A Reference Framework for the Restoration of Riparian Vegetation in the Western Cape, South Africa, Degraded by Invasive Australian Acacias

A reference framework for the restoration of riparian vegetation in the Western Cape, South Africa, degraded by invasive Australian Acacias

N Prins, PM Holmes and DM Richardson*

Institute for Plant Conservation, Botany Department, University of Cape Town, Private Bag, Rondebosch 7700, South Africa * Corresponding author, e-mail: [email protected]

Received 6 May 2004, accepted in revised form 28 May 2004

Riparian vegetation, which normally differs structurally tified four prospective plant communities: 1) a Nivenia and compositionally from surrounding vegetation, has corymbosa-Brachylaena neriifolia community; 2) a been degraded in many parts of the fynbos biome by Leucadendron salicifolium – Berzelia lanuginosa com- several species of invasive Australian Acacia. munity; 3) a Cliffortia ruscifolia – Metrosideros angusti- Systematic clearing of dense stands of these alien trees folia community; and 4) a Kiggelaria africana – was initiated in 1995, and information is urgently need- Brabejum stellatifolium community. These formed a ed to guide the restoration of riparian habitats. A prob- continuum with only the Leucadendron and Kiggelaria lem is that degradation of these communities is so communities separating in ordination space. Soil pH dif- advanced and widespread that in many cases managers fered between the latter two communities, reflecting dif- do not know what species to use in restoration, or what ferent geology. It was found that many riparian special- kinds of target communities to aim for. This study’s ist species are relatively widespread. For the study area, aims were to provide baseline information on riparian it is concluded that where information on the historical plant community structure and composition from non- composition of riparian communities is lacking, target transformed habitats. Species and environmental data communities for restoration can be defined from pris- were recorded from 76 sites located along the headwa- tine communities with similar geology, and secondly, ter systems of six rivers in the southwestern part of the altitude. In all cases the target community will comprise Western Cape province. Analysis of the data applying a large proportion of widespread, predominantly multivariate classification (TWINSPAN) and ordination resprouting, riparian species. (Detrended Correspondence Analysis) techniques iden-

