Invasive alien woody plants of Natal and the north-eastern Orange Free State

The frequency and abundance of invasive alien woody plants were recorded along roadsides and at watercourse crossings in 87% (152/175) of the quarter degree squares in the study area. The survey yielded BO species of which the most prominent species (in order of prominence) in roadside and veld habitats were: Chromolaena odoruta, Solatium mauritianum, Psidium guajava, Rubus spp., Acacia meamsu and Lantana camara The most prominent species (in order of prominence) in streambank habitats were: Acacia dealbata, A. meamsii and Salix babylonica. The greatest intensity of invasion was recorded in the Natal midlands and in the coastal belt of southern Natal, including the metropolitan areas of Pietermaritzburg and Durban. There was relatively little invasion in the north-eastern lowlands of Natal but the potential for expansion is great. Little invasion was recorded in the north-eastern Orange Free State except along some watercourses.


INTRODUCTION
Rainfall varies from 300 to 600 mm in some of the drier river valleys and in the north-eastern interior region to over 1 500 mm along the coast and on some of the peaks in the Drakensberg.Rain falls mostly in summer from November to March but on the coast up to one third of the rain falls during the winter period (Haigh & Wilhelmij 1973;Francis 1977).There are two major indigenous vegetation types or biomes, Savanna and Grassland, in the study area (Rutherford & Westfall 1986) (Figure 1).Twenty Acocks Veld Types (Acocks 1988) occur in the study area and have been grouped into six veld type categories for the purposes o f this survey.These are temperate grassland, moist subtropical grassland, dry subtropical grassland, mistbelt grassland, all falling within the Grassland Biome; tropical bush and savanna, and tropical forest (Table 1 and Figure  2) are classified under the Savanna Biome.Temperate grassland occupies the highest and coldest parts of the study area at elevations of 1 500 m to greater than 3 000 m. Rainfall ranges from 7 0 0 -1 000 mm per annum.Moist subtropical grassland occurs on the eastern slopes and foothills of the Drakensberg from about 1 350 m to 2 150 m.Rainfall ranges from 7 5 0 -1 500 mm per annum.Dry subtropical grassland lies at elevations o f between 900 and 1 500 m and rainfall ranges from 6 0 0 -1 000 mm per annum.This grassland has been invaded by woody thomveld species especially in its drier parts (Edwards 1967).Mistbelt grassland occupies rolling misty country at elevations of between 450 and 1 350 m.Rainfall ranges from 7 5 0 -1 300 mm per annum.Tropical bush and savanna occupies the hot and dry river valleys but with minor changes to the abundance ratings (see next subheading).The presence and abundance of all naturalized alien trees, shrubs and conspicuous climbers were recorded for each veld type category, habitat type (roadsides and adjoining veld, and streambanks) and quarter degree square traversed by road.Although the objective of the survey was to record woody species, other large non-woody and succulent species were included rather than lose valuable information.

The climate of the coastal belt and lowland interior is
Recordings o f roadside and veld invaders were made from a moving vehicle whereas recordings of streambank invaders were made at watercourse crossings.Abundance estimates o f roadside and veld invaders were based on frequency of encounter within road transects of five to ten kilometres in length.Abundance estimates of streambank invaders were based on estimates at specific sites.
The width o f road transects and length of watercourses scanned for invaders varied according to local conditions.Usually no more than 50 m of veld and 100 m o f stream bank habitat were scanned on either side of the road for invaders.Species occurring beyond these ranges and along watercourses which were not crossed were recorded as present in the given habitat type and veld type category but were not included within the formal recordings.
Although Henderson & Musil (1984) suggested the use of a single road transect length of ten kilometres, a variable transect length of between five and ten kilometres was found to be more practical.This applied particularly in the highly dissected veld type categories in the midlands and lowlands of Natal.In these regions it was often not possible to accommodate ten kilometre transects and where there was much invasion a shorter transect length proved to be more manageable.Since all road transects were plotted on maps the survey is easily repeatable despite the use o f variable transect lengths.
Seventeen quarter degree squares were selected for more intensive surveying (Figure 1).These intensive sites were selected primarily to ensure that representative parts of each veld type category and geographical region in the study area were well sampled.They may also be used at a later date for a quick resurvey of the study area to assess any changes that may have taken place.In each of the seventeen sites abundance recordings of roadside and veld invaders were made along a total road length of approximately 30 km (six transects each five km long).Recordings were made at virtually all watercourse crossings.Herbarium specimens o f all invader species which were flowering or fruiting, were collected.Survey routes and road transects were plotted on 1:250000 maps (general survey area) and 1:50000 maps (intensive sites) before a field trip was undertaken.Wherever possible two or more road transects were plotted per quarter degree square.In most instances road transects were not contiguous but were separated by a distance of between five and ten kilometres.This approach was adopted mainly due to time considerations since abundance estimates along road transects can be very time-consuming depending on the intensity of invasion.In addition, it was necessary to have a breathing space between transects not only to avert eye strain but to free one's attention from the immediate roadside and to observe invasion further afield.Road transects along national roads and other routes with heavy traffic were kept to a minimum.All road transects to be sampled were plotted before a field trip was undertaken to ensure non-selective recording in the field.Recordings were made at most bridges over watercourses but some were omitted because o f time constraints and traffic considerations.

