The vegetation of the northeastern Orange Free State, South Africa: physical environment and plant communities of the Ea land type

The research was carried out in the Ea land type of the northeastern Orange Free State, with the objective of reclassifying and refining Acocks’s veld types. TWINSPAN classification results were further refined by Braun-Blanquet procedures. The 100 relev^s distributed over the Ea land type resulted in the recognition of four major vegetation types which may be divided into nine plant communities. The communities were hierarchically classified, described and ecologically interpreted. DECORANA ordination was used to determine vegetation/environmental gradients and relationships.


INTRODUCTION
The Grassland Biome of South Africa covers approxi mately 27% o f the country. As a result of intensive agricultural practices and urbanization, together with in dustrialization, the deterioration o f the grassland led to concern amongst decision-makers, resulting in the launch o f the Grassland Biome Project (Mentis & Huntley 1982). This project aims at developing a better knowledge and understanding o f the grasslands of South Africa to permit efficient land-use planning, utilization, conservation and management. To reach these goals, it is necessary to re classify Acocks's (1988) Veld Types. This reclassification means a more detailed identification, description and map ping of the present grassland types (Scheepers 1986). The northeastern Orange Free State was identified as an area for which little or no phytosociological data exist. This study will also contribute to the syntaxonomic synthesis presently being undertaken by the Botany Department of the University of Pretoria (Bredenkamp et al. 1989;Kooij 1990;Fuls e t al. 1992a, b;Bezuidenhout 1988).

STUDY AREA
The lota! study area is situated in the northeastern comer o f the Orange Free State, i.e. between 29° 00' and 29° 47' E longitude and 27° 00' and 28° 00' S latitude, bor dering Transvaal and Natal (Figure 1). It covers approxi mately 5 600 km2 and comprises five land types, namely land types A , B, C, E and F (Land Type Survey Staff 1984), which can be further subdivided (Figure 2). One third of the area (184 000 ha) is covered by the Ea land type which is dealt with in this report. The other land types are to be discussed in detail in later papers. A land type is an area which is uniform with respect to terrain form, soil pattern and climate. Towns situated in the area are Vrede and Memel (Land Type Survey Staff 1984). According to Acocks (1988), the study area represents six veld types: Patchy Highveld to Cymbopogon-Themeda  I 5 3 P a tc h y H ig h v e ld to C v m to p o q o n -T h e m e d a V e ld T r a n s itio n i 1 5 4 T u r f H ig h v e ld to H ig h la n d S o u r v e ld T r a n s itio n l + 4 .1 5 6 H ig h la n d S o u rv e ld to C y rto o p o q o n -TheroedaT r a n s itto n E « A * V * J 6 5 S o u th e r n T a l l G r a s s v e ld

Beaufort Group
The Beaufort Group covers more than 80% of the study area. It can be subdivided into three subdivisions, namely the Lower, Middle, and Upper Beaufort Beds. The argillaceous rocks are massive or blocky weathered. The mudstones are greenish grey, blue-grey or red. Cross bedded sandstones are common.

Clarens Formation
The Clarens Formation, previously known as Cave Sandstone, is a massive, fine-grained rock type, which reaches a thickness of up to hundreds of metres. Under weathering conditions, this formation features fantastic shapes in the form of pillars and caves. Exposed surfaces are white or cream-coloured, whereas its base is pink or deep red. A few isolated hills in the central and southern part of the study area have the characteristic shapes of this formation as described above. The main components are subangular to rounded grains o f quartz and subordinate feldspar. The accumulation of sandstone is probably of aeolian origin, being re-arranged later by flowing water (Du Toit 1954).

Elliot Formation
Purple and red mudstones and shales, together with red sandstones and thick beds o f yellow and white feldspathic sandstones are characteristic o f this formation (Du Toit 1954). It is well distinguished by its prevailing coloura tion as can be seen on the slopes o f the few koppies in the study area.

Molteno Formation
This formation is distinguished by the typical grey and blue colouration of the shales, and the coarse grain and 'sparkling' appearance of the dominating sandstones. Between Harrismith and Memel the Molteno Formation occurs only as a single thin grit, covered by the Elliot Formation and underlain by the Beaufort Group (Du Toit 1954).

Alluvium, sand, calcrete
Alluvium and sand are more recent by-products of erosion, most probably originating from the Beaufort Group (pers. obs.). A narrow strip of these deposits, together with calcrete, occurs in the Seekoeivlei area, which is drained by the Klip River (Department of Mineral and Energy Affairs 1984).

Dolerite
The dolerites intruded the sediments o f the Karoo Sequence during the last stages of the Drakensberg volcanicity. These intrusions are either horizontal, evenly inclined or undulating sheets (SACS 1980). The dolerite dykes are restricted to the eastern part of the study area.

