Competitive interactions between grass and succulentshrubs at the ecotone between an arid grasslandand succulent shrubland in the Karoo

Ndafuda Shiponeni • Nicky Allsopp •

Peter J. Carrick • M. Timm Hoffman

Received: 26 January 2010 / Accepted: 9 November 2010 / Published online: 25 November 2010

� Springer Science+Business Media B.V. 2010

Abstract Nearest-neighbour analysis was used to

examine the competitive interactions between Stipa-

grostis brevifolia, a C4 perennial grass, and two leaf

succulent shrubs, Ruschia robusta and Leipoldtia

pauciflora, at the ecotone between semi-arid grass-

land and succulent shrubland in the Karoo. The root

distribution in the soil was also compared to assess

the degree of overlap in the potential use of soil

resources. Regressions between the combined sizes of

interspecific, nearest-neighbour species and the dis-

tance between them showed significant positive

correlations for S. brevifolia and R. robusta, which

suggest the presence of competition. We infer from

individual species regressions that the grass exerted a

stronger competitive force on the shrub R. robusta

than R. robusta on the grass. There was also evidence

for strong intraspecific competitive relationships

within S. brevifolia and R. robusta. There was no

evidence of competition between S. brevifolia and

L. pauciflora or among L. pauciflora individuals.

S. brevifolia had the deepest root system, and was

recorded at depths of 70 cm. Most of this root mass

occurred between 10 and 40 cm. Ruschia robusta

roots were recorded as deep as 55 cm, but more than

90% was found in the top 20 cm of the soil, creating a

degree of overlap with the vertical root distribution of

S. brevifolia. A clear separation in rooting depths

occurred between S. brevifolia, and L. pauciflora

which had only 3% of the total root mass below

10 cm. The partial overlap in the vertical root

distribution between S. brevifolia and R. robusta

may account for the observed competitive relation-

ship, but each species dominates in a different layer,

potentially minimising the net competition between

S. brevifolia and R. robusta. Our findings demonstrate

the possibility of a two-layer water-obtaining strategy

in a semi-desert ecosystem, where the succulent

shrubs seem to be playing the typical ‘‘grass’’ role

described in most models of water partitioning

between grass and woody plants. The stronger

competitive effect of S. brevifolia on R. robusta at

all the sites is of significance to species dynamics,

and might be related to winter/summer rainfall

dynamics at the climatic transition.

N. Shiponeni � P. J. Carrick � M. T. Hoffman

Plant Conservation Unit, Botany Department, University

of Cape Town, PO Box X3, Rondebosch 7701,

South Africa

Present Address:N. Shiponeni (&)

Department of Biological Sciences, University

of Namibia, Private Bag 13301, Windhoek, Namibia

e-mail: nshiponeni@unam.na

N. Allsopp

ARC-Livestock Business Division, Private Bag X17,

Bellville 7535, South Africa

Present Address:N. Allsopp

South African Environmental Observation Network,

SANBI, Private Bag X7, Claremont 7735, South Africa

123

Plant Ecol (2011) 212:795–808

DOI 10.1007/s11258-010-9864-0

Keywords Climatic ecotone � Nearest-neighbour

analysis � Plant competition � Resource partitioning �Root system � Vertical separation in rooting depth

Introduction

Species may segregate into different communities or

they may co-exist, according, in part to ccompetition

among them (Rosenzweig 1981; Scholes and Archer

1997; Gordon 2000). The principle of niche differ-

entiation through resource partitioning, both in space

and time is at the root of many competition theories

explaining co-existence between plants (Yeaton et al.

1977; Walker and Noy-Meir 1982; Knoop and

Walker 1985; Cody 1986; Sala et al. 1989). The

vegetation along the ecotone between Bushmanland

grassland and the succulent shrubland of Namaqua-

land, in the Karoo region of South Africa is

characterised by segregated grassland and leaf

succulent shrubland communities, but also commu-

nities in which grasses and succulent shrubs co-occur

(Fig. 1). While the distribution of perennial C4 grass

in the Karoo is generally associated with deeper and

sandy soils (Lloyd 1989; Carrick 2001), the grass

species S. brevifolia and certain leaf succulent shrub

species appear to break this segregation, forming

inter-mingled communities with no clear distinction

in soil properties (Shiponeni 2007), and creating

tension zones of possible competitive interactions at

the ecotone.

There is enough evidence in the Karoo to support

the occurrence of competition in arid and semi-arid

environments as observed among perennial shrubs

(Yeaton and Esler 1990; Esler and Cowling 1993;

Carrick 2003) and between perennial shrubs and

annuals (Cunliffe et al. 1990; Yeaton et al. 1993).

However, studies on competitive interactions

between grass and shrubs in general, and succulent

shrubs in particular are lacking in the Karoo. In the

eastern Karoo, where perennial grasses and dwarf

shrubs co-exist, Hoffman et al. (1990) have proposed

that the abundance of grasses regulates the abundance

of dwarf shrubs through competition. Species or

growth forms are also characterised by several

different patterns of root distribution resulting in

separation in the use of soil resources sufficient to

permit co-existence (see Fowler 1986). In the

Succulent Karoo, Carrick (2003) ascribed evidence

of competition between the two leaf succulent

Aizoaceae shrubs, Leipoldtia schultzei and Ruschia

robusta, and the absence of competition between

these shrubs and a non-succulent Asteraceae shrub

Hirpicium alienatum, to vertical separation in rooting

morphologies. The shallow nature of the root systems

of leaf succulent shrubs is well documented (Esler

and Rundel 1999; Midgley and van der Heyden 1999;