Introduction

There exist no formal published descriptions or classification Bokkeveld Group shales. The downstream coastal plain of riparian vegetation in the Western Cape. Much of the cur- zone contrasts with the upstream counterparts in having rent information is restricted to specific rivers in unpublished lower-gradients, finer sediments and less confined chan- reports (e.g. Boucher 1998) and theses (e.g. Sieben 2003). nels. The dominant geomorphological process here is depo- Some studies have included riparian habitats within a larger sition, in contrast to the mountain stream zone where it is terrestrial matrix (e.g. Boucher 1978a, Macdonald 1988, erosion (Davies and Day 1998). Taylor 1996), but riparian zones have mostly received less Riparian vegetation in the fynbos biome is normally dis- attention owing to the narrow band they form within the ter- tinctive from the surrounding fynbos vegetation, even though restrial matrix. it occurs under similar macroclimatic conditions (Boucher The geomorphology of most Western Cape rivers is char- 1978a). The predominant vegetation type of riparian zones acterised by the Cape Fold Belt mountains that dominate the in the Western Cape has been variously named closed area, rising to an elevation of about 2 000m. These moun- scrub fynbos (Campbell 1986, Cowling and Holmes 1992, tains comprise rocks of the Table Mountain Group series Cowling et al. 1997), hygrophilous mountain fynbos (Taylor (mainly sandstones) that yield predominantly nutrient-poor 1978) and broad sclerophyllous closed scrub (Kruger 1978). substrata. Owing to the short distances between mountains This vegetation is described as being similar to forest and and coast, headwater stream systems dominate the land- thicket in its relatively high cover of mesophyllous non-pro- scape and in some rivers comprise the predominant system. teoid woody plants, but dissimilar in its high cover of From the steep mountainous terrain, the rivers flow through restioids and presence of Ericaceae (Cowling and Holmes a foothill zone, often on Cape Granite Suite soils, onto the 1992). The above describes the predominant structural type, coastal plains, the latter comprising the Malmesbury or yet other types, from tall herbland to forest, may occur in the 2 Prins, Holmes and Richardson riparian zone (Kruger 1978). Afromontane forest may devel- non-transformed habitats. The intention was to use the op in areas of steep topography and boulder screes that results to define benchmarks for restoring degraded riparian afford protection from fires. sites in the Western Cape. Riparian vegetation in downstream systems of the coastal forelands has in most cases been transformed by agriculture Materials and Methods and no Western Cape river remains undisturbed between the foothills and the sea (Brown 1998). It is not certain what Sampling sites the historical vegetation assemblages would have been in these downstream systems. It is likely that the interplay Riparian vegetation was sampled along six rivers and their between substratum type (clay versus sandy soils), climate tributaries in the Western Cape (Table 1). In all cases sam- and fire frequency would have determined whether fynbos, pling was confined to the headwater systems of the moun- renosterveld, Acacia karroo thicket or forest vegetation tains or foothills. In the case of the Palmiet and Rooiels types would have predominated in downstream riparian Rivers, where mountains abut the sea and there is no zones. coastal plain, some samples were located at low altitude. Invasion by alien trees, particularly Australian Acacia (e.g. Most of the sampling sites were located in pristine riparian Acacia mearnsii, A. longifolia and A. saligna) and vegetation. However, a few sites did contain low densities of Eucalyptus species (especially E. camaldulensis; Forsyth et alien species, which were either still present or had recently al. 2004), has had a large negative impact on riparian vege- been cleared. Sample sites in the Palmiet River differed tation throughout the Western Cape (Richardson et al. from the others in that areas upstream of sampling were not 1992). These invasive species displace the indigenous veg- pristine, but were impacted by deciduous fruit orchards and etation and may reduce water flows and alter sediment forestry (predominantly pine) and dense alien invasions. dynamics and channel form, thus also directly affecting the This is owing to the Palmiet River and its tributaries travers- aquatic systems. Dense alien stands may obstruct the flow ing the fertile Elgin Basin before entering the southern por- of water during flooding, leading to the erosion of the water- tion of the Kogelberg Nature Reserve. courses and the conversion of well-defined rivers into diffuse systems of shallow channels, that in low-flow periods may Sampling procedures be further colonised by aliens (Hoffmann and Moran 1988). In response to the rapid spread of these thicket-forming The mature vegetation in 76 plots was sampled between invasive trees, the ‘Working for Water’ programme was initi- April and September 1999 (Table 1). To allow for comparison ated in 1995 to control invasive aliens in order to safeguard with other vegetation studies in the fynbos biome, a rectan- water production and water quality (van Wilgen et al. 1998). gular (10 x 5m) plot was chosen (Boucher 1978b, Taylor Control is effected by means of appropriate mechanical, 1984, McDonald 1993). The long side of the plot was placed chemical and biological methods. parallel to the riverbank, thus capturing a 5-m band of ripar- Riparian vegetation that has been invaded for several ian vegetation. In most cases this was adequate to capture decades may not rapidly recover following alien clearing both wet and dry bank lateral zones, incorporating the lower operations, owing to the elimination of indigenous vegetation shrub and upper shrub zones (Boucher 1998). The intention and the depletion of its propagule bank (Galatowitsch and was to sample the generalist riparian community, as Richardson, submitted). In areas where a local propagule opposed to the neighbouring terrestrial community, and not source is not available, or is depauperate, it will be neces- to disaggregate the various riparian vegetation lateral zones, sary to restore riparian vegetation following alien clearance. some of which were very narrow or absent. In order to re-introduce appropriate species, it is necessary In each plot, percentage cover values were assigned to to predict the likely pre-invasion vegetation assemblage (i.e. each species from visual estimates of their projected canopy the restoration target community). cover and species densities determined from counts of the The objective of this study was to provide baseline infor- total numbers of individuals present. mation on riparian plant community structure from a range of Average soil depth was estimated by hammering a steel

