Spatial and temporal distribution of cyanobacterial soil crusts in the Kalahari:
implications for soil surface properties
Thomas, A.D.1 and Dougill, A.J.2
– Department of Environmental and Geographical Sciences, Manchester Metropolitan
University, John Dalton Building, Chester Street, Manchester, M1 5GD, U.K. Tel. +44 (0)
161 247 1568. Fax. +44 (0) 161 247 6318. Email: a.d.thomas@mmu.ac.uk
Corresponding Author.
1
– School of Earth and Environment, University of Leeds, Leeds, LS2 9JT, UK. Tel. +44 (0)
113 343 6782. Fax. +44 (0) 113 343 6716. Email: adougill@env.leeds.ac.uk
2
Abstract
Keywords
1
1. Introduction
Extensive research reviews exist assessing the role of surface vegetation on dryland
geomorphological processes (e.g. Thornes, 1990; Bullard, 1997), in particular on the role that
surface vegetation has in controlling aeolian erosion (Tsoar & Møller, 1986; Lancaster &
Baas, 1998). Specifically for the Kalahari region of Southern Africa, research has focused on
the link between surface vegetation cover and dune mobility in the arid southwest of
Botswana (Wiggs et al., 1995) and on the controlling influence of bush cover on nebkha
formation in the mixed farming areas in the dry sub-humid Molopo Basin (Dougill &
Thomas, 2002). Much less attention has been paid to aeolian erosion processes in the more
extensive semi-arid savanna rangelands that typify much of Botswana, Eastern Namibia and
Northern South Africa (Thomas, 2002), or to the possible controlling influence of
cyanobacteria assemblages that colonise biological soil crusts on the soil surface (Dougill &
Thomas, 2004). Biological soil crusts have been shown in other dryland regions to decrease
surface erodibility significantly due to a number of contributory factors. Biological crusts
bind sand particles together (Belnap et al., 2003). Crusts also provide notable elasticity to the
soil surface that reduces the force of grain impact (McKenna Neuman & Maxwell, 2002). In
addition, crusts also provide an enhanced nitrogen source that increases the surface vegetation
cover (and thus landscape roughness) in ecosystems that are typically N-limited (Scholes &
Walker, 1993). This paper aims to report on recent research findings from a range of
locations in the semi-arid Kalahari that focus on the impact of biological soil crusts on spatial
and temporal patterns of soil surface properties and their role in regulating aeolian erosion
processes.
Biological soil crusts are present in all arid and semi-arid regions (Belnap & Lange, 2003)
and comprise cyanobacteria, algae, lichens, mosses, microfungi and other bacteria (Belnap et
al., 2003). Fundamental to understanding the geomorphological significance of
cyanobacterial soil crusts is a comprehension of their spatial distribution and its temporal
variation, as well as their morphological influence on surface erodibility.
Several factors are recognised as influencing crust distribution and development, especially
substrate, vegetation type and cover, and disturbance levels (Belnap et al., 2003). It has been
shown that vegetation and biological crust cover are inversely proportional due to the effects
of competition for light (Malam Issa et al., 1999) and nutrients (Harper & Belnap, 2001). It is
also generally accepted that trampling, as a result of grazing, damages biological crust
surfaces (e.g. Eldridge, 1998), thus in areas of intense grazing, the spatial distribution of
crusts will be limited. Zaady & Bouskila (2002) describe disturbance as the key factor in
2
determining biological crust development in areas where physical conditions are relatively
constant. Given the spatial homogeneity of the Kalahari, in terms of altitude, relief and
surface water (Thomas & Shaw, 1993), it is reasonable to impart a significant role to grazing
disturbances in affecting the distribution of cyanobacterial soil crusts. In this context,
Berkeley et al. (2005) have shown that on a localised scale the canopies of woody plants
represent quasi-discrete environments, in which the response of crusts to disturbance regimes
is altered. This displays the need to base analyses of spatial patterns of crust distribution and
dynamics on three scales of study: a landscape scale to investigate the role of different land
uses; a farm scale to assess differences in grazing intensity; and a localised ‘plot’ scale to
analyse inter-relationships between crust and surface vegetation communities.