A bundance ratings
Minor changes were made to the abundance ratings for roadside and veld habitats used in the Transvaal survey (Henderson & Musil 1984).The two ratings below the old rating '1' were removed.These ratings became obsolete as a result of the standardization o f road transect length from 5 -1 0 km.The abundance ratings for roadside and veld habitats and streambank habitats are given in Table 2.

Sam pling level envisaged and achieved
The sampling level envisaged was at least 60% o f the total quarter degree squares at an average o f 33 km per square which was achieved in the Transvaal (Henderson & Musil 1984).The sampling level achieved in this survey was 87% (152 of the total 175 quarter degree squares) at an average o f 38 km travelled per square.An average of  The selection of abundance rating 5 as the cut-off point square for abundance estimates of roadside and veld in-is arbitrary but one at which a species can be regarded vaders.
The veld type coverage in terms of quarter degree squares and road transects sampled, kilometres travelled and watercourse recordings made, is given in Table 3.

Prominence value
Prominence is used here in preference to the term Importance defined by Henderson & Musil (1984).The prominence value, which has been derived from Curtis' Importance Value (Mueller-Dombois & Ellenberg 1974), is a measure of the prominence (in terms of frequency and abundance) of a species in a vegetation category relative to all other species in the same category.Other aspects, such as rate o f spread and difficulty of control, which should be taken into account when assessing species importance, are not included here, hence the preferred use of the term prominence.
In streambank habitats the prominence value for a species x in veld type category y was calculated as follows: frequency of species x in veld type y scoring 5, 6 or 7 ------------------------------------------------------x 100 sum frequency of all species in veld type y scoring 5. 6 or 7 prominence value = + frequency of species x in veld type y sum frequency of all species in veld type y * each abundance rating was expressed in numbers of individuals/groups recorded per transect (see Table 2).To be both conservative and consistent the minimum number was used in each instance, e.g. an abundance rating of 5 over ten kilometres = 50 and an abundance rating of 5 over five kilometres = 25.** mean no. of individuals/groups per 10 km converted to rating (see Table 2).
as locally prominent (see definition in Table 2).A formula using all abundance ratings would be preferable if each rating could be converted to an absolute value.
In roadside and veld habitats the prominence value for a species x in veld type category y was calculated as follows: total abundance* of a species x in veld sum of the abundances* of all species in veld type y prominence value = + frequency of a species x in veld type y ----------------------------------------------------------x 100 sum frequency of all species in veld type y The highest prominence values in a given category which add up to approximately 160 points out of a total of 200 are printed in bold in Tables 6 and 7.The cut-off point o f 160 points is arbitrary but represents 80% o f the summed prominence values.

Mean species abundance rating in roadside and veld habitats (see Table 7)
The mean abundance rating** o f species x in veld type category y was calculated as follows: mean no of total no. of individuals/groups of species x in individuals/ ve^ y groups per tota] distance along which species x was rated 10 km jn vej(j type y

Mean abundance o f invaders p er km in roadside and veld habitats (see Table 5 and Figure 5)
The mean abundance o f invaders per kilometre in veld type category y/quarter degree square z was calculated as follows: total abundance* of all species in veld type y/quarter degree square z mean abundance = - ------------------------------------------------------------------total kilometres rated for abundance estimates in veld type y/quarter degree square 7.In the Grassland Biome there was a progressive increase in the severity of invasion in terms o f percentage crossings heavily invaded with decreasing elevation from the montane region (temperate grassland) to the mistbelt.The rivers in mistbelt grassland and tropical forest south of Durban were the most invaded in terms of percentage crossings heavily invaded and percentage crossings invaded.The rivers o f tropical bush and savanna were the least invaded (Table 4).