Table 1: Location of sampling plots

River Map Reference Position No. plots sampled Locality range of plots Altitudinal range plots (m) Wit 3319CA Bainskloof Pass 20 33°34’25”S 19°08’60”E– 260–570 Bainskloof 33°37’30”S 19°06’50”E Steenboks 3319CA Bainskloof Pass 7 33°32’20”S 19°07’10”E– 340–440 Bainskloof 33°32’50”S 19°08’20”E Molenaars 3319CA Du Toitskloof Pass 21 33°42’25”S 19°14’00”E– 260–480 Bainskloof 33°43’80”S 19°06’60”E Rooiels 3418BD Kogelberg Nature Reserve 5 34°18’15”S 18°49’60”E– 40–50 Hangklip 34°18’00”S 18°50’50”E Palmiet 3418BD Kogelberg Nature Reserve 9 34°15’50”S 18°55’35”E– 40–100 Hangklip 34°16’70”S 18°59’60”E Eerste 3318DD Jonkershoek Nature Reserve 14 33°58’45”S 18°56’20”E– 240–380 Stellenbosch 33°59’70”S 18°58’50”E South African Journal of Botany 2004, 70: x–x 3 rod into the soil until reaching an impenetrable layer, at five Community (acronym: Leucadendron Community) char- random points within each plot. A soil sample was taken acterised by the high frequency and dominance of from each plot by removing a few trowelfuls in the upper Metrosideros angustifolia and Cannomois virgata. The 50mm immediately below the litter layer and mixing them community occurs on deep, acidic soils at lower altitudes together. The samples were air-dried and the pH measured underlain by sandstone. It was broadly distributed, occur- in 30g sample of soil suspended in 75ml of deionised-dis- ring in five of the six rivers examined, except the tilled water using a pH meter (WTW 320 pH meter, Molenaars River, but is most common in the Palmiet River Germany). The proportions of coarse, medium and fine sand system. It contrasts with the other communities in that the in each soil sample were determined using the Bouyoucos riparian tree Podocarpus elongatus is usually absent. particle size method (Bouyoucos 1962). The percentage 3.The Cliffortia ruscifolia-Metrosideros angustifolia cover of surface rocks in each plot was estimated visually. Community (acronym: Cliffortia Community) characterised by the presence of Cliffortia ruscifolia, frequent dominance Vegetation and environmental analyses of Metrosideros angustifolia and Diospyros glabra, and absence of Prionium serratum. The community is promi- Vegetation from the 76 samples was classified using nent at higher altitude on soils of intermediate depth and TWINSPAN (DOS version; Hill 1979), and the groupings val- pH, underlain by sandstone and present in all rivers sam- idated against a Detrended Correspondence Analysis (Hill pled. and Gauch 1980). Alien species and rare unidentifiable 4.The Kiggelaria africana-Brabejum stellatifolium species were excluded from the ordination. Community Community (acronym: Kiggelaria Community) had no identities (designations) were based wherever feasible on characteristic species, although in the Eerste River it is two diagnostic species, a differential and a dominant species characterised by the presence of Kiggelaria africana and (Taylor 1996). Canonical Correspondence Analysis (CCA), Olea europaea subspecies africana with Brabejum stellat- was also applied to assist the interpretation of plot and ifolium and Metrosideros angustifolia occurring frequently species patterns in relation to environmental variables (ter as dominants. This community is most common on deep- Braak 1991). Environmental data incorporated in the analy- er, moderately acidic soils underlain by granites and is sis included altitude, slope, rock cover, soil depth, percent- present in all rivers sampled. age coarse, medium and fine sand, soil pH and total vege- The results of the DCA ordination are graphically illustrat- tation cover. Geological properties (Table 2) were extracted ed in Figure 1. The eigenvalues of axes 1 and 2 of the DCA from an electronic geological dataset (Council for for the sites were 0.67 and 0.52 respectively. No community Geosciences) integrated in the Protea Atlas Project using was tightly clustered in the ordination, and only two of the the plot locality data in ARCMAP. proposed communities were separated in the ordination along axis 1, namely the Leucadendron and Kiggelaria Statistical analyses Communities, with the other communities appearing as intermediates. The Nivenia Community was compositionally Measured environmental parameters were tested for signifi- more closely related to the Leucadendron Community but cant differences between delineated communities by apply- the Cliffortia Community’s compositional attributes were dif- ing a non-parameteric Kruskall Wallis H-test. Significantly fuse. different means were separated by applying Dunn’s test. Axes 3 and 4 of the ordination contributed no further clus- tering of the communities suggesting a community continu- Results um bounded by the Leucadendron and Kiggelaria Communities. Riparian vegetation classification and ordination Vegetation-environment relationships No distinctive communities emerged from the initial TWINSPAN classification, which comprised a high propor- Only three of the environmental variables measured, name- tion of common indigenous taxa (Table 2). Subsequent ly soil pH, altitude and soil depth, differed significantly manipulations of rows and columns delineated four prospec- between some of the proposed plant communities (Table 3). tive communities with 15 residual plots remaining as outliers The Nivenia and Leucadendron Communities were associ- (Table 2). These communities included: ated with lower soil pH than the Kiggelaria community; this 1.The Nivenia corymbosa-Brachylaena neriifolia a consequence of an underlying geology of sandstone in the Community (acronym: Nivenia Community) which com- first two communities compared with predominantly granite prises two unique species, Nivenia corymbosa and in the third (Table 2). The Leucadendron Community was Pentaschistis pallescens, and three community domi- found at significantly lower altitudes than the other commu- nants, namely Brachylaena neriifolia, Cannomois virgata nities. The Nivenia Community occurred on shallower soils and Metrosideros angustifolia. The community is located than the Leucadendron and Kiggelaria Communities. on shallow soils with a low pH underlain by sandstone The data set that incorporated all environmental factors (Table 3) and confined to the Wit River. This concurs with produced eigenvalues for the first two CCA axes of 0.35 and the reported restriction of (Nivenia corymbosa to the 0.27. The biplot indicating site positions and vectors for the Bainskloof-Tulbagh area (Goldblatt and Manning 2000). nine environmental factors is shown in Figure 2. Altitude and 2.The Leucadendron salicifolium-Berzelia lanuginosa pH had the highest canonical coefficients for Axis 1 (–0.66 4 Prins, Holmes and Richardson ------1--1------5---3------1132412------22----21------2----1--53----2 ------1------1------1------1------1------2------1------2------3--1------2------55433344312--- --3-24541-14-5 2---211------22----1------1-1-----5------52------32----4- -1---2---2---- -2----1------1------1-- 2------2------2------3------4------2------4-2------1------2--1--- -2------2------2------1------4------3-4------4-2 ------3------4------3------4 ------1-1------3------21413142- 1-34----1 ------s------1------21-----2 ------13------1------3------3------

