Shrubland Restoration cesses.

Shallow and deep topsoil treatments yielded

high plant density, richness, and projected canopy cover,
but canopy cover was higher in deep topsoil plots

Following Woody Alien throughout the trial. Fertilizer addition increased can-
opy cover in untopsoiled and shallow topsoil plots via
an increase in alien annual species. Fertilizer addition
Invasion and Mining: ultimately may lead to increased native vegetation
cover in untopsoiled areas, but as it increased proteoid

Effects of Topsoil Depth, mortality on deep topsoil plots, it is not recommended

for sites where topsoil is available. A species-rich and
structurally representative fynbos community may be

Seed Source, and restored on topsoiled areas provided that the native
disturbance regime is simulated and seeds of major
structural guilds not present in the soil seed bank are
Fertilizer Addition included in the seed mix.

Key words: invasive alien plants, soil disturbance, soil-

Patricia M. Holmes1,2 stored seed banks, fynbos vegetation, seedling recruit-
ment, establishment and growth.

Abstract

Invasion by woody alien plants, construction, and

Introduction
mining operations are among the major disturbances
degrading vegetation in the Cape Floristic Kingdom,
South Africa. The aim of this study was to assess
whether native fynbos shrubland vegetation could be
T he Cape Floristic Kingdom of South Africa boasts
among the world’s highest concentrations of plant
species, of which about 80% are endemic (Cowling et al.
restored following dense alien invasion and distur-
1996, 1997). Fynbos vegetation is a major component of
bance by mining. An area supporting dense alien
the Cape Floristic Kingdom (Rebelo 1996) and occurs on
trees was cleared and topsoil was stripped and stock-
nutrient-poor substrata from the mediterranean-climate
piled to simulate mining disturbance. A field trial in-
west to the aseasonal temperate east.
vestigated the effects of topsoil depth, seed mix appli-
Remaining fynbos vegetation has been degraded by
cation, and fertilizer on native species recruitment
factors such as alien plant invasion, habitat fragmenta-
and vegetation development over a three-year period.
tion resulting from agricultural and urban develop-
Soil-stored seed banks contributed 60% of the species
ment, and inappropriate management practices (Rebelo
recruited, indicating that areas invaded for three de-
1992a, 1992b). In addition, disturbances of shorter dura-
cades have good restoration potential. The addition of
tion such as construction and mining activities have de-
a fynbos seed mix, which included serotinous over-
graded fynbos vegetation (Romoff 1986; Levitt 1997). In
story species, improved both the richness and struc-
the mountains, invasive alien trees and shrubs domi-
tural composition of the vegetation. Most species sown
nate thousands of hectares of natural vegetation, signif-
in untopsoiled plots established, but survival and
icantly altering communities and reducing diversity
growth was low compared to topsoil plots. Poor growth
and water run-off (Richardson et al. 1992; Le Maitre et
in combination with a lack of soil seed bank species,
al. 1996). At least 26% of natural vegetation in the
indicate that restoring a diverse and functional cover
mountains has been invaded at some stage (Richardson
of indigenous vegetation on subsoil is not possible in
et al. 1997). The realization that woody alien vegetation
the short-term. Soil amelioration is required to im-
uses more water than indigenous vegetation led to the
prove rooting conditions and initiate ecosystem pro-
initiation in 1995 of a multi-million dollar government
program to control invasive plants in catchment areas
(van Wilgen et al. 1997). The ultimate success of this
1 Institute for Plant Conservation, University of Cape Town,
program depends on a sustainable cover of fynbos veg-
Private Bag, Rondebosch 7700, South Africa
2 Address correspondence to P. M. Holmes, 23 Dreyersdal
etation being restored following alien clearing.
In the lowlands, a large proportion of the vegetation
Road, Bergvliet, 7945, Cape Province, South Africa, email
[email protected] already has been transformed. Remaining vegetation
continues to be threatened as a result of increasing hu-
© 2001 Society for Ecological Restoration man pressures on the land: directly by agricultural and

MARCH 2001 Restoration Ecology Vol. 9 No. 1, pp. 71–84 71

Shrubland Restoration Following Alien Invasion and Mining

urban expansion and alien plant invasions, and indi- (4) Does the addition of fertilizer promote establish-
rectly by habitat fragmentation and inappropriate man- ment of fynbos vegetation?
agement (Wood et al. 1994).
There is an increasing awareness of the need to re-
Methods
store vegetation that has already been degraded and to
lessen the impact of future developments (Anonymous Study Site
1997). Although little restoration research has been
done in fynbos, the recruitment dynamics, community The restoration trial was set up on unmined land at the
ecology, and functional ecology of fynbos have been Serina Kaolin Pty. Ltd. mine site at Noordhoek, on the
relatively well studied (Cowling 1992) and based on Cape Peninsula, South Africa (34⬚05⬘40⬙S 18⬚22⬘20⬙E).
this, protocols for restoration have been outlined The vegetation at the site originally was Proteoid Fyn-
(Holmes & Richardson 1999). For example, it is known bos (shrubland comprising ⬎10% non-sprouting pro-
that many fynbos species have long-lived propagules teoid cover; Cowling & Holmes 1992), but was planted
stored in the soil, which can survive moderate-to-long to Pinus pinaster during the late 1950s (Jeremy Wiley per-
alien plant invasions (Holmes & Cowling 1997a, 1997b). sonal communication, and aerial photographs). Planting
Germination of soil-stored fynbos species may be stim- is a process analogous to invasion as young trees are
ulated by direct and indirect fire-related cues (e.g., heat planted into burnt or young fynbos vegetation. The dense
pulse, smoke compounds, and post-fire diurnal temper- alien overstory was dominated by P. pinaster, with occa-
ature regime; Keeley & Bond 1997). Thus, if appropriate sional individuals of Eucalyptus diversicolor (karri) present.
germination cues are applied, persistent seed banks The understory comprised scattered Acacia saligna (Port
may be used to assist in restoring native vegetation. Jackson willow), A. cyclops (red eye), Hakea drupacea
A proposal to mine kaolin near the boundary of the (sweet hakea), and H. sericea (silky hakea) (all invasive
Cape Peninsula National Park (an area of international alien species) and was devoid of indigenous species. A
conservation importance; Trinder-Smith et al. 1996) few indigenous species survived in gaps alongside ve-
precipitated the need to investigate optimal methods hicle tracks. The site is situated on a gentle east-facing
for restoring native vegetation following opencast min- slope. The topsoil is brown sandy loam, derived from
ing. In the past, alien grasses had been used to rehabili- Table Mountain Sandstone colluvial material, which
tate mined areas. As the proposed mine site supported overlies heavier subsoil with granite-derived clay be-
dense alien vegetation (predominantly Pinus pinas- neath. Although these clay subsoils have a limited dis-
ter[cluster pine]) it provided a testing ground for resto- tribution, sandy loam topsoils are fairly common on
ration, both following alien clearing and disturbance by lower mountain slopes in the western Cape and often
mining operations. Mining was simulated by stripping, support dense alien stands. The climate is Mediterra-
stockpiling, and respreading topsoil. Findings from this nean, with hot, dry summers and cool, wet winters. An-
study will assist in forming guidelines for the restora- nual rainfall was 670 mm during the first year of the
tion of mountain catchment areas following the clearing trial.
of dense stands of alien trees, as well as for the restora-
tion of fynbos areas following disturbance by mining Seed Collecting
and construction activities.
In view of the possibility of topsoil being unavailable In January 1995, fynbos seed was collected from indige-
or in short supply following mining, a secondary aim nous vegetation within a five-kilometer radius of the
was to compare restoration without topsoil as well as trial site. This ensured that any introduced seed would
with two different topsoil depths. One of the aims of be of local origin. Seed was handpicked from plants
revegetation following mining is to provide quick plant (predominantly the canopy-storage species, e.g., Protea
cover to control soil erosion and create an aesthetically lepidocarpodendron [black-beard sugarbush] and Leuca-
pleasing landscape. Toward this end, different seed dendron xanthoconus [sickle-leaf conebush) [Proteaceae])
mixes and a fertilizer treatment were tested for their ef- and suction-harvested from under bushes and in the lit-
ficiency in providing indigenous cover and diversity on ter layer where the smaller-seeded species drop their
the various substrata. seed. The species composition of the suction-harvested
The following key questions were addressed: component was not directly assessed. The seed was
sun-dried for a week to release seeds from serotinous
(1) Can fynbos be restored on untopsoiled substrata? cones; and then sorted into paper bags ready for the
(2) Is topsoil depth an important factor in fynbos resto- trial. The insecticide Karbadust (Efekto, Silverton, South
ration? Africa) was sprinkled into the bags to kill potential seed-
(3) Which seed mix application best restores Proteoid eating insects. Bags were stored in a cold room (4⬚C) in
Fynbos on the different substrata? order to minimize loss of viability until required for the

