188 S. Arr. J. Bot. 1996,62(4): 188- 195

Successional pathways in disturbed coastal dune forest on the coastal dunes in north-east KwaZulu-Natal, South Africa

G.P. von Maltitz, G.F van Wyk* and D.A. Everard

Division of Water, Environment and Forestry Technology, Box 395, Pretoria, 0001 Republic of South Africa ~Oiv i sion of Water, Environment and Forestry Technology, Futu lulu Forestry Research Sialion, Private Bag X7066, Mtubaluba, 3935 Republic of Soulh Africa

Received 29 September 1995: reviJed 27 AI'ril1996

The floristic composition and structu re of secondary dune forest, established on old lands and in grassland. was Investigated on the coastal dunes of the north-eastern coast 01 KwaZuru-Natal. Two distinctly diNeren! pathways from disturbance to forest were identified. One is typifie d by an even-aged stand of Acacia karroa, the other by bush clumps in a grassland matrix. The bush clumps are initiated around a single , normally Syzygium corda tum, which then acts as a nurse fo r the establishment of other woody species. The species composition and structure of the bush clumps were the same, regardless of whether the area where it established was identified as an old field or grassland. The bush clumps had a substantially higher density of both woody and woody species than did the A. karroo stands.

Keywords: Disturbances, forest development, grasslands, KwaZulu-Natal, secondary succession.

°To whom correspondence should be addressed.

Introduction Study site The dune systems of the north-eastern KwaZulu-Natal coast have The Eastern Shores (28° 16'S;32°29'E) study si te is a strip of land probably been subjected to some form of human di sturbance bounded in the west by Lake S1. Lucia and in the east by the since their formmion c 10 000 years ago (Hall 1981; Maud Indian Ocean. The area in which plots were located extended 1990). Since the arrival of Early Iron Age people, approximately from a few kilometres north of the St. Lucia estuary mouth to just 1700 years BP, there has been slas h-and-bu m agriculture wi thin north of Cape Vidal. The study was limited to the high coastal the dune forest communities (Hall 1980). In the Eastern Shores dunes that run parallel to the coast. area (east of Lake St. Lucia), the disturbance regimes changed aft!:!r {he earl y 1970s when the Department of Forestry removed Geology and soils the last of the subsistence farmers. The result of this removal is The dune systems of the Zu luland coast are at the southem (hat many areas that were previously cultivated or fire-main­ extremity of the Mozambique coastal plain. The geology and tained grasslands are now reverting to forest. An idl!al opportu­ development of the dunes are well described in Hobday (1979). nity to investigate successional processes from old fields and/or During the last glacial period (Wurm Glaciation), which had a grasslands to forest therefore ex ists along this portion of coastal midpoint approximately 18000 years ago, the sea-level de sce n~ dunes, as the vegetation covering these dunes was generally pro­ ded to approximately 100 m below present sea-level (Tinley tt:::cted fro m nrt::: and other disturbances. This was done with the 1985). Large areas of sands were exposed on the coastal plain. intenti on for il 10 revert to forest (Venter 1972, 1976; Weisser The present dune cordon was presumably built up (possibly on 1978: Weisser & Marques 1979; Weisser & Muller 1983; von the base of previous Pleistocene deposits) as one of a series of Maltitz el al. 1990). Approximate time-lapse si nce the fields dunes derived from these exposed sand deposits during a period have been abandoned can be calculated from aerial photographs when there were hi gh wind velocities (Hobday 1979; Tinley that have been taken periodically since 1937. 1985). After the stabilization of the sea-level at around present A successional pathway involving Acacia kanvo, as an initial levels, parabOlic blowout dune-building processes (Tinley 1985) serial stag t::: establishing on old lands, has bl!cn well documented could have resulted in the growth of the coastal du nes to their on the Zulu land coastal dune systems (Venter 1972, 1976; present form. The precise dynamics of coastal changes are deter­ We isser 1978: We isser & Marques 1979; Breen 1979; We isser & mined by local coastal processes and vary over time. From an Muller 1983: Mentis & Ellery 1994). However, this is by no ecological point of view, the Zulu land coastal dunes must be means the only or most common successional pathway from dis­ ,regarded as a young formation, possibly less than 10000 years turbances to forest within this dune system (Weisser & Muller old (Maud 1990). 1983: von Maltitz et (II. 1990). Almost no data are available on Studies on the coastal dune soils have shown that the nutrients what species are involved in these different pathways, how the are concentrated in the topsoil of the dunes (D.R. MacDevette pathways change over time, or their consequence to species com­ pers commun.; van Daa len el 01. 1986). Clays and nutrients are position and diversi ty. These issues were Ih l! target of our study. higher in the A horizon than in the B horizon. The deve lopment An un derstanding of the rates of successional change, the pos­ of water-repellent soil s may be a significant factor in coastal dun e~sys tem ecology. sible different pathways and what initiates these, and the cons~­ quences of different pathways, is important, not only for the conservat ion of the area, but also to help in determin ing rehabi li ­ Climale tation programmes for other arl!as of dune forest that are severl!ly Gradients of rainfall occur along the Zululand coast from around disturbed. 1 170 mm at Mtunzini to I 300 mm at Cape St. Lucia to I 100 S. Afr. J. Bot. 1996,62(4) 189