Carrick 2003), but the root systems of perennial grass

in the broader Karoo region have not been studied. It

is thus not clear how species within this growth form

interact with other growth forms such as succulent

shrubs, which together with non-succulent shrubs are

the most abundant components of these communities

(Cowling et al. 1994). Midgley and van der Heyden

(1999) have proposed that perennial grasses and

succulent shrubs might be competing for water in the

same vertical zone, due to the shallow-rooted nature

of succulent shrubs. Understanding how species or

growth forms interact can be useful in understanding

and predicting vegetation dynamics or responses

particularly at climatic transitions and under the

influence of grazing. In this study, we investigate the

nature of competitive interactions between semi-arid

Fig. 1 Subset of grass and shrubs distribution at the

transition zone between Bushmanland grasslands and Nam-

aqualand shrublands, showing tension zones of mixed grass/

shrubs communities (extracted from Shiponeni (2007), based

on nonparametric decision-tree classification using ASTER

images)

796 Plant Ecol (2011) 212:795–808

123

perennial grass S. brevifolia, and two leaf-succulent

shrubs, Ruschia robusta and Leipoldtia pauciflora

(Aizoaceae) as a way of understanding co-existence

and potential vegetation change across the grassland/

shrubland ecotone. Vertical distribution in root sys-

tems was also studied to explain the nature of

possible belowground interactions between the two

growth forms.

The occurrence of plant competition is often

inferred from studies of spatial patterns (Fowler

1986). In this study, evidence of competition was

inferred from the relationship between distance and

size of neighbouring plants. This technique, the

nearest-neighbour analysis, first described by Pielou

(1960), has been widely used to study plant interac-

tions in arid and semi-arid areas (Yeaton and Cody

1976; Yeaton et al. 1977; Welden et al. 1988; Yeaton

1990; Esler and Cowling 1993; Carrick 2003). It is

based on the premise that significant positive linear

correlations between the sum of neigouring plant

sizes, and the distance that separates them, indicates

competition, while the lack of such correlations is

interpreted as the absence of competition between

the two plants. Many studies have also compared the

distribution of roots in the soil in order to explain the

degree to which species are likely to interact (Cable

1969; Pelaez et al. 1994; Briones et al. 1996; Casper

and Jackson 1997; Nobel 1997; Carrick 2003).

Methods

Study area

The study was carried out at four sites (Burdensputs,

Kougoedvlakte, Vaalputs and Goegap Nature Reserve)

stretching a distance of approximately 80 km from the

farm, Burdensputs (30� 240S; 18� 340E) near Kliprand

in the southeast, up to Goegap Nature Reserve (29�420S; 17� 590E) near Springbok in the northwest along

the ecotonal border between Bushmanland and Nam-

aqualand, in the Northern Cape Province of South

Africa. The area forms part of a zone separating the

predominantly winter rainfall Succulent Karoo and the

predominantly summer rainfall Nama-Karoo biomes.

Winter rainfall is a consequence of coastal low

pressure systems and summer rain comes as convective

storms (Cannon 1924; Desmet and Cowling 1999).

Rainfall patterns at the sites are depicted in Fig. 2.

Goegap Nature Reserve to the most north-western part

of the study area has a rainfall peak from April to

September, and it forms part of the winter rainfall

Namaqualand (Rosch 2001), whereas the other sites

have no marked seasonal pattern in rainfall and appear

to receive both winter and summer rainfall.

The Burdensputs and Kougoedvlakte farms are

privately owned and are grazed by domestic livestock

using commercial management practices. Vaalputs

Radioactive Waste Disposal Facility comprises three

commercial farms which were taken out of produc-

tion when these became a nuclear waste disposal site

from 1983. It is currently stocked with small indig-

enous antelope species. The Goegap Nature Reserve

area has been under conservation since 1960, and is

also stocked with indigenous antelope species. As

with other semi-arid and arid regions, lack of

moisture results in less weathering and leaching

giving rise to soils that have weakly developed

structure with freely drained soils and little organic

matter (Lloyd 1985a, b; Watkeys 1999).

We focused on ecotonal communities, where both

grass and shrubs co-occur. The grass component is

represented by one of the Bushmanland perennial C4

grass species, S. brevifolia. The shrub species co-

occuring with grass at the ecotone belong to a group

of leaf succulent plants in the family Aiozaceae,

which formerly constituted a separate family, the

0

5

10

15

20

25

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Rai

nfal

l (m

m)

De Riet GoegapBurdensputs Vaalputs

*

Fig. 2 The average monthly rainfall at four sites where mixed

grass-succulent communities occur: Burdensputs (134 mm;

38%; 1993–2003), De Riet (116 mm; 41%; 1984–2006),

Vaalputs (131 mm; 53%; 1987–2006), and Goegap (150 mm;

33%; 1977–2001). The names of the sites are followed by the

mean annual rainfall, coefficient of variation (%CV) and period

of length of record, respectively. *De Riet is the farm adjacent

to and in close proximy to the Kougoedvlakte site

Plant Ecol (2011) 212:795–808 797

123

Mesembryanthemaceae, and which are colloquially

referred to as ‘mesembs’. At each of the four sites,

a community co-dominated by S. brevifolia and

R. robusta was selected for sampling, but at Goegap

Nature Reserve Leipoldtia pauciflora also co-existed

with S. brevifolia and a community of this combina-

tion was sampled.

Vegetation cover and the nearest neighbours

Vegetation composition and projected canopy cover

in each of the communities were estimated using a

line intercept method (Canfield 1941), to obtain the

percentage cover and the relative abundance. Skeletons

of dead shrubs were frequent in these communities, and

thus were also recorded. For the nearest-neighbour

sampling, we walked along a 1 m-wide transect and

whenever one of the studied species was encountered

and its nearest neighbour was either a conspecific or one

of the other study species, then the distance between

them, two perpendicular diameters (one of which was

the longest possible) and height of each individual were

recorded. At least 50 pairs of each interspecific and

intraspecific combination of plants were sampled within

each community.

