These figures are somewhat greater in magnitude than those obtained for soil of the Highlands study site.It would be reasonable to suppose that the toxic effects of Al are countered either edaphically (Norrish & Rosser 1983), or physiologically within heathland and fynbos systems, where this element is liable to be common (Hesse 1971).
The observed downhill movement of topsoil during the rainy season at the Highlands site implies that the process of soil creep responsible for the formation of this duplex soil is still in progress.However, root pene tration, together with some activities of the soil fauna, may be acting to ameliorate and stabilize the soil in local patches.

Synthesis
The data which describe phenomena of the Highlands study site are valuable to the ongoing study by providing a base-line for the investigation of ecosystem functions.The immediate objective, however, is to place that infor mation in a general descriptive context which relates to other Mountain Fynbos systems.This is attem pted in Table 3.As human demands inevitably increase with time, con servation and effective utilization of natural resources such as Mountain Fynbos vegetation will depend greatly on the extent to which managers are able to identify and predict responses o f ecosystems to the impacts of exploitation.Classification of ecosystem attributes is an important step in establishing a means to extrapolate knowledge of specific sites to larger managerial units.Treating the Highlands site as a test case, we have seen above that hopes for the development of a classification which encompasses the functional complexity of Moun tain Fynbos are justified.This is especially true consider ing the large body of information which has accumulated over the past decade under the co-ordination of the Fyn bos Biome Project of the CSIR(see Moll & Jarman 1984).
FIGURE 3.-Comparison of weekly precipitation totals record ed at Highlands (combined forest station and study site data), and the Elgin Experimental Farm during 1986.
Soil at the site was duplex, a category found through out the south-western Cape (Schloms et al. 1983).It comprised a dense underlying stratum of saprolitic shale with a shallow (1 5 0 -8 0 0 mm) colluvial overburden of predominantly quartzitic material.The top stratum con sisted of an orthic A horizon, a leached E horizon, and a basal stone-line (commonly 150 mm thick) o f quartz and sandstone rock fragments.In places, the topsoil contained more fine shale-derived material, while in others the sandy surface layer was missing entirely, leaving a lithosolic A/E horizon.

FIG
FIG URE 5. -Frequencies with which the three species Leuca dendron xanthoconus, Chondropetalum hookerianum and Erica cristata occurred at Mountain Fynbos sample plots in the Cape Hangklip a rea and their degree of distributional overlap.Drawn from the data o f Boucher (1978).

-Total monthly precipitation during 1986, and long-term averages for sites in Highlands and Elgin, Legend symbols are as follows: HS86 = Highlands study site, 1986; EEF86 = Elgin Experimental Farm, 1986; HFLT = Highlands Forest Station (1938-1984); and EEFLT = Elgin Experimental Farm (1963-1983). 24 h period in February 1986 when the data logger sys tem at the study site was not functional. An accumula tion type rain gauge at the study site confirmed rainfall in excess of 100 mm for the month o f February.
FIG URE 1.-The derived climate diagram for the Highlands study site.The broken line depicts mean m onthly air tem p erature, while the solid line is total m onthly precipitation (after Muller 1982).

TABLE 3 .
-A sum m ary description o f the Highlands study site with regard to the clim ate, vegetation and soil Linder (1985) et al. (1985)S, B.H.A., ELLIS, F. & LAMBRECHTS, J.J.N .1983.Soils o f the Cape coastal platform .In H.J. Deacon, Q.B. Hendey & J.J.N .Lam brechts, F ynbos palaeoecologya prelim inary synthesis.Scientific Programmes R eport No. 75.CSIR, Pretoria.STOCK, W.D. 1985.A n investigation o f nitrogen cycling pro cesses in a coastal ecosystem in the south-w estern Cape Province, South Africa.Ph.D. thesis, University o f Cape EATHER BUREAU 1985.Unpublished data.D epartm ent o f E nvironm ent Affairs.WHITE, R.E.1979.Introduction to the principles and practice o f soil science.Blackwell, Oxford.W HITTAKER, R.H., NIERING, W.A. & CRISP, M.D. 1979.Structure, pattern, and diversity o f a mallee com m unity in New South Wales.Vegetatio 39: 6 5 -7 6 .WITKOWSKI, E.T.F.& MITCHELL, D.T. 1987.Variations in soil phosphorus in the fynbos biom e, South Africa.of the species occurring at the Highlands study site in the Caledon D istrict The alphabetical arrangem ent o f Bond &G oldblatt (1984)is used here for ease o f access, b u t nom enclature and authorship are according toGibbs Russell et al. (1985)andGibbs Russell et al. (1987), except for the Restionaceae, whereLinder (1985)has been used.Ceratocaryum decipiens (N.E.Br.) Linder C hondropetalum hookerianum (Mast.)Pillans Elegia filacea Mast.