PHYSIOGRAPHY
The study area is part of the inland plateau region or highveld ( Figure 1) and consists of plains with moderate relief to closed hills and mountains with moderate and high relief (Kruger 1983;Mentis & Huntley 1982). The alti tude is between 1 500 and 2 000 m with some peaks reaching heights of up to 2 200 m. There is a clear gradient in the physiography of the area from south to north. Three broad divisions can be distinguished.
The southern and eastern parts are characterized by isolated hills and mountains with moderate and high relief. The middle part is depicted by lowlands, hills and moun tains with a moderate to high relief. This part can further be described as strongly undulating irregular land, gradually changing over into plains with moderate relief. These plains are slightly irregular, undulating, with occasional hills scattered over the area. This region is most suitable for cultivation purposes, whereas the rest of the area is more suited to cattle farming.  used according to the Soil Classification Working Group (1991). The Glenrosa and Mispah Forms are restricted to terrain units 1 and 2 and very often occur together to form a complex. These soils are shallow (< 2 0 0 mm) and have a low clay content (15-20% ) and are mostly not arable.

Terrain units 3 and 4 are characterized predominantly by pedocutanic, lithocutanic and yellow-brown apedal B horizons. Soils are relatively deep ( > 350 mm) and have a higher clay content ( > 35 %). Although the Ea land type
is generally more suitable for crop production than the rest of the study area, large areas are non-arable because of the high clay content of the soils. Terrain unit 5 is characterized by either rock and alluvium or the vertic Rensburg and Arcadia soil forms. The last two forms have a high clay content (> 5 5 % ) and are deep ( > 5 0 0 mm).

Rainfall
The study area is situated in the summer rainfall zone with an average annual rainfall of 750 mm. Precipitation takes place mostly in the form of thunderstorms, between November and March. Midsummer droughts occur towards the end of December until middle of January (Department of Agriculture and Water Supply 1986). Rainfall data for weather stations at Frankfort and Standerton are given in Figure 6 (Weather Bureau 1986). Table 1. The frost period extends from April to October, which means a frostfree period of approximately 150 days a year (Weather Bureau 1986).  (1985 & 1987). Structural classification was according to Edwards (1983). The following habitat data were recorded in each sample plot: geology, topography, terrain unit, slope and aspect, rockiness, soil types and erosion.

1988; Land Type Survey Staff 1984). Sampling of the different terrain units was done on a subjective basis. Five
To derive a first approximation o f the vegetation types, two-way indicator species analysis (TWINSPAN) (Hill 1979a) was applied. This was further refined by Braun-Blanquet procedures (Behr & Bredenkamp 1988;Bredenkamp et al. 1989). The results obtained are presented in a phytosociological table (T&ble 2). Detrended correspondence analysis (DECORANA) (Hill 1979b) was applied to the floristic data set to determine vegetation gradients and illustrate vegetation/environmental rela tionships.

DESCRIPTION OF COMMUNITIES
The Ea land type is characterized mainly by the con stant presence, mostly with high cover-abundance values of Themeda triandra, and Eragrostis curvula and E. plana (species group L, Table 2). The number o f species recorded in the relev^s varies between 15 and 25, with an average of 19 species.

Artemisia afra-Rhus dentata shrubveld
This shrubveld is situated on moderate to steep scarps (3 0°-9 0°) (terrain unit 2) facing rivers and streams ( Figure 7). Shallow, rocky soils of the Glenrosa and Mispah Forms are typical of this terrain type (Figure 8). The average rock size is more than 500 mm in diameter, covering more than 20% of the surface. Clear signs of erosion can be observed, which are ascribed mainly to the steepness o f the slopes. The utilization of the vegeta tion by dassies (Procavia capensis Pallas 1766) is apparent in some areas.
The vegetation is characterized by species group C (Table 2) and the diagnostic species include the shrubs Ar temisia afna, Rhus dentata and Diospyros austro-africana. Bidens p ilosa and Hibiscus trionum are weeds and are often associated with disturbed areas. Dominant woody species are the diagnostic shrub species. Con spicuous and dominant grasses include Themeda triandra, Eragrostis curvula and E. plana of species group L, and Aristida congesta and A. junciformis of species group I ( Table 2). The herbaceous layer of terrain unit 2 is more conspicuous and better developed than those of other terrain units.

