Flora of the Zuurberg National Park . 1 . Characterization of major vegetation units

The distribution of major vegetation units or veld types in the Zuurberg National Park, situated on the eastern limits of the Fynbos Biome, is presented. Structural and floristic criteria are used to describe and map five basic units, namely Afromontane Forest, Subtropical Thicket, Mountain Fynbos, Grassy Fynbos and Grassland.


INTRODUCTION
The Zuurberg National Park represents one of the largest conservation areas incorporating Grassy Fynbos, a vegetation type characteristic of the eastern limits of the Fynbos Biome.An intricate mosaic o f vegetation types is present, reflecting the rugged topography, variety of aspects and different microclimates.The biogeographi cal complexity of the eastern Cape is well known and is a result o f the convergence o f four major phytochoria (Goldblatt 1978;Gibbs Russell & Robinson 1981;Cowling 1983aCowling , 1983bCowling , 1984;;Lubke et al. 1986).This diversity represents a major challenge in terms o f con servation, since management measures taken for one plant community may not be suitable for another.The topography is very rugged due to the erosion of softer shale bands from between alternating layers of quartzites, but there are no peaks or steep cliffs.Height above sea level varies between 250 and 970 m.The climate is temperate with a mean annual rainfall of ± 722 mm.Table 1 shows rainfall figures recorded at the office (Lot 16) between 1931 and 1962.The mean monthly figures clearly show that spring and autum n maxima are expe rienced.Unlike other fynbos areas, winter months are the driest.Thunderstorms commonly occur during the summer months, when lightning fires may also be ex pected.Soil texture and soil depth vary considerably as a result of the geological and topographical diversity.
The soils of the Zuurberg are generally more fertile and finer textured than soils of the Cape Folded Belt to the west (Campbell 1983;Cowling 1984).

METHODS
Ground patrols and an aerial reconnaissance by heli copter were undertaken to interpret aerial photographs and to become familiar with the terrain, vegetation and plant species.Herbarium specimens were collected on several visits during 1985, 1986 and 1987.A checklist of all the plant species recorded, complete with their author names, is presented in part 2 of this series (Van Wyk et al. 1988).
Vegetation units of the study area were visually iden tified according to vegetation structure and they corres pond roughly with the 'veld type' concept of Acocks (1953) or the rank of class (Cowling 1984).A map (Figure 2) was drawn from 1:50 000 aerial photographs.A series of colour slides taken by helicopter from differ ent angles at low altitude was used to verify the boun daries between vegetation units.The area covered by each unit (Table 3) was estimated from the 1:50 000 map by a randomly positioned 2 mm grid.Descriptive data for each of the five major vegetation units were obtained from 64 sample quadrats distributed as shown in Table 2.All the quadrats were permanently marked by 1,2 m iron fencing standards in each corner (5 x 10 m plots), one at the centre of each short end (4 x 25 m plots) or one at the centre (400 m 2 circular plots).Localities were selected so as to include most of the variation in each vegetation type.Plot size varied between 50 and 400 m2 (Table 2).
The following information was recorded in each plot: all identifiable species present, Braun-Blanquet cover values for each species (r = < 1 % projected canopy cover; 1 = <5% ; 2 = 6 -25% ; 3 = 2 6 -5 0 % ; 4 = 5 1 -7 5 % ; 5 = >75% ), total projected canopy cover of all species, height of different strata (grass layer, shrub layer, canopy height and height of emergents) and, for Forest and Thicket plots, also diameter at breast height of all indi vidual trees (if more than 100 mm).In view of the tremen dous variability of the vegetation, the sample size was inadequate for a detailed phytosociological classification.
It does, however, provide sufficient information to cha racterize the major vegetation units.For descriptive pur poses, species were classed into growth forms as shown in Tables 5 -1 4 .Dominant and characteristic species of each vegetation unit were chosen as follows: Characteristic (diagnostic) species: species with a fidelity value of 80% or more.
Dominant species: species with a mean Braun-Blanquet cover value of at least 0.80.
In these calculations, single occurrences (species pre sent in only one plot) were excluded.Forest and Thicket plots were considered separately from the Mountain Fyn bos, Grassy Fynbos and Grassland plots.This seemed reasonable as only a few species were common to both subdivisions, and of these very few qualified as character istic or dominant.
A summary of floristic and structural characteristics of the major vegetation units as recorded in 64 sample plots is given in Table 4. Species richness (expressed as species per m 2 of plot area) varied between 0,13 (Forest) to 0,92 (Mountain Fynbos).These values are dependent on quadrat area, so that only the figures for Grassland, Grassy Fynbos and Mountain Fynbos are directly com parable.The high figure for Thickets compared to Forest agrees with previous findings that Afromontane Forests are poorer in species than Thickets in the east ern Cape (Cowling 1983b).When distinct differences in structure (Table 4) are considered in conjunction with diagnostic and dominant species (Tables 5 to 14), each of the major units is clearly distinguishable.