Assessments of the spatial patterns and temporal dynamics of biological crust cover are vital
to improving geomorphological assessments of drylands due to the direct and indirect impacts
that crusts have on reducing the erodibility of the soil surface (e.g. Van der Ancker, 1985;
Danin et al., 1989; Yair, 1990; Zobeck, 1991; Belnap & Gillette, 1998; Rice & McEwan,
2001; McKenna Neuman & Maxwell, 2002; Belnap et al., 2003; Eldridge & Leys, 2003;
Goosens, 2004). Erodibility (the degree to which a surface is susceptible to erosion)
assessment has generally received less attention than erosivity (the potential to erode the
surface), partly due to the difficulty of measuring surface erodibility (Geeves et al., 2000).
This difficulty in assessment is exacerbated by the control of inter-related textural,
mineralogical, chemical, hydrological and biological surface characteristics on erodibility,
with each of these factors highly variable in both space and time. There is a recognised need
to improve the information available on soil surface conditions, such as cohesive strength and
roughness, that control erodibility to enable wind erosion models to be improved to
incorporate the evolution of soil surface conditions (Sokolik & Toon, 1996; Shao & Leslie,
1997; Chappell et al., 2005). Such advances are essential if dust and erosion models are to be
linked to global climate change and regional biogeochemical cycle models.
Dougill & Thomas (2004) have previously documented a cyanobacterial soil crust cover of
between 19 - 40 % at a range of regularly disturbed, communal grazing land, sites on Kalahari
Sands. They identified three morphologically distinct crusts: a weakly consolidated crust with
no surface discolouration (type 1); a more consolidated crust with a black or brown speckled
surface (type 2); and a crust with a bumpy surface with an intensely coloured black/brown
surface (type 3). This paper builds on this preliminary analysis by addressing the following
research objectives:
3
1. To determine the controls on the distribution of soil crusts on Kalahari Sands at
regional, farm, plot and canopy scales.
2. To quantify recovery of cyanobacterial crust cover after removal of disturbance
impacts.
3. To examine the implications of cyanobacterial crusts for soil surface roughness,
strength, and total nitrogen content as key factors affecting surface erodibility.
2. Study Area
Kalahari Sands are ancient wind-blown deposits that cover an area of over 2.5 million km2 of
Southern Africa, including over 80 % of Botswana (Thomas & Shaw, 1991). Kalahari Sand
soils typically consist of over 95 % fine sand sized sediments and are predominantly deep,
structureless and nutrient deficient (Dougill et al., 1998). The Kalahari includes areas of both
active sediment movement and stable surfaces, with much research focusing on the
importance of surface vegetation cover (Bullard, 1994; Wiggs et al., 1994, 1995) and climate
change (Thomas et al., 1997; Knight et al., 2004) in affecting sediment mobility, notably in
the arid south west corner of Botswana. The importance of soil surface characteristics,
especially the development of biological soil crusts, has been largely overlooked. Preliminary
research on the presence of crusts has largely focused on their role in affecting nitrogen
cycling with a number of studies detailing crust presence in Kalahari rangelands of western
Bostwana (Skarpe & Hendrikson, 1987; Aranibar et al., 2003, 2004; Dougill & Thomas,
2004; Berkeley et al., 2005). All these studies focused on highly disturbed communal
rangeland sites and have estimated crust covers of between 19 and 40 %. No empirical
estimates exist of crust cover in lightly disturbed sites, such as National Parks or Wildlife
Management Areas that still cover large tracts of the Kalahari, where crust cover can be
expected to be greater.
Pastoral land use across the Kalahari has been made possible in the last fifty years by
boreholes that allow access to deep groundwater reserves (Sporton & Thomas, 2002).
Concentrated grazing pressure close to boreholes adds a new environmental gradient to an
otherwise relatively homogenous environment (Thomas & Shaw, 1993). This leads to
localised sediment mobility (evidenced by rippled sand surfaces and sediment accumulation
around bush species) in a spatially confined ‘sacrifice zone’ close to boreholes (Perkins &
Thomas, 1993). This zone is typically followed by a more extensive ‘bush encroached zone’
of between 2 and 8 km radius where woody shrubs, typically A. mellifera, dominate (Dougill
et al., 1999; Moleele et al., 2003). The Kalahari has a highly seasonal summer rainfall regime
with very high inter-annual variability that increases with declining mean annual rainfall, with
4
inter-annual variability of over 50 % found in sites in the arid south-west of Botswana where
mean annual rainfall is less than 200 mm (Bhalotra, 1987). The study sites investigated here
(Figure 1) form a transect across the semi-arid Kalahari between Tsabong (c. 300 mm mean
annual rainfall; 45 % inter-annual variability) and Okwa (c. 400 mm mean annual rainfall; 40
% inter-annual variability). The rainfall regime over the past seven years for Tsabong is
shown in Figure 2 notably displaying the below average rainfall experienced since 2002 when
sampling and crust recovery studies have been undertaken.