Roots excavation

The root systems of three isolated adult plants of

S. brevifolia and R. robusta were excavated at

Kougoedvlakte, and three L. pauciflora plants were

excavated at Goegap Nature Reserve. Excavation was

done by hand using a pointed wooden stick to loosen

the soil which was then hand removed, leaving the

root system intact. Excavation progressed by extract-

ing the soils, and recording the root architecture and

lateral extent. Diagrams were made and photographs

were taken as excavation progressed. At the recorded

depths (see below), roots were collected in separate

paper bags, washed of excess sand, then dried and

weighed. In the grass community, it was particularly

difficult to find isolated adult grass plants because the

grass density was high, and young adults were thus

excavated. The root data for S. brevifolia were

collected at the following depth categories: 0–5,

5–10, 10–20, 20–30, 30–40, 40–50 and 50–60 cm.

Excavation continued up to 65 or 70 cm where a

hardpan develops and further excavation was diffi-

cult; but roots from below 60 cm were only used for

total root mass value. All roots of S. brevifolia were

classified as fine roots (diameter \2 mm). Ruschia

robusta and L. pauciflora root data were collected at

these depths: 0–5, 5–10, 10–20, 20–30, 30–40 and

40–50 cm, and were grouped into fine roots (with

diameter \2 mm) and coarse roots (with diameter

[2 mm). The taproot of R. robusta extended below

50 cm, but it was cut at this depth because bedrock

prevented further excavation. The taproot of one

plant was successfully excavated in its entirety, but

this extra data was not included in the analysis. To

determine whether the root depths of S. brevifolia

and R. robusta vary along the ecotone, confirmatory

rapid excavations of plants were also done at the

northern part of the study area, next to Goegap Nature

Reserve.

Data analysis

Variation in representing the size of the plants is

determined from the morphology and shape or the

growth form of the individual species (Welden et al.

1988). In this study, the sizes of both the grass and

the succulent shrub individuals were considered as

the area of an elipse calculated from the canopy

diameters using the formula for the area of an elipse:

A ¼ p ab=4ð Þ2;

where A is the area of the plant, a and b are the two

diameters (Welden et al. 1988).

According to Kolmogorov–Smirnov and Liliefors

tests, size data were not normally distributed, and,

therefore, were square root transformed. Size data

were then standardised by dividing the size of each

plant by the mean size for that species, in the

particular pair combination, in order to minimise the

effect of different sizes of plants in each pair. Linear

correlations between the sums of transformed and

standardised size data for each pair and the distances

between pairs were calculated. In inter-specific

combinations, the relative contribution of each spe-

cies to the interaction was tested by regressing the

individual sizes of plants of the two species against

distance. Significant positive correlations were inter-

preted as an indication of competition among plants.

The proportions of roots in different depth classes

were described using graphs and schematic presen-

tations. The ratio of mean rooting diameter to mean

798 Plant Ecol (2011) 212:795–808

123

canopy diameter, and the root mass fraction (root dry

mass/total dry mass) between the three species were

compared using a one-way ANOVA followed by a

Scheffe’s test for post hoc comparison where

necessary.

Results

Vegetation cover and composition

The total vegetation cover in the communities ranged

between 36% at Burdensputs and 72% at Goegap

Nature Reserve (Table 1). The co-dominance between

the grass S. brevifolia and the succulent shrubs

R. robusta or L. pauciflora varied between the sites,

but the grass and succulent shrub comprised more than

80% of the total vegetation cover in each community.

Skeletons of dead shrubs made up the third most

abundant cover class at all sites.

The nearest neighbours

Intra- and inter-specific nearest-neighbour pairings

of S. brevifolia and R. robusta in all communities

showed significant positive linear correlations

between nearest-neighbour distances and the sums

of sizes of the two neighbours (Figs. 3, 4). Intra-

specific comparisons yielded stronger correlations

than inter-specific comparison. For inter-specific

pairs, the correlation between the size of R. robusta

and the distance from S. brevifolia was stronger,

than the correlation between S. brevifolia and the

distance from R. robusta. There was no relationship

between the sizes of L. pauciflora plants and the

distances between them (r = -0.001; P = 0.99), or

between L. pauciflora and S. brevifolia (r = 0.031;

P = 0.80).

Root distribution

The vertical distribution of roots and root architecture

of each species are schematically represented in

Figs. 5, 6 and 7. S. brevifolia had the deepest roots,

extending below 60 cm in the soil. The top 10 cm of

the root system of S. brevifolia was made up of a

dense tuft of thick and tough non-fibrous roots. These

roots contributed 30% to the total root mass, but they

did not spread horizontally (Fig. 5). The root system

then became more fine and fibrous below 10 cm,

spreading both horizontally and vertically throughout

the excavation depth. More than half of the total root

mass for S. brevifolia was found between 10 and

40 cm deep. Rapid comparative excavations near

Goegap Nature Reserve and Dabeep farm confirmed

this general root architecture for S. brevifolia.

Ruschia robusta at Kougoedvlakte had a taproot,

recorded as deep as 55 cm. The roots of R. robusta

were primarily coarse, with about 70% of the root

mass greater than 2 mm in diameter. More than 90%

of the total root mass occurred in the top 20 cm in the

form of secondary roots that spread horizontally,

often extending more than a meter from the stem

(Fig. 6). In all three, R. robusta plants excavated at

Kougoedvlakte, a coarse lateral root split from the

taproot at 30 cm. The diameter of the taproot then

reduced and continued down vertically into the

cracks in the bedrock. Since the ground hardened at

about 40 cm, excavation was difficult beyond this

depth and was discontinued at 50 cm in all but one

excavation (taproots were cut at this depth). Only one

of the three excavated R. robusta plants was followed

up to the end of the root sytem, measured at 55 cm.