Afrom ontane Forest
Forests comprising tall evergreen trees with canopy heights of 10 to 14 m and emergents of up to 21 m occur on south-facing slopes and in some valley bottoms.Forest types on northern slopes and in alluvial valley bottoms with canopy heights of 2-9 m and emergents of up to 12 m are grouped with the next unit (Sub tropical Thicket).The distinction was not made on the basis of structure only.We also used the almost total absence of a herbaceous ground layer and the presence of typical Afromontane species (White 1978) such as Podocarpus falcatus and Diospyros whyteana.Despite a strong Tongoland-Pondoland influence, there are pro nounced flonstic differences between Afromontane For est and Subtropical Thicket in the eastern Cape (Cowling 1984).In the results of our survey, 24 tree species have fidelity values of more than 80% (present in less than    5 and 6.

Subtropical Thicket
Thicket as defined here comprises a variable assem blage of communities dominated by thorny and/or suc culent shrubs.They occur on dry north-facing slopes in higher parts and on all aspects in lower-lying south ern parts of the study area.Most of them comprise what Lubke et al. (1986) describe as Valley Bushveld.This term was used by Acocks (1953) but is no longer useful because it incorporates too wide a range of types (Cowling 1984).At least three basic types were includ ed in our sample: Kaffrarian Thicket Closed, non-succulent shrubland to low forest com munities dominated by evergreen, sclerophyllous trees and shrubs with a high cover of stem spines and vines (Cowling 1984;Everard 1987).Campbell (1985) would classify much of the taller thicket of this type as Eastern Forest & Thicket.

Kaffrarian Succulent Thicket
This type occurs in dry areas and is characterized by a high proportion of succulents, a great diversity in growth form and a strong Karoo-Namib floristic influence (Cow ling 1984;Campbell 1985;Everard 1987).A variation of this type, similar to Addo Bush and Sundays River Scrub (Acocks 1953), occurs in southern parts such as the northern slopes at Superbus (see Figure 2).The lat ter has a canopy height of no more than 2 -3 m and is dominated by Schotia afra.Putterlickiapyracantha, Phyllanthus verrucosus and Euphorbia ledienii.Portulacaria afra dominates in some parts, particularly on steep slopes in the western parts of the Park.

Combretum caff ru m -Acacia caffra Thicket
This type has a very limited distribution along river beds.Since the dominant species (Combretum caffrum and Acacia caffra) are deciduous, it is not included in the Subtropical Thicket concept and we group it here provisionally.It shows similarity to the tropical thickets found in the valleys of Natal and Transvaal.
Diagnostic and dominant species of Subtropical Thicket are listed in Tables 7 and 8.

Mountain Fynbos
Mountain Fynbos has the highest species richness (Table 4) and covers only an estimated 5% of the Park  3).It is most common on wet southern slopes, but also occurs on sandy soils in protected low-lying areas.Mountain Fynbos as defined here is largely synony mous to the Mountain Fynbos of Taylor (1978), and Kruger (1979) and the Mesotrophic Proteoid Fynbos of Campbell (1985), but it also includes some communities that may represent later serai stages of the next unit (Grassy Fynbos).Panicum maximum 9 0,83 100 (1985) did not include mature fynbos of the Zuurberg in his sample, presumably because it was thought to be absent (Campbell op.cit., page 7).
Widdringtonia nodiflora did not occur in any of our fynbos plots, but it is highly characteristic of our con cept of Mountain Fynbos and should be added to the lengthy list of diagnostic species in Table 9.A large number of species are locally dominant.The list of dominant species (Table 10) has therefore been limited to those with a mean cover value of 1,00 or more and, except for species with very high cover values, a pre sence of more than 50% (present in at least eight of the 15 Mountain Fynbos plots).