mm at Sodwana. Rainfall decreases from the coast as on!! moves mining the species composition in these forests. inland. Winds, which are predominately from the north-east and Successional gradients in the dune vegetation were described south-west, are a vital faclor in determining coastal dune pro­ by Moll (1972), Van Daalen et al. (1986), MacDevelle & Gordon cesses. Gale-force winds occur mostly from September to ( 1989) and Weisser (1991). Detailed work on succession follow­ December, with wind speeds greater than 50 km h-I occurring on ing disturbance was done by Weisser & Marques (1979) and an average or 12 days per year in Durban (Tinley 1985). Tropical Weisser & Muller (1983). Most or these studies rocused on the cyclones, oflen accompanied by extremely high wind speeds and interpretation of aerial photographs to determine the rates and high rainfall, have tremendous effects on the dune systems within sequence of changes in the dune vegetation over the last 40 to 50 one day (such as the large coastal erosion at Mabibi following years in an effort to deriw a satisfactory classification of areas cyclone Claude in 1969) (MacDevette 1989a). Although these for nature conservation. In tht! warm, moist tropical and subtropi­ catastrophic cyclones occur infrequently al ong the coast, they cal climate experienced along the Zululand coast the rate of appear to be a significant disturbance factor in the dune systems. succession and change is relatively rapid (Weisser & Marques Eastern Shores is just south of the 18°C isoline and hence coin­ 1979; Weisser & Muller 1983; Pammenter et af. 1985). cides roughly with the southern boundary or Koppen's Tropical Savanna climatic type (von Bruggen & Appleton 1977). Methods Archaeology and history Aerial photographs taken from 1937 to 1985 were used to identify Stone Age peoples were present in the mountains bordering the areas that had reverted from either old tields or grassland to ­ Mozambique Coastal Plain and presumably also in the coastal land s. Areas consi sting of pure A. karroo stands were easily identi­ fied on the photographs, and sires wen;: chosen to give a mix of A. areas from at least 100 000 years ago (Beaumont et at. 1978). !wrroo and non-A. !wrmo stand s. The approximate age since Hence man's presence on the coastal plain pre-dates the forma­ started to establish in the sites was determined using a sequence of tion of the present-day coastal dune cordon. The influence of ae rial photographs (1937. 1960. 1970. 1985). From these, the last humans on the natural system presumably only started with tht! period before trees established was identified, and hence the approx­ arrival of iron smelting and shifting cultivation practices in the imate age of the established trees. The sites were located in the field coastal dunes by Early Iron Age peoples (Feely 1980; Hall and two or three monitoring plots were located in each site. 1981). The arrival of callie in approximately I 000 AD would Plots were located in abandoned fields (topsoil disturbance) which have been an additional influence on the vegetation due to the reverted to A. karroa woodlands and in bush clumps which origi~ increased frequency of fire initiated by the local inhabitants to nated in abandoned fields or in grasslands (no topsoil disturbance). provide favourable grazing. Three different-aged stands were sampled (Table J). A total of 28 Work done by Hall (1980) on Iron Age sites at Eastern Shores plots were sampled. of which II were old lands that had developed indicates that a relatively large number of people lived there. The into stands of A. karma, nine were identified as old lands which had environment was favourable for human occupation as there was developed into bush clumps (with species other than A. karma) and fresh water close at hand, an abundant supply of seafood, game eight were grasslands that developed into bush clumps. animals, good agricultural sites, an abundant variety and supply Circular plots with a diameter of 1l.3 m (400 m2) were located of plants for herbal, medicinal, building and craft work purposes around a central large tree. In bush clumps, the largest tree was cho­ as well as good grazing for livestock (Cunningham 1985). These sen but. in the even-aged stands of A. karma. a random mature indi­ people may have exercised a significant selective influence on vidual was used as the central tree. The species, di stance from the the plant and animal populations. Certain sites were more central tree, height and diameter at breast height (DBH) (if greater favoured for agriculture, and certain species of trees used for than 50 mm) were recorded for all woody species (excluding seed­ medicinal purposes or food were not cut when land was cleared. lings less than 0.25 m) within the 11.3-m radius from the central tree. These factors undoubtedly played a significant role in determin­ Shannon and Simpson's diversi ty indices (Odum 1983) were cal­ ing the pre se nt~day species composition and vegetation patterns. 2 principly through slash-and~bum agriculture and burning for culated for the 400-m plots. This was done to investigate differences in diversity between the three areas sampled. and to ascertain whe­ grazing. ther these variables changed over time. The 1937 series or aerial photographs show that the Eastern Differences in species composition between the three treatments Shores coastal dunes were relatively sparsely covered by dune and age-groups sampled were investigated by classification and ordi­ woodland and forest, the rest of the dunes being covered by nation of the dam set. A polythetic divisive classification was carried grassland (Flight 117 37, scale 1:28 ODO). Parabolic blowouts out using TWINSPAN (Hill 1979). The number of plants per species (called drirt sands by the then Department or Forestry) occurred was used for the input matrix. Cut-off levels were set to 0, 10.20,30. at various places along these dunes, many of which were later 40.50 and 60. The TWINS PAN defaults were used for all other set­ stabilized by the planting of Casuarina equisetifolia. tings. Detrended correspondence analysis. (using the CANOCD pro­ gram) was used to inductively investigate the data (ter Braack 1988). Vegetation To investigate differences in structure between the three lreat­ The Zululand coastal dunes are covered by Acock's veld type Id. ments, the plant and species densities per 50_m2 area of concentric Dune forest (Acocks 1953). The natural vegetation in the area circles radiating out from the central tree were calculated (i.e. from the tree to a radius of 3.99 m, and from between the intervals 3.99. consists of a mosaic of mature dune forest, secondary dune forest 5.65.6.91,7.98,8.92,9.77, 10.96 and 11.30 m). The densities were and grassland. Gradients from east to west across the dune cor­ converted to frequencies and these were compared between path­ don and descriptions of the dune vegetation along the Zululand ways, using non-parametric analysis of variance. Simple linear coast have been presented in detail by Venter (1976), Ward regression was used to test for trends in mean densities at different (1980), Weisser (1980), Tinley (1985), MacDevelle (1989a, b), distances from the central tree. The total densities of plants and spe­ MacDevette & Gordon (1989) and von Maltitz et al. (1990). The cies per 4OO_m2 plots were compared using one-way analysis of var­ extent of sea-spray exposure, the exposure to strong winds, the iance and the Tukey multiple range tt!st. These tests wt!re done disturbance from dune slumping and windfalls, and the distur­ irrespective of the age of the plot. as most of the successional age bance from human innuence appear to be major factors deter- classes had too few replicates to allow for statistical comparison. 190 S. Afr. 1. Bot 1996.62(4)