1111 5674454476 745 2 32 11142176675 665555425627765374133 626133222333442 ^ sssssssss sssssssssssssg sssgssssssssssssg ggggggssgggssgggsgsgg gsgsgsssssggssg ATwinspan Cape. Numbers are cover values generated in vegeation samples from the Western of riparian classification table Plot number 234535671 18457089568142 90789642866939709 154567213236738323152 450069278017014 Geology Community 1 Nivenia corymbosa Tetraria thermalis Pentaschistis pallescens Community 2 Leucadendron salicifolium Metalasia densa Osmitopsis asteriscoides Brunia albiflora Erica hirtiflora Erica muscosa Pennisetum macrourum Erica curvirostris Erica sphaeroidea Leucadendron laureolum Cliffortia graminifolia Erica cf. armata Leucadendron gandogeri Ischyrolepis subverticillatu Erica perspicua Erica gnaphaloides Muraltia heisteria Cullumia setosa Cliffortia atrata Protea cynaroides Elytropappus glandulosus Pelargonium candicans Cliffortia polygonifolia Elegia asperifolia Dipogon lignosus Euryops abrotanifolius Psoralea aphylla Erica plukenetii Plecostachys serpyllifolia Community 3 Cliffortia ruscifolia Erica species Cunonia capensis Table 2: Table South African Journal of Botany 2004, 70: x–x 5 ------3------4------2------4----5 ------3--1------2------2------12------4------2------24------3------12------2------4-15---1------1---34------2--- 5------1552-2------3------3-3------113------312-5----3----3------2---5------2455112------1------3------433------3212------1---4------2------1------4-----4------2------2------3------2------1------4------3------5544--353 ------2--- 231----4---3--341 ---2---122----3---4------1------1-----2-----1 --1-3------31-1-2------221------32------21-----241------43------11------1- 4------3------2--2------4-----2------1-- ---1-----1---- -2---3------3------1------4------4------3---5------1------1 ------1 ------243342323 -2-2522-145-5- -25542-54155-3355 -45-4-541444--3152555 25434554525555- 23514---4 -3--1----41-35 2--553---5--55555 55555455552--55335214 -555544555513-4 -13-1-111 12--2----22-4- 1--242212321---2- 1-321-322-1--131-31-- -4324532113---- . africana 4-14222-1 433555234334-- -22----2---1------2------subsp Rhus laevigata Elegia species Phylica spicata Struthiola myrsinites Lichtensteinia lacera Blechnum capense Hypodiscus aristatus Rapanea melanophloeos Cynodon dactylon Clutia pulchella Shared species to all recognized communities Erica caffra Blechnum species Chironia linoides Cullumia ciliaris Pseudobaeckea africana Stoebe cinerea Shared species to all plots Metrosideros angustifolia Brabejum stellatifolium Diospyros glabra Passerina vulgaris Erica longifolia Olea europaea Hartogiella schinoides Oftia africana Protasparagus rubicundus Tribolium uniolae Restio bifidus Restio species Adenandra villosa Cynanchum africanum Heeria argentea Otholobium decumbens Phylica imberbis Polygala myrtifolia Elegia cuspidata Community 4 Kiggelaria africana Shared species communities 1-3 Berzelia lanuginosa Table 2 cont. Table 6 Prins, Holmes and Richardson 42323-4-5 5553555--23-5- 3-2--51425-4--34------1-43-2-----5-3------3-1------4 ----51321 -----22------2-1---224-3------1--1------32------1-14353-1 -34----5-21--- -512---32--5-232- 2------4---3----3 ------1-333--5 454423435 ------21-24--3 431-4--35-54--23- 5-2--42-54221-3---3-- 53444------2--- -513---35 ---54------1------33-3---554------1-- 5-2---5------3---- -43------2------1-1-----5 1----3------35-5-25------3---33------1------1- ---3--1-----43-4------31--355-1-2---3 ----33-21-----5 2-2-4------1 5----2---15-----2 1--531---21-424------4------2------2------4------33------2------1--- --1------1------3-1------123----- 1-----2-- 551--1-2231------13-23---3451 ------11-1-2-2-43------4 --42-22------255-5------3------1------2-42------2------4344--45------3-4-3 33------3-----4------4---3------3------2------2------3------1--- 2------3------43-4------4 -----4------5------3------1------1----2 ------4------1------2-2------1------1------2------1------1------1------1 ------5------5------1------2------1------2------21 -----2------1------3------1------4------1------1------1------3------3------2------1------2------4------1 ----1------Cannomois virgata ^ geology: s = sandstone, scree over sandstone & alluvium sandstone; g granite granite. Psoralea pinnata Elegia capensis Brachylaena neriifolia Prionium serratum Poaceae species Rhus angustifolia Podocarpus elongatus Podalyria calyptrata Stoebe plumosa Pteridium aquilinum Restio cf leptostachyus Dryopteris inaequalis Indeterminate Rubus pinnatus Senecio species Osteospermum ciliatum Maytenus acuminata Tritoniopsis antholyza Protea repens Myrsine africana Anaxeton laeve Aristea juncea Hymenolepis parviflora Ficinia filiformis Halleria elliptica Halleria lucida Helichrysum cymosum Hypocalyptus species Juncus punctorius Phragmites australis Cassytha ciliolata Stoebe incana Aristida junciformis Gnidia oppositifolia Cyperaceae species Tetraria ustulata Table 2 cont. Table South African Journal of Botany 2004, 70: x–x 7

