**4. Results**

#### **4.1 Pedo-geomorphological characteristics of the lunette dunes**

To explore the distribution pattern of selected soil resources in the lunette dune-pan environment, correlation analysis (Table 3) was used to establish the relationships between soil variables at different sampling depths (SD) and the distance from the pan fringes. It was observed that only sodium indicated a significant negative correlation (*r =0.991, P =0.009*) at SD 0–20 cm in the Tshube lunette dune at = 0.01. Aluminium and organic carbon also exhibited negative correlations (*r* = -*0.980, P = 0.020* and *r = -0.958, P = 0.042* respectively) with distance at = 0.05. At SD 40-60 cm, sodium (*r = -0.958, P = 0.042*) and EC (*r = - 0.985, P = 0.015*) showed negative significant relationships with distance at = 0.05. It was observed that at SD 80-100 cm all soil variables indicated negative relationships with distance except sand fraction, but the relationships were not significant at both = 0.01 and 0.05. ECEC was the only soil variable that showed significant and negative relation with distance (*r = -0.998, P = 0.002*) at SD 130-150 cm and = 0.01. Furthermore, all other soil variables indicated negative relationships with distance except sand fraction, phosphorus and pH. Magnesium (*r = -1, P = 0.011*), manganese (*r = -0.999, P = 0.033*) and phosphorus (*r = 0.999, P = 0.029*) were the only soil variables that exhibited significant relationships with distance at = 0.05 in relation to SD 180-200 cm (Table 3) in the Tshube lunette dune.

In Leremela lunette dune, none of the selected soil variables showed a significant relationship with distance from the pan fringes (SP1) to the slip face slope (SP 4) at SD 0-20 cm and = 0.01 and 0.05 (Table 3). However, all variables displayed negative relationships with distance except sand fraction, aluminium and manganese. Magnesium (*r = 0.984, P = 0.016*) was the only soil variable that indicated positive significant relationship with distance at = 0.05 in relation to the SD 40-60 cm. With the exception of sand fraction, manganese and phosphorus, all other soil variables were negatively related to distance at SD 40-60 cm. From SD 60-200 cm, soil variables and distance were not significantly related at = 0.01 and 0.05.

Investigating Soils, Vegetation and Land Use in

\* *Correlation is significant at the 0.05 level (2-tailed).*

\*\**Correlation is significant at the 0.01 level (2-tailed). SD: Sampling Depth*

Table 3. Correlation analysis of sampling depth and distance from the pan fringes to SP4.

a Lunette Dune-Pan Environment: The Case of Sekoma Lunette Dune-Pan Complex, Botswana 239

All selected soil variables did not show significant relationships with distance at SD 0-20 cm in Kebuang lunette dune. Furthermore, all soil variables were negatively related to distance except sand fraction and aluminium at SD 0-20 cm. Potassium (*r = -0.984, P = 0.016*) and EC (*r = -0.964, P = 0.036*) were the only soil variables that showed negative significant relationships with distance at = 0.05 with respect to SD 40-60 cm. At SD 80-100 cm, the relationships between all soil variables and distance were not significant at both = 0.01 and 0.05. In addition, all soil variables were negatively related to distance except sand fraction and aluminium. Only calcium (*r = -1, P = 0.013*) displayed a perfect negative relationship with distance at = 0.05 and SD 130-150 cm sampling depth. At = 0.01 and 0.05 significant levels, all selected soil variables were not significantly related to distance at SD 180-200 cm in Kebuang lunette dune (Table 3). It was also observed that all the relationships were negative except for sand fraction and pH.

#### **4.2 Plant species distribution patterns and community composition**

In Detrended Correspondence Analysis (DCA) diagram, each site point lies at the centroid of the points of the species that occurs at the sampling site (Hill, 1979). Therefore, Figure 4 mirrors the approximate plant species distribution patterns and plant community composition in the lunette dune-pan environment. On the basis of Figure 4, Inferences were made about the species that were likely to be found at a particular sampling site. Sites that were close to the point of the species were likely to exhibit high density of that particular species, and the density of a species was expected to decrease with the increase in distance from its location.

Two main plant communities were identified in the lunette dune-pan environment. The first one was dominated by *Acacia mellifera* and the other by *Grewia flava* (Figure 4). *A. mellifera*  community was dominant particularly at the sampling points that were located on the slip face of the lunette dunes, and between the lunette dune-pan complex and the settlement area. *G. flava* community was predominated the wind ward slope.

Fig. 4. Plant species distribution in the lunette dune-pan environment (*scale = 1; multiplier = 100*).

