Ecology and population biology of Euphorbia perangusta ( Euphorbiaceae ) in the Transvaal , South Africa

The conservation status of Euphorbia perangusta R. A. Dyer, an endangered plant restricted to the Marico District o f the Transvaal, South Africa, and adjoining parts o f Bophuthatswana was determined. The distribution, habitat and population dynamics o f E. perangusta are discussed. The monitoring o f the largest known population has revealed that this population has declined rapidly since the onset of a drought in 1983. The major cause of this decline appears to be the destruction of the plants by porcupines which feed on E. perangusta during droughts. It appears that, during droughts, E. perangusta is restricted to rocky ridges because there is an increase in porcupine damage on more accessible populations. The species is also subjected to other factors which reduce flower formation in a large proportion of plants. If these factors continue to operate, the species could become extinct in the near future. Conservation recommendations are discussed.


INTRODUCTION
Euphorbia perangusta R. A. Dyer was first described (Dyer 1938) from material collected in the M arico Dis trict of the western Transvaal.The plants are distinctive in that the primary branches rarely re-branch and that they have very thin wing-like angles (Figure 1).The species is threatened by collecting, small mammal dam age and trampling by cattle (White et al. 1941;Fourie 1982).This has resulted in a serious decline in the number of plants, causing it to be listed as an endangered species (Fourie 1986).
Fourie (1982) briefly described the habitat of E. p e r angusta Dyer (1938) and White et al. (1941) state from notes made by the discoverer of the species that E. p e r angusta appeared not to set seed even though large numbers of flowers were formed.

D istribution
All distribution records for E. perangusta were ex tracted from the literature and the National Herbarium in Pretoria, using the method proposed by Hall et al. (1984).The recorded localities were visited and the species searched for and positively identified in the field.New populations in the Transvaal (excluding Bophutha tswana) were searched for in the manner described by Fourie (1986) in June 1985.Populations were defined as stands further than two kilometres apart.

H abitat
The physiognomy of the vegetation associated with each population was recorded.The vegetation structure was classified according to Edwards (1983).Average annual precipitation was obtained from a local land owner and tested for reliability against W eather Bureau records (W eather Bureau 1986) of rainfall stations in the District.Altitude to the nearest 20 m above sea level and aspect for each new population found was recorded from South Africa 1:50 000 topocadastral maps.Soil texture was recorded on site for each population using the sau sage method (National Working Group For Vegetation Ecology 1986).The geology associated with each popu lation was determined using a recent 1:250 000 geologi cal map (Visser 1984).
In addition to the above, presence of all trees, shrubs and herbs growing within 3 m of twenty randomly se lected E. perangusta plants were recorded, i.e. plants directly associated with E. perangusta.The species at each population were recorded.The Czekanowski coef ficient (Bray & Curtis 1957), was used to show the per centage similarity between sets of these species.The formula used was 2w --------x 100 where W = number of associate species in common; A = number of associates for one population, and B = number of associates for another population.

M onitoring
In June 1979 the largest population (no. 1) of E. p e r angusta was first visited and a total count of all plants done.The state of each plant, i.e. whether vegetative or flowering, was recorded.
In June 1985 a total count of all plants in all popula tions found was done and the state of each plant, i.e. whether vegetative or flowering, was recorded.
A permanent monitor plot measuring 30 m x 50 m was established centrally within the largest known E. perangusta (no. 1) population so that changes over time in the population and the vegetation associated with the population could be recorded.The four corners of the plot were permanently marked by piles of white painted rocks.Monitoring was carried out during the last week of June 1985 and 1986.
The monitor plot was subdivided into sixty 5 x 5 m blocks and the exact locality of each individual E. p e r angusta plant within these blocks was mapped.This pro cedure allows for individual plants to be easily relocated in successive years and facilitates the recording of infor mation for each plant in the plot.This procedure is es sential if a full life-table analysis is to be made as an aid to understanding the population biology of E. p e r angusta.It is possible that short-lived seedlings that were recruited into the population may have been missed by monitoring only once a year.The classes of factors affecting the plant, land use practices and disturbances to the habitat were determined by the percentage of E. perangusta plants affected and divided into the following: (i) modifies but does not de stroy habitat or affects but does not kill plant and (ii) destroys habitat and/or plant or has potential to do so.
Photographs of the monitor plot from a fixed photo point were taken in June 1985 and June 1986 to visually record change in the vegetation over time.

D istribution
Prior to this study E. perangusta was known from only two populations (White et al. 1941); one of which is in Bophuthatswana.During this study four additional populations were found in South Africa.Figure 2 shows the known distribution of E. perangusta in the Trans vaal.In South Africa the species is restricted to the Marico District of western Transvaal.

H abitat
All populations occur within Acocks (1975) Veld Type No. 13-Other Turf Thornveld, are at altitudes of 1 050-1 150 m and receive an annual average precipita tion of 450 mm.Four of the five known Transvaal popu lations are found on the southern or south-eastern slopes of quartzite ridges.Table 1 shows the approximate dis tances between the five known E. perangusta popula tions in the Transvaal.The plants grow in partial to full sunlight, are wedged between rocks in well drained, gritty sand and vary in height from 50-900 mm.The vegetation associated with E. perangusta populations on the ridges is a Croton gratissim us var.C roton gratissim u s var.gratissim us
Table 2 presents the floristic composition of the vege tation associated with the five E. perangusta popula tions.