Figure 2. Mean annual rainfall at Berry Bush Farm, Tsabong September 1997 to August 2004
(Source: farmers personal rainfall record).
Annual Rainfall (mm)
600
500
400
1934 -1988 average
rainfall for Tsabong
300
200
100
0
`97 - `98 - `99 - `00 - `01 - `02 - `03 `98 `99 `00 `01 `02 `03 `04
Y ear
3. Research Design and Methods
Research was undertaken during July 2003 and November 2004 at five locations of differing
land use, and a total of 10 sites all on Kalahari Sands across Kgalagadi and Ghanzi Districts,
in Western Botswana (Figure 1). Disturbance was quantified at each study site using a
disturbance index based on counts of cattle track frequency and dung density (as per Dougill
& Thomas, 2004 and Berkeley et al., 2005). At each site, a 30 m x 30 m grid was established.
The grid was crossed at 10 m intervals in two perpendicular directions(N-S and E-W). Cattle
tracks and dung were counted along each of these gridlines, cattle tracks being defined as well
established ‘routes’, and a dung ‘count’ made of single or collections of dung pats within an
arbitrary 0.5 m either side of the gridline.
Crust cover data were estimated at four scales. To assess the impact of regional variations in
land use, 5 study locations were sampled (Figure 1). These encompassed a range of grazing
intensities from lightly-grazed National Park (Mabuasehube) and Wildlife Management Zone
(Okwa) sites to more intensely-grazed communal rangelands (Tshane, Tshabong and
Draihoek). Multiple study sites were assessed at Mabuasehube (two), Tshane (two) and
5
Tsabong (four) to account for differences in the vegetation community assemblages found
across these sites in relation to variability on the ‘farm’ scale of approximately 10 km caused
by both natural landscape heterogeneity and grazing disturbance gradients (leading to the 10
sites as listed in Table 1). Within each site, localised measurements of crust-vegetation
relationships were taken using a 1 m x 1 m quadrat sampled at regular intervals of 2 m along
fourteen 30 m transects aligned N-S in a 30 m x 28 m grid, such that a total of 240 quadrats
were examined at each site. Percentage cover was estimated for each cyanobacterial soil crust
type (according to the morphological classification system of Dougill & Thomas, 2004),
unconsolidated soil, litter and cover of each vegetation species within each of the quadrats
and spatial patterns of cover displayed using Surfer software. On an even finer spatial scale,
detailed crust associations with the two most common shrub species in the Southern Kalahari
(Acacia mellifera and Grewia flava) were also examined. The canopy dimensions of three
shrubs of each species were measured and crust cover estimated in a series of 0.5 m x 0.5 m
quadrats, adjacent to one another along a line extending from the bowl to the canopy edge in a
northerly and southerly direction to account for any orientation-controlled differences in
cover.
In addition to spatial pattern studies, permanent monitoring plots were established to assess
recovery of crust cover through time after the removal of disturbance impacts. Two 5 m x 5
m plots were established in July 2003 at a commercial ranch site (Berry Bush farm)
neighbouring the Tsabong sample sites. These plots were located at a severely disturbed site
close to the water point and farm compound where no crust cover was found. One plot was
fenced to exclude all grazing animals and to compare crust cover characteristics of the
ungrazed plot to the neighbouring unfenced plot where grazing was still possible. This
contrast was not as marked as expected as cattle were removed from the farm shortly after
plot establishment in favour of camel and game. Sixteen 1 m x 1 m quadrats were marked
with pins in each plot (leaving small walkways between qudrants of plot for access) and
surveyed for crust type and vegetation cover in each plot in July 2003 and in November 2004
to assess the changes in crust cover and morphology following the removal of disturbance.
Properties – OM/TN;
Chlorophyll
Strength and Roughness – Tom
6
Results and Analysis
4.1 Controls on the distribution of soil crusts at a variety of spatial scales
Spatial analyses of biological crust distribution were conducted on landscape, farm, plot and
individual bush canopy scales and are initially presented separately prior to discussion of their
implications on soil surface properties and erodibility.