Table 3: Mean values and standard deviations of environmental parameters associated with delineated riparian communities. Values with similar superscript letters not significantly different at P ≤ 0.05

Environmental parameter Community acronym H-statistic Nivenia Leucadendron Cliffortia Kiggelaria (P-value) No. of samples 9 14 17 21 Altitude (m) 324 ± 72.0ab 200 ± 188a 388 ± 14.6b 298 ± 28.1ab 10.3 (0.016) Slope (o) 12.1 ± 12.7 11.8 ± 13.1 14.5 ± 13.9 10.4 ± 6.70 1.04 (0.792) Rock cover (%) 9.78 ± 15.0 17.6 ± 17.7 17.5 ± 17.6 16.9 ± 24.5 2.80 (0.424) Soil depth (m) 0.13 ± 0.14a 0.51 ± 0.46b 0.34 ± 0.36ab 0.46 ± 0.42b 9.98 (0.019) % coarse sand 38.4 ± 15.4 31.5 ± 9.99 36.7 ± 20.8 31.7 ± 21.0 1.40 (0.705) % medium sand 10.5 ± 8.00 24.0 ± 14.9 15.4 ± 14.4 16.4 ± 15.1 5.85 (0.119) % fine sand 8.11 ± 5.21 7.29 ± 7.25 8.81 ± 7.36 9.33 ± 7.17 3.10 (0.377) pH 3.96 ± 0.39a 3.91 ± 0.58a 4.50 ± 0.88ab 4.64 ± 0.39b 22.6 (<0.0001) Vegetation cover (%) 90.4 ± 10.3 89.6 ± 8.65 85.9 ± 15.2 87.9 ± 9.82 0.524 (0.914)