The whole study area
Five hundred and seventy eight watercourse crossings were sampled in which 90 species were recorded, with up to 13 species in one sample (Table 4).Invaders were present at 81.7% of all crossings and 24,9% of all crossings were heavily invaded (Table 4).

A nalysis according to veld type
More alien species were recorded in the Savanna Biome than in the Grassland Biome.The most species were recorded in tropical forest and the least in temperate grassland.

Frequency
Only eight species were recorded at 10% or more crossings in the whole study area (Table 6).Acacia dealbata and Salix babylonica were the most frequently encountered species.
In the Grassland Biome the most frequently recorded species were Acacia dealbata, Salix babylonica and Acacia m eam sii.Ricinus communis was by far the most frequent invader in the Savanna Biome followed by Psidium guajava, Chromolaena odorata, Solanum mauritianum and Melia azedarach which were almost equally frequent.

The highest percentage frequency recorded was 73 % for
Salix babylonica in temperate grassland.

Prominence
The most prominent invader in the study area was Acacia dealbata, followed by A. m eam sii and Salix babylonica (Table 6). A. dealbata was very common to abundant (i.e.scored a 5, 6 or 7) at 10% (60/578) of all river crossings in the study area.A. m eam sii and Salix babylonica were very common to abundant at 4% (24/578) and 3% (19/578) respectively (Table 6).
In the Grassland Biome A. dealbata was by far the most prominent invader followed by Salix babylonica and A. m eam sii.Salix babylonica was prominent in temperate grassland, A. dealbata most prominent in moist sub tropical grassland and both Acacia spp.were prominent in dry subtropical grassland and mistbelt grassland.
In the Savanna Biome Chromolaena odorata, Lantana camara and Ricinus communis were the most prominent invaders.The first two species were most prominent in tropical forest and although they were not the most frequently recorded species they formed dense stands more often than did any other species.Ricinus communis rarely formed dense stands, its prominence being largely attributable to its high percentage frequency.Sesbania punicea and Melia azedarach were the most prominent species in tropical bush and savanna.

R oadside and veld habitats
The whole study area One hundred and fifty two quarter degree squares and 350 road transects were sampled in which 116 species were recorded.Up to 44 species were recorded per quarter degree square.Invaders were recorded in 93,7% o f all transects sampled and 30,6% of all transects were heavily invaded (see Table 5).

Analysis according to veld type
Tropical forest was the most heavily invaded in terms of species diversity and abundance o f invaders (Table 5).Mistbelt grassland was the next most heavily invaded.Within these veld type categories the Pietermaritzburg, Durban and particularly the south coast were the most heavily invaded (Figures 4 & 5).Temperate grassland was the least invaded.

Frequency
The most frequently recorded species in the study area were Melia azedarach, Acacia m eam sii, Solanum mauritianum and Ricinus communis (Table 7).Only about 14 spp.were recorded in more than 10% of all transects.
In the Savanna Biome Ricinus communis, Melia azedarach and Psidium guajava were the most frequently recorded invaders, followed by Solanum mauritianum, Chromolaena odorata and Lantana camara.Opuntia ficusindica was the most frequent invader in tropical bush and savanna.

Prominence
Chromolaena odorata, despite the fact that it was almost entirely confined to the coastal belt, scored the highest prominence value in the study area (Table 7).It had the highest mean abundance rating (7) o f all species in the study area (Table 7).The next most prominent species in order (mean abundance ratings in brackets) were Solanum mauritianum (5), Psidium guajava (5), Rubus spp., mainly R. cuneifolius (5), Acacia m eam sii (4) and Lantana camara (5).
In the Grassland Biome Rubus spp., Solanum m auri tianum and Acacia m eam sii were the most prominent species.In the Savanna Biome Chromolaena odorata, Psidium guajava and Lantana camara were the most prominent species.
Acacia dealbata deserves mention as the second most prominent invader after Rubus spp. in moist subtropical grassland.Opuntia ficus-indica rated second most prominent after Lantana camara in bush and savanna.Melia azedarach rated second most prominent after Acacia m eam sii in dry subtropical grassland.