Rapid root excavations of R. robusta in the northeast

of the study area showed similar structure in the first

20 cm, but lacked the taproot, and thus the rooting

system was shallower. Leipoldtia pauciflora had the

Table 1 Vegetation canopy

cover (%) at the five

communities sampled for

nearest-neighbour analysis,

and percentage cover for the

most abundant species

a Dead shrubs made up the

third most abundant cover

and thus was included

Site Total S. brevifolia R. robusta L. pauciflora Dead shrubsa

Burdensputs 36 16 15 – 5

Kougoedvlakte 53 39 9 – 5

Vaalputs 48 16 23 – 10

Goegap

Ruschia 70 46 22 – 5

Leipoldtia 72 24 – 42 3

Plant Ecol (2011) 212:795–808 799

123

A

0

1

2

3

4r = 0.58***; y = 1.0 + 0.02x

Burdensputs n = 80

0

1

2

3

4r = 0.53***; y = 1.15 + 0.02*x

Kougedvlakte n = 52

0

1

2

3

4 r = 0.71***; y = 1.13 + 0.02x

Vaalputs n = 52

0

1

2

3

4r = 0.69***; y = 1.19 + 0.02x

Goegap n = 70

B

0

1

2

3S. brevifolia: r = 0.23*; y = 0.74 + 0.01xR. robusta: r = 0.46***; y = 0.36 + 0.01x

Burdensputs

0

1

2

3

S. brevifolia: r = 0.38*; y = 0.63 + 0.01xR. robusta: r = 0.46***; y = 0.49 + 0.01x

Kougoedvlakte

0

1

2

3S. brevi fo l ia : r = 0 .42**; y = 0.72 + 0.01xR. robusta : r = 0.60***; y = 0.41 + 0.02x

Vaalputs

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 10 20 30 40 50 60 70 80 90 100

0 20 40 60 80 100

0 20 40 60 80 1000

1

2

3S. brevi fo l ia : r = 0.40***; y = 0.77 + 0.01xR. robusta : r = 0.65***; y = 0.42 + 0.02x

Goegap

Distance (cm) between nearest neighbours

Squ

are

root

tran

sfor

med

and

sta

ndar

dize

d si

zes

(are

a) o

f nea

rest

nei

ghbo

urs

Fig. 3 Correlations between the sizes and the distance

between nearest neighbour plants in inter-specific comparisons

using a combined sizes, and b individual species sizes where

each line shows the effects of one species on the other species.

*P \ 0.05, **P \ 0.001, ***P \ 0.0001. In b straight line,

Stipagrostis brevifolia; dashed line, Ruschia robusta

800 Plant Ecol (2011) 212:795–808

123

A B

0

1

2

3

4

r = 0.73***; y = 0.57 + 0.03x

Burdensputs n = 52

0

1

2

3

4r = 0.62***; y = 0.46 + 0.03x

Kougoedvlakte n = 59

0

1

2

3

4

r = 0.53***; y = 1.30 + 0.02x Vaalputs n = 52

0 20 40 60 80 1000

1

2

3

4

r = 0.70***; y = 0.84 + 0.02xGoega p n = 5 2

0

1

2

3

4

5r = 0.64***; y = 0.65 + 0.03x

Burdensputs n = 51

0

1

2

3

4

5

r = 0.62***; y = 0.46 + 0.03xKougoedvlakte n = 50

0 20 40 60 80 100

0 10 20 30 40 50 60 70 80 90 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 1000

1

2

3

4

5r = 0.72***; y = 0.77 + 0.03x

Vaalputs n = 52

0 20 40 60 80 1000

1

2

3

4

5 r = 0.78***; y = 0.87 + 0.037xGoegap n = 50

Distance (cm) between nearest neighbours

Squ

are

root

tran

sfor

med

and

sta

ndar

dize

d si

zes

(are

a) o

f nea

rest

nei

ghbo

urs

Fig. 4 Correlations between the size and the distance between

nearest neighbour plants between intra-specific comparisons

using combined sizes between a Stipagrostis brevifolia pairs,

and b Ruschia robusta pairs. ***P \ 0.0001

Plant Ecol (2011) 212:795–808 801

123

shallowest root system and only 3% of the total root

mass was found below 10 cm, and none below 20 cm

(Fig. 7). The roots were also predominantly fine, with

about 80% of the total dry root mass less than 2 mm

in diameter.

The ratios of mean rooting diameter to mean canopy

diameter were significantly different among the three

species F(2,6) = 67.65, P \ 0.0001. Ruschia robusta

had a dramatically greater relative root diameter

(5.77 ± 0.36, mean ± 1 SE) than S. brevifolia

(2.19 ± 0.21) P = 0.0002, and L. pauciflora (2.08 ±

0.24) P = 0.0001. The root mass fraction (root dry

mass/total dry mass) on the other hand was not

significantly different F(2,6) = 2.38, P = 0.17 among

R. robusta (0.26 ± 0.01), S. brevifolia (0.19 ± 0.023,

mean ± SE), and L. pauciflora (0.32 ± 0.07).