Hyperthelia dissoluta-Eragrostis curvula shrubveld
This shrubveld is situated on steeper (4 0°-90°) slopes and displays patches of bare soil with a relatively high degree of erosion (Figures 7 & 8). Diagnostic species include the dominant grass species Hyperthelia dissoluta, the forbs Clutia natalensis, Garuleum woodii, Sutera polelensis, the xerophytic fern Pellaea calomelanos and the grass M elica racemosa (species group A, Table 2). Themeda triandra, Eragrostis curvula and E. plana are among the most constantly present companion grass species occurring in this community. An average of 15 species was recorded per sample plot.

Hyparrhenia hirta-Diospyros lycioides shrubveld
This shrubveld is characterized by species group B (Table 2) and can be further distinguished from the

Aristida bipart ita Berkheya pinnatifida Hermannia depressa S etaria sphacelata Ha Iafr ida densiflora Indigofera obscura Cymbopogon piurinodis Aristida Ju n e i f o r m is Aristida congest a Oenothera tetraptera S olanum panduriform e
Sp e c ies group J

Conyza sumatrensis P s e u d o g n a p h a /ium o /igandrum C ynodon dactylon Paspalum diIatatum Eragrost is micrantha H e!ichrysum dregeanum Cephalaria scabios a
Sp e c ies group K

Paspalum distichum Cyperus /ongus Nariscus congestus Bidens bipinnata Chenopodium ambrosioides A r gyrolobium pauciflorum Juncus exserptus S i u m repandum Dev err a b u r c h e //ii Gerber a ambigua Cymbopogon excavatus Cyperus marginatus Diplachne fusea Schoenopiectus decipiens
Sp e c ies group L

Hyperthelia dissoluta-Eragrostis curvula shrubveld by the presence of conspicuous and constantly present grass species of species group I (Table 2). Diagnostic species include the tall and dominant grass species Hyparrhenia hirta, the shrubs Diospyros lycioides, Grewia occidentalis and Heteromorpha trifoliata and the weedy Zinnia peruviana and Monsonia angustifolia (species group B, Table 2). A further characteristic of this community is the presence of serai and pioneer species listed in species group I, for example the grasses Aristida junciformis, A. congesta and A. bipartita.
An average o f 23 species was recorded per sample plot.

Themeda triandra-Elionurus muticus grassland
This plant community is to a great extent similar to grasslands described by Bezuidenhout (1988) and Kooij (1990), except for the prominence of Eragrostis plana and other species typical of the moister eastern grasslands of the Orange Free State. This community compares well to the Elionurus muticus-Themeda triandra alliance de scribed by Fuls et al. (1992a). This grassland type covers more than 60% of the Ea land type enclosed by the study area. A great diversity of soil forms occur, including Mispah, Glenrosa, Mayo, Inhoek, Westleigh, Swartland, Clovelly, Oakleaf, Glencoe, Bonheim and Arcadia. Soil depth varies from 100 to more than 700 mm. Deeper soils (> 5 0 0 mm) display higher clay contents (35-55% ). Most of these soils have been ploughed, and cultivated lands re place this grassland community to a great extent (Figure 7). (Table  2). Other prominent species are Themeda triandra, Eragrostis plana and E. curvula, while the asteraceous forbs Helichrysum rugulosum and Berkheya pinnatifida are conspicuously present.

Two variations can be distinguished, namely the Vemonia oligocephala-Trachypogon spicatus variation and the Harpochloa falx-Trachypogon spicatus variation.
These two variations are distinguished by the presence of species groups E and F respectively (Table 2). No clear differences in the habitat can be found to explain the occurrences of the two variations, but soil moisture regime seems to be decisive in the delimitation of the two variations.

Microchloa caffra-Elionurus muticus grassland
This grassland covers the largest part of the Ea land type within the study area with respect to the other plant communities. It occurs on a wide range of soil types, for example vertic Arcadia to orthic Glenrosa Forms. This grassland occurs on crests, slopes and plains (Figure 7). The terrain as a whole is gently undulating with slopes of 0°-8°. Overgrazing, especially by sheep, often results in patches of bare soil, which are prominent in this grass land. The reason for patch-overgrazing is the dispropor tionate utilization o f the veld, resulting in patches being over-utilized (Fuls 1992). These patches are gradually retrograding until bare patches of soil develop. (Table 2), indicating a higher soil moisture content with respect to the former variation.