72 Restoration Ecology MARCH 2001

 Shrubland Restoration Following Alien Invasion and Mining

trial. One day before sowing, the suction-harvested Sixteen meter square (4⫻4) plots were demarcated us-
seed mix was given a smoke treatment to enhance ger- ing logs pegged across the slopes, so one plot took up
mination response (Brown 1993). the entire width of each substratum strip, and between
Portions of the hand-collected seed and the suction- four and nine plots fitted along the length of each strip.
harvested seed were combined to give the fynbos mix Logs also helped to stabilize the soil against surface
(F) used in the trial. For the fynbos annual (FA) supple- wash.
ment, fynbos seed harvested from a firebreak area was Seed was applied (12–13 April 1995) by broadcasting
added to F. Firebreak areas support more short-lived over the entire plot, as evenly as possible. A thin layer
species and should yield more annual and short-lived of wood-chip mulch (3–6 cm thick) was applied imme-
perennial seeds than the mature veld. For the fynbos diately over this to prevent the seed from blowing away
commercial (FC) supplement, the commercially avail- and to ensure good contact between soil and seed. The
able indigenous seed of Cynodon dactylon (Bermuda first application of fertilizer was done on 25 August
grass, Poaceae), Pelargonium capitatum and Geranium in- 1995 following seed germination. Owing to the known
canum (both sprawling perennial forbs, Geraniaceae) sensitivity of fynbos species to fertilizer addition, espe-
were purchased and added to F. Cynodon was sown at a cially phosphorus (Witkowski 1991), a 0.5% solution of
rate of 300 seeds/m2 (half the recommended rate for the fish emulsion Seagro (Seagro, Cape Town, South
pastures), and Pelargonium and Geranium each at two Africa) was used in the first year of establishment at a
seeds/m2. rate of 1.2 L/m2. Undiluted Seagro contains 5.5% N,
0.75% P, and 1.6% K and trace elements. On 15 April
1996, the ultra-slow-release fertilizer Horticote 7:1:2 (27)
Experimental Design (Ocean Agriculture, Johannesburg, South Africa) was
applied at a low rate of 100g per m2 (100 g yields 18 g N,
The three main factors accommodated within the exper-
2.6 g P, 6.6 g K, 1.2 g Mg, and small amounts of Fe, Mn,
imental design were topsoil depth, seed treatment and
Zn, Cu, B, and Mo). This ensured slow release of nutri-
fertilizer. Substratum strips were prepared for the dif-
ents over a 9–12 month period.
ferent topsoil treatments and the other factors were
nested within these. Topsoil depth treatments were un-
topsoiled (0 mm), shallow (100 mm), and deep (300 Monitoring
mm). Seed treatments were control (con: no seed
Five separate censuses of seedling recruitment were
added), fynbos mix (F), fynbos plus annual supplement
done 3, 6, 12, 18, and 36 months following sowing. In
(FA), and fynbos plus commercial supplement (FC).
each of the 105 plots, four 1 m2 permanent quadrats
Fertilizer was applied at one concentration level only,
were positioned, two on each side of a transect bisecting
and was applied to half of the seed treatment plots, ex-
the plots, alternately above and below the transect.
cluding the seed control plots. The four different seed-
Counts of all individuals in each growth-form guild
ing treatments and the fertilizer treatment were allo-
were made at the 3-, 6-, 12-, and 18-month censuses. At
cated to the plots at random (total number of plots 105,
the 18-month census, the numbers of species represent-
five replicates for each seed/fertilizer/topsoil treat-
ing each growth-form guild and the numbers of alien
ment).
plants, were also recorded. Estimates of total projected
plant cover were recorded for each quadrat at each of
the censuses.
Site Preparation
At 36 months, projected plant cover for each species
At the end of summer (March 1995) a 0.2 ha area of was estimated for the entire plot, and species richness
dense alien vegetation was cleared for the trial. The was recorded in one single 1 m2 quadrat placed at the
alien trees were felled, logs removed, and finer branches center of each plot. All species that flowered during the
chipped for use as mulch. Roots and stumps were bull- trial were collected and identified.
dozed out of the soil and then the topsoil was stripped In February 1997, ten months after the application of
and stockpiled to simulate disturbance by mining. Top- the ultra-slow-release fertilizer, samples were taken
soil was stockpiled until the site was cleared (a maxi- from fertilized and unfertilized plots in each topsoil
mum of seven days). Substratum strips (4 m wide and depth treatment (five replicates each), as well as three
between 16 and 36 m long) were established downslope samples from the adjacent undisturbed soil under
in a west-to-east orientation on the cleared site. Untop- pines. Samples were analyzed for pH, resistance, and
soiled strips comprised the heavier subsoil layer (to nutrient concentrations, using standard techniques of
simulate the clay overburden substratum); shallow top- the Soil Science Division (Department of Agriculture:
soil blocks had 100 mm of the topsoil reapplied; and Western Cape, Private Bag X1, Elsenburg 7607, South
deep topsoil blocks had 300 mm of the topsoil re-applied. Africa).

MARCH 2001 Restoration Ecology 73

Shrubland Restoration Following Alien Invasion and Mining

Alien Plant Control

In September 1994, the biocontrol fungus (Uromycla-

dium tepperianum) was inoculated onto alien Acacia sali-
gna in the vicinity of the restoration trial site. This fun-
gus has been shown to weaken plants and reduce their
reproductive output (Morris 1999). Weeding of alien in-
vasive trees and shrubs recruiting from the soil seed
banks (Acacia species) and the seed rain (Pinus, Eucalyp-
tus and Hakea species) was done in January 1996, to pre-
vent these species from outgrowing and shading the
fynbos species. Alien seedlings were uprooted where
damage to indigenous species could be avoided; other-
wise, they were cut below ground level with secateurs
to prevent resprouting. Following this major weeding
intervention, very little follow-up control was required.

Data Analysis

Two-way analysis of variance (ANOVA; BMDP Pro-

gram 2V, Dixon 1992) was used to investigate the main
treatment effects and the interactions between depth
and seed, and depth and fertilizer treatments at various
stages in the development of the vegetation. Earlier
analyses indicated that the blocking factor (substratum
strip) influenced the variance less than the experimental
factors, and it was excluded to simplify the analyses.
Cover values and counts of plant density were square-
root-transformed to satisfy ANOVA homogeneity of
variance requirements.
Guild structure (by plant cover) in three-year-old
vegetation was investigated using contingency tables
(BMDP Program 4F, Dixon 1992). The null hypothesis
of independence between guild structure and topsoil
and seeding treatments was tested by chi-square. Ad-
justed standardized deviates (Haberman 1973) exceed-
ing 3.0 in absolute value were taken to indicate cells
with unusually large deviations from the expected. In-
digenous species were assigned to growth form, leaf
size, seed dispersal, seed storage, regeneration, and pol-
lination classes following Holmes & Cowling (1997a).

Figure 1. Indigenous plant density during the restoration trial

following different non-fertilized seeding treatments: (a) un-
Results
topsoiled, (b) shallow topsoil, (c) deep topsoil (n ⫽ 20). No
seed addition (con), fynbos mix (F), F plus annuals (FA), F
Recruitment Pattern and Development of Plant Cover
plus commercial seed (FC).
Two-thirds of seedling recruitment occurred during the
first three months after sowing, and highest densities
were recorded at six months, when both perennial and a slight rise in plant density owing to the renewed ap-
ephemeral (annual and geophyte) species were present pearance of some ephemerals.
(Fig. 1). Plant density decreased over the first summer In topsoil plots, plant cover developed rapidly over
as ephemerals died back and some perennial mortality the first 18 months (Fig. 2). By three years, dense vege-
occurred. Mortality was most pronounced on untop- tation (90% projected canopy cover) had developed on
soiled plots (Fig. 1). During the second winter there was deep topsoil plots and moderately dense vegetation

74 Restoration Ecology MARCH 2001

 Shrubland Restoration Following Alien Invasion and Mining

(Fig. 2). Within each topsoil depth, positive correlations

between plant cover and density were all highly signifi-
cant (r ⬎ 0.33, p ⬍ 0.001).