All selected soil variables did not show significant relationships with distance at SD 0-20 cm in Kebuang lunette dune. Furthermore, all soil variables were negatively related to distance except sand fraction and aluminium at SD 0-20 cm. Potassium (*r = -0.984, P = 0.016*) and EC (*r = -0.964, P = 0.036*) were the only soil variables that showed negative significant relationships with distance at = 0.05 with respect to SD 40-60 cm. At SD 80-100 cm, the relationships between all soil variables and distance were not significant at both = 0.01 and 0.05. In addition, all soil variables were negatively related to distance except sand fraction and aluminium. Only calcium (*r = -1, P = 0.013*) displayed a perfect negative relationship with distance at = 0.05 and SD 130-150 cm sampling depth. At = 0.01 and 0.05 significant levels, all selected soil variables were not significantly related to distance at SD 180-200 cm in Kebuang lunette dune (Table 3). It was also observed that all the relationships were

In Detrended Correspondence Analysis (DCA) diagram, each site point lies at the centroid of the points of the species that occurs at the sampling site (Hill, 1979). Therefore, Figure 4 mirrors the approximate plant species distribution patterns and plant community composition in the lunette dune-pan environment. On the basis of Figure 4, Inferences were made about the species that were likely to be found at a particular sampling site. Sites that were close to the point of the species were likely to exhibit high density of that particular species, and the density of a species was expected to decrease with the increase in distance from its location. Two main plant communities were identified in the lunette dune-pan environment. The first one was dominated by *Acacia mellifera* and the other by *Grewia flava* (Figure 4). *A. mellifera*  community was dominant particularly at the sampling points that were located on the slip face of the lunette dunes, and between the lunette dune-pan complex and the settlement

Axis1 00 -100 0 100 200 300 400

Fig. 4. Plant species distribution in the lunette dune-pan environment (*scale = 1; multiplier = 100*).

S27

S10

Mokw

S1

S11

S7 S8

S16 S17 S18

S29

S28

S30

A. flec

G. flav

S19 S20 S21S22

S36

A. erio

S24 S25S26

S31 S32 S34 S33 S35

S14

S23

S12 S13

S15

A. mell A. hebe

negative except for sand fraction and pH.

**4.2 Plant species distribution patterns and community composition** 

area. *G. flava* community was predominated the wind ward slope.

Axis2

R. tenu

S2 S3 S5 S4

S9



Z. mucr

0

B. albi

100

S6

E. rigi

R. tric

200

300


\* *Correlation is significant at the 0.05 level (2-tailed).*

\*\**Correlation is significant at the 0.01 level (2-tailed). SD: Sampling Depth*

Table 3. Correlation analysis of sampling depth and distance from the pan fringes to SP4.

Investigating Soils, Vegetation and Land Use in

lunette dune-pan environment (Figure 5).

carbon, ECEC, and phosphorus (Figure 6).

a Lunette Dune-Pan Environment: The Case of Sekoma Lunette Dune-Pan Complex, Botswana 241

The CANOCO programme excluded pH, electrical conductivity, aluminium, and manganese from the CCA for SD 0-20 cm because they exhibited negligible variance (Figure 5). Silt-clay was also eliminated from the analysis because the programme detected collinearity when fitting the variables. *Acacia hebeclada*, *Gardenia volkensii*, S13, S14, S15 and sand fraction were not displayed in the diagram because they introduced polarity in the data points. This suggested that the distribution of *A. hebeclada* and *G. volkensii* were not associated with the distribution of the selected soil variable. Furthermore, it was evident that S13, S14, and S15 were not inhabited by the common plant species in the lunette dune environment. It was observed that *A. mellifera* was closely associated with sites that had high level of calcium, ECEC and organic carbon. These were sampling sites that were mostly located on the slip face of the lunette dunes. The ranking of selected soil variables at SD 0-20 cm on the basis of their significance was ECEC, organic carbon, calcium, potassium, sodium, magnesium and phosphorus in descending order of significance for plants existing in the

Shallow solum (depth >20cm) that existed at S2, S13, S14, S26, and S27 resulted in the exclusion of the sampling sites from the CCA for SD 40-60 cm (Figure 6). S15, S23, S33, *G. volkensii*, *A. hebeclada*, sand, silt-clay were rejected because they polarized data points. EC, pH, aluminium and manganese indicated negligible variance and were therefore removed from the CCA. It was observed that *A. mellifera* still dominated sampling sites that were relatively fertile at SD 40-60 cm, and the distribution of other species were insignificantly influenced by the distribution of selected soil variables at SD 40-60 cm. Figure 6 suggests that up to the depth of 60 cm, *A. mellifera* had a competitive edge over *G. flava* with respect to soil nutrients. The significant soil variables at SD 40-60 cm were potassium, organic