Patterns of invasion
Most invasion in terms o f species diversity and abundance of invaders was recorded in the coastal belt and adjacent midlands.Within this zone most invasion occurs in and around towns and cities particularly Pietermaritz burg and Durban.The coastal belt is heavily invaded south o f Durban as well as in the north around Mtunzini and Lake St Lucia (Figure 5).The north-eastern coastal plains and bushveld north of Lake St Lucia have little invasion except along the perennial rivers such as the Pongola.3 and 4, indicating the severity of invasion in streambank and roadside and veld habitats respectively, shows similar patterns except that in the uplands there is more severe invasion of the streambank habitat than o f roadside and veld habitats.This pattern of streambank invasion in the uplands is mainly the result o f Acacia dealbata invasions (Figure 6A) and, to a lesser extent, that o f Salix babylonica (Figure 7K).

Sam pling
As mentioned previously (Henderson & Musil 1984) the sampling method has its limitations, such as the under sampling of certain habitats which are inaccessible by road, and the less distinctive species.The results reflect only the situation along road verges, which are highly disturbed sites, and a small strip of veld and watercourse visible from the road.Despite these limitations the method has proved successful and econom ical in obtaining information that otherwise would be unobtainable.
One recommendation with regard to the abundance ratings is that the seven point scale used for streambank habitats be revised or replaced with a cover-abundance scale such as used in the Braun-Blanquet method of vegetation analysis (see Mueller-Dombois & Ellenberg 1974).The present scale is unsatisfactory as it cannot cope      o .
f N fN     The aforementioned seven most prominent species in the study area are not likely to greatly increase their distribution ranges.Solanum mauritianum can be expected to invade the coastal plains north of Lake St Lucia.Chromolaena odorata, Psidium guajava and Lantana camara are all expected to expand their distributions in the same region.Further expansion by all seven species will probably occur in marginally suitable areas.All species can be expected to increase in density within their ranges.Ricinus communis has always been regarded as an introduced weed from elsewhere in Africa.However, recent archaeological diggings in the eastern Cape have unearthed seeds o f this species from a grass-lined storage pit dating back more than 1 200 years (Brink 1988).This suggests that, if indeed introduced, primitive huntergatherers were the agents (Brink 1988).This is in sharp contrast to the majority of our foreign weeds which have been introduced since the colonization o f the Cape 300 years ago.Whether it is classified as an alien or indigenous plant, R. communis is not generally regarded as an important weed (e.g.Macdonald 1983) because it is a pioneer plant, colonizing disturbed sites, and eventually giving way to longer-lived species.

Melia azedarach and
M elia azedarach on the other hand is a long-lived tree and can grow to a considerable size (1 2 -2 3 m in height and a spreading canopy) (Poynton 1972).Add to this its production o f vast quantities o f berries which are dispersed by birds and water, its high germinability, its hardiness to drought and cold, rapid growth and its response to felling by coppicing profusely, then it deserves recognition as an important invader.It is most important in streambank habitats where its efficient downstream dispereal enables it to invade protected areas far from source areas (Macdonald 1983).In this survey it rated as the most prominent tree species in tropical bush and savan na.It has virtually reached the limits o f its distribution in the study area but can be expected to increase in abun dance especially along river banks in the Savanna Biome.