Discussion

Competitive relationships

Results of nearest-neighbour analyses in this study

suggest that competitive interactions occur between

adjacent individuals within R. robusta–S. brevifolia

communities, as indicated by positive significant

correlations in both intra-specific and inter-specific

nearest-neighbour comparisons. The relative strengths

A % root mass

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Stipagrostis brevifolia

B

Soil depth (cm

)

0-55-10

10-2020-30

30-4040-50 50-60

Soil depth (cm

)

30 80 90 110 120 130 150 16010 20 40 50 100 1400

Horizontal distance (cm)

010

2030

4050

6070

70

Fig. 5 a Mean (±1 SE)

root mass at different soil

depths and b the plant

architecture showing the

root system for Stipagrostisbrevifolia, at

Kougoedvlakte, n = 3

802 Plant Ecol (2011) 212:795–808

123

of intra- and inter-specific competition are relevant to

species co-existence and community stability (Scholes

and Archer 1997). Thus, the higher ratios of intra-

specific to inter-specific competitive relationships

reported in this study between R. robusta and S. brev-

ifolia facilitate co-existence between the two species

along this ecotone. Conspecific individuals are expected

to compete strongly with each other, owing to the

similarities in the individual’s resource requirements

(Yeaton and Cody 1976; Turkington and Harper 1979).

Both intra- and inter-specific competitive relationships

have been reported previously among species of

Stipagrostis genus in the Namib desert (Yeaton 1990).

What brings about lack of intra-specific interactions

among L. pauciflora in the S. brevifolia–Leipoldtia

pauciflora community, on the other hand is not clear and

might require further investigations. Previous studies on

other leaf succulent shrub species like R. robusta,

L. schultzei (Carrick 2003), and L. constricta (Cunliffe

et al. 1990), in the Succulent Karoo have reported

evidence of intra-specific competition.

The distribution of S. brevifolia roots recorded in

the soil profile in this study is comparable to the

root distributions recorded for other grass species

from other deserts. Two grass species in semi-arid

Argentina have root systems mainly distributed in

the top 10–40 cm of the soil profile (Pelaez et al.

1994). In the Chihuahuan desert, the perennial

bunch grass, Hilaria mutica, had roots through the

entire excavated profile of 75 cm, and 70% of these

occurred between 5 and 30 cm (Briones et al.

1996). Ruschia robusta in this study has displayed

A % root mass

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

coarse roots fine roots

Ruschia robusta

B

Soil depth (cm

) 0-5

5-1010-20

20-3030-40

40-50 50-60

30 10 20 40 50 80 90 110 120 130 150 160100 1400

Horizontal distance (cm)

0 1 0 20

3040

5060 70

Soil depth (cm

)

Fig. 6 a Mean (±1 SE)

root mass at different soil

depths and b the plant

architecture showing the

root system for Ruschiarobusta, at Kougoedvlakte,

n = 3. In a: coarse roots

[2 mm, fine roots \2 mm

Plant Ecol (2011) 212:795–808 803

123

differential rooting depth, with a taproot at

Kougoedvlakte and the absence of a taproot near

Goegap Nature Reserve. However, even where a

taproot is present, most of the roots of R. robusta

occurred in the first 20 cm of the soil. Carrick

(2003), observing similar distribution at Paulshoek

in Namaqualand, described the small proportion of

the root mass extending beneath 20 cm soil layer

in R. robusta, as unusually deep among the leaf

succulent Aizoaceae (‘mesembs’).

The shallow root structure for both R. robusta and

L. pauciflora recorded in this study confirms the

perception of succulents in the Succulent Karoo as

being extremely shallow rooted (Esler and Rundel

1999; Midgley and van der Heyden 1999; Carrick

2003). Shallow roots and succulent leaf tissue are

reported as the two common morphological adapta-

tions of succulent perennial shrubs in a winter rainfall

Namaqualand area enabling them to compete for

water uptake and store water (Von Willert et al. 1992;

Cowling et al. 1999). Shallow roots enable the

‘mesembs’ to take advantage of the frequent small

winter rainfall events which only penetrate the top

few centimetres of the soil (Cowling et al. 1994; Esler

and Rundel 1999; Carrick 2001). Leipoldtia paucifl-

ora with the most shallow rooting system in the

% root mass

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Leipoldtia pauciflora

coarse roots fine roots

B

A

Soil depth (cm

)

0-55-10 10- 20 20-30

30-40 40-5050-60

Horizontal distance (cm)

010

2030

4050

6070

Soil depth (cm

)

30 80 90 110 120 10 20 40 50 130 150 160100 1400 70

Fig. 7 A Mean (± SE) root

mass at different soil depths

and b plant architecture

showing the root system for

Leipoldtia pauciflora, at

Goegab Nature Reserve,

n = 3. In a: coarse roots

[2 mm, fine roots \2 mm

804 Plant Ecol (2011) 212:795–808

123

current study is confined to the north western part

of the study area, where the rainfall is typical of

the winter rainfall Succulent Karoo biome. The

morphological plasticity of R. robusta roots may

thus be an adaptation that has contributed to its wider

distribution at the ecotone, where rainfall is less

seasonal and less predictable relative to the main

Succulent Karoo biome. Another mesemb species that

displays a taproot is Ruschia spinosa, which is

widespread in the summer rainfall Nama-Karoo biome

(P. J. Grubb and P. J. Carrick, unpublished data).

Distribution of vertical root structure has been

widely advocated to explain co-existence and com-

petition between species (Walter 1971; Yeaton et al.

1977; Knoop and Walker 1985; Fowler 1986; Briones

et al. 1996). The relative dominance of roots in the

soil between the species is presented in a hypothetical

illustration in Fig. 8 to explain the observed compet-

itive relationships. Leipoldtia pauciflora with most of

its roots in the top 10 cm is expected to utilise water

in this layer, facilitating co-existence with S. brevi-

folia which has most of its absorbing roots between

10 and 40 cm. Lack of competition between

S. brevifolia and L. pauciflora can thus be ascribed

to this clear separation in the vertical root systems.