Effect of Topsoil Depth on Plant Density and Cover

The ANOVAs run for the different censuses yielded

similar results; those for the 18-month census are pre-
sented (Table 1). Most of the variance attributed to the
topsoil depth factor is associated with the untopsoiled
treatment. Plots without topsoil had significantly lower
indigenous plant density and cover, and lower alien
plant density, compared to topsoil plots across all seed
and fertilizer treatments (Figs. 1–4).
Differences between the shallow (100 mm) and deep
(300 mm) topsoil treatments took longer to develop.
There were few differences at six months, but by 18
months density was higher in most of the shallow top-
soil plots (Fig. 1). Plant cover without fertilizer addition
was higher in deep than in shallow topsoil plots by 18
months, and this difference increased after three years
(Fig. 2).

Effect of Seed Treatments on Plant Density and Cover

The addition of seed significantly increased both plant

density and cover over the unseeded controls (Table 1;
Figs. 1 & 2). Without seed addition, there was virtually
no plant recruitment in untopsoiled plots. In topsoil
plots, the soil seed bank contributed one-third of the
seedlings during the main recruitment phase of the first
winter (Fig. 1). In topsoil plots at three years, differ-
ences in plant density between control and seed mix

Table 1. Results of ANOVAs for indigenous plant density,

projected canopy cover, and herbaceous alien density, 18
months after sowing.
Indigenous Density Canopy Cover Alien Density
Source df F F F
Figure 2. Projected plant canopy cover during the restoration
trial following different non-fertilized seeding treatments: (a) Experimental factors topsoil depth (depth) and sowing
untopsoiled, (b) shallow topsoil, (c) deep topsoil (n ⫽ 20, ex- treatment (seed)
cept for 36 months, where n ⫽ 5). No seed addition (con), fyn- Depth 2 236.4*** 754.4*** 49.4***
Seed 3 54.0*** 44.1*** 15.1***
bos mix (F), F plus annuals (FA), F plus commercial seed (FC).
Depth ⫻ seed 6 4.5*** 0.8ns 3.0**
Error 408
Experimental factors topsoil depth (depth) and fertilizer
(fert)
Depth 2 157.2*** 818.5*** 40.4***
(75% projected canopy cover) on shallow topsoil plots. Fert. 1 0.1ns 94.8*** 65.5***
Depth ⫻ fert. 2 1.1ns 40.8*** 2.8ns
However, on the sown untopsoiled plots, cover de-
Error 354
creased during the first summer owing to plant mortal-
ity, and remained relatively low (20%) after three years ns ⫽ not significant. *p ⬍ 0.05. **p ⬍ 0.01. ***p ⬍ 0.001.

MARCH 2001 Restoration Ecology 75

Shrubland Restoration Following Alien Invasion and Mining

Figure 4. Effect of seed source on native growth-form distribu-

tion: (a) by plant density at the peak of recruitment (6 months;
n ⫽ 20), (b) by species richness at 18 months (n ⫽ 20), and (c)
by percentage projected canopy cover at 36 months (n ⫽ 5).
No seed addition (con), fynbos mix (F), F plus annuals (FA), F
plus commercial seed (FC).

Effect of Fertilizer Treatment on Plant Density, Cover, and

Figure 3. Effect of fertilizer addition on (a) alien plant density Soil Variables
and (b) percentage projected canopy cover, at 18 months for
the different topsoil depths (n ⫽ 20). Fynbos mix (F), F plus Application of the ultra-slow-release fertilizer did not
annuals (FA), F plus commercial seed (FC). alter indigenous plant density, but caused a large in-
crease in alien herbaceous plant density after the sec-
ond winter (Table 1; Fig. 3a). The relative increase in
treatments were still evident, whereas differences in alien density was highest in untopsoiled plots.
cover were insignificant (Figs. 1 & 2). Fertilizer addition had no effect on plant cover in the
Differences among the three seeded treatments were deep topsoil plots, but caused an increase in cover in
relatively small. On shallow topsoil, the fynbos-plus- shallow topsoil and untopsoiled plots after the second
annual seed mix performed better than the other two winter (Fig. 3b). Most of this increase could be attrib-
seed mixes in the early stages of restoration, in terms uted to an increase in herbaceous, predominantly alien,
both of indigenous density and cover (Figs. 1 & 2). plant recruitment.
However, there was little benefit in this treatment after Although signs of stress (red leaf coloration) and die-
the first year. The grass Cynodon dactylon, which was back in proteoid shrubs were observed following the
sown as part of the commercial seed mix, did not ger- ultra-slow-release fertilizer application, no negative
minate well, and its establishment was poor. The peren- long-term impacts on total plant density or cover were
nial forbs, sown as part of the commercial seed mix, found. In deep topsoil plots, proteoid mortality was sig-
contributed little to plant density or cover. nificantly higher in fertilized plots (␹2 ⫽ 17.4, p ⬍ 0.0001),
At the six-month census, seed treatment had no effect but there was no such effect in the other two soil treat-
on herbaceous alien density: 97% of seedlings derived ments.
from the topsoil or seed rain. The significant effect of Fertilizer addition caused little change to soil nutrient
seed treatment on alien density at 18 months (Table 1) status and pH measured ten months after the ultra-
relates to an indirect effect on open space for recoloni- slow-release fertilizer application (Table 2). Resistance
zation by these aliens. decreased significantly in fertilized untopsoiled plots,

76 Restoration Ecology MARCH 2001

 Shrubland Restoration Following Alien Invasion and Mining

Table 2. Soil pH, resistance, and nutrient concentrations 10 months after fertilizer application in fertilized and unfertilized plots at
the restoration trial and in neighboring undisturbed soil under pines (mean ⫾ SD, n ⫽ 5, except for undisturbed soil n ⫽ 3). Units
are ohms (resistance), mg/kg (Na, P, K), me% (Ca, Mg), and % (C). Results of 1-way ANOVAs and post-tests for unfertilized soil
are presented.
ANOVA
Topsoil Undisturbed (Unfertilized Only)
Depth (Pines) 0 mm 0 mm 100 mm 100 mm 300 mm 300 mm F(3,17)
Fertilizer No No Yes No Yes No Yes Tukey-Kramer

pH 4.4 ⫾ 0.1 4.8 ⫾ 0.3 4.5 ⫾ 0.4 4.7 ⫾ 0.2 4.6 ⫾ 0.1 4.5 ⫾ 0.1 4.6 ⫾ 0.2 1.79ns
Resistance 1,840 ⫾ 125 1,986 ⫾ 838 882 ⫾ 203a 2,540 ⫾ 832 2,266 ⫾ 369 3,910 ⫾ 499 2,128 ⫾ 654 8.60** pines,0,
100 ⬍ 300
Na 78 ⫾ 19 72 ⫾ 28 86 ⫾ 15 67 ⫾ 30 44 ⫾ 10 33 ⫾ 4 36 ⫾ 12 3.46*
P 3.3 ⫾ 0.6 2.0 ⫾ 0.7 1.2 ⫾ 0.4 2.8 ⫾ 1.6 3.8 ⫾ 1.6 2.4 ⫾ 0.9 3.2 ⫾ 1.1 1.05ns
K 59 ⫾ 16 25 ⫾ 6 28 ⫾ 7 33 ⫾ 18 46 ⫾ 16 40 ⫾ 11 41 ⫾ 10 4.40* 0 ⬍ pines
Ca 4.4 ⫾ 1.6 0.7 ⫾ 0.2 0.7 ⫾ 0.1 3.0 ⫾ 0.6 2.7 ⫾ 0.7 3.1 ⫾ 0.8 2.4 ⫾ 0.8 14.6*** 0 ⬍ pines,
100,300
Mg 1.3 ⫾ 0.5 0.4 ⫾ 0.1 0.4 ⫾ 0.1 0.9 ⫾ 0.2 0.8 ⫾ 0.2 0.9 ⫾ 0.2 0.7 ⫾ 0.3 7.97*** 0 ⬍ pines,
100,300
C 2.2 ⫾ 0.4 0.4 ⫾ 0.1 0.5 ⫾ 0.1 1.6 ⫾ 0.8 2.0 ⫾ 0.5 1.8 ⫾ 0.7 2.1 ⫾ 0.4 6.57*** 0 ⬍ pines,
100,300
a Significantly different to unfertilized equivalent (p ⬍ 0.05, Welch’s approximate t-test); ns ⫽ not significant. *p ⬍ 0.05. **p ⬍ 0.01. *** p ⬍ 0.001.