Table 1 A summary of the plot type, plot location and his­ Table 2 The mean number (and SD) of trees and tree tory of the 28 plots. The central tree around which Ihe plot species in 400-m' plots for the three different pathways. was located, and its diameter at breast height (DBH) are Different letters in the range test indicate that the groups given. The latest date before the area was invaded by trees are significantly different (ANOVA and Tukey multiple range and whether the area was identified as a grassland of old test) field at that stage is indicated. All the plots have been allo­ Range cated to one of eight classes depending on the approximate Pathway II mean SD test F p age, whether the area was an old field or grassland, and whether the area went to a bush clump or A. karroa patch. Trees A represents A. karroa stands, L represents fields Ihat went A. kllrroo pathway I I 103.6 42. 1 a to non-A. karroa stands and G represents grasslands that went to bush clumps. 1 = between 1937 and 1960, 2 = Grassland to bush R 220.2 81.9 b between 1960 and 1970 and 3 = post 1970 Fiekls to bush 9 232.5 19.0 b 15.9 <0.0001

Plot No. CCnlrai tree species DBH Date Class No. trees Species Species

Amda karroo 17.7 1960 A2 48 II A. karroo pathway II 17.9 4.5

2 Acacia karmo 32.5 1960 A2 86 17 Grassland to bush 8 33.8 4.7 d

3 Acacia karmo 48.0 1960 A2 95 16 Fields to bush 9 33.7 4.6 d 39.4 <0.0001

4 Syzygiutn (.'ordatum 35.0 1960 G2 2 15 30 5 GardlZ ia livingstonei 27.3 1960 G2 339 36 Bush-clump plots had a greater number of species than any of 6 SyzygiulI1 (..'orriatllm 63.0 1960 G2 326 41 Ihe A. karroo plots. Only Ihree bush clump plots had fewer tree species than any of the A. karroa plots (Figure I ). Within the A. 7 ApodyteJ dimidiata 30.5 1937 Ll 280 40 karroa plots there was a general trend for the older plots to have 8 Syzygiutn nmialum 45.0 1937 LI 191 36 more species and more individual trees. This trend was not as pronounced in the bush-clump plots. The bush clump plots from 9 Trichilia emetica 25 .1 1960 L2 156 27 the two pathways were intermixed in terms of the numbers of 10 Syzygium cordatum 21.4 1960 L2 210 36 plants and species. II Albizia adiallihifolia 27.5 1960 G2 158 35 The species diversity (as given by the Shannon and Simpson's diversity indices) (Odum 1983) was generally higher for the 12 Apodyresdimidiata 29.6 1937 LJ 183 35 bush-clump plots than for the A. karroo plots (Figure 2). There of 13 S),zygium cordalum 35.5 1937 Ll 233 33 was no clear trend increased diversity with increased age (as indicated by the data labels in Figure 2). 14 S)'z)'gium C/Jrdat!lm 26.0 1970 G3 174 34 The TWINSPAN classification separated the A. karroa plots 15 Syzygium (."(Jfc!alum 32.3 1937 Ll 273 37 from all other plots as the first main division (See Appendix 1 for full species-by-stand matrix). No further logical divisions could 16 S)'z)'g iwn cordal/lm 30. 1 1937 G 1 218 33 be found in the A. karroa plots. 17 Syzygiwn cordatum 31.7 1960 G2 96 25 Further divisions of the bush-clump plots seemed to be marc dependent on the age and position of the stands than on whether 18 Syz)'giwn cordatum 19.1 1960 G2 236 36 the area was identified as an old field or grassland. There was a 19 ApodyteJ dimidiala 30.5 1937 L 1 341 26 clear division in species that was common to the bUSh-clump

20 Apodytesdimidiara 41.2 1937 LI 226 33

21 At·acia karmo 29.9 1937 AI 164 22 45 ,------, G2 22 Acacia karr"(l(1 52.2 1937 Al 91 18 40 r L1 23 Acado karn)() 44.5 1937 AI 162 24 G2 ~ 35 G2d3~1 L2 G2 ~ GU.1 24 Acada karmll 46.2 1937 Al 117 21 '0 L1 ~ 30 G2 a. Acacia karrt}(} 154 19 ~ 25 39.1 1937 Al L2 L1 '0 25 G2 A1 26 Awciakarmo 29.5 1970 A3 88 20 "~ A1 .c 20 M3 A1 27 Acado karmll 25 .8 1970 A3 95 20 E A1 :f 15 4i2 28 Acacia karmo 42.2 1970 A3 40 9 A2 10 A3 Results The mean number of trees and species within the 400· m2 plot 20 60 100 140 180 220 260 300 340 380 around the central tree was considerably lower in the A. karroa Number of trees pathway plots than in either of the bush-clump pathway plots. Figure 1 A comparison of the number of trees and species in the Differences in mean values were highly significant between the plots of the different pathways. The codes are as follows: A = A. kar­ A. karroa and bush-clump pathways, but not significant between roo pi OlS ; L :: fields to bush clumps; G = grassland to bush clumps; the two bush-clump pathways (Table 2). I = 1937 to 1960; 2 = 1960 to 1970; 3 = post 1970. s. Mr. J. Bot. 1996,62(4) 191