Salix
FIGURE 1.-The study area.Grass land ^nd Savanna Biomes, in tensive sites and survey routes.
FIGURE 2. -The six broad veld type categories in the study area (af ter Acocks 1988).
frequency of occurrence; I = % crossings heavily invaded; P = prominence value; * = species occurring in the given category but not included in a formal recording at a watercourse crossing; bold numbers highest prominence values i n a given category which add up to ± 80% of the summed prominence values (see text)frequency of occurrence; I = % crossings heavily invaded; P = prominence value; * = species occurring in the given category but not included in a formal recording at a watercourse crossing; bold numbers = the highest prominence values i n a given category which add up to ± 80% of the summed prominence values (see text)frequency of occurrence; I = % crossings heavily invaded; P = prominence value; * = species occurring in the given category but not included in a formal recording at a watercourse crossing; bold numbers -the highest prominence values i n a given category which add up to ± 80% of the summed prominence values (see text).
frequency of occurrence; A = mean abundance rating; P = prominence value; * = species occurring in the given category but not included in a formal recording in a road transect; bold numbers = the highest prominence values i n a given category which add up to ± 80% of the summed prominence values (see text).
frequency of occurrence; A = mean abundance rating; P = prominence value; * = species occurring in the given category but not included in a formal recording i n a road transect; frequency of occurrence; A -mean abundance rating; P = prominence value; * = species occurring in the given category but not included in a formal recording in a road transect; bold numbers = the highest prominence values in a given category which add up to ± 80% of the summed prominence values (see text).
frequency of occurrence; A = mean abundance rating; P = prominence value; * = species occurring in the given category but not included in a formal recording in a road transect; bold numbers = the highest prominence values in a given category which add up to ± 80% of the summed prominence values (see text) FIGURE 3 .-Invasionin streambank habitats in terms of the intensity o f invasion o f watercourse crossings and species diversity per quarter degree square.withall situations and especially where only one species is present.The term 'commonest species' has not been defined but has been subjectively evaluated in terms of cover and density.
Ricinus communis are widespread but most common in the coastal belt (Figures 6 0 & 71).They are considerably less abundant than the species already m entioned and therefore received lower prominence rankings.
babylonica is a widespread streambank invader in the upland grasslands (Figure7K).It is unable to reproduce sexually as only female trees have been introduced into South Africa, but reproduces vegetatively from branches which are torn from trees by floodwaters and deposited downstream (Poynton 1973).It does not form dense stands like the Acacia spp.but a single large tree probably achieves the same basal area as 10, 20 or more Acacia trees.It can also effectively exclude the growth o f other species under its canopy (pers.obs.).In places such as along the Umzimvubu River in south-western Natal, it forms continuous stands which stretch for many kilometres.Salix lasiandra, referred to byImmelman (1987), is less widely distributed (Figure7L) than S. babylonica but it too can form continuous stands in places.It is particularly abundant along the Wilge River near Harrismith.It is more widely distributed and abundant in southern Natal than shown in Figure 7L but the data are not available as it was mistaken for an indigenous Salix species during this part of the survey.Sesbania punicea was the fourth most prominent streambank invader in the study area.It was locally abundant along watercourses in the interior and along roadsides in the humid coastal belt (Figure 7N ).Three species of herbivorous weevils have been imported for biological control of this weed.Already one o f the w eevils has made an outstanding contribution to the biological control o f S. punicea and, in combination with the two other species, is expected to halt the invasive spread o f this plant in South Africa (Hoffmann & Moran 1988).Opuntia spp.are w idely distributed in the study area but occur mainly as scattered plants.Opuntia ficus-indica (Figure 7B) and O stricta are locally abundant in the Tugela River valley and elsewhere in tropical bush and savanna.Opuntia vulgaris occurs as widely scattered plants in the coastal belt.All these species have been the subjects o f biological control campaigns and their numbers have been greatly reduced.They are no longer considered a threat but may still form localized infestations (Zimmermann et al. 1986).Cassia didym obotrya (Figure 61) and Caesalpinia decapetala (Figure 6H) are fairly widespread in the coastal belt and mistbelt o f Natal.They are locally abundant, particularly in disturbed sites around kraals, villages, along roadsides and riverbanks.Cassia didymobotrya is a central African plant and although very abundant in places it is mainly a