but no significant changes were detected for any other soiled plots, both from seed drift during the initial sow-
measured soil variable. ing and localized dispersal of early-maturing species.
Soil variable comparisons between the unfertilized There was no significant difference in richness among
plots and the undisturbed soil around the restoration site seeded, topsoil plots at quadrat (1 m2) or plot (16 m2)
(currently under pines) indicated significantly higher scales (Table 3). Among topsoil plots, seed controls had
resistance for deep topsoil samples, but no other differ- the lowest richness; on average, seed treatments con-
ences. Subsoil from the untopsoiled plots was lower in tributed an additional three species per 1 m2 or five spe-
many essential plant nutrients (e.g., K, Ca, Mg) and carbon cies per plot. Topsoil contributed a minimum 60%, and
than the undisturbed soil and topsoil plots (Table 2). the seed mix 15%, of all species recorded at the restora-
tion trial site (Table 4).
Vegetation structure differed between untopsoiled
Effects of Treatments on Species Composition and Guild Structure
and topsoil plots and between control and seeded treat-
Species richness was significantly lower in untopsoiled ments. Density, richness, and cover of all growth forms
compared to topsoil plots across all seed treatments were lower for untopsoiled plots (Fig. 4). Proteoid
(Table 3). Over the three-year monitoring period, an av- shrubs (with canopy-stored seeds) were virtually ab-
erage of ten species colonized the unseeded, untop- sent in control plots, as they do not recruit from soil

Table 3. Species richness at quadrat (1 m2) and plot (16 m2) scales in three-year old vegetation following
different topsoil and seeding treatments (mean ⫾ SD, n ⫽ 5); no seed addition (Control), fynbos mix (F), F
plus annuals (FA), F plus commercial seed (FC). Results of 1-way ANOVAs for seed treatments are
presented.
0 mm 100 mm 300 mm
Seed Treatment 1 m2 16 m2 1 m2 16 m2 1 m2 16 m2

Control 0.2 ⫾ 0.4a 10.2 ⫾ 3.6a 9.4 ⫾ 1.5a 25.8 ⫾ 4.7 9.2 ⫾ 2.4a 22.8 ⫾ 5.3
F 6.2 ⫾ 1.9b 16.6 ⫾ 2.9ab 13.4 ⫾ 1.3b 31.2 ⫾ 5.9 12.6 ⫾ 1.1ab 30.0 ⫾ 3.7
FA 6.0 ⫾ 1.4b 16.2 ⫾ 3.8ab 11.8 ⫾ 0.4ab 30.2 ⫾ 3.4 12.6 ⫾ 1.3ab 30.0 ⫾ 6.7
FC 6.2 ⫾ 1.3b 17.0 ⫾ 0.7ab 11.0 ⫾ 1.0ab 29.2 ⫾ 3.9 11.8 ⫾ 2.9ab 29.0 ⫾ 2.0
F ⫹ fertilizer 4.6 ⫾ 1.3b 16.6 ⫾ 4.3ab 12.8 ⫾ 1.1b 27.4 ⫾ 5.6 11.6 ⫾ 1.5ab 28.8 ⫾ 5.8
FA ⫹ fertilizer 6.6 ⫾ 2.9b 18.4 ⫾ 4.2b 13.0 ⫾ 1.9b 32.0 ⫾ 5.0 14.0 ⫾ 2.1b 30.0 ⫾ 4.8
FC ⫹ fertilizer 5.8 ⫾ 1.9b 16.6 ⫾ 4.8ab 13.8 ⫾ 2.6b 35.2 ⫾ 4.4 13.6 ⫾ 2.1b 27.8 ⫾ 4.2
F(6,28) 8.34*** 2.54* 5.05** 2.10ns 3.12* 1.41ns
ns ⫽ not significant. *p ⬍ 0.05. **p ⬍ 0.01. ***p ⬍ 0.00. Within columns different letter superscripts denote a significant difference (Tukey-

Kramer, p ⬍ 0.05).

MARCH 2001 Restoration Ecology 77

Shrubland Restoration Following Alien Invasion and Mining

Table 4. The number of indigenous species* recorded at the had a relatively low proportion of shrub (especially eri-
restoration trial site by growth form and propagule source. The coid) cover and a relatively high proportion of grami-
third column includes species deriving from both the soil and noid cover compared to deep topsoil plots. Species with
seed mix, and those species which could not be assigned with soil-stored seeds, especially ericoid shrubs and those
certainty to either source (mostly ephemeral species which had
dispersed by ants, were under-represented in untop-
died back before species counts were done in plots).
soiled plots. Short-lived seeder species with wind-dis-
Source Soil Seed Mix Both/Unknown Total persed seeds and insect-pollinated flowers typified the
Proteoid shrub 0 6 0 6 untopsoiled plots. Among seed treatments on topsoil
Ericoid shrub 24 5 5 34 plots, unseeded controls had a relatively low propor-
Other shrub 22 6 5 33 tion of graminoid cover and a relatively high propor-
Graminoid 20 2 8 30 tion of shrub (especially ericoid) cover. On topsoil plots,
Forb 4 2 17 23 the guild structure of unseeded controls was character-
Geophyte 15 0 1 16
Total 85 21 36 142 ized by an over-representation of species with soil-
stored seeds, particularly shrubs with ericoid leaves
* Species list available from author. and passive or ballistic seed dispersal.
Restored plant communities on seeded topsoil plots
resembled Proteoid Fynbos in general structure, with a
seed banks. In topsoil plots, richness was significantly well-developed proteoid shrub overstory layer and
lower for control than sown plots for all growth forms mixed ericoid shrub-graminoid understory layer at
except geophytes (Fig. 4; Table 5). Richness of growth- three years (Fig. 5).
forms did not differ among the different seeded treat-
ments (Table 5). Graminoid density (predominantly
Costs and Benefits of the Various Treatments
Poaceae and Cyperaceae) was higher in the fynbos-
plus-annual treatment in the untopsoiled and 100 mm The costs incurred in the restoration trial were extrapo-
topsoil plots (Table 5). By three years, geophytes and lated to give general estimates for different aspects of
forbs contributed insignificant cover (Fig. 4). the restoration program (Table 6). Costs of initial alien
Analysis of the different native guild components in- tree clearing were excluded, as this would be a prereq-
dicated significant differences among topsoil depth and uisite before any mining or vegetation restoration oper-
seed treatments for all the guilds tested (Appendix 1). ations. Earthmoving costs also were excluded, as this
Among topsoil depth treatments, shallow topsoil plots forms part of general mining operation costs.