3.5 ~------, the A. karroo plots where Diospyros natalellSis, Sideroxylon i,lerme, Erythrina lysistemon, Deil1bolfia oblollgifolia, Ziziphus )( I11llcronata , ClausellCl anisata and Pavetw revo!uta. ., G3 '0 G~P-1i.1,1 L1 No species occurred exclusively in any of the plots of either of , ~ 3 La..1 the bush-clump pathways. Apodytes dimidiata, Brachy!aena dis­ ~ G2 color and Tricalysia sOllderiana were found in all the plots of .,'"~ G2iiJd A1 bush-clump pathways. Other common species of the bush-clump ,2: 2.5 pathways included Allophyius natalensis, Bridelia cathartica, '0 A1A1A2 Canthium illerme, Carissa bispinosa, Dovyalis longispina, Ery­ 'c:'" throxylum emargill(lwm. Euclea lIataiellsis, May tenus sellegalen­ 0 c: sis, Psychotria capensis, Psydrax obovntum, Rhus nataiensis, c: 2 f- .c:'" A2 Strychnos spinosa, Syzygium cordatum and Xylotheca kraussii. en The detrended correspondence analysis clearly separated the A3 , A. karroo plots from the rest of the plots on its first axis (Figure 1.5 j). The second axis separated the younger bush-clump plots from 0 5 10 15 20 th< older ones. Th< grassland 10 bush, and old fields 10 bush plols Simpson's diversity index we re not separated by the ordination. Figure 2 Species diversity in the different pathways as given by Figure 4 shows the mean density of trees as the distance from the Shannon and Simpson's diversity indices. The codes are as fol~ the central tree increases. Each band has an equal area of 50 rn2. lows: A::; A. karma plots: L::; fi elds to bush clumps; G ::; grassland to The two bush-clump pathways have very similar trends. In both, bush clumps; I = 1937 10 1960; 2 = 1960 to 1970; 3 = pOSl. 1970. the highest density of trees was near the central tree and decreased rapidly as one moves away from the centre. In both instances a highly significant multiplicative regression (the palhways as opposed to Ihe A. karma pathway. Forty·eight spe· equivalent of a linear regression on log-transformed data) was fit­ cies were unique to the bush-clump pathways, whereas only 15 ted. The A. karroo plots exhibited no clear trend of changes in species were unique to the A. karroo pathway. Of the 48 species density as one moves away from the central tree. The data did not unique to the bush-clump pathways, 46 of these species where fit either a linear or mUltipli cative regression model. In all found in the old lands thaI had reverted to bush clumps. Thirty in stances, the mean density of trees was greater in the two bush­ species were common to both bush-clump and A. karma path­ clump pathways than in the A. karroo pathway. ways. There were 61 species that occurred in both lands and The density of species as one moves away from a central tree grasslands Ihal had developed into bush clumps. A further 15 showed very si milar trends to the density of trees (Figure 5). In species occurred in old-field bush clumps that did not occur in theA. karroo plots there was a weak positive correlation between grassland bush clumps. Only three species occurred in grassland increase in species density (p:::; 0.07) and distance from the cen­ bush clumps that did not occur in bush clumps on old lands. tral A. karroo tree. At all distances the mean species densi ty of No species were common to all plots, but Drypetes nata!ensis, the A. karroa plots was lower than either of the bush-clump path­ Grewia oceidelltaJis, Oehlla lIata/ilia and Scuria myrtilla were ways. common in all the successional pathways described. There was a high proportion of {he trees in the old A. karroo Species COmmon to all A. karmo plots were Celtis a/ricGlla, sites that were in the greater-than-4-m class (Table 3). A decline Diospyros inhacaellsis and A. karmo. Other common species in in the density of large A. karroo trees from a mean of 11 in the posl·1970 plols and 12.3 in Ihe 1960 10 1970 pi DIS, 10 only 4,8 in the pre-1960 plots was observed. Similar trends to the number of 250 species in height classes were observed for (he number of trees in

200 L2

150 80 L2 N 100 f- G2 '" ., 50 G3 '" 60 <{''"" G2 G2 ; A2 "","" g 0 0 'll\' 7." - '0', ~ 40 ·50 L1L1 .,'"c: '0 ·100 f- .." 20 , , , I ., ·150 " , :;; -200 ·100 o 100 200 300 400 500 DCA axis 1 0 0 2 4 6 8 10 Figure 3 First and second axes of a detrended correspondence Distance from centre (in 50 m~ units) analysis of 28 A. karroo and bush-clump plots mea sured at Eastern Shores. The A. karroo and bush-clump pathways are clearly sepa­ rated by the first axis. The second axis separates the bush-clump plots • Acacia pathway .Grassland to bush clumps and is seemingly related to the time since dis turbance. The codes are * Fields to bush clumps as Follows: A :::; A. karroo plots; L:::; fields to bush clumps: G :::; grass­ land to bush clumps; I = 1937 to 1960; 2 = 1960 to 1970; 3 = post· Figure 4 The mean density of tree individuals per 50 m2 of area 1970. radiating out from the central plant. 192 S. Afr. 1. Bot. 1996.62(4)

25 Table 3 The mean number of trees in different DBH VI classes (of trees with DBH > 50 mm) found in different- .<:;" 20 aged A. karmo and bush-clump pathways Co" VI DBH classes (mm) ~ 15 0 Pathway type and age < 100 <200 < 300 > 301 Total .~ VI c: 10 A. karma 1937- 1960 36 19 11 7 72 "C" A. komI(} 196 1- 1970 21 8 5 8 42 c: .. 5 A. karroo posl- I ~no 14 3 6 5 28 :;;" Grassland to bush 1937- 1960 17 10 5 2 34 0 0 2 4 6 8 10 Grassland to bush 1961-1970 24 t1 \0 3 48 Distance from centre (in 50m' units) Grassland to bush post- 1970 10 6 2 2 20