In the CCA for SD 80-100 cm (Figure 7), EC, pH, aluminium and manganese were excluded from the analysis due to their negligible variance. S1, S2, S13, S14, S15, S25, S26, and S27 were not included in the CCA because the solum at the sampling sites was shallow (depth >60cm). Sand and silt-clay were also excluded from the diagram as they indicated collinearity. It is also worth noting that *G. volkensii*, *A. hebeclada* and *Acacia erioloba* were not displayed because they caused polarity of data points. It was observed that the density of *A. mellifera* was positively related to phosphorus and potassium in the soil and negatively related to other variables at SD 80-100 cm. On the other hand, the density of *G. flava, Ehretia rigida* and *Rhigozum trichotomum* were closely linked to the distribution of ECEC, calcium, and organic carbon in the soil. This may suggest that the competitive capacity of *A. mellifera*  for soil nutrients diminished with increase in soil depth. The most significant soil variables

The challenge of shallow solum associated with the pan fringes continued to cause exclusion of some sampling points in the CCA. Consequently, S1, S2, S5, S13, S14, S15, S25, S26, and S27 were excluded for SD 130-150 cm (Figure 8). *G. volkensii* and *A. hebeclada* were also eliminated from the analysis due to data polarity. Sand and silt-clay were disregarded for collinearity. EC, pH, aluminium and manganese were excluded due to negligible variance. It was noted that the number of soil variables that had positive relationships with the distribution of *A. mellifera* continued to decrease with an increase in soil depth. Only phosphorus showed positive relationship with *A. mellifera* distribution in the lunette dunepan environment. Contrarily, *G. flava* distribution had positive relationships with ECEC, sodium and calcium in the soil. This may indicate that *G. flava* gained a competitive

at SD 80-100 cm were iron, phosphorus, ECEC, sodium and calcium (Figure 7).

#### **4.3 Vegetation - Environment relationships**

Canonical Correspondence Analysis (CCA) diagrams show the interrelationships between vegetation and selected soil variables that were observed in the lunette dune-pan environment (Figures 5-9). The diagrams display points that represent species and sampling sites, and arrows that symbolize soil variables. The species and sampling points mutually portray the dominant patterns in community composition to the extent that these could be elucidated by the selected soil variables (ter Braak, 1988). The species points and the arrows of the soil variables mutually depict the plant species distribution along each of the soil variable (ter Braak, 1988). It is worth noting that only the direction and the relative length of the arrows convey essential information (ter Braak, 1995). The length of an arrow representing a soil variable was considered to be equal to the rate of change in the score as inferred from Figures 5-9, hence a measure of how much the plant species distribution differ along that soil variable (Gauch, 1982). As a result, important soil variables were represented by longer arrows (Figures 5-9). The species points that are positioned on the edge or very close to the edge of a CCA diagram are often considered to be rare species (ter Braak, 1995), and such species are usually considered to be very insignificant in CCA. Consequently, plant species of that nature were excluded in the analysis.

Similar researches that have been carried out in the past on vegetation-soil associations focused on the 0-20 cm top soil layer (e.g. Moleele & Perkins, 1998; Moleele, 1999; Smet & Ward, 2006). However, it is widely acknowledged that different plant species exhibit various rooting systems as well as different responses to various environmental factors along environmental gradients (Gauch, 1982). For instance, potassium, phosphorus and sodium gradients may not necessarily be the same at SD 0-20 cm and 180-200 cm, and some plant species may not be able to access essential soil resources from the depth of 200 cm and beyond. To this end, an attempt was made to examine the effect of changes in soil properties due to soil depth variation on the vegetation-soil interrelationships. The assessment was premised on the inferences from Figures 5-9. Worth highlighting is the scale of the diagrams (Figures 5-9); 1 unit in the plot corresponds to 1 unit for the sites, to 1 unit for the species and to 10 units for the soil variables.

Fig. 5. Relationships between soil properties (SD 0-20 cm) and plant species distribution.