pioneer plant.Flowering specimens are invariably host to a caterpillar which may help to control the spread o f this species.Caesalpinia decapetala is a vigorousgrowing, exceedingly thorny woody shrub or climber that can form a dense canopy that smothers the existing vege tation and excludes other species.It is particularly trouble som e as an invader o f forest margins.It is easily overlooked during a roadside survey when not flowering and thus is likely to be more widely distributed and abun dant than shown.Less widespread species which are locally abundant and can form dense stands include: Arundo donax (Figure 6G), Pereskia aculeata (Figure 7C), Pinus elliottii (Macdonald & Jarman 1985), Schinus terebinthifolius (Figure 7M) and Solanum seaforthianum in the coastal belt; Tithonia diversifolia (Figure 7P), Montanoa hibiscifolia (Figure 6P) and Leucaena leucocephala (Figure 6N ) around Durban; Cardiospermum grandiflorum and Cestrum laevigatum in the Pietermaritzburg, Durban and coastal areas and Populus x canescens (Figure 7E) along streambanks in the Grassland Biome.U lfx europaeus and Cytisus scoparius have been known to form dense stands in moist subtropical grassland at the Highmoor Forest Estate (National Herbarium specimens collected by D. Edwards in 1961).Arundo don ax, the giant reed, has invaded streambank habitats largely unnoticed because it is similar in appearance to the indigenous reeds (Phragmites spp.).In the Natal coastal belt, where it forms dense stands in places and flowers prolifically, It can be easily recognised by its large inflorescence with dense ascending branches.Suitable habitat for this species occurs along river banks throughout the Savanna Biome.It can be expected to expand its distribution in this region.Schinus terebinthifolius is a popular ornamental and hedge plant in the coastal belt and produces large numbers of berries.There have been observations of seed predation by insects which has probably prevented it from becoming a serious weed (S.Neser pers.com m .).It should nevertheless be regarded as a potentially important invader since it has becom e a pest in other parts of the world, can thrive at the outer limits of vegetation exposed to salt spray and has even invaded mangroves in Florida, USA (Morton 1978; Toufexis 1985).Dense stands of this species were observed in the Umgeni River in Durban North just upstream o f the Beachwood Mangroves Nature Reserve.If seed is made available S. terebinthifolius may even threaten the swamp forests of Maputaland in north-eastern Natal.This habitat is of very limited extent and according to Macdonald & Jarman (1985) is the least invaded of any habitat in Natal.However several species are being cultivated in these swamps, namely Mangifera indica, Carica p a paya, Musa sp. and Ananas sp.M. indica is naturalized in the swamp forest at Kosi Bay.Concern has been expressed about the invasiveness of Litsea glutinosa in the coastal lowlands o f Natal (Macdonald & Jarman 1985).According to records in the National Herbarium it has been much planted in Durban, regenerates easily from seed and it was recorded as naturalized in the Eshowe District as far back as 1937.It is regarded as a weed on the island o f Mauritius (Holm et al. 1979).Pereskia aculeata is potentially a very serious invader o f coastal forests.It is a very thorny vigorous-growing climbing cactus which can smother and kill the trees it overtops.Its current widespread distribution in the conservation areas o f the coastal lowlands in northern Natal is mainly thought to be the result o f previous intentional plantings around kraals and burial sites (Macdonald & Jarman 1985).Natural spread from these sites o f previous introduction has in many localities been surprisingly limited given its bird-dispersed fruit.But the rate at which it can be spread by frugivorous animals and by vegetative reproduction, is potentially rapid (Macdonald & Jarman 1985).Campbell (1988) reports that the seed is geared for rapid germination in a range o f habitats but that the soil seed bank is likely to deteriorate rapidly, with the occurrence of either germination or seed death.P. aculeata is a difficult weed to control as any part o f the plant which survives treatment can reproduce vegetatively and restart the thicket.Control methods involving slashing, poisoning and burning can result in the total destruction o f all vegetation in the affected area (Bruton 1981).However, in several KwaZulu nature reserves infestations have been hand-cleared with as little disturbance as possible.Hand-weeding and herbicide treatments are necessary in follow-up operations (Macdonald & Jarman 1985).R elation o f invasion to environm ental factorsAlien invasion is related to indigenous veld type categories and broad climaticfactors.There is a general trend for more invasion in terms of species diversity and abundance with decreasing elevation from the cold upland grasslands to the warm coastal belt.Most invasion occurs in the humid to subhumid coast (tropical forest) and mistbelt (mistbelt grassland) where there is little or no incidence o f frost.There is considerably less invasion, particularly in roadside and veld habitats, in the colder and drier veld types.The limited invasion of the northern coastal