• Acacia pathway .Grassland to bush clumps Field to bush 1937-IY60 17 5 7 4 34 * Fields to bush clumps Field to bush 1961-1970 14 8 4 3 28 Figure 5 The mean number of tree species per 50 m2 of area radi­ ating out from the central plant. trees senesee, other species dominate the area (von Mahitz et al. 1990). A. different DBH classes. Only 40 karroo seedlings were Bush clumps form in a different manner to the A. karroa obserwd, of which only one was in a bush dump. patches. Bush c lumps start around individual trees that have established within the grassland or old fields, or that were left Discussion standing when the fi eld was cleared. Syz.ygium cardatum is often The two bush-clump pathway, differ floristically and structurally the first species, but other species. such as Garcinia livings/ollei from the A. karroa pathway. There is no clear distinction and Phoenix reclinata may also be involved. These species have between species composition and structure between bush clumps fleshy fruits that are attractive to birds and animals. they are fire formed on old fie lds and those on grasslands. Although the bush tolerant, have rapid growth rates and mature early. Once the ini­ clumps on old fields had more unique species than the bush tial tree has established, it provides shade, increased moisture clumps in grasslands, they all occurred in low density and in only and protection from fire . This enables other tree species to estab­ a few of the plots. li sh. The initial tree attracts birds which act as seed-dispersal agents by bringing in seeds from other species (Tinley 1985). Structurally. bush clumps and A. karroa patches are very dif­ More than 80% of the speci es in this study are pott:!ntially bird ferent. This arises presumably from the way in which the two dispersed. The decline in density and size of trees as one moves pathways are initiated. A. karroo patches are always character­ away from [he central plant supports this hypothesis. As the bush ized by a near even-aged stand of A. karroo trees (Breen 1971). clumps develop, the grass density below decreases and this pre­ Observations of newly established A. karmo stands, including vents fires from entering the bush clump, which allows fire-sen­ those that have establi shed on old rehabilitated dune mining sitive species (Q establish. These bush clumps in time become areas (Camp & Weisser 1991), suggest that A. karroo trees estab­ bigger and eventually merge, forming fore st patches. The rate at lish soon after a disturbance event. A. karroa has characteri stics which this happens presumably depends on temporal fire fre­ typicnl of a weed species, in that seeds can persist in the soil seed quency and intensity. bank for many years, due to the hard seed coat. Seed germination A. karma seeds seemingly do not attract frugivorous birds and of mnny Acacia species is stimulated by a disturbance and most birds encountered in these areas were insectivores. The increased light levels (Milton 1982; Smith & Shackelton 1988) A. karroo patches, therefore, do not promote dispersal in the same and all indications are that thi s is true for A. karroo. Once estab­ way as the bush clumps, which could account for the fewer spe­ lished, A. karroa plants grow as an even-aged stand. A process of cies present in this pathway. self-thinning occurs, and after about 25 years, trees in the stand start to senesce and die. The A. karroa individuals are character­ The occurrence of two floristically distinct successional path­ ized by large multi-ste mmed trees with an umbrella-shaped can­ ways in a relatively homogeneous area such as the coastal dune opy. The area below the canopy is very open with a paucity of cordon could be ascribed to differences in micro-sites, the pres­ tree seedlings. The seed lings and saplings that do establi sh are ence of a fire-maintained herbaceous community before the not clumped around the A. karroo trees. In addition, species den­ onset of woody encroachment or the relative recem introduction sity increases away from the central tree. Although there is a of A. kllrmo.lO the area. large number of A. karroa seeds present in the soil beneath the canopy, it is not common to find A. karroa seedlings or saplings Implications for restoration of dune forest communities under adu lt trees (Smith & Walker 1983; Smith & Goodman The existence of more than one successional pathway from dis­ 1986, 1987). As the stand starts to senesce, large gaps are formed turbance to forest, and the fact that the different pathways are in the canopy. These gaps are seldom recolonized by A. kanvo both tloristically and structurally very different have distinct seedlings, but rather by seedlings and saplings of other species. implications for any rehabilitation programme on di sturbed dune particularly Celtis ajricana, Tec1ea germrdii, Diaspyros natalen­ vegetation in the Zulu land area. If a rehabilitation programme sis, Diuspyros inhacnellSis and Clausena allisara (see also Smith uses onl y one or the other of the successional processes then & Goodman 1987). Even at this stage, the ge neral appearance of ·there will be a decrease in species and habitat diversity. It is pas· the sub~c anopy is very open. The lack of A. karma trees within sible that many of the species that only occur in the alternative the mature dune forest suggests that, once the mature A. karroo successional pathway could become locally extinct. Although S. Mr. 1. Bot. 1996. 62(4) 193