ed with low values of soil pH and altitude were Erica per- (a) Kiggelaria Nivenia Leucadendron spicua, E. sessiliflora, Cliffortia graminifolia and Brunia albi- 4 Cliffortia flora, and those with high levels of these environmental Kiggelaria parameters were Heeria argentea, Myrsine africana and Unclassified Cynanchum africanum. 3 Discussion 2 Four longitudinal plant communities were delineated for the mountain and foothill zones of the six rivers examined 1 between Bainskloof and Kleinmond. Noteworthy, was the Leucadendron large number of unclassified plots and community overlap in the ordination based exclusively on floristic composition. This pattern suggests a plant community continuum rather AXIS 2 (b) Wit than a series of unique communities with distinct bound- Molenaars aries. The most distinctive of these is the Leucadendron 4 Palmiet salicifolium-Berzelia lanuginosa Community, but even this Eerste community shares many of its common riparian species with Steenbok 3 those of the other communities. Rooiels Riparian vegetation in the Western Cape mountain and foothill zones comprises a mixture of terrestrial fynbos ele- 2 ments as well as non-fynbos plants that are adapted to the specific ecological conditions present close to the river 1 (Sieben 2003). Many of the riparian specialist species are nevertheless relatively widespread in their distribution, occurring in most riverine systems in the Western Cape. Fynbos shrublands also contain a number of generalist 123456 species that may result in communities forming a continuum, AXIS 1 for example, as was found in Swartboschkloof where community boundaries were difficult to interpret (McDonald Figure 1: Detrended correspondence analysis ordination of ripari- 1987). Generalist fynbos and riparian species, together with an vegetation plots indicating positions of (a) plant communities and many infrequently-occurring species that have limited value (b) rivers in distinguishing communities (47% of species occurred in fewer than 5% of plots), likely explain the poor definition of communities. and –0.57 respectively) and medium sand and soil depth for The interplay between substratum, topography and fire Axis 2 (0.63 and 0.62 respectively). The CCA resulted in lit- are important in determining both riparian community struc- tle separation of sites from the different rivers (Figure 3), ture and its lateral extent. Fynbos elements occurring in the with the exception of six plots from the Palmiet River that riparian zone are adapted to regular fires and most of the grouped on the positive side of Axis 1. These comprised common riparian scrub species are resprouters that can sur- mostly Leucadendron plots that differed environmentally and vive periodic fires. By contrast, the forest species, such as floristically from those in the other rivers sampled. These Kiggelaria, Olea and Rapanea, require long fire-free inter- results indicate that the Leucadendron community is associ- vals for establishment before they are large enough to with- ated with lower soil pH and altitude. Species highly correlat- stand fires. Where the river channel is deep, with steep 8 Prins, Holmes and Richardson

CCA case scores 71s 4.2 76r DEPTH

3.3

FSAND 63j 2.5 VEGCOV 24m SLOPE 64j43p 46p 1.7 73r 69s 77r52j 75r 74r 0.8 57j61j 49p pH 56j27m62j 53j 45p 10w30m 66s67s 42m26m60j36m65j 9w 17w 8w 68s 20w15w 51p 58j54j55j 70s 21m 3w 1w16w6w 50p -4.2 -3.3 -2.5 -1.7 59j -0.8 23m18w4w13w 0.82w 1.7 2.5 3.3 4.2 28m25m38m 11w 72s 19w 5w 47p 29m 22m -0.8 39m 31m 12w 48p 35m14w 33m44p -1.7 37m CSAND 7w ROCK 40m ALT 41m 32m-2.5