For S. brevifolia and R. robusta, the degree of partial

overlap in the vertical root distribution may account

for the observed evidence of competition between

them. Stronger competitive impacts of S. brevifolia

on R. robusta may be explained by the fact that

R. robusta has only about 6% of its root mass beneath

20 cm, while a greater proportion of the root system

of S. brevifolia extends deeper down the soil profile.

As each species dominates in a different layer, the net

competition by S. brevifolia on R. robusta in the

communities where they co-occur is minimised,

facilitating their co-existence over a wider area along

the ecotone. These results reveal the possibility of co-

existence between the succulent shrubs and grass,

which has been largely overlooked in the Karoo.

Although untested, it has been proposed that

succulent shrubs and grasses compete for water in

the same horizontal plane (Midgley and van der

Heyden 1999). Similarly, Desmet (2007) contends

that ‘mesembs’ and grasses never co-dominate in

plant communities, as both functional groups share a

similar, shallow root structure.

The competitive relationship for soil resources

between grass and succulent shrubs in these commu-

nities is unique and contrasts with savanna compe-

tition-based models where grass is proposed to obtain

water from the top layer and shrubs mainly from

deeper soil layers (Walter 1971; Walker and

Noy-Meir 1982; Knoop and Walker 1985; Sala

et al. 1989; Kochy and Wilson 2000; Snyman

2005). The succulent shrubs are shallower rooted

comparing to grass and are thus expected to utilise

water mainly from the top layer. Ruschia robusta,

with a high proportion of root mass in the first 20 cm

of the soil is expected to predominantly utilise water

in the upper layers of the soil, and this reduces the net

competition from the deeper rooting grass. A thor-

ough analysis of water relations between grass and

succulent shrubs is, however, required, to improve

our understanding of species co-existence between

the two growth forms. Furthermore, grasses are

generally hypothesised to have higher root mass

fraction compared to trees and non-succulent woody

plants (Waisel et al. 1991; Kochy and Wilson 2000;

A B

Predominantly R. robusta

Strong overlap

S. brevifolia

0

60

50

40

30

20

10

Predominantly S. brevifolia

Soil depth (cm

)

Predominantly L. pauciflora

Predominantly S. brevifolia

S. brevifolia

0

60

50

40

30

20

10

Fig. 8 Hypothetical model of relative dominance of roots in a

soil profile of Stipagrostis brevifolia and a Leipoldtia paucifl-ora and b Ruschia robusta. The clear vertical separation in root

proportion in a avoids competition, whereas the partial overlap

in b allows for competition between the species

Plant Ecol (2011) 212:795–808 805

123

Snyman 2005), but a contrasting observation indi-

cated that succulent shrubs have greater root mass

proportion than grass, as shown for L. pauciflora and,

R. robusta in this study, and L. schultzei in a previous

study (Carrick 2003). We demonstrate that succulent

shrubs do fit the simple two-layer hypothesis but they

operate in the inverse way to that proposed for

savanna grass–woody systems.

In addition to differential root distribution dis-

cussed in this study as a mechanism that facilitates

species co-existence at the ecotone, the grass and

succulent shrub species dominate biomes character-

ised by summer and winter rainfall, respectively.

This may suggest that a certain degree of temporal

segregation in plant growth and water uptake among

the species is likely, as previously hypothesised

(Midgley and van der Heyden 1999). Variations in

seasonal rainfall patterns have profound effects on the

abundance of grass and Karoo shrubs in the eastern

Karoo (Hoffman et al. 1990; Hoffman and Cowling

1990; Bousman and Scott 1994; O’Connor and Roux

1995). A detailed investigation of the phenological

patterns of grass and shrubs at the ecotone will yield

useful information on this.

Competition and community dynamics

The asymmetric competitive relationship between

S. brevifolia and R. robusta is consistent in all the

communities studied. While a small competitive

difference can be sufficient to cause rapid extermina-

tion of the less competitive species (Hardin 1960), the

persistence, and therefore co-existence, of species in

any community depends on the sum of its fitness and

abundance throughout its life phases. The very high

number of seeds produced by R. robusta, and the

extreme drought tolerance of its seedlings (Hoffman

et al. 2009) may lead to far higher numbers of

R. robusta seedlings establishing successfully, than

those of S. brevifolia, and this may compensate for its

relatively poor competitive ability as an adult, and

hence mitigate for co-existence in the community.

Unfortunately, the nearest-neighbour method, as used

here, primarily measures the impact of competition on

the growth of neighbouring adult plants, which is

distinct from the impact of competition on the

survival, or from the impact of competition of adults

on seedlings, both of which influence persistence and

co-existence more directly than growth.

From the observed nature of the competitive

interactions, and patterns in rooting depths, it is

inferred that the stronger competition from grass

indicates reduced soil water availability to R. robusta

individuals. In the presence of S. brevifolia, the results

likely indicate a reduction in water availability in the

upper soil layer, where most of the root mass of

R. robusta is concentrated. This competitive dynamic

will interact with rainfall patterns across the ecotone,

such that changes in winter- or summer-rainfall

patterns, associated with climate change, may influ-

ence the overall fitness of the two ecotonal growth

forms in the study area by influencing their establish-

ment, survival and mortality. The winter-rainfall

shrublands in the west receives primarily small frontal

rainfall events that do not percolate deeply, while the

summer-rainfall grasslands in the east receive primar-

ily larger convective rainfall events that penetrate the

soil more deeply (Carrick 2001). While there is

conjecture concerning trends, and future predictions,

of a shift in the magnitude or distribution of the two

rainfall systems, as a result of anthropogenic climate

change (MacKellar et al. 2007; Hoffman et al. 2009), a

concomitant shift in the co-existence and distribution

of the dominant grass and leaf succulent shrubs with

such a shift in rainfall patterns is consistent with the

results of this study. This interpretation is in line with

previous non-mechanistic postulations, e.g. that of

Esler and Rundel (1999) who suggest that an increased

summer rainfall could lead to a rapid increase in grass

cover and thus an increased competitive pressure from

grasses.