Table 5. ANOVA and post-test results by growth-form guild for density (plants/m2) at 6 months, richness (species/m2) at 18
months, and projected canopy cover (%) at 36 months; no seed addition (Con), fynbos mix (F), F plus annuals (FA), F plus
commercial seed (FC).
Tukey-Kramer Tukey-Kramer Tukey-Kramer
Guild F (p ⬍ 0.05) F (p ⬍ 0.05) F (p ⬍ 0.05)

Density F(df: 3,79) 0 mm topsoil 100 mm topsoil 300 mm topsoil

Proteoid 38.2*** Con ⬍ F,FA,FC 60.9*** Con ⬍ F,FA,FC 53.5*** Con ⬍ F,FA,FC
Ericoid 16.5*** Con ⬍ F,FA,FC 5.9** Con ⬍ F 5.7** Con ⬍ FA,FC
Graminoid 20.8*** Con ⬍ F,FA,FC; F⬍FA; FC⬍FA 6.4*** Con ⬍ FA; F⬍FA 3.4* Con ⬍ FC
Forb 11.8*** Con ⬍ F,FA,FC 9.6*** Con ⬍ F,FA,FC 7.7*** Con ⬍ FA,FC
Geophyte 10.8*** Con ⬍ F,FC; FA⬍F,FC 2.4ns 1.2ns
Richness F(df: 3,79)
Proteoid 26.5*** Con ⬍ F,FA,FC 42.6*** Con ⬍ F,FA,FC 55.0*** Con ⬍ F,FA,FC
Ericoid 5.70** Con ⬍ FC 3.40* F ⬍ FA 3.4* Con ⬍ FA
Other shrub 3.36* Con ⬍ FC 2.74* 4.6** Con ⬍ FC
Graminoid 12.2*** Con ⬍ F,FA,FC; F⬍FA 5.19** Con,F ⬍ FA 4.6** Con ⬍ FA
Forb 9.9*** Con ⬍ F,FA,FC 13.4*** Con ⬍ F,FA,FC; F⬍FA 7.5*** Con ⬍ F,FA,FC
Geophyte 10.2*** Con ⬍ F,FA,FC 1.12ns 2.2ns
Cover F(df: 3,19)
Proteoid 2.76ns 5.04* Con ⬍ F,FA 7.73** Con ⬍ FA,FC
Ericoid 4.75* Con ⬍ F,FC 1.27ns 1.29ns
Other shrub 2.66ns 0.31ns 0.73ns
Graminoid 4.6* Con ⬍ FA,FC 0.66ns 1.54ns
Forb 0.92ns 0.18ns 0.94ns
ns ⫽ not significant. *p ⬍ 0.05. **p ⬍ 0.01. *** p ⬍ 0.001.

78 Restoration Ecology MARCH 2001

 Shrubland Restoration Following Alien Invasion and Mining

useful in accelerating restoration in untopsoiled areas,

via the promotion of herbaceous species and the initia-
tion of nutrient-cycling processes. Woody alien species
recruited well and would have formed a dense stand
overtopping the fynbos vegetation, had a follow-up
program not been implemented. Regular follow-up
control of woody alien species is therefore an essential
component of restoring fynbos vegetation.

Discussion

Despite the complete absence of fynbos vegetation be-

fore the initiation of restoration, it has been possible to
restore topsoiled areas to a locally species-rich and
Figure 5. General view of the Noordhoek restoration trial site, structurally representative fynbos community. Casual
with three-year-old restored Proteoid Fynbos vegetation in observations at the trial site indicate that the indigenous
the foreground and alien invasive vegetation in the back- fauna also is returning, including reptiles, amphibians,
ground (Pinus pinaster dominant). and invertebrates.
The trial has demonstrated that restoration succeeds if
an ecologically sound conceptual framework is followed
By 1996 the fungus (Uromycladium tepperianum) was (Holmes & Richardson 1999). In most systems, key fac-
widespread on Acacia saligna around the restoration site, tors contributing to a conceptual framework include the
and by 1998 the infected plants were heavily galled. As a processes leading to degradation, and the recruitment
result of the successful spread of this biocontrol agent, dynamics, community structure, ecosystem function,
the reproductive output of this alien and the amount of and disturbance regime of the native ecosystem (Hobbs
follow-up control required are expected to decline. & Norton 1996; Primack 1996; Dobson et al. 1997; Palmer
The most cost-effective way of promoting fynbos res- et al. 1997; Holmes & Richardson 1999). Invasive alien
toration was to apply the general fynbos seed mix trees had completely altered the structure and function-
which included the major structural component miss- ing of the vegetation at the site. By removing the alien
ing from the soil-stored seed banks (overstory proteoid trees and controlling their recruitment, fynbos species
shrubs). The fynbos annual supplement added a small were able to re-establish, predominantly from in situ per-
benefit to restoration during the first year on untop- sistent seed banks, but also from sown seed.
soiled and shallow topsoil plots, but incurred nearly Recruitment in fynbos vegetation is almost entirely
twice the cost in seed-collecting time. Fertilizer pro- coupled to fire (Le Maitre & Midgley 1992), and germi-
vided no benefit in deep topsoil areas. However, it was nation cues are directly or indirectly linked to fire (Kee-

Table 6. Costs and benefits associated with various treatments applied in the restoration trial. Benefits are
rated in a relative way: N ⫽ none or detrimental, ⫹ small benefit, ⫹⫹ moderate benefit, ⫹⫹⫹ large benefit.
Cost
Treatment man-days/ha US$/ha Benefit

Site preparation (earthworks excluded, as it is part of mining operation)

Biocontrol a – 167 ⫹⫹
Seed mixes
Fynbos mix 40 – ⫹⫹⫹
Fynbos annuals 30 – ⫹
Commercial seed – 448 N
Fertilizer
Fish emulsion ⬍1 146 N
Horticote ⬍1 1,587 ⫹b
Follow-up alien control
Pine & Hakea 7.5c – ⫹⫹⫹
Acacia & Eucalypt 27.5 – ⫹⫹⫹
a Rust fungus Uromycladium tepperianum which was inoculated onto Acacia saligna in September 1994. Biological control generally works

by lowering the reproductive output of the aliens and ultimately the amount of follow-up control required.
b Of small benefit to untopsoiled areas only.
c Would have been unnecessary had mulch been prepared without pine and hakea seed.

MARCH 2001 Restoration Ecology 79

Shrubland Restoration Following Alien Invasion and Mining

ley & Bond 1997). By clearing the site in late summer pose a real threat to the conservation of biodiversity
and giving suction-harvested seed a smoke treatment (Rebelo 1992b). This study has demonstrated that fyn-
before sowing, many of the natural post-fire conditions bos areas supporting dense aliens for three decades
were simulated. Smoke and smoke extracts have been may be restored to native vegetation; therefore, such ar-
shown to stimulate germination in a wide range of spe- eas should never be written-off for nature conservation
cies from fynbos and other mediterranean-climate purposes. Although many fynbos species have long-
shrublands (De Lange & Boucher 1990; Dixon et al. term persistent seed banks (Holmes & Cowling 1997a,
1995; Brown & Van Staden 1997; Keeley & Fothering- 1997b) which can survive average-to-long fire intervals
ham 1998). Smoke treatment has been used successfully (15–45 years, van Wilgen et al. 1992), these seed banks
in stimulating recruitment in fire-prone vegetation fol- eventually decline in the absence of seed rain. The in-
lowing bauxite mining in Western Australia (Roche et tensity of restoration intervention required is propor-
al. 1997). The only natural germination cue omitted was tional to the duration and the number of fire-cycles of
a heat pulse, which occurs during fire and is known to dense alien invasion, and invaded sites should be cleared
stimulate some species with soil-stored seeds (Jeffery et as soon as possible. In order to maximize recruitment
al. 1988; Musil & de Witt 1991; Kilian & Cowling 1992). from persistent fynbos seed banks, clearing treatments
Seeds of these refractory species could persist in the should be planned to maximize germination of soil-
dormant state at the trial site. stored seeds. The risk of a damaging, high-intensity fire
Despite three decades of dense pine invasion, the soil through dense, alien slash should be minimized either
seed banks contributed 60% of all species recruiting, by removing large, felled biomass or by burning the
representing 70% of species present in three-year-old aliens standing. In areas of extensive alien invasion, seed
restored vegetation. Many fynbos species have long- of local serotinous species should be sown after burning
term, persistent seed banks (Holmes & Cowling 1997a, the site. Regular follow-up control is essential to pre-
1997b), and this result emphasizes the important contri- vent the reestablishment of an alien stand.
bution of topsoil in restoring such vegetation (Putwain &
Gillham 1990; Ward et al. 1997; Holmes & Richardson
1999). The relatively high level of recruitment recorded Implications for Restoration After Mining and Construction
in topsoiled, unseeded areas suggests that soil distur-
bance warrants further investigation as a method of Can Fynbos Be Restored on Untopsoiled Substrata? The few
stimulating germination. species that established on subsoil derived from the
Restored community structure and composition may seed mix. Those sown species which established rela-
be compared to that of a native community to assess tively consistently (e.g., Protea lepidocarpodendron, Berze-
whether restoration objectives have been met (Dobson lia lanuginosa, and Athanasia crithmifolia) are able to
et al. 1997; Palmer et al. 1997; Holmes & Richardson grow in heavier soil.
1999). The aim of returning Proteoid Fynbos to the site Plant cover and vegetation development were re-
was attained on topsoil plots. All major functional tarded in untopsoiled plots, indicating that restoration of
groups were also present, although resprouting guilds a fully diverse and functional cover of fynbos on subsoil
may be under-represented compared to undisturbed or mine overburden substrata is unlikely to be achieved
native communities. Structural and functional composi- in the short-term. In addition to providing soil-stored
tion is relevant in assessing most restoration outcomes seed banks, topsoil has better water-absorbing and hold-
in floristically complex regions such as the Cape Floris- ing capacity than subsoil, and contains important plant
tic Kingdom. The extremely high turnover in species growth nutrients, and the micro-flora and fauna essential
along environmental and geographical gradients (Cowl- for nutrient-cycling processes (Brady 1974). Thus, even a
ing et al. 1992; Simmons & Cowling 1996) renders floris- thin layer of topsoil would greatly ameliorate subsoil
tic comparisons between restored and native communi- conditions by enabling nutrient-cycling processes to be
ties problematic. initiated. Proteaceae survival and growth were lower on
a rehabilitated sand mine than in nearby scrub heath
communities in Western Australia. A major problem at
Implications for Restoring Alien-Invaded Sites
the rehabilitation site was impedance between surface
The transformation of fynbos vegetation to dense and subsurface soils, which prevented normal taproot
stands of alien trees is analogous to the conversion of development (Enright & Lamont 1992).
native forests to agricultural and then degraded land: If possible, some topsoil should be spread over the
without intervention, an “extinction debt” of species is site. If topsoil is not available, subsoil or overburden
created (Dobson et al. 1997). Because of the localized should be ripped to reduce compaction and increase wa-
distribution of many rare and endemic species in the ter penetration. Ecosystem processes should be initiated
Cape Floristic Kingdom (Cowling et al. 1992), invasions by applying mulch and sowing non-invasive, fast-grow-