only two distinct pathways were identified in this study, it is pos­ CUNNINGHAM. A.B. 1985. The resource value of indigenous plants to sihle that other pathways also exist and this should be further rural people in a low agricu ltural potential area. Ph.D. thesis. Uni v. of investi gated. Rehabilitation programmes have had great success Cape Town. in initiating and using the A. kanno pathway (Camp & Weisser FEELY, J. M. 1980. Did Iron Age man have a rolc in the history of Zulu­ 1991 : Mentis & Ellery 1994). Further research is however land's wilderness landscapes? S. Afr. J. Sci. 76: 150--- 152. needed to determine how to initiate non-A. karroo pathways. HALL. M. 1980. Enkwazini. an Iron Age site on the Zululand coast. The bush-clump pathways have many positive features over Anll. Natal Mus. 24( 1): 97-1 11 . the A. karroo pathway. HALL. M. 198 1. Settlement patterns in tile Iron Age of Zu lul and. Cam­ These include: bridge Monographs in African Archaeology 5. BAR International Series 119. Oxford. 1 ) A soil-binding herbaceous layer is an carly stage o f the path­ way. HILL, M.O. 1979. TWINSPAN - A FORTRAN program fo r arranging multivariate data in an ordered two-way table by classification of indi­ 2) Because most of tht.! tree species involved have fru it that viduals and attributes. Cornell University. Ithaca. N.Y. attracts birds. rhis wi ll lead to seed dispersal from the existing HOB DAY, D.K. 1979. Geological evolution and geomorphology of the mature fort!sl into the areas to be reclaimed. Zu luland coastal plain. In : Lake Sibaya. ed. B.R. Allanson, pp. 1-2 1. 3) The pathway rapid ly Icads to a greater d iversity of species W. Junk. The Hague. thil n thl.! A. karma pathway. MACDEVETTE. D.R. 1989a. The vegctation and conservation of Ihe 4) Then! is greater habitat diversity (bush clumps in a grassland Zululand coastal dunes. In: Nalal indIgenous forests: A preli minary matrix) and this diversity wi ll favour both bird and animal collection of reports on indigenous forests in Nata l. ed. 1.0. Gordon. species. pp. 2 1-32. Nalal Parks Board. Pielermaritzburg. 5) The pathway promotes species not present in the A. karroa MACDEVETTE, D.R. 1989b. A study of coastal dune gradients in Sod­ pathway. wa lla State Forest. Natal, Soulh Africa. In : Natal indigenous forests: A preliminary collection of reports on in digenous forests in Natal, ed. Conclusions I.G. Gordon. pp. 33-66. Natal Parks Board. Pietermaritzburg. MACDEVElTE. D.R. & OORDON, 1.0. 1989. A report on a quantita­ At least two tloristicall y and structu rall y disti nct successional tive analysis of the Mlalazi fores t. In: Natal indigenous forests: A pre­ pathways from old fields and grasslands to forest can be distin­ liminary collecti on of repoTts on indigenous forests in Natal. ed. loG. guished on the coastal dunes of northem KwaZulu-Natal. One Gordon, pp. 87-100. Natal Parks Board . Pietermaritzburg. pathway is in iti ated by A. karroa, which grows as an even-aged MAUD. R.R. 1990. Geology. geomorphology and soils of the Natal stand with a low level of establishment of other species within north coast. In: Dune forest dynamics in relation to land-use practlces, the stand. The second pathway is associated wi th bush dumps. eds. D.A. Everard & G. P. von Ma ltilz, pp. 40-44. Occ::ls ional report These were identified as occurring on bOlh old fields and in 49. FRD. Pretoria. grasslands, but flori stically and structurally, no distinction could MENTIS . M.T. & ELLERY. W.N . 1994. Post-mining rehabilitation of be found based on the origin. This might be because grasslands dunes on the north-east coast of South Africa. S. Afr. J. Sci. 90: 69-74. arc always an intermediate stage in the succession. The bush MILTON, S.J. 1982. Effects of shading on nursery grown Acacia seed­ clumps arc centred around an initial nucleus trec. Species and lings. S. Afr. J. BOl. 48: 245- 272. plant density decrease as one moves away from the central tree, MOLL. E.J . 1972. A preliminary account of the dune communities at and both species and planl density are far higher than in the A. Pennington Park. Mtun zini. Natal. 8utha/ia 10: 6 15- 626. karmo plots. ODUM. E.P. 1983. Basic ecology. Holt-Saunders, Philadelphia. At thi s stage, no clear understanding exists as to what initiates PAMMENTER, N.W., BERJAK. M. & MACDONALD. lAW. 1985. the different pathways. The initial presenc.e of A. karroa seeds in Regeneration of a Natal coastal dune forest after fire. S. Afr. 1. Bot. 51: the di sturbed area seems to be a prerequisite for the A. karroo 453-459. pathway. The bUSh-clump pathway may only occur where there SMITH. T.M . & WALKER. B. H. 1983. The role of competition in the is no A. karma seed. or it is possible that a factor such as fire or a spacing of savanna trees. Pmc;. Grassl. SOL". Srh. Afr. 18: 159-164. dense grass cover could inhibit the A. karroo seed from gennina­ SM ITH. T.M. & GOODMAN. P.S. 1986. The effect of competition on ting. A feature o f the bUSh-clump pathway is th <.\t animals play an the structure and dynamics of Acacia savannas in Southern Africa. J. important role in seed dispersal. Because A. karmo does not have Ecoi. 74: 1031-1044 neshy seeds. they do not promote seed dispersal in the same way. SM ITH , T.M . & GOODMAN. P.S. 1987. Successional dynamics of an Further research is required to understand fully what in itiates Acacia nj[otrca-Ew.:/ea divinorum savanna in southern Africa. 1. £1:01. the different pathways and what the long-term consequences of 75: 603-6 10. these pathways wi ll be to the mature dune forest that results from SM ITH , T.M. & SHACKELTON. S.E. 1988. The effects of shad ing on the di lTcrcnt pathways. the establishment and growth of Acacia torrj/i,l" seedlings. S. Afr. 1. Bot. 54: 375-379. References TER BR AAC K, C.J .F. 1988. CANOCO - a FORTRAN program for ACOCKS. J.P.H. 1953. Veld lypes of South Africa. Mem. bot. Surv. S. canonical community ordination by [partialJ (detrendedJ [canonical] AI' 28: 1-128. correspondc:nce analysis. principal components analysis and redun­ BEAUMONT. B.. DE VILLI ERS, H. & VOGEL J. 1978. Modern man dancy analys is (version 2. 1). Report LWA -8802. Agricultural Mathe­ in suh -S aharan Africa prior to 49000 years b.p.: A rev iew and evalua­ matics Group, Wageningen. lion WIth particular reference to Border Cave. S. Afr. 1. Sd. 74: 409- TINLEY, K.L. 1985. Coastal dunes of South Afnca. S. Afr. nat. St...·i Pmg. 4 19. Rep. No. 109. Ecosystem Programmes. FRD, CS IR, Pretoria. BR EEN. C. M. 1971. An account of the pl ant ecology of the dune fores t VAN DAALEN, J.C GELDENHUYS. C.J .. FROST, P.G. H. & MOLL. at Lake Sibaya. Trans. R. S/u..:. S. Afr. 39: 223-234. E.J . 1986. A rapid su rvey of forest succession at Mlalazi Nature BREEN. C.M. 1979. The dune forest: its structure and maintenance. In : Reserve. Occasional Repor! No II. Ecosystem Programmes. FRD, Lake Sibaya, ed. B.R. Allanson. pp. 21-33. W. Ju nk, The Hague. CSIR, Pretoria. CAMP. P. & WEISSER. P.J . 199 1. Dune rehabilitation. nora and plant VENTER. H.J.T. 1972. Die plantcgroei va n Ric hards-Baai. M.Sc. thesis. succession after mi ni ng at Richards Bay. South Africa. In : Dune forest Univ. of Preloria. dynamics in relation to land-usc practises. cds. D.A. Everard & G.P. VENTER, H.J .T. 1976. An ecological stl!dy of the dune forest at Maphe­ von Malt itz. pp. 106-123. Occasional repon 49, FRD. Pretoria. .l ane. Cape St Lucia Zululand. S. Afr. J. BOl. 42(2): 211-230. 194 S. Afr. 1. Bot. 1996. 62(4)