This study indicated that both intra- and inter-

specific competition are important for co-existence

between R. robusta and S. brevifolia in these com-

munities. Competitive relationships (observed in

R. robusta–S. brevifolia combinations) as well as

lack thereof (in L. pauciflora–S. brevifolia combina-

tion) were explained by the vertical distribution of the

root systems in the soil. The root distribution between

the grass and the shrubs described in this study

demonstrate the possibility of co-existence of two

growth forms according to the two-layer hypothesis

for water acquisition among grasses and trees

proposed in the savanna. However, in this arid

environment, the upper layer is occupied by the roots

of a succulent shrub which appear to have specialised

in capitalising on extremely small rainfall events, and

the lower layers occupied by grass roots. Our findings

806 Plant Ecol (2011) 212:795–808

123

make a contribution to arid ecology in general, and

specifically to community ecology in the Karoo

where the possibility of co-esistence between the

succulent shrubs and grass has been largely over-

looked or misinterpreted. The results also show how

understanding the nature of competitive interactions

at a climate ecotone can be used to predict how the

vegetation may respond to climate change and

provide ecological insights in understanding observed

or expected changes in species distribution along the

ecotone.

Acknowledgments This study was funded under the

umbrella of the BIOTA Southern Africa project (www.

biotaafrica.org), by the German Federal Ministry of Educa-

tion and Research under promotion number 01 LC 0024A. We

thank the Mazda Wildlife Vehicle Fund for the use of a

courtesy vehicle during the course of the study. Rainfall data

for Goegap Nature Reserve and Vaalputs Radioactive Waste

Disposal Facility were supplied by their respective offices,

while Burdensputs and De Riet rainfall data were obtained

from the landowners. We gratefully acknowledge the land-

owners in Namaqualand for kindness and permission to work

on their land, the management of Vaalputs Radioactive Waste

Disposal Facility for permission to work on Vaalputs farm, and

the Northern Cape Department of Nature Conservation for

permission to undertake research on Goegap Nature Reserve.

Special thanks to Mariana Lot, the para-ecologist at Paulshoek,

for field assistance. The following people are thanked for

assistance in one way or the other during field trips, Lee

Simons, Azieb Woldensae, Sebataolo Rahlao, Elizabeth Cla-

assen and Vonkie Claassen.

References

Bousman B, Scott L (1994) Climate or overgrazing—the pal-

ynological evidence for vegetation change in the eastern

Karoo. S Afr J Sci 90:575–578

Briones O, Montana C, Ezcurra E (1996) Competition between

three Chihuahuan desert species: evidence from plant

size-distance relations and root distribution. J Veg Sci

7:453–460

Cable DR (1969) Competition in semidesert grass-shrub type

as influenced by root systems growth habits and soil

moisture extraction. Ecology 50:27–38

Canfield RH (1941) Application of the line intercept method in

sampling range vegetation. J For 39:388–394

Cannon WA (1924) Vegetation of more arid portions of

Southern Africa. Carnegie Institution of Washington,

Washington

Carrick PJ (2001) Shrub community dynamics in a South

African semi-desert. PhD thesis, Magdalene College,

University of Cambridge

Carrick PJ (2003) Competitive and facilitative relationships

among three shrub species, and the role of browsing

intensity and rooting depth in the Succulent Karoo, South

Africa. J Veg Sci 14:761–772

Casper BB, Jackson RB (1997) Plant competition underground.

Annu Rev Ecol Syst 28:545–570

Cody ML (1986) Structural niches in plant communities. In:

Diamond J, Case TJ (eds) Community ecology. Harper

and Row, New York, pp 381–405

Cowling RM, Esler KJ, Midgley GF, Honig MA (1994) Plant

functional diversity, species-diversity and climate in arid

and semiarid Southern Africa. J Arid Environ 27:141–158

Cowling RM, Esler KJ, Rundel PW (1999) Namaqualand,

South Africa—an overview of a unique winter-rainfall

desert ecosystem. Plant Ecol 142:3–21

Cunliffe RN, Jarman ML, Moll EJ, Yeaton RI (1990) Com-

petitive interactions between the perennial shrub Lei-

poldtia constricta and an annual forb, Gorteria diffusa.

S Afr J Bot 56:34–38

Desmet PG (2007) Namaqualand: a brief overview of the

physical and floristic environment. J Arid Environ 70:

570–587

Desmet PG, Cowling RM (1999) The climate of the karoo—a

functional approach. In: Dean WRJ, Milton SJ (eds) The

Karoo: ecological patterns and processes. Cambridge

University Press, United Kingdom, pp 3–16

Esler KJ, Cowling RM (1993) Edaphic factors and competition

as determinants of pattern in South-African Karoo vege-

tation. S Afr J Bot 59:287–295

Esler KJ, Rundel PW (1999) Comparative patterns of phenol-

ogy and growth form diversity in two winter rainfall

deserts: the Succulent Karoo and Mojave Desert ecosys-

tems. Plant Ecol 142:97–104

Fowler N (1986) The role of competition in plant communities in

arid and semiarid regions. Annu Rev Ecol Syst 17:89–110

Gordon CE (2000) The coexistence of species. Rev Chil Hist

Nat 73:175–198

Hardin G (1960) The competitive exclusion principle. Science

131:1292–1298

Hoffman MT, Cowling RM (1990) Vegetation change in the

semiarid eastern Karoo over the last 200 years—an

expanding Karoo—fact or fiction. S Afr J Sci 86:286–294

Hoffman MT, Barr GD, Cowling RM (1990) Vegetation

dynamics in the semiarid eastern Karoo, South-Africa—

the effect of seasonal rainfall and competition on grass

and shrub basal cover. S Afr J Sci 86:462–463

Hoffman MT, Carrick PC, Gillson L, West AG (2009)