Landscape scale across Kalahari region …
Table 1: Study Site Morphological, Vegetation and Disturbance Characteristics
Cyanobacterial Crust Cover
(%)
Vegetation Cover
(%)
Disturbanc
e
Index
Site (n = 240 at each site)
BC
1
BC2
BC3
Total
Shrub
Grass
Tshane 1 (T1)
40
4
0
44 (36)
27 (40)
2 (0.2)
13.2
Tshane 2 (T2)
42
2
0
44 (38)
26 (40)
5 (6)
7.5
Mabua 1 (M1)
75
10
1
86 (17)
7 (23)
94 (20)
0.0
Mabua 2 (M2)
64
24
7
95 (8)
1 (9)
17 (11)
0.1
Okwa WMZ (O1)
80
5
1
86 (22)
4 (19)
22 (22)
0.0
Tsabong Mixed (T1)
26
39
2
68 (44)
22 (30)
-
1.6
Tsabong Bush Enc. (T2b)
24
36
4
64 (45)
26 (28)
-
2.1
Tsabong Pan (T2p)
21
48
15
84 (29)
-
2.4
Tsabong Sacrifice Zone (T2s)
11
0
0
11 (18)
2 (12)
-
15.4
Draihoek Bush Enc. (D1)
55
8
1
64 (35)
32 (44)
0.2 (0.9)
8.2
Within site / farm scale – impact of disturbance investigated in sites with similar ecology and
parent material at Tshane and Tsabong sites.
7
Figure 2 – Crust cover in relation to disturbance index (add site labels to show within site
variability at Tsabong and Tshane in relation to disturbance).
120
Crust Cover %
100
80
60
40
20
0
0
3
6
9
12
15
18
Disturbance Index
Running ANOVA using a Post-Scheffe’s test should account for the fact that all data may not
be normally distributed (as per our JAE revisions).
Plot scale assessments in 30 m x 28 m plots were used to assess the relationships between
crust cover and surface vegetation.
8
Surfer Plots and Vegetation Patterns
30
30
25
25
100
90
100
100
90
90
80
80
70
70
60
60
50
50
40
40
30
30
20
20
10
10
0
0
80
20
20
70
60
15
15
50
40
10
10
30
20
5
5
10
0
0
0
0
25
5
10
15
20
25
0
5
10
15
20
25
Figure 3. Total Cyanobacterial Crust Cover within 30m x 28 m plots at a) Okwa Wildlife
Management Zone and b) Tshane communal grazing area
30
30
25
25
100
90
100
100
90
90
80
80
70
70
60
60
50
50
40
40
30
30
20
20
10
10
0
0
80
20
20
70
60
15
15
50
40
10
10
30
20
5
5
10
0
0
0
0
25
5
10
15
20
25
0
5
10
15
20
25
Figure 4. The relationship between different vegetation types and cyanobacterial crust cover
within 30m x 28 m plots at a) Okwa Wildlife Management Zone and b) Tshane communal
grazing area
For veg canopy bits refer to key findings summarised from Berkeley et al., 2005. – if there’s a
more simplified way of us presenting key results here this would be beneficial.
9
4. 2 Recovery of cyanobacterial crust cover after disturbance
Link to plot monitoring.
Table 2:
Cyanobacterial Crust Cover (%)
Site (n = 240 at each site)
Vegetation
BC1
BC2
BC3
Total
Shrub
Grass
August 2003
0
0
0
0
8
0
November 2004
77
13
0.3
90.3
11
2
August 2003
0
0
0
0
4
0
November 2004
93
6
0
99
5
1.4
Open Plot
Closed Plot
Chlorophyll data (response to wetting) – refer to last paper (JAE)
10
4. 3 Implications of cyanobacterial crusts for soil surface roughness, strength, and total
nitrogen content
All Sites
Unconsolidated
BC1
BC2
BC3
Organic
Matter %
Ammonium
µg/g-1
Total N
µg/g-11
Phosphate
µg/g-1
pH
Strength
(kg cm-2)
n= 278
n = 145
n = 98
n = 92
n = 192
n = 66
1.14 (0.93)
0.92 (0.49)
1.04 (0.4)
1.68 (0.53)
76.4 (48.7)
101.3 (52.3)
80.6 (44.1)
89.4 (51.8)
183.1 (191.1)
172.9 (89.2)
341.3 (221.3)
652.2 (856.5)
2.7 (1.2)
2.8 (1.1)
3.1 (1.5)
2.5 (0.9)
5.9 (0.4)
5.8 (0.5)
5.8 (0.6)
5.9 (0.3)
0
0.56 (0.8)
0.73 (0.8)
1.50 (0.9)
Figure 5 – Roughness Measurements for each morphologically distinct crust type (needs y
axis to be labelled as Roughness measure in some way).