Themeda triandra-Eragrostis plana dry/wet grassland
This grassland represents a transitional zone between relatively dry and wet grasslands, separating the relatively drier communities on the higher-lying terrain units from the relatively moister communities on the lower-lying terrain units (Figure 7). The soils are deep (> 5 0 0 mm) and moist (Figure 8). Overgrazing in this grassland can be observed to a lesser extent in the form of bare soil patches, but rather as dense patches of Eragrostis plana, invading the disturbed areas.
The transitional grassland is differentiated by the presence of species groups H and I, and the absence of species group G (Table 2)

Eragrostis plana-E ragrostis cunula wet grassland
This grassland represents the vegetation falling within moist to wet areas, including footslopes and drainage lines (Figure 7). The areas adjacent to drainage lines, display gentle slopes (0°-5°) and subsequently show few signs of erosion. Rivers and streams in contrast, often used by cattle and sheep as drinking places, frequently show serious signs of erosion. Seasonal and perennial water pans occur widely scattered over the area, often attracting a variety o f birds. These pans are also accessible to cattle and sheep. Since most o f the rivers and streams flow throughout the year, they may be used by stock at any time.

Eragrostis cun'ul a -Setaria sphacelata moist grassland
This grassland represents the vegetation found in moist areas adjacent to drainage lines, and is transitional to grass land (Figure 7). This is indicated by the presence o f spe cies groups H and I, which are differential species for this community. Soils are deep ( > 5 0 0 mm) and without any surface rocks. These areas are not suitable for cultivation due to the high clay content ( > 55%) o f the soils. Coverabundance values for the diagnostic and dominant species are relatively high, indicating a dense vegetation cover. (Table  2). An average of 17 species was recorded per sample plot.

Eragrostis p la n a -Paspalum distichum wet grassland
This grassland represents the vegetation found along rivers, streams and pans (Figure 7). Riverbeds and streambeds are degraded to a large extent, often displaying bare rock surfaces and alluvium (Figure 8). Soils found here are of the Rensburg Form, being deeper than 500 mm. Few widely spaced pans do occur in the area; they are restricted to depressions. Trampling effects by cattle and sheep are noticeable especially on the margins of these pans, where the animals normally drink. The vegetation in these marginal zones is clearly disturbed.  Figure 9 represents the distribution of all 100 releves along the first and second axes of a DECORANA ordination. No discontinuities are observed and by inspec tion it is clear that the vegetation communities are distributed along a moisture/trophic gradient. This gradient is extracted by the first axis of the DCA ordination. McDonald (1987) found in his study on the vegetation of the Swartboschkloof that soil moisture played a secondary role next to the major role being played by soil geology. In his study on the vegetation of the mire Northern Kisselbergmosen, SE Norway, Okland (1990) found that depth to the water table had a major influence on the dis tribution pattern of the vegetation. The communities found under extreme conditions, namely the dryland and wet land communities, occur on the periphery of the diagram. The dryland community is represented by the Artemisia a fra -R h u s dentata shrubveld, occurring on dry, sandy, dystrophic soils to the top left of the diagram.

A decrease in species diversity can be observed, if the species-richness o f this community is compared with that of other vegetation types. This grassland is characterized by species group K (
The wetland community, Eragrostis pla n a -Eragrostis curvula grassland, occurs on wet, clayey, eutrophic soils to the right of the diagram. The Themeda triandra-Elionurus muticus and Themeda triandra-E ragrostis plana grasslands are situated in an intermediate position. Figure 9, concerning the last two grassland types. A gradient along the second axis can also be observed. Communities at the top of the axis occur on scarps and slopes, whereas the bottom part represent communities occurring on undulating terrain. Figure 10 represents an ordination of only the Themeda triandra-Elionurus muticus grassland. The releves of variations 2.2.1 and 2 .2 .2 are more or less situated to the right of the releves of community 2.1. The fact that no clear discontinuity can be observed emphasizes the in distinct, gradual change from dryland to wetland commu nities associated with the gradually undulating terrain. Releves to the right of the diagram are situated on wetter lower-lying bottomland areas with deep, clayey vertic soils. The left part of the diagram represents drier higher-lying areas with shallow, sandy orthic soils. N o gradient can be observed along the second axis.

CONCLUSION
The application of Braun-Blanquet procedures to refine the results of the TWINS PAN classification was success ful. Four major vegetation types were identified, which are further subdivided into nine plant communities. These units do exist in practice and can be incorporated in veld management programmes.
It is of the utmost importance that the farmer considers each unit on its own and that management programmes take into account the characteristics of each unit (see Eckhardt et al. submitted). Communities, which are considered as having conservation value, are those occur ring on steep slopes ( 4 0°-9 0°) adjoining wet grass land communities. The Hyperthelia dissoluta-E ragrostis curvula shrubveld and Eragrostis p lan a-P aspalu m If farmers are to benefit practically from the results obtained in this study, it is advisable to bring to their attention the diagram presented in Figure 8. This diagram serves as an important key for the delimitation o f vegeta tion types and habitat units, which are to be managed in accordance with the characteristics o f each.