Population sim ilarity
Table 3 shows the percentage similarity between the five populations based on associated vegetation using the Czekanowski coefficient.The habitat of E. perangusta is similar in all popula tions except for the population growing on the flat, sandy plain.Similarity values between sets of associates (Table 3) were high for the four populations occurring on the rocky ridges.Population no. 5 showed a low similarity between sets of associates with each of the other four populations.
The difference in the species composition of the asso ciated vegetation of the population growing on the flat, sandy plain may be due to the deeper, alluvial soil in that area.The occurrence of Terminalia sericea, a tree spec ies associated with deep, sandy soils (Coates Palgrave 1983; Palmer & Pitman 1972), the taller, more closed structure of the associated vegetation and the taller E. perangusta plants within this population, support this view.

Num ber o f plants in each population
Table 4 presents the number of plants for each popula tion of E. perangusta as determined in June 1985.It shows that the populations on the rocky ridges (nos 1 -4 ) comprise more plants than the population (no.5) on the flat, sandy plain.

Population biology
Figure 3 shows the population dynamics of the largest known E. perangusta population (no. 1) as determined from a survey undertaken in June 1979 and monitoring over the period 1985-1986.Figure 3 shows that this population is declining dramatically and could soon be come extinct.angusta by an unidentified moth species.

Distribution
From this study it appears that E. perangusta is en demic to the Marico District of the western Transvaal in South Africa and western Bophuthatswana.The highly localized distribution range of the species within the Transvaal was intensively and repeatedly searched and it is unlikely that any additional populations remain undis covered in the Province.
The distribution o f E. perangusta extends marginally into Bophuthatswana; with one population (White et al. 1941) being recorded for that region.It is possible that a few small populations remain undiscovered in Bophu thatswana and adjacent parts of Botswana. 1990 1985

Years
The greatest factor causing the death of E. perangusta plants are porcupines which expose and eat the subterra nean stems and roots (White et al. 1941; Fourie 1982; this study).The shortage of available food since the start of the drought has forced the porcupines to utilize alter native sources; one source being the roots and stems of E. perangusta plants.Porcupine damage has increased since the start of the drought and has resulted in a rapid decline in the number of plants in the population being monitored.
Porcupine damage was most evident in population no. 5 located on the deep, sandy plain.It is obviously easier for the porcupines to expose the roots and stems of plants growing in deep sand than those growing wedged between rocks.The majority of plants on the rocky ridges that were damaged by porcupines grew in relati vely large pockets of soil.

Flower formation
The sudden decline in flowering from 1985 to 1986 may be due to reduced rainfall in 1986.Flower forma tion appears to be suppressed when there is a shortage of water during the period of active growth (O ctober-June) of the plants.
The fewer flowers formed in 1986 resulted in a lower potential for seed set and consequently, the number of seedlings that could be recruited into the population.

Population biology
Mortality E. perangusta numbers confirm that it is an en dangered species and is in danger of becoming extinct if the factors causing its decline are allowed to continue operating.The results show that plant mortalities have increased dramatically since the start of a drought in June 1983.

Factors affecting the taxon (i)
Trampling: the five Transvaal populations were subjected to trampling by cattle.Although normally nonfatal, trampling has the effect of breaking off emergent branches and exposing the subterranean stems.The drought has forced cattle to wander onto the rocky ridges in search of food.This increased exposure of E. p eran gusta plants to trampling, and the subsequent breaking The following information was recorded for each E. perangusta plant within the monitor plot: a, plant dead or alive; b, state of plant-vegetative or flowering; and c, factors affecting the plant.Additional information re corded included (a) precipitation for year of monitoring; (b) land use practice; (c) vegetation structure; (d) domi nant associated vegetation; and (e) disturbances to the habitat of the population.

FIGURE 2 .
FIGURE 2.-The known distribution of Euphorbia perangusta R. A. Dyer in the Transvaal.
gratissim us domi nated woodland with a tall (3-5 m), open structure (Ed wards 1983).The fifth population of plants was found growing ap proximately 40 m from a quartzite hill on a flat, sandy plain.No surface rocks were present.The plants grow in the shade of trees in deep (>1 m) alluvium soil and vary in height from 0 ,2 -1,2 m.The vegetation associated with this population is a Peltophorum africanum and Term inalia sericea dominated woodland with a tall ( 5 -1 0 m), closed structure (Edwards 1983).

Figure 3
Figure 3 shows that the total population (no. 1) de clined from 135 plants in 1979 to 108 in 1986.5,18 % of the plants in the monitor plot died between June 1979 and June 1985, while 15,26 % died between June 1985

FIGURE 5 .
FIGURE 5 .-Thedestruction of emergent branches of Euphorbia p e r angusta by trampling.