belt relative to that in the south of the province is interpreted as being a result of less disturbance and fewer plantings o f alien species.Localized infestations, for example around Lake St Lucia, indicate that this region is vulnerable to invasion.According to Liggitt (1983) Chromolaena odorata is spreading rapidly in the north where up to 2000% increases in vegetation cover have been recorded in the Dukuduku plantations during a time span of five years.The distributions of some species correspond well with broad climatic zones.For example Chromolaena odorata is virtually restricted to the frost-free and moist coastal belt, although, being wind-dispersed, it has a potentially wide distribution.Members o f the Rosaceae, such as Rubus, Rosa, Prunus, Pyracantha and Cotoneaster spp., are most evident in the cold high-lying grasslands.This may be attributable to a dormancy mechanism in their seeds which is terminated by cold winter temperatures (Dean et al. 1986).Watercourses have played an important role in the dispersal of species and in particular those which otherwise have a limited dispersal range.For example Acacia dealbata, A. m eam sii, Sesbaniapunicea and Caesalpinia decapetala have rather immobile and hard seeds but which are readily transported by water.The abrasion which the seeds receive along their journey may well promote ger mination.Watercourses have also enabled the long-range vegetative dispersal o f species such as Salix babylonica and Opuntia spp.The spread of suckering species such as Popufas x canescens and Robinia pseudoacacia is also promoted by stream flow.Some o f the important invader species are dispersed by birds enabling them to invade relatively undisturbed sites and far afield from parent plants.Notable species include Lantana camara, Solanum mauritianum, Melia azedarach, Pyracantha spp., Psidium guajava, Rosa spp.and Rubus spp.In the grassland regions bird-dispersed species are clearly associated with perching sites, such as fence lines, rocky outcrops, bush clumps and plantations.In the mistbelt Solanum mauritianum and Rubus cuneifolius form dense thickets in the understoreys o f plantations.CONCLUSION The intensity of alien plant invasion is expected to increase in all parts of the study area and particularly in the coastal and mistbelts o f Natal.Top priority should be given to the control of invaders, especially Chromolaena odorata, in north-eastern Natal [see Macdonald & Jarman (1985), where a possible control strategy is detailed] which is an important conservation area and where the potential for expansion is great.Shortly after completion of this survey, in September/ October 1987, Natal experienced devastating widespread floods.Most o f the major river valleys as well as the floodplains along the coast were severely affected.Vast tracts o f riverine vegetation that used to grow within the flood line were swept away.The long-term consequences o f the floods remain to be seen but in the short-term it can be expected that an explosion of pioneer and other fast-growing plants will occur.In the coastal belt Chromolaena odorata can be expect ed to rapidly invade floodplains, river banks and any other exposed land.Ricinus communis is likely to greatly increase in abundance especially in riverbeds as it did in the Pongola River after the floods caused by cyclone Demoina in January 1984.The floods are also expected to promote the downstream spread o f water-dispersed species such as Acacia dealbata, A. m eam sii, Melia azedarach and Salix babylonica.The floods, together with the prospect of a wetter climatic cycle ahead (Tyson 1986), could result in the spread of species into areas which previously were too dry and inhospitable.In this respect there is a danger that A. dealbata and C. odorata will become more widely established and abundant in bush and savanna.An assessment o f the ecological consequences of streambank invasion, particularly pertaining to water usage and soil conservation, is considered to be extremely important in the motivation for the control o f streambank invaders.Acacia dealbata and A. m eam sii, although declared invaders, have spread uncontrolled throughout the Grassland Biome from the upland grasslands along the Drakensberg escarpment down into the mistbelt.In the case of A. m eam sii it has even spread into the coastal belt.Apart from a few notable exceptions most of the important alien woody invader species in Natal and the rest of South Africa (Henderson & Musil 1984; Macdonald & Jarman 1984; Macdonald e ta l. 1986; Stirton 1978) have been cultivated either on a grand scale in plantations, or as barrier plantings, cover/binders, shelterbelts and ornamentals in gardens.This raises the issue o f screening alien plant species for potential invasiveness before they are allowed to be propagated on a grand scale.This applies particularly to the establishment o f plantations of species new to the region belonging to the genera Acacia and Pinus, many species of which have becom e serious invaders in South Africa.

TABLE 5 .-Statistics for roadside and veld habitats in each veld type category, biome and the study area Mean abundance** Veld type category Total no.
* 1 or more species scored 5 or more.** See data treatment-formulae used.