Average Standard Deviation
1.2
1
0.8
0.6
0.4
0.2
0
BC1
n = 76
BC2
n = 64
BC3
n = 60
Unconsolidated
n = 12
11
Discussion
The research findings presented here provide a first assessment of the spatial variability of
biological soil crusts on both local and regional scales for the Southern Kalahari. These
findings display that the spatial extent of crust cover is far more extensive than suggested in
previous analyses that focused solely on regularly disturbed communal rangeland sites (e.g.
Skarpe & Henriksson, 1987; Aranibar et al., 2003; Dougill & Thomas, 2004). Previous
taxonomic analyses of these crusts (Thomas & Dougill, in press) have shown that they are
dominantly comprised of only two main cyanobacteria species, Microcoleus and Scytonema.
Both these species are capable of binding fine sand particles together, thus decreasing surface
sediment mobility and explaining the lack of major aeolian erosion events, or dust losses,
observed from the Kalahari sandveld (TOMS style RS references??). This is exacerbated by
the large portions of Botswana (approximately 46 %) that has protected status as either
National Park or Wildlife Management Area. Our findings show that these areas typically
have a cyanobacterial mean crust cover of over 90 % that can explain the low levels of
surface erosion observed. The Kalahari, appears notably different to many other drylands
globally where with low disturbance levels where crusts become dominated by lichens and
mosses (e.g. refs??). In particular, surface cohesivity and strength is provided to Kalahari
soils by Microcoleus species that typically occur in bundles surrounded by a polysaccharide
sheath. This filamentous sheath material forms a web of cohesive material in the upper soil
layer following cycles of wetting and drying (Belnap et al., 2003).
On a localised (site) scale, crust distribution is related directly on a species-specific basis to
the surface vegetation cover, notably bush and tree canopies that determine the spatial
patterns of surface disturbance by grazing animals. This has important implications on the
patterns of potentially erodible sediment source areas (in open areas between bush canopies)
and stable zones (notably under A. mellifera canopies) where crusts can continue to develop
and thus significantly increase their surface strength, roughness and total-N content further.
This localised spatial heterogeneity of crust cover offers a potential mechanism that can
contribute to the rapid expansion of bush encroachment by increasing the nitrogen content
and water retention capacity of soils under A. mellifera canopies (Hagos & Smit, 2005).
There are similarities in such a spatial heterogeneity explanation of bush encroachment to
hypotheses suggested for the SW USA (e.g. Schlesinger et al., 1990; Titus et al., 2002).
However, unlike these Mojave Desert examples where sediment redistribution by water
erosion leads to ‘islands of fertility’ under bush canopies where sediment is deposited, the
increase in soil resource heterogeneity in the Kalahari appears to occur without significant
12
erosion, even on such a localised scale. Indeed, when significant localised wind erosion does
occur it leads to the formation of small nebkha dunes as seen in the Molopo Basin (Dougill &
Thomas, 2002) and around major settlements such as Tshabong and Tshane. Crust formation
in such cases is likely to be restricted by both the burial of the crust and the surface disruption
by saltating sand grains (McKenna Neuman & Maxwell, 2002). Our findings suggest that
erosion rates in the Kalahari rarely reach levels such that the redevelopment of the
cyanobacterial soil cover is threatened. Further research is required to assess the role of
crusts in suppressing erosion rates in more arid sites in south-west Botswana, where the crests
of linear dunes appear active only when surface vegetation cover is below an approximately
14 % threshold (Wiggs et al., 1995).
The ability of crusts to redevelop rapidly following disturbances is clearly portrayed by the
recovery of crust cover on the experimental plots established in this study. This recovery has
occurred despite the proximity of plots to sparsely vegetated surfaces and the low rainfall at
the site since their establishment and offers a plausible explanation for the resilience of
Kalahari soils that has often been inferred in past studies (e.g. Perkins & Thomas, 1993;
Dougill et al., 1998).
Erodibility – roughness, strength, total-N acting together to reduce surface erodibility.
Need for further field process-based research.
Conclusions
Acknowledgements
Usual Research permit reference and help in field from Andy B, Angela C and Jill and Keith
Thomas, Berry Bush Farm, Tshabong.
References
I’m happy to do these next Friday if aiming for mid-April submission.
Tables
Figures
13