A homogeneity index based on species diversity in Sour Bushveld

The co-efficient of variation (CV) for the linear regressions of species-area curves, on a linear species number/log area scale, for the samples in a community, appears to indicate the homogeneity of the community. A CV of less than 15% appears to indicate homogeneous vegetation while a CV greater than 15% seems to be indicative of vegetation heterogeneity, as correlated with extremes in recorded environmental factors in Sour Bushveld.

The co-efficient of variation (CV) for the linear regressions of species-area curves, on a linear species number/log area scale, for the samples in a community, appears to indicate the homogeneity of the community.A C V of less than 15% appears to indicate homogeneous vegetation while a C V greater than 15% seems to be indicative of vegetation heterogeneity, as correlated with extremes in recorded environmental factors in Sour Bushveld.

IN T R O D U C T IO N
The plant ecology of the farm Groothoek, Thabazimbi District, was studied in order to supply data on the Sour Bushveld (Acocks, 1975) on natural resource inventory for the Department of Agriculture's Natural Resources Classification.The study included classification of the vegetation according to the Zurich-Montpellier approach (Werger, 1974) as well as alpha and beta diversities according to Whittaker et al. (1979).Alpha diversity refers to the number of species in a community (Whittaker, 1972), whereas the rate of increase in species number with increasing area is also characteristic of a community and is indicative of dispersion (Whittaker et al., 1979).
Both species number and rate of increase can be illustrated by means of a linear regression of the species-area curve on a linear species-number/log area scale (Whittaker et al., 1979).
Many attempts have been made to quantify vegetation homogeneity (Raabe, 1952;Curtis, 1959;Dahl, 1960 andMoravec, 1971), however, there has been no satisfactory objective method (Mueller-Dombois & Ellenberg, 1974).In this study, the relative homogeneity of the plant communities is assessed after the classification of the vegetation.

S T U D Y A R E A
The study area is the farm Groothoek 278KQ situated in the south-western Waterberg area of the Transvaal between southern latitudes 24° 28' and 24° 31' and eastern longitudes 27° 32' and 27° 39'.The original farm Groothoek, which has subsequently been subdivided into a number of small farms covers approximately 4 000 ha.
The study area varies in altitude from 1 050 to 2 080 m.The soils are mainly of the Mispah Form, Mispah Series.Also locally present are: Shortlands Form, Bokuil Series; Hutton Form, Middelburg Series; Westleigh Form, Sibasa Series and Kroonstad Form, Slangkop Series.The soils are derived from sandstone conglomerate and shale of the Waterberg Group as well as post-Waterberg diabase.The vegetation is described by Acocks (1975) as Sour Bushveld, with an outlier of North-Eastern Mountain Sourveld above 1 600 m altitude.

M E T H O D S Analysis
A total of 170 quadrats, each measuring 10 m x 20 m, was distributed on a stratified random sampling basis throughout the study area for classifying the vegetation.Species diversity data (Whittaker et al., 1979) were obtained by recording species presence in nested quadrats within each 10 m x 20 m quadrat.Five 1 m x 1 m, one 2 m x 5 m and one 10 m x 10 m subsamples within each 10 m x 20 m quadrat were used.All permanently recognizable species were recorded in each quadrat.
Habitat data recorded at each sample site included altitude, geological formation, aspect, slope, soil depth, soil form and series, geomorpho logy, vegetation structure and estimated percentage surface rock cover.Soil samples were collected for later chemical analysis.

Synthesis
The vegetation was classified into communities and sub-communities with the aid of the PHYTO TAB program (Westfall et al., 1982) and the condition of the vegetation with respect to grazing was assessed (Westfall et al., 1983).The species diversity for each quadrat was determined by the linear regression of the species area curve, on a linear species-number/log area scale.Species num ber is plotted on the ordinate with the log of the area, in square metres, on the abscissa.The log area scale, therefore, reads: 0 (1 m2); 1 (10 m2); 2 (100 m2) and 2,3 (200 m2).The species number for 1 m2 was taken as the average of the five 1 m2 subsamples for each quadrat.
The species number was extrapolated to log 3 (1 000 m2), because this range can be extrapolated to allow standard comparisons (Whittaker et al., 1979).The co-efficient of variation (CV) was calculated for the variation in species number at log 3 of the quadrats in each community.The communities were then listed according to their CV and comparisons were made with recorded environ mental factors.

RESULTS
The listing of communities according to their quadrat CV is given in Table 1 together with the environmental factors which have a sufficiently wide variation within a community to possibly account for the high CV.The following factors appear to be significant ( Community 8.1 does not have a single recorded habitat factor which could account for the high CV.However, a combination of habitat factors or possibly a single factor, not recorded, could be responsible.Community 2 is situated on the escarpment in the west of the study area.Although regarded as kloof forest the landform is hetero geneous with cattle able to graze in places.The heterogeneity of the landform, although difficult to quantify as extreme, could account for the high CV in Community 2.   It appears, therefore, that a CV of 15% is an acceptable boundary between heterogeneous and homogeneous vegetation in Sour Bushveld, because no environmental heterogeneity (extremes in envi ronmental factors) could be detected for the communities with a CV of less than 15%.The CV gradient represented in Table 1 is difficult to corroborate with the recorded environmental fac tors, but personal observation confirms the order of the communities regarding homogeneity.