80 Restoration Ecology MARCH 2001

 Shrubland Restoration Following Alien Invasion and Mining

ing annuals and indigenous legumes and hardy peren- During the first year, on both untopsoiled and shal-
nial species. After germination, organic or low concen- low topsoil plots, the fynbos-plus-annual seed mix in-
trations of ultra-slow-release fertilizer should be applied creased the density of graminoids compared to the
to improve the soil nutrient status and promote the de- general fynbos mix. No other growth-form differed
velopment of plant cover. In this way primary succes- among the seed treatments on any of the substrata. In
sion may be accelerated (Dobson et al. 1997). Inorganic most situations a general fynbos mix containing the lo-
fertilizers that contain high levels of phosphate should cal serotinous species and a variety of non-serotinous
be avoided, as fynbos species are particularly sensitive species adequately complements the soil-stored seed
to this nutrient (Stock & Allsopp 1992). Alien herba- bank in restoring fynbos diversity, structure, and
ceous species are promoted by fertilizer addition, but cover. In situations where it is essential to have rapid
should decline once a dense perennial cover develops. development of plant cover (for example, if topsoil is
of poor quality or soil erosion is a threat), it may be ap-
Is Topsoil Depth an Important Factor in Restoring Fynbos? propriate to collect additional seed from a fire-belt
Shallow and deep topsoil treatment differences were of area to boost the graminoid component. However, the
a much lower magnitude than those between topsoiled result may not justify the additional expense of doing
and untopsoiled treatments for all the measured vari- this: nearly twice the seed-collecting time. The results
ables. Thus, replacing a little topsoil is far better than no indicate a high correlation between plant density and
topsoil, when restoring a site to fynbos. cover. Therefore, to ensure good initial cover, it is nec-
Plant density was generally higher in the shallow essary to sow at a reasonably high rate (40–70 seeds/m2,
topsoil treatment, which suggests that a larger propor- depending on the predicted soil-stored component).
tion of its soil seed bank was stimulated to germinate Seed of the commercially available cultivar of Cynodon
compared to the deep topsoil treatment. Differences in dactylon (Bermuda grass) failed to establish well and
density were more marked at 18 months than at six therefore did not fulfil its potential role of providing
months, indicating that there was more recruitment early plant cover.
during the second winter in shallow than in deep top- The importance of topsoil in successfully restoring
soil. This may relate to the lower canopy cover in shal- fynbos at the restoration trial site emphasizes the need
low topsoil, which could impact both seed germination to manage this resource extremely carefully. If possible,
and recruitment. topsoil should be spread on another area immediately
Plant canopy cover remained highest in deep topsoil after stripping to minimize the death of buried indige-
throughout the three-year trial. The deeper topsoil nous propagules. It has been shown that even short pe-
probably provided a more favorable growing environ- riods of topsoil stockpiling may result in death of plant
ment for the establishing plants, including a longer propagules (Hargis & Redente 1984).
growing season than shallow topsoil. Therefore, the
more topsoil returned to a restoration site, the faster Does Fertilizer Addition Promote Fynbos Establishment? On
will be the development of plant cover. However, care deep topsoil plots the addition of fertilizer increased
should be taken when stripping topsoil to ensure that mortality in proteoid shrubs that have sensitive cluster
the layer containing the viable seed bank (0–100 mm in roots, and imparted no benefit in terms of increased
fynbos) is stockpiled separately from deeper layers be- plant cover. Vegetation cover increased following fertil-
fore respreading, to prevent the seed bank being di- izer addition in untopsoiled and shallow topsoil plots.
luted (Putwain & Gillham 1990). This increase in plant growth, coupled with a lack of
nutrient enrichment in the soil, suggests that nutrients
Which Seed Mix Best Restores Fynbos on the Different Sub- were assimilated by the vegetation. As most of the plant
strata? Seed addition is essential for any indigenous re- cover increase was attributable to alien herbaceous spe-
cruitment to occur on untopsoiled sites, because subsoil cies, fertilization is not recommended as a means of
and overburden are devoid of native seeds. On topsoil promoting fynbos establishment where topsoil is avail-
plots, however, the seed mixes improved the richness able. However, in areas devoid of topsoil, such as
and structural composition of the vegetation. Proteoid eroded areas or soil slips, the application of ultra-slow-
shrubs form an overstory layer in many fynbos commu- release fertilizer, together with mulch and indigenous
nities and are an important structural component of the seed, may accelerate nutrient-cycling processes and the
vegetation (Cowling & Holmes 1992). Furthermore, this re-establishment of vegetation cover.
overstory layer has been shown to maintain community
diversity in the long-term (Vlok 1996). As proteoids are
Future Management
predominantly serotinous, and are eliminated by alien
invasions and mining operations, they must be sown To ensure a sustainable ecosystem in the long-term, it is
back into the community. important to reinstate the native disturbance regime. To