Canonical Correspondence Analysis (CCA) diagrams show the interrelationships between vegetation and selected soil variables that were observed in the lunette dune-pan environment (Figures 5-9). The diagrams display points that represent species and sampling sites, and arrows that symbolize soil variables. The species and sampling points mutually portray the dominant patterns in community composition to the extent that these could be elucidated by the selected soil variables (ter Braak, 1988). The species points and the arrows of the soil variables mutually depict the plant species distribution along each of the soil variable (ter Braak, 1988). It is worth noting that only the direction and the relative length of the arrows convey essential information (ter Braak, 1995). The length of an arrow representing a soil variable was considered to be equal to the rate of change in the score as inferred from Figures 5-9, hence a measure of how much the plant species distribution differ along that soil variable (Gauch, 1982). As a result, important soil variables were represented by longer arrows (Figures 5-9). The species points that are positioned on the edge or very close to the edge of a CCA diagram are often considered to be rare species (ter Braak, 1995), and such species are usually considered to be very insignificant in CCA. Consequently,

Similar researches that have been carried out in the past on vegetation-soil associations focused on the 0-20 cm top soil layer (e.g. Moleele & Perkins, 1998; Moleele, 1999; Smet & Ward, 2006). However, it is widely acknowledged that different plant species exhibit various rooting systems as well as different responses to various environmental factors along environmental gradients (Gauch, 1982). For instance, potassium, phosphorus and sodium gradients may not necessarily be the same at SD 0-20 cm and 180-200 cm, and some plant species may not be able to access essential soil resources from the depth of 200 cm and beyond. To this end, an attempt was made to examine the effect of changes in soil properties due to soil depth variation on the vegetation-soil interrelationships. The assessment was premised on the inferences from Figures 5-9. Worth highlighting is the scale of the diagrams (Figures 5-9); 1 unit in the plot corresponds to 1 unit for the sites, to 1 unit for the species

Axis1 .5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

A. erio

A. mell

<sup>P</sup> %OC

Ca

ECEC

Axis2



R. tric


Mg Na

Fig. 5. Relationships between soil properties (SD 0-20 cm) and plant species distribution.


0

1

2

A. flec

B. albi

R.tenu

Mokw

K

E. rigi G. flav

Z. mucr

3

**4.3 Vegetation - Environment relationships** 

plant species of that nature were excluded in the analysis.

and to 10 units for the soil variables.

The CANOCO programme excluded pH, electrical conductivity, aluminium, and manganese from the CCA for SD 0-20 cm because they exhibited negligible variance (Figure 5). Silt-clay was also eliminated from the analysis because the programme detected collinearity when fitting the variables. *Acacia hebeclada*, *Gardenia volkensii*, S13, S14, S15 and sand fraction were not displayed in the diagram because they introduced polarity in the data points. This suggested that the distribution of *A. hebeclada* and *G. volkensii* were not associated with the distribution of the selected soil variable. Furthermore, it was evident that S13, S14, and S15 were not inhabited by the common plant species in the lunette dune environment. It was observed that *A. mellifera* was closely associated with sites that had high level of calcium, ECEC and organic carbon. These were sampling sites that were mostly located on the slip face of the lunette dunes. The ranking of selected soil variables at SD 0-20 cm on the basis of their significance was ECEC, organic carbon, calcium, potassium, sodium, magnesium and phosphorus in descending order of significance for plants existing in the lunette dune-pan environment (Figure 5).

Shallow solum (depth >20cm) that existed at S2, S13, S14, S26, and S27 resulted in the exclusion of the sampling sites from the CCA for SD 40-60 cm (Figure 6). S15, S23, S33, *G. volkensii*, *A. hebeclada*, sand, silt-clay were rejected because they polarized data points. EC, pH, aluminium and manganese indicated negligible variance and were therefore removed from the CCA. It was observed that *A. mellifera* still dominated sampling sites that were relatively fertile at SD 40-60 cm, and the distribution of other species were insignificantly influenced by the distribution of selected soil variables at SD 40-60 cm. Figure 6 suggests that up to the depth of 60 cm, *A. mellifera* had a competitive edge over *G. flava* with respect to soil nutrients. The significant soil variables at SD 40-60 cm were potassium, organic carbon, ECEC, and phosphorus (Figure 6).

In the CCA for SD 80-100 cm (Figure 7), EC, pH, aluminium and manganese were excluded from the analysis due to their negligible variance. S1, S2, S13, S14, S15, S25, S26, and S27 were not included in the CCA because the solum at the sampling sites was shallow (depth >60cm). Sand and silt-clay were also excluded from the diagram as they indicated collinearity. It is also worth noting that *G. volkensii*, *A. hebeclada* and *Acacia erioloba* were not displayed because they caused polarity of data points. It was observed that the density of *A. mellifera* was positively related to phosphorus and potassium in the soil and negatively related to other variables at SD 80-100 cm. On the other hand, the density of *G. flava, Ehretia rigida* and *Rhigozum trichotomum* were closely linked to the distribution of ECEC, calcium, and organic carbon in the soil. This may suggest that the competitive capacity of *A. mellifera*  for soil nutrients diminished with increase in soil depth. The most significant soil variables at SD 80-100 cm were iron, phosphorus, ECEC, sodium and calcium (Figure 7).