Drought, climate change and vegetation response in the

succulent karoo, South Africa. S Afr J Sci 105:1–7

Knoop WT, Walker BH (1985) Interactions of woody and

herbaceous vegetation in a southern African savanna.

J Ecol 73:235–253

Kochy M, Wilson SD (2000) Competitive effects of shrubs and

grasses in prairie. Oikos 91:385–395

Lloyd JW (1985) A plant ecological study of the farm Vaal-

puts, Bushmanland, with special reference to edaphic

factors. MSc thesis, University of Cape Town, South

Africa

Lloyd JW (1989) Discriminant-analysis and ordination of

vegetation and soils on the Vaalputs radioactive waste

disposal site, Bushmanland, South-Africa. S Afr J Bot

55:127–136

Plant Ecol (2011) 212:795–808 807

123

MacKellar NC, Hewitson BC, Tadross MA (2007) Namaqua-

land’s climate: recent historical changes and future sce-

narios. J Arid Environ 70:604–614

Midgley GF, van der Heyden F (1999) Form and function in

perennial plants. In: Dean WRJ, Milton SJ (eds) The

Karoo: ecological patterns and processes. Cambridge

University Press, Cambridge, pp 303–313

Nobel PS (1997) Root distribution and seasonal production in

the northwestern Sonoran Desert for a C-3 subshrub, a C-4

bunchgrass, and a CAM leaf succulent. Am J Bot 84:

949–955

O’Connor TG, Roux PW (1995) Vegetation changes (1949–71)

in a semiarid, grassy dwarf shrubland in the Karoo, South-

Africa: influence of rainfall variability and grazing by

sheep. J Appl Ecol 32:612–626

Pelaez DV, Distel RA, Boo RM, Elia OR, Mayor MD (1994)

Water relations between shrubs and grasses in semiarid

Argentina. J Arid Environ 27:71–78

Pielou EC (1960) A single mechanism to account for regular,

random and aggregated populations. J Ecol 48:575–584

Rosch H (2001) The identification and description of the

management units of the Goegap Nature Reserve. Koedoe

44:17–30

Rosenzweig ML (1981) A theory of habitat selection. Ecology

62:327–335

Sala OE, Golluscio RA, Lauenroth WK, Soriano A (1989)

Resource partitioning between shrubs and grasses in the

Patagonian steppe. Oecologia 81:501–505

Scholes RJ, Archer SR (1997) Tree-grass interactions in sav-

annas. Annu Rev Ecol Syst 28:517–544

Shiponeni NN (2007) Spatio-temporal distribution of grass and

shrubs at the ecotone between an arid grassland and

succulent shrubland: ecological interactions and the

influence of soils. PhD thesis, University of Cape Town,

South Africa

Snyman HA (2005) Rangeland degradation in a semi-arid

South Africa. I: influence on seasonal root distribution,

root/shoot ratios and water-use efficiency. J Arid Environ

60:457–481

Turkington R, Harper JL (1979) Growth, distribution and neigh-

bor relationships of Trifolium repens in a permanent pasture.

1. Ordination, pattern and contact. J Ecol 67:201–218

Von Willert DJ, Werger MJA, Brinckmann E, Ihlenfeldt HD,

Eller BM (1992) Life strategies of succulents in deserts:

with special reference to the Namib Desert. Cambridge

University Press, Cambridge

Waisel Y, Eshel A, Kafkafi U (1991) Plant roots, the hidden

half. Marcel Dekker, New York

Walker BH, Noy-Meir I (1982) Aspects of stability and resil-

ience of savanna ecosystems. In: Huntley BJ, Walker BH

(eds) Ecology of tropical savannas. Springer, Berlin,

pp 577–590

Walter H (1971) Ecology of tropical and subtropical vegeta-

tion. Oliver and Boyd, Edinburgh

Watkeys MK (1999) Soils of the arid south-western zone of

Africa. In: Dean WRJ, Milton SJ (eds) The Karoo: eco-

logical patterns and processes. Cambridge University

Press, Cambridge, pp 17–26

Welden CW, Slauson WL, Ward RT (1988) Competition and

abiotic stress among trees and shrubs in northwest Colo-

rado. Ecology 69:1566–1577

Yeaton RI (1990) The structure and function of the Namib

dune grasslands–species interactions. J Arid Environ

18:343–349

Yeaton RI, Cody ML (1976) Competition and spacing in

plant communities—northern Mojave desert. J Ecol 64:

689–696

Yeaton RI, Esler KJ (1990) The dynamics of a Succulent

Karoo vegetation—a study of species association and

recruitment. Vegetatio 88:103–113

Yeaton RI, Travis J, Gilinsky E (1977) Competition and

spacing in plant communities—Arizona-upland-associa-

tion. J Ecol 65:587–595

Yeaton RI, Moll EJ, Jarman ML, Cunliffe RN (1993) The

impact of competition on the structure of early succes-

sional plant- species of the Atlantic coast of South Africa.

J Arid Environ 25:211–219

808 Plant Ecol (2011) 212:795–808

123