Grassy Fynbos
Grassy Fynbos covers the largest proportion of the surface area of the Park (Table 3) and occurs on all plateau tops and also on gentle southern and northern slopes in higher-lying areas.Diagnostic and dominant species of Grassy Fynbos are listed in Tables 11 and 12. Campbell (1985) distinguished between three sub series of Grassy Fynbos, namely Dry, Mesic and Meso trophic.He classified the dominant vegetation of the Zuurberg as Sundays Mesic Grassy Fynbos but also men tioned the lack of good differential characters.The pre sence of proteoids over 1 m tall, the less than 40% cover of Ericaceae, the 10-50% cover of restioids and the 30-90% cover of grasses are used as differentiating fea tures by Campbell.Two types of his Mesotrophic sub- series also occur in the Zuurberg, namely Mannetjiesberg Mesotrophic Grassy Fynbos and Grahamstown Mesotrophic Grassy Fynbos.In the study area, these two types have a much more limited distribution than Sundays Mesic Grassy Fynbos.
Our concept of Grassy Fynbos is much wider than that of Campbell.We also include Suurberg Grassland and much of Hankey Grassland, both of which approach Acocks's (1953) Dohne Sourveld (Campbell op. cit.).Hankey Grassveld shows two extremes.A sourveld with Tristachya leucothrix, Merxmuellera stricta and nume rous fynbos elements and a sweetveld with grasses such as Themeda triandra and Heteropogon contortus with out fynbos elements (Campbell op.cit.).The sourveld is here included under Grassy Fynbos and the sweetveld under Grassland.In the study area, there is a much great er discontinuity in the distribution of fynbos elements than in those characters used by Campbell to distinguish between Grassy Fynbos and Grassland.For practical reasons, we have therefore used the presence of fynbos elements to differentiate between Grassy Fynbos and Grassland.

Grassland
R.A. Lubke (unpublished data) recognized, in the northern part of the Sundays River area (Figure 1), two major grassland communities, namely Festuca costata Tussock Grassland and Themeda triandra-Tristachya leiicothnx Grassland.The latter was provisionally sub divided by him into Bobartia orientalis Grassland and Trachypogon spicatiis Grassland, further subdivided into a, Heteropogon contortus Grassland and b, Setaria sphacelata Grassland.
We have taken a much narrower view and the Grass lands of the study area are here considered to include only those areas where Restionaceae, Ericaceae and Proteaceae are totally absent.Campbell's (1985) criteria for recognizing Grassland are difficult to use because of the gradual decrease of fynbos elements along the tran sition from Grassy Fynbos to Grassland.Our concept therefore includes only part of Campbell's Hankey Grassland and seems to be identical to Lubke's Setaria sphacelata Grassland.As such it is perhaps the most uni form vegetation unit of all and occurs mostly on steep northern and western slopes.What variation there is, appears to be the result of soil depth and rockiness.Some species (Acacia karroo, Diospyros lycioides and Aloe ferox for example) are restricted to deep soils on lower northern slopes, while succulents such as Euphor bia polygona are locally dominant only in very rocky areas.
Diagnostic and dominant species are listed in Tables 13 and 14.Very few of the dominant species have high fidelity values, so that most of the diagnostic species are forbs or succulents and not grasses.Elionurus muticus and Brachiaria serrata are very common but only Setaria sphacelata var.torta and Aristida diffusa subsp.burkei appear to be characteristic of Grassland as defined here.The dominant grasses are also present in Grassy Fynbos, where their cover values are scarcely lower.