D IS C U S S IO N
The variations in species number (diversity) and the rate of increase in species with increasing area (dispersion) provide a simple method of determining homogeneity in a floristically defined community.The rate of increase in species with increasing area partly determines the species number at log 3 so that both parameters, diversity and dispersion, are taken into account.
In the study area the main environmental factors influencing vegetation homogeneity are grazing, slope, aspect, soil depth, surface rock cover and soil electrical resistance as an expression of soluble salts.Notably lacking in influence is soil form, where three communities in the homogeneous range have two different soil forms each.Communities 5 and 4 are structurally and floristically similar, the difference being that Community 4 is diagnosed by mainly weedy species.This is confirmed by the lower composition score and higher CV for Community 4. The difference is attributed to overgrazing in Community 4.
The homogeneity index, as expressed by the CV for the communities, is of particular importance when interpreting the composition scores.For example, Community 18 has a high composition score, but is relatively heterogeneous.Grazing management should, therefore, take the heter ogeneity of the vegetation into account.The open woodlands have markedly higher composition scores and are generally less heterogeneous than the closed woodlands.This could be significant in timeous control of woody encroachment.Community 8.2, for example, although predominantly open, is closed in places.
The homogeneity index can also indicate sam pling adequacy.In the floristic classification, Community 17a is represented by two quadrats only and is considered a variation of Community 17b.Community 16 is represented by 12 quadrats of which three are heavily infested with Stoebe vulgaris.The potential vegetation is grassland without S. vulgaris, however, and the high CV indicates the infestation.Community 2 is represented by eight quadrats.This community could consist of several communities, because of the topographic range, if studied in greater detail than in the present study.It is noteworthy that sampling adequacy in this study is suspect with a CV of greater than 20%.The remaining communities are considered to be adequately sampled.

C O N C L U S IO N S
The homogeneity index based on species diversity appears to be a simple, yet sensitive method for quantifying homogeneity in floristically defined communities, in Sour Bushveld.The homogeneity index can be of value in assessing veld condition and determining sampling adequacy.Further data will, however, have to be analysed to determine the adequacy of the suggested 15% criterion between homogeneous and heterogeneous vegetation in vegetation other than Sour Bushveld.
The authors thank Dr J.C. Scheepers for comments and suggestions.
R.H. WESTFALL*, N. VAN ROOYEN** and G.K. THERON** A B S T R A C T

T
A B L E 1. -H o m o g e n e ity o f the plant c o m m u n itie s o n the farm G r o o th o e k , T h a b a z im b i D istrict, according to t h c C V fo r q ua drat diversity (e x tra pola ted to 1 0 0 0 m -), and correlated w ith extrem es in recorded e n v iron m en tal c o n d itio n s Veld co ndition assessment, com position Exteme habitat conditions recorded in com m unities grazing * The m in im u m soil depth recorded is 40 m m A range greater than 179 m m from the m in im u m depth appears to influence homogeneity, therefore the range o f 4 0 -8 0 m m does not appear to be significant.

Table 1
(iv) The aspect of Community 18, being on the summit of a mountain, has the greatest variation in the study area.
o r a v e c , J., 1971.A simple method for estimating homotoneity of sets of phvtosociological releves.Folio Geobot.Phytotax.Praha 6: 147-170.M u e l l e r D o m b o is .D .& E l l e n b e r g , H ., 1974.Aims and methods of vegetation ecology.New York: John Wiley & Sons.R a a b e .E .W ., 1952.Uber den " Affinitatswert" in der Pflanzensoziologie.Vegetatio 4: 53-68.W e r g e r , M .J .A ., 1974.O n concepts and techniques applied in the Zurich-Montpellier method of vegetation survey.e s t f a l l , R .H .D e d n a m .G ., V a n R o o y e n , N .& T h e r o n .G .K ., 1982.P H Y T O T A B -A program package for Braun-Blanquet tables.Vegetatio 49: 35-37.W e s t f a l l .R .H ., V a n R o o y e n , N .& T h e r o n , G .K ., 1983.Veld condition assessments in Sour Bushveld.Proc.Grassld Soc.sth.Afr.In press.
D a h l , E., 1960.Some measures of uniformity in vegetation analysis.Ecology 41: 805-808.M W W h it t a k e r .R .H ., 1972.Evolution and measurement of species diversity.Taxon 21: 213-251.W h it t a k e r .R .H ., N ie r in g , W .A .&C r is p , M .D ., 1979.Structure, pattern and diversity of a mallee community in N ew South Wales.Vegetatio 39: 65-76.