VON BRUGGEN. A.C. & APPLETON. c.c. 1977. Studies on the ecol· ecology of Maputaland, cds. M.N. Bruton & K. H. Cooper. Wi ldl ife ogy and systematics of the terrestrial molluscs of the Lake Sibaya area Society of South Africa, Durban. of Zululand. Soulh Africu. Zoo/. Verh. 154; \-44. WEISSER, P.J . 1991. Vegetation dynamics on the Zulul and dun es: The VON MALTITZ. G.P.. VAN DAALEN. J.C .. MACDEVETTE. D.R. & big piclUre , facts, thoughts and some ideas for fU lUre research. In: JACO BS, C.J . 1990. The woody dune vegetation at Eastern Shores. Dune forest dy namics in relation to land-use practises, eds. D.A. Ever­ Internal report, South African Forestry Research Institute. George. ard & G.P. von Mallitz., pp. 72-80. Occasional report 49, FRO, Pre­ toria. WA RD , c.J. 1980. The plant ecology of lhc Isi pingo Beach area. Natal. WEISSER, P.J . & MARQUES. F. 1979. Gross vege tation changes in the South Africa. Mem. bot. SUr\!. S. AIr. No. 45. dun e area between Richard Bay and the Mfolozi River. 1937-1974. WEISSER, PJ.1 978. Changes in areas of grasslands on the dunes BlI1halia 12: 7 11 - 721. between RichurcJs Bay and the Mfolozi River, 1937 to 1974. P m!,", WEISSER. PJ. & MULLER, R. 1983. Dune vegetation dynamics from Gr.ul. Soc. 5th. Afr. 13: 95- 97. 1937 to 1976 ill the Ml alazi- Richards Bay area of Natal, South WEISSER. P.J. J 980. The dune fore st of Maputaland. In: Studies on the Africa. Bothalia 14: 661 -627.

Appendix 1 Species·by·stand matrix of the 28 sampled A. karroa and bush·clump plots. See Tabl e 1 for explanation of plot codes