DISCUSSION
The forests of the Zuurberg have floristic elements in common with both the Amatola and Alexandria Forests and are similar in species composition to forests in the Watersmeeting Nature Reserve (Bathurst), the Fort Grey Nature Reserve (East London) and the Groendal Wilder ness Area north of Uitenhage (Geldenhuys 1985).They differ from the Knysna and Tsitsikamma Forests in species composition, notably the absence of Ocotea bullata and Trichocladus crinitus and the presence of species of Pondoland-Tongaland affinity.(Geldenhuys 1985).
Thicket communities of the Zuurberg are very vari able, probably as a result of topographic, rainfall and edaphic gradients.In terms of structure and species composition, most of the thicket agrees with Everard's (1987) Xeric Kaffrarian Thicket.Only a very small part of the Kaffrarian Succulent Thicket of the study area is similar to Addo Bush (Acocks 1953) or Spekboomveld (Archibald 1955), the dominant vegetation of the Addo Elephant National Park.Everard (1987) classified the latter as one of two suborders of Kaffrarian Succulent Thicket, namely Xeric Succulent Thicket.His other suborder, Mesic Succulent Thicket, seems floristically similar to some of the thickets of the Zuurberg.
The Mountain Fynbos of the study area was not in cluded in the classification of Campbell (1985) but it has the differentiating features of his Mesotrophic Proteoid Fynbos.Protea lorifolia and P. repens are the dominant canopy species and Grassy Fynbos is present as understorey.In view of the limited and localized dis tribution and the high species richness, this vegetation unit should receive special attention when management Judged by diagnostic species, our concept of Grassy Fynbos appears to be similar to that of Cowling (1984), who argued that it is not a recently derived vegetation type as Acocks (1953) proposed.The presence of regional endemics {Erica demissa, E. pectinifolia, Podalyria burchellii and Protea foliosa for example) and the resprouting ability of virtually all the species found in Grassy Fynbos indicate that it should be recognized as a distinct vegetation type.Campbell (1985) proposed that the Grassy Fynbos (Eastern Fynbos) of the Zuurberg and Grahamstown areas should be included in the Fynbos Biome and perhaps also in the Cape Floristic Region.
The abundance of grasses was discussed by Cowling (1984).He suggested that high temperatures during the growing season (the high proportion of summer rain) in crease the competitive advantage of C4 grasses, although the more fertile and finer-textured soils (Campbell 1983) also need to be considered.Too frequent fires may lead to an increase in grassiness by removing the shading effect of the overstorey.It is possible that longer intervals be tween fires will result in an increase of Mountain Fynbos in certain areas.

CONCLUSIONS
In the Zuurberg National Park several totally differ ent and unrelated vegetation types occur in close proxi mity.The dynamics of the boundaries between the types and between communities need to be studied in more detail to explain the intricate mosaic of vegetation.Com munity boundaries may be determined at least partly by an equally intricate mosaic of soil types.Campbell (1983) has reported distinct edaphic gradients in the mountains of the Fynbos Biome.Another major ecological factor seems to be the natural fire cycle, in which the warmer, drier northern slopes tend to burn at more frequent intervals than the wet southern slopes.Fire is considered to be the major disturbance factor in fynbos biome com munities (Cowling et al. 1987).
If the present-day patchy distribution of plant com munities has been shaped by edaphic factors and a natural fire regime, little seems to be gained by interfering with the natural cycle.It is indeed impractical to divide such complex communities into conventional 'burning blocks' and bum them according to a rigid schedule.Instead, management should try to allow lightning fires to run their natural course, and exclude man-made fires from outside the Park.Further research is required to assess this policy, focusing on post-fire succession and the effects of fire on the characteristic and dominant species in each vegetation unit.It is also im portant to gain infor mation on the extent to which edaphic factors are re sponsible for community boundaries.

FIGURE
FIGURE 2.-Distribution of major vegetation units i n the Zuurberg National Park.

TABLE 2 .
-Num ber o f sample quadrats.All were perm anently m arked to double as long term m onitoring plots

TABLE 3
Shrubs are rare, succulents are virtually absent, and eight of the 10 fern species recorded occur exclusively in Forest.Diagnostic and dominant species are listed in Tables

TABLE 5 .
-Diagnostic species o f A from ontane Forest grouped by grow th form.Single occurrences are excluded * Sum o f Braun-Blanquet cover estim ates divided by presence; for Forest and Thicket plots a cover o f less than 5% was recorded as 0,50.

TABLE 7
.-Diagnostic species o f Thicket com m unities grouped TABLE 8 .-Dom inant species o f T hicket com m unities grouped by grow th form . Single occurrences are excluded by grow th form

TABLE 9 .
-Diagnostic species of M ountain Fynbos com m uni ties grouped by grow th form.Single occurrences are ex cluded

TABLE 10 .
-D om inant species o f M ountain Fynbos com m uni ties grouped by grow th form.Criteria for entry are specified in the te x t

TABLE 11 .
-Diagnostic species o f Grassy Fynbos grouped by grow th form . Single occurrences are excluded

TABLE 13 .
-Diagnostic species o f Grassland grouped by growth form . Single occurrences are excluded

TABLE 14 .
Cowling 1984)pecies o f Grassland grouped by growth form from the plateau areas may be part ly due to frequent fires in the past (see the successional model ofCowling 1984), but shallower and more fertile soils could also be limiting factors.Mature stands of Protea lorifolia are almost exclusively found on sandy soils.
The absence of seed-regenerating Protea species (P.lori folia and P. repens)