MARCH 2001 Restoration Ecology 81

Shrubland Restoration Following Alien Invasion and Mining

do this, invasive plants must be controlled and the site Dobson, A. P., A. D. Bradshaw, and A. J. M. Baker. 1997. Hopes
burned on a 15–45 year cycle (depending on local con- for the future: restoration ecology and conservation biology.
Science 277:515–522.
ditions; van Wilgen et al. 1992). Without burning, the Enright, N. J., and B. B. Lamont. 1992. Survival, growth and water
vegetation eventually will become moribund and a po- relations of Banksia seedlings on a sand mine rehabilitation
tential fire hazard. If fire is excluded, secondary succes- site and adjacent scrub-heath sites. Journal of Applied Ecol-
sion to coastal thicket or Afromontane forest could oc- ogy 29:663–671.
cur (Cowling & Holmes 1992). However, biodiversity Haberman, S. J. 1973. The analysis of residuals in cross-classified
tables. Biometrics 29:203–229.
would be lost in this process and it is recommended Hargis, N. E., and E. F. Redente. 1984. Soil handling for surface
that areas successfully restored to fynbos be managed mine reclamation. Journal of Soil & Water Conservation 39:
for fynbos conservation. Burning the site also would 300–305.
provide a test of the ecological sustainability of the res- Hobbs, R. J., and D. A. Norton. 1996. Towards a conceptual
toration in the long-term (Palmer et al. 1997). framework for restoration ecology. Restoration Ecology 4:
93–110.
Holmes, P. M., and R. M. Cowling. 1997a. Diversity, composition
Acknowledgments and guild structure relationships between soil-stored seed
banks and mature vegetation in alien plant-invaded South
I wish to thank Janet Barker of the Environmental Evalua- African fynbos shrublands. Plant Ecology 133:107–122.
tion Unit at the University of Cape Town for facilitating Holmes, P. M., and R. M. Cowling. 1997b. The effects of invasion
by Acacia saligna on the guild structure and regeneration ca-
the study, and John Butterworth of Serina Mines Pty. Ltd. pabilities of fynbos shrublands. Journal of Applied Ecology
for supporting the research, arranging the site prepara- 34:317–332.
tion and research finance, and monitoring rainfall at the Holmes, P. M., and D. M. Richardson. 1999. Protocols for restora-
site. John Jordaan, Fiona Powrie, and Lee Jones gave ad- tion based on recruitment dynamics, community structure
vice on seed collecting, seed storage, and fertilizer addi- and ecosystem function: perspectives from South African
fynbos. Restoration Ecology 7:215–230.
tion, and Jaana Ball and Fatima Parker assisted with mon- Jeffery, D. J., P. M. Holmes, and A. G. Rebelo. 1988. Effects of dry
itoring. I thank Tony Rebelo, Dave le Maitre, and an heat on seed germination in selected indigenous and alien le-
anonymous reviewer for commenting on the manuscript. gume species in South Africa. South African Journal of Bot-
any 54:28–34.
LITERATURE CITED Keeley, J. E., and W. J. Bond. 1997. Convergent seed germination
in South African fynbos and Californian chaparral. Plant
Anonymous. 1997. White paper on the conservation and sustain- Ecology 133:153–167.
able use of South Africa’s biological diversity. South African Keeley, J. E., and C. J. Fotheringham. 1998. Smoke-induced germi-
Government Gazette. Vol. 385, Number 18163. nation in California chaparral. Ecology 79:2320–2336.
Brady, N. C. 1974. The nature and properties of soils. 8th edition. Kilian, D., and R. M. Cowling. 1992. Comparative seed biology
Macmillan Publishing Co. Inc., New York. and co-existence of two fynbos shrub species. Journal of
Brown, N. A. C. 1993. Promotion of germination of fynbos seeds Vegetation Science 3:637–646.
by plant-derived smoke. New Phytologist 123:575–583. Le Maitre, D. C., and J. J. Midgley. 1992. Plant reproductive ecol-
Brown, N. A. C., and J. Van Staden. 1997. Smoke as a germination ogy. Pages 135–174 in R. M. Cowling, editor. The ecology of
cue: a review. Plant Growth Regulation 22:115–124. fynbos: nutrients, fire and diversity. Oxford University
Cowling, R. M., editor. 1992. The ecology of fynbos: nutrients, fire Press, Cape Town.
and diversity. Oxford University Press, Cape Town. Le Maitre, D. C., B. W. van Wilgen, R. A. Chapman, and D. H.
Cowling, R. M., and P. M. Holmes. 1992. Flora & vegetation. Pages McKelly. 1996. Invasive plants and water resources in the
23–61 in R. M. Cowling, editor. The ecology of fynbos: nutri- Western Cape Province, South Africa: modelling the conse-
ents, fire and diversity. Oxford University Press, Cape Town. quences of a lack of management. Journal of Applied Ecol-
Cowling, R. M., P. M. Holmes, and A. G. Rebelo. 1992. Plant di- ogy 33:161–172.
versity and endemism. Pages 62–112 in R. M. Cowling, edi- Levitt, M. 1997. Putting fynbos back: the rehabilitation of Du
tor. The ecology of fynbos: nutrients, fire and diversity. Ox- Toitskloof. Veld & Flora 83:81–83.
ford University Press, Cape Town. Morris, M. J. 1999. The contribution of the gall-forming rust fun-
Cowling, R. M., P. W. Rundel, B. B. Lamont, M. K. Arroyo, and M. gus Uromycladium tepperianum (Sacc.) McAlp. to the biologi-
Arianoutsou. 1996. Plant diversity in mediterranean-climate cal control of Acacia saligna (Labill.) Wendl. (Fabaceae) in
regions. Trends in Ecology & Evolution 11:362–366. South Africa. African Entomology Memoir No. 1:125–128.
Cowling, R. M., D. M. Richardson, and P. J. Mustart. 1997. Fyn- Musil, C. F., and D. M. de Witt. 1991. Heat stimulated germina-
bos. Pages 99–130 in R. M. Cowling, D. M. Richardson, and tion in two Restionaceae species. South African Journal of
S. M. Pierce, editors. Vegetation of southern Africa. Cam- Botany 57:175–176.
bridge University Press, Cambridge. Palmer, M. A., R. F Ambrose, and N. L. Poff. 1997. Ecological the-
De Lange, J. H., and C. Boucher. 1990. Autecological studies on ory and community restoration ecology. Restoration Ecology
Audouinia capitata (Bruniaceae). I. Plant-derived smoke as a 5:291–300.
germination cue. South African Journal of Botany 56:700–703. Primack, R. B. 1996. Lessons from ecological theory: dispersal, es-
Dixon, K. W., S. Roche, and J. S. Pate. 1995. The promotive effect of tablishment and population structure. Pages 209–233 in D.
smoke derived from burnt native vegetation on seed germina- A. Falk, C. I. Millar, and M. Olwell, editors. Restoring diver-
tion of Western Australian plants. Oecologia 101:185–192. sity: strategies for reintroduction of endangered plants. Is-
Dixon, W. J. 1992. BMDP statistical software manual. University land Press, Washington D. C.
of California Press, Berkeley. Putwain, P. D., and D. A. Gillham. 1990. The significance of the

82 Restoration Ecology MARCH 2001

 Shrubland Restoration Following Alien Invasion and Mining

dormant viable seed bank in the restoration of heathlands. ogy of fynbos: nutrients, fire and diversity. Oxford Univer-
Biological Conservation 27:1–16. sity Press, Cape Town.
Rebelo, A. G. 1992a. Preservation of biotic diversity. Pages 309–344 Trinder-Smith, T. H., R. M. Cowling, and H. P. Linder. 1996. Pro-
in R. M. Cowling, editor. The ecology of fynbos: nutrients, filing a besieged flora: endemic and threatened plants of the
fire and diversity. Oxford University Press, Cape Town. Cape Peninsula, South Africa. Biodiversity and Conserva-
Rebelo, A. G. 1992b. Red Data Book species in the Cape Floristic tion 5:575–589.
Region: threats, priorities and target species. Transactions of van Wilgen, B. W., W. J. Bond, and D. M. Richardson. 1992. Eco-
the Royal Society of South Africa 48:55–86. system management. Pages 345–371 in R. M. Cowling editor.
Rebelo, A. G. 1996. Fynbos biome. Pages 62–74 in A. B. Low and The ecology of fynbos: nutrients, fire and diversity. Oxford
A. G. Rebelo, editors. Vegetation of South Africa, Lesotho University Press, Cape Town.
and Swaziland. Department of Environmental Affairs and van Wilgen, B. W., P. R. Little, R. A. Chapman, A. H. M. Gorgens,
Tourism, Pretoria. T. Willems, and C. Marais. 1997. The sustainable develop-
Richardson, D. M, I. A. W. Macdonald, P. M. Holmes, and R. M. ment of water resources: history, financial costs, and benefits
Cowling. 1992. Plant and animal invasions. Pages 271–308 in of alien plant control programmes. South African Journal of
R. M. Cowling, editor. The ecology of fynbos: nutrients, fire Science 93:404–411.
and diversity. Oxford University Press, Cape Town. Vlok, J. H. J. 1996. The role of overstorey proteoid shrubs in main-
Richardson, D. M, I. A. W. Macdonald, J. H. Hoffmann, and L. taining species richness in a southern Cape mountain fynbos
Henderson. 1997. Alien plant invasions. Pages 535–570 in R. community. M.Sc. Thesis. University of Natal, Pietermar-
M. Cowling, D. M. Richardson, and S. M. Pierce, editors. itzburg, South Africa.
Vegetation of southern Africa. Cambridge University Press, Ward, S. C., J. M. Koch, and C. D. Grant. 1997. Ecological aspects
Cambridge. of soil seed-banks in relation to bauxite mining. I. Unmined
Roche, S., J. M. Koch, and K. W. Dixon. 1997. Smoke enhanced jarrah forest. Australian Journal of Ecology 22:169–176.
seed germination for mine rehabilitation in the Southwest of Witkowski, E. T. F. 1991. Growth and competition between seed-
Western Australia. Restoration Ecology 5:191–203. lings of Protea repens (L.)L. and the alien invasive Acacia sali-
Romoff, N. 1986. Revegetation of disturbed areas in the fynbos gna (Labill.)Wendl. in relation to nutrient availability. Func-
biome – the current status. Veld and Flora 72:46–48. tional Ecology 5:101–110.
Simmons, M. T., and R. M. Cowling. 1996. Why is the Cape Pen- Wood, J., A. B. Low, J. S. Donaldson, and A. G. Rebelo. 1994.
insula so rich in plant species? An analysis of the indepen- Threats to plant species diversity through urbanization and
dent diversity components. Biodiversity & Conservation 5: habitat fragmentation in the Cape Metropolitan Area, South
551–573. Africa. Pages 259–274 in B. J. Huntley, editor. Botanical di-
Stock, W. D., and N. Allsopp. 1992. Functional perspective of eco- versity in southern Africa. Strelizia 1, National Botanical In-
systems. Pages 241–259 in R. M. Cowling, editor. The ecol- stitute, Pretoria, South Africa.