Plot code LGGLGG LLLGLLL GL GG AAAA AAAAAAA 2 2 2 2 2 3 1111111 2 1 2 2 2 1 1 1 1123233 Plot number 1 1 1 1 1 1 1 1 2 1 222 2 2 222 o 7 8 9 4 235 6 8 0 9 6 7 4 5 3 5 3 4 1227186 Species Antidesma venosum 1 · 2 1 1 1 · 1 · · · · 1 · · 1 · · · · · · · · · · Chrysanthemoldes monififera 1 1 2 · 1 · · · · · · · · · · · · · · - · · · · · · · Sapium integerrlmum 1 · 1 1 1 1 · · · · · 1 · · · · · · · · · · · · · · · · A/bizia lJdisnfhifolilJ · 1 1 1 1 1 1 · 1 1 · · · · 1 · · · · · · · · · · · · · Rhus nebu/oss 1 1 2 4 1 · 2 · 1 1 1 · · · · · · · · · · · · Erythroxylum emargnatum 1 · · · 1 1 2 2 1 1 1 2 2 1 · · · 1 · 1 · · · · · · Mimusops caffrs 3 1 1 2 2 2 2 4 3 4 2 3 2 2 2 1 2 1 1 · 1 · · · · · · Catunaregum spinosa · · · · 1 1 · 1 1 · · 1 · · · · · · · · · · · · · · Strychnos ge"ardii 1 · · 1 · · 1 1 1 1 · · · · · · · · · · · · · Tarenna littoralis · · 1 1 · 1 · 2 1 · · · · · · · · · · · · · · Eugenia capensis 1 1 1 1 1 1 2 2 3 1 1 2 4 1 1 1 1 · · - · - · · · · Kraussia noribunda 3 2 3 1 1 1 1 2 4 3 1 1 2 1 1 1 1 · · · · · · · · · · · Maytenus senegafensis 1 2 1 1 1 · 1 2 1 1 1 2 2 1 1 1 1 · · · · · · · · · · Maytenus nemorosa 1 1 1 · · 1 1 1 1 1 1 1 1 1 · · · · · · · · · · · Sc/erocarya biffea 1 · · · 1 1 · 1 · · · · 7 · · · · - · · · · - - · · · Strychnos spinosa 2 2 1 · 1 1 · 1 1 1 1 1 · 1 1 1 1 · · · · · · · · · · · Syzygum cordalum 1 1 1 · 1 1 1 1 1 1 1 · · 1 · 1 · · · · · · · · · · · · Xylotheca krauss/ana 1 1 2 1 2 2 1 1 1 1 1 1 · 2 2 1 1 · · · · · · · · · · Canthium inenne 3 1 1 2 1 2 1 · 1 1 1 · 1 1 1 2 · · · · · · 1 · · · · discolor 1 2 1 2 1 2 2 2 4 3 1 3 2 2 2 4 4 · · · · · · · · · · Garcinis livingstonei 1 1 1 · · 2 1 · · 1 · 2 · 2 1 1 2 · · · · · · · · · Apodytes dimidiala 1 1 1 1 1 1 2 1 3 1 2 1 2 7 3 3 5 -- - · · · · 1 · Cassins sudeifonnis · · · · · · · 1 1 2 1 1 1 3 · 1 · · - · · · · · · Plsclroniella armata · · · · · 1 1 1 · · 1 · · 1 · 1 1 · · · · · · · · · · Strychnos madagascariensis · · 1 · 1 1 1 · · · · 1 1 2 · · · · · · · · · · · Vepris undulata · · · 1 · 1 1 · 1 1 · · 1 1 · 1 2 · · · · · · · · · · · Carissa bispinosa 1 1 · 1 1 1 1 1 1 2 1 2 1 1 1 1 · · 1 1 · · · · · · - Ozoroa obovata · 1 1 · · · 1 · 1 · 1 · 1 1 · · · · · · 1 · · · · Rhus nata/ensis 1 · 1 1 1 · 1 1 1 1 1 1 1 1 1 1 1 · · · · · - · · 1 1 Tricslysls sondarians 2 1 3 2 3 2 2 3 3 2 4 3 4 4 4 6 6 · 1 1 1 · · - - · · · Psychotria capensis 4 · 7 2 1 1 1 · 1 1 2 1 1 2 4 1 2 2 · · · - · 1 · · 1 · Psydrax obovara 1 · 1 1 · · 1 1 1 1 1 1 1 1 1 1 1 · · · · · · · · · · Euclea natalensis 1 1 1 1 2 2 2 3 1 1 2 7 2 3 2 3 1 1 · 1 · · · · · · · Eugenia natah'ta 1 · · · · 1 1 · 1 1 · 1 1 · · · · · · 1 · · · · · Dovyafis longispina 1 1 · 1 1 1 1 1 1 1 · 1 2 1 1 1 · · · · · 1 1 · · · Trichilia emetica 1 · 1 1 1 · · · · · · · · 1 1 · · · 1 · · · · · · · Bride/ia cathartica 1 · 1 1 1 1 1 1 1 · 1 · 1 · · 1 1 · · · · - · · 1 AHophy/us natafensis 1 · · 1 1 1 1 · 1 · · 1 1 1 1 · 1 · · · · 1 1 1 · · · Peddiea africana · · 1 · · · 1 2 · 2 1 2 3 3 2 1 1 1 1 · 1 1 · 1 1 1 · · Scuo'a myrtina 1 1 1 · · · 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 · 1 · · · Ochna natalifia 1 · 1 1 2 1 1 2 1 1 1 · 1 1 1 1 2 · 1 1 1 1 2 · 1 · 1 1 Ficus burff·davyi · 1 · 1 1 · 1 · · 1 1 · · · · · · · 1 · 1 · · · · · 1 Drypetes na(alensis · · · · 1 1 1 1 · 1 1 1 · · 1 · · 1 2 1 1 · · 1 1 · · 1 Monanthotaxis caffra · · · · · · 1 · · · 1 · · · 1 · · · · · · 1 1 · · · · ClerodendrtJm glabrom · 1 1 · 1 · · · · 1 · · · 1 · · · · 1 1 1 1 1 · · Grewia occidenta/is · 1 1 · · 1 1 1 1 1 · · · 1 · · 1 1 1 · 1 1 1 1 1 · 1 Cettis africana · · · · · · · · · · · · 2 2 2 2 1 1 1 1 1 · 1 Diospyros inhacalJnsis · · · · · · · - 1 · 1 1 1 6 3 3 2 1 2 1 1 1 1 Sideroxylon inerme · · 1 · · · · · 1 1 · 1 1 1 1 1 1 1 1 · 1 Teclea geffardii · · · · 1 1 · 1 1 · · I 1 1 · · 2 2 2 1 2 2 2 2 1 1 2 Erythrococca berberideB · · · · · - · · · · · · · · · · · 1 2 2 3 1 · 1 · · 1 S. Mr. J. But. 1996.62(4) 195

Appendix 1 Continued

Plot code LGGLGG L L L G L L L G L GG AAAA AAAAAAA 2 2 2 2 2 3 1 1 1 1 1 1 1 2 1 2 2 2 1 1 1 1123233

Plot number 1 1 1 1 1 1 1 1 1 2 1 222 2 2 2 2 2 o 7 891 4 2 356809 6 7 4 5 3 5 3 4 12 2 7186 Species

Oeinbollia obfongifolia - 1 2 1 1 1 - Ziziphus mucronata 1 1 1 1 - 2 - Clausena anisaia 2 - 1 4 1 1 1 Pavetta reva/uta 1 1 2 1 1 - Zantho)(yJum capense 1 1 1 1 - 1 Acacia kSffOO 4 1 2 2 3 2 3

o a a a a a 0000000 a a a a 1 1 1 1 1 1 1 1 1 a a a a a a a a 0 0 0 0 0 1 1 1 1 o 00 0 1111111 o 0 0 0 0 0 1 1 1 1 1 1 1 o 0 1 1 o 0 1 1 0000001