-3.3

-4.2

Vector scaling: 5.00 AXIS 1

Figure 2: Canonical correspondence analysis biplot showing plot positions and vectors at a scaling of 5.0. Plot numbers are indicated, followed by a lower case initial representing the first letter of the river name (an exception is ‘j’ indicating the Eerste River at Jonkershoek) banks or adjacent boulder scree, fires may skip across the The most important environmental determinants of ripari- riparian zone allowing taller scrub or forest to develop on the an community composition in this study were soil pH and banks. Depending on local conditions, forest communities depth, followed by altitude and soil texture. Soil pH largely may be flanked in the back dynamic zone by scrub or fynbos reflected the underlying geology, with lowest values associ- communities. Where the topography is flat and the substra- ated with sandstone-derived parent materials and the high- tum is not rocky, fires are more likely to penetrate the ripari- er values associated with granite parent materials. In a study an zone, thus promoting the establishment of fire-adapted of riparian vegetation in the Hottentots Holland Mountains, elements. Sieben (2003) also found that plant communities on different A finer definition of riparian plant communities may have parent materials (shales and granites versus sandstones) been achieved by sampling the entire lateral gradient from were quite distinctive, with altitude being the next most the wet bank zone through to the back dynamic zone using important environmental determinant. The Kiggelaria a transect approach (Boucher 1998, Sieben 2003). Riverine africana-Brabejum stellatifolium Community was most likely vegetation is influenced by the flood regime, with frequency to occur over granite parent materials. A number of of inundation and stream power being the most important Afromontane Forest elements, including Rapanea determinants of lateral zonation (Sieben 2003). However, in melanophloeos and Olea europaea subspecies africana in the headwater streams of the mountains it was found that this community indicate that its existence also requires pro- the lateral zones were less obvious than in foothill and lower tection from regular fynbos fires and drought-season access reaches (Sieben 2003). The focus of this study was rather to to the water table (Sieben 2003). identify broad riparian communities associated with any Altitude was not sampled over a high proportion of the major longitudinal environmental factors. potential range in this study, and thus was not a particularly South African Journal of Botany 2004, 70: x–x 9

Intermediate as, for example, following the clearance of dense wattle that (a) Leucadendron has dominated a riparian site for many decades, a site 5.0 assessment should include a scan of upstream riparian Kiggelaria communities and neighbouring terrestrial plant communities. D Vectors Both are potential sources of suitable propagules for restor- 2.5 ing the site. If the adjacent vegetation is relatively intact, the disturbed site may rapidly recover through natural re-colonisation. The advantage of natural re-colonisation, apart from pH a large saving in costs, is that the local gene pools will be -5.0 -2.5 2.5 5.0 maintained. Alternatively, if the surrounding landscape is highly degraded, the re-introduction of species should be via seed or propagated material from sites whose environmen- A -2.5 tal characteristics most closely resemble those of the community being restored. Particular emphasis should be placed -5.0 on re-introducing the common and generalist riparian species, particularly those adapted to fire, as these are most likely to quickly re-establish appropriate vegetation structure (b) Wit Molenaars and resilient plant cover. If any specialist species are totally 5.0 Palmiet eliminated from a river system, then some consideration D Eerste should be given to their re-introduction from the nearest Steenbok alternative source. However, such attempts should be prag- 2.5 Rooiels matic. For example, Afromontane forest elements should not Vectors be introduced in areas that are unlikely to be protected from fynbos fires by deep gorges or other natural features. pH -5.0 -2.5 2.5 5.0 Acknowledgements — We thank the National Research Foundation and the University of Cape Town for financial support and S Galatowitsch and A Rebelo for useful comments on the manuscript. A -2.5 References -5.0 Boucher C (1978a) Cape Hangklip area. II. The vegetation. Bothalia 12: 455–497 Figure 3: Canonical correspondence analysis biplots showing posi- Boucher C (1978b) Cape Hangklip area. I. The application of asso- tions of (a) the Leucadendron and Kiggelaria communities and (b) ciation-analysis, homogeneity functions and Braun-Blanquet rivers. The important environmental vectors, namely soil pH, alti- techniques in the description of south-western Cape vegetation. tude and soil depth, are indicated on the figures by the symbols pH, Bothalia 12: 293–300 A and D, respectively Boucher C (1998) Flora and Vegetation. In: Brown CA, Day E (ed) Starter Document: Assessment of the Instream Flow Requirements for the Palmiet River and the Freshwater useful indicator of plant community composition. Requirements for the Palmiet Estuary. IFR. Unpublished report Nevertheless, the Leucadendron salicifolium-Berzelia lanug- prepared for Southern Waters, Cape Town, pp 101–135 Bouyoucos GJ (1962) Hydrometer method improved for making inosa community appeared to be associated with lower-alti- particle size analysis of soils. Agronomy Journal 54: 464–465 tude sites on deep acid sands. Brown C (1998) The ecological status of Western Cape rivers inves- tigated: shock survey findings. African Wildlife 52: 27–28 Conclusions and Recommendations Campbell BM (1986) A classification of the mountain vegetation of the Fynbos Biome. Memoirs of the Botanical Survey of South It is concluded that the restoration of degraded riparian Africa 50: 1–121 communities in the mountain and foothill zones of the Cowling RM, Holmes PM (1992) Flora and vegetation. In: Cowling Western Cape should be guided by site geology (i.e. granite RM (ed) The Ecology of Fynbos: Nutrients, Fire and Diversity. versus sandstone parent materials) and soil physical and Oxford University Press, Cape Town, pp 23–61 chemical properties (pH, texture and depth) with altitude of Cowling RM, Richardson DM, Mustart PJ (1997) Fynbos In: Cowling RM, Richardson DM, Pierce SM (eds) Vegetation of Southern secondary importance. In all cases, the restoration target Africa. Cambridge University Press, Cambridge, pp 99–130 plant community should comprise a large proportion of gen- Davies BR, Day JA (1998) Vanishing Waters. University of Cape eralist riparian scrub species, such as Metrosideros angusti- Town Press, Rondebosch, South Africa, pp 460 folia, Brabejum stellatifolium, Diospyros glabra, Brachylaena Forsyth GG, Richardson DM, Brown PJ, Van Wilgen BW (2004) A neriifolia, Psoralea pinnata, Cannomois virgata and Elegia rapid assessment of the invasive status of Eucalyptus species in capensis, which are widespread in all communities delineat- two South African provinces. South African Journal of Science (in ed. Common fynbos elements from neighbouring terrestrial press) vegetation also form part of most riparian plant communities. Galatowitsch S, Richardson DM (submitted) Riparian forest recov- It is recommended that in planning a restoration operation ery after alien clearing in headwater streams of the Western Cape South Africa. Biological Conservation 10 Prins, Holmes and Richardson