MARCH 2001 Restoration Ecology 83

Shrubland Restoration Following Alien Invasion and Mining

Appendix 1. The association between biological guilds and (a) topsoil depth and (b) seed treatments in three-year-old restored
fynbos. The relative percentages of each guild, based on cover values, are listed with adjusted standardized deviates (ASD) in
parentheses. ASD absolute values exceeding 3.0 indicate cells with unusually large deviations from the expected. Seed treatments
are: Control (no seed added), F (fynbos mix), FA (F & annuals), FC (F & commercial seed); Reseeder-sl is short-lived (⬍5 yr), -ll is
long-lived (⬎4 yr).

(a) Topsoil depth: all plots

Growth form ␹2 ⫽ 81*** Regeneration mode ␹2 ⫽ 221***
Geophyte Graminoid Forb Shrub Resprouter Reseeder-sl Reseeder-ll
0 mm 0.1 (1.0) 14.2 (⫺1.2) 0.3 (0.9) 85.4 (1.0) 0 mm 10.9 (⫺3.4) 12.6 (10.8) 76.6 (⫺2.9)
100 mm 0.1 (1.4) 19.5 (8.3) 0.3 (2.3) 80.1 (⫺8.6) 100 mm 19.0 (9.3) 5.3 (0.9) 75.7 (⫺8.8)
300 mm 0 (⫺2.0) 12.7 (⫺7.4) 0.1 (⫺2.8) 87.2 (7.9) 300 mm 12.1 (⫺7.1) 3.3 (⫺7.3) 84.6 (10.3)
Leaf size ␹2 ⫽ 781*** Seed storage mode ␹2 ⫽ 200***
Leptophyll Nanophyll Microphyll Mesophyll Canopy Soil None
0 mm 27.7 (⫺22.6) 32.5 (22.9) 28.4 (11.8) 11.4 (0) 0 mm 51.4 (13.4) 48.5 (⫺13.4) 0.1 (0.1)
100 mm 68.5 (5.1) 8.1 (⫺3.2) 12.3 (⫺3.7) 11.1 (⫺0.6) 100 mm 26.7 (⫺7.9) 73.1 (7.8) 0.1 (1.0)
300 mm 69.6 (8.7) 5.9 (⫺10.7) 12.8 (⫺3.5) 11.6 (0.6) 300 mm 31.3 (⫺0.3) 68.6 (0.4) 0.1 (⫺1.0)
Seed dispersal mode ␹2 ⫽ 294*** Pollination mode ␹2 ⫽ 141***
Wind Vertebrate Ant Ballistic Passive Bird Insect Wind
0 mm 59.6 (13.5) 0.1 (0.1) 1.7 (⫺15.2) 2.3 (0.8) 36.4 (⫺0.9) 0 mm 7.2 (⫺4.9) 77.9 (11.8) 14.8 (⫺9.3)
100 mm 35.7 (⫺4.5) 0.1 (1.0) 23.4 (3.4) 2.0 (0.4) 38.7 (1.5) 100 mm 12.8 (0.8) 58.0 (⫺2.1) 29.2 (1.8)
300 mm 36.7 (⫺3.6) 0.1 (⫺1.0) 24.1 (5.8) 1.8 (⫺0.9) 37.3 (⫺0.9) 300 mm 13.2 (2.2) 56.8 (⫺5.0) 30.0 (3.8)
(b) Seed treatments: topsoil plots only
Growth form ␹2 ⫽ 24*** Regeneration mode ␹2 ⫽ 17*
Geophyte Graminoid Forb Shrub Resprouter Reseeder-sl Reseeder-ll
Control 0.0 (0) 17.5 (⫺3.3) 0.9 (1.8) 81.6 (3.0) Control 18.4 (⫺2.0) 12.8 (0.1) 68.8 (1.7)
F 0.0 (0) 26.7 (2.5) 0.2 (⫺0.9) 73.1 (⫺2.4) F 23.6( 1.3) 12.4 (⫺0.3) 64.0 (⫺1.0)
FA 0.0 (0) 20.7 (⫺2.1) 0.3 (⫺0.6) 79.1 (2.2) FA 18.3 (⫺2.7) 13.3 (0.5) 68.4 (2.0)
FC 0.0 (0) 26.6 (2.3) 0.4 (0) 73.1 (⫺2.2) FC 25.8 (3.0) 12.4 (⫺0.3) 61.8 (⫺2.4)
Leaf size ␹2 ⫽ 227*** Seed storage mode ␹2 ⫽ 54***
Leptophyll Nanophyll Microphyll Mesophyll Canopy Soil None
Control 93.7 (12.6) 3.3 (⫺7.7) 2.5 (⫺6.2) 0.5 (⫺4.5) Control 7.2 (⫺5.9) 92.6 (5.9) 0.2 (0)
F 56.2 (⫺5.5) 25.5 (6.4) 14.9 (2.4) 3.5 (⫺2.6) F 16.3 (⫺0.5) 83.6 (0.6) 0.1 (⫺0.7)
FA 63.4 (⫺1.0) 11.7 (⫺4.1) 14.1 (1.7) 10.8 (6.5) FA 23.7 (5.8) 76.0 (⫺5.8) 0.3 (0.4)
FC 58.0 (⫺4.1) 23.0 (4.2) 13.6 (1.2) 5.4 (0) FC 16.5 (⫺0.3) 83.2 (0.2) 0.3 (0.3)

Seed dispersal mode ␹2 ⫽ 92*** Pollination mode ␹2 ⫽ 19**

Wind Vertebrate Ant Ballistic Passive Bird Insect Wind
Control 15.0 (⫺7.3) 0.2 (0) 22.4 (0.5) 10.5 (4.2) 51.8 (4.3) Control 14.3 (⫺0.5) 58.5 (3.2) 27.1 (⫺3.1)
F 33.7 (3.2) 0.1 (⫺0.7) 19.6 (⫺1.7) 4.8 (⫺1.9) 41.8 (⫺0.5) F 14.4 (⫺0.7) 49.9 (⫺1.1) 35.7 (1.7)
FA 29.2 (⫺0.2) 0.3 (0.4) 18.6 (⫺2.3) 6.3 (0.1) 45.7 (2.0) FA 15.9 (0.7) 53.4 (1.1) 30.7 (⫺1.8)
FC 33.6 (2.9) 0.3 (0.3) 26.2 (3.6) 5.0 (⫺1.6) 35.0 (⫺4.9) FC 15.5 (0.3) 47.6 (⫺2.6) 37.0 (2.5)

***p ⬍ 0.05. ***p ⬍ 0.01. ***p ⬍ 0.005.

84 Restoration Ecology MARCH 2001