Goldblatt P, Manning J (2000) Cape Plants. A conspectus of the Richardson DM, Macdonald IAW, Holmes PM, Cowling RM (1992) Cape Flora of South Africa. Strelitzia 9: 744 Plant and animal invasions. In: Cowling RM (ed) The Ecology of Hill MO (1979) TWINSPAN – A FORTRAN Program for Arranging Fynbos: Nutrients, Fire and Diversity. Oxford University Press, Multivariate Data in an Ordered Two way Table for Classification Cape Town, pp 271–308 of the Individuals and the Attributes. Cornell University, Sieben EJJ (2003) The Riparian Vegetation of the Hottentots Department of Ecology and Systematics, Ithaca, New York Holland Mountains, SW Cape. Unpublished PhD dissertation, Hill MO, Gauch HG (1980) Detrended correspondence analysis: an University of Stellenbosch, South Africa improved ordination technique. Vegetatio 42: 47–58 Taylor HC (1978) Capensis. In: Werger MJA (ed) Biogeography and Hoffmann JH, Moran VC (1988) The invasive weed Sesbania Ecology of Southern Africa. Junk, The Hague, pp 171–229 punicea in South Africa and prospects for its biological control. Taylor HC (1984) A vegetation survey of the Cape of Good Hope South African Journal of Science 84: 740–742 Nature Reserve. II. Descriptive account. Bothalia 15: 259–291 Kruger FJ (1978) South African heathlands. In: Specht RL (ed) Taylor HC (1996) Cederberg vegetation and flora. Strelitzia 3: 75 Ecosystems of the World. Heathlands and Related Shrublands. ter Braak CJF (1991) CANOCO – A FORTRAN Program for Elsevier, Amsterdam, pp 19–80 Canonical Community Ordination by Partial Detrended Canonical McDonald DJ (1987) Ordination by detrended correspondence Correspondence Analysis (version 3.12). Rep. Agricultural analysis (DCA) of the vegetation of Swartboschkloof, Mathematics group, Wageningen Jonkershoek, Cape Province. Bothalia 17: 121–129 Van Wilgen BW, Le Maitre DC, Cowling RM (1998) Ecosystem serv- McDonald DJ (1988) A synopsis of the plant communities of ices, efficiency and equity: South Africa’s Working for Water Swartboschkloof, Jonkershoek, Cape Province. Bothalia 18: Programme. Trends in Ecology and Evolution 13: 31–36 233–260 McDonald DJ (1993) The vegetation of the southern Langeberg, Cape Province. 1. The plant communities of the Boosmansbos Wilderness Area. Bothalia 23: 129–151

Edited by CF Musil