The challenge of shallow solum associated with the pan fringes continued to cause exclusion of some sampling points in the CCA. Consequently, S1, S2, S5, S13, S14, S15, S25, S26, and S27 were excluded for SD 130-150 cm (Figure 8). *G. volkensii* and *A. hebeclada* were also eliminated from the analysis due to data polarity. Sand and silt-clay were disregarded for collinearity. EC, pH, aluminium and manganese were excluded due to negligible variance. It was noted that the number of soil variables that had positive relationships with the distribution of *A. mellifera* continued to decrease with an increase in soil depth. Only phosphorus showed positive relationship with *A. mellifera* distribution in the lunette dunepan environment. Contrarily, *G. flava* distribution had positive relationships with ECEC, sodium and calcium in the soil. This may indicate that *G. flava* gained a competitive

Investigating Soils, Vegetation and Land Use in

a Lunette Dune-Pan Environment: The Case of Sekoma Lunette Dune-Pan Complex, Botswana 243

iron in the soil. This therefore gave impetus to the observation that *G. flava* gained a competitive advantage over *A. mellifera* for soil nutrients as soil depth increased. Phosphorus, sodium, potassium, iron, calcium and ECEC were identified as significant soil variables in relation to plant species distribution in the lunette dune-pan environment.

Axis1 -3 -2 -10 1 2 3

R.tenu

Axis2

3

S29


Axis1 -4 -3 -2 -1 0 1 2 3

Mokw

E. rigi

Ca ECEC

G. flav

Fe

R. tric R.tenu

Mg

Fig. 9. Relationships between soil properties (180-200cm depth) and plant species



A. erio

A. flec

%OC

Z. mucr

A. mell

B. albi


0

1

2

Fig. 8. Relationships between soil properties (130-150cm depth) and plant species



G. flav

S28

S30

S3 S4 S6

0

S16 S17 S18

S19

1

S7 S8

2

%OC

Axis2

A. erio

S10 S11 S12

S20 S21 S22

A. flec

Mg

A. mell

S23

P

P

S24

S31 S32 S34 S35 S33 S36

Fe

3

B. albi

S9

Mokw

E. rigi

Z. mucr

Ca

K

Na

K

distribution.

distribution.

R. tric

Na

ECEC

4

advantage over *A. mellifera* for soil nutrients as soil depth increased. Significant soil variables that had impacts on plant species distribution in the lunette dune environment at SD 130-150 cm were iron, sodium, calcium and ECEC.

Fig. 6. Relationships between soil properties (SD 40-60 cm) and plant species distribution.

Fig. 7. Relationships between soil properties (80-100cm) and plant species distribution.

Due to the shallowness of the solum, S1, S2, S3, S4, S5, S13, S14, S15, S25, S26, and S27 were not included in the CCA for SD 180-200 cm (Figure 8). Furthermore, *G. volkensii*, *A. hebeclada*, EC, pH, aluminium, manganese, sand and silt-clay were excluded in the CCA owing to negligible variance. It was observed that the number of soil variables that had positive relationships with the distribution of *A. mellifera* still continued to decrease with an increase in soil depth. In addition, similar to SD 130-150 cm, only phosphorus indicated positive relationship with *A. mellifera* distribution in the lunette dune-pan environment. In contrast, *G. flava* distribution had positive relationships with ECEC, sodium, calcium, potassium and

advantage over *A. mellifera* for soil nutrients as soil depth increased. Significant soil variables that had impacts on plant species distribution in the lunette dune environment at

.0 -0.5 0.0 0.5 1.0 1.5 2.0

Fig. 6. Relationships between soil properties (SD 40-60 cm) and plant species distribution.

Axis 2

2

1

S10 S11

Mokw

%OC

S5

S9

Axis 1 -3 -2 -10 1 2 3

S19 S20

R.tenu

S29

S36

Na

Fe

K

S34

S21 S22


Fig. 7. Relationships between soil properties (80-100cm) and plant species distribution.

Due to the shallowness of the solum, S1, S2, S3, S4, S5, S13, S14, S15, S25, S26, and S27 were not included in the CCA for SD 180-200 cm (Figure 8). Furthermore, *G. volkensii*, *A. hebeclada*, EC, pH, aluminium, manganese, sand and silt-clay were excluded in the CCA owing to negligible variance. It was observed that the number of soil variables that had positive relationships with the distribution of *A. mellifera* still continued to decrease with an increase in soil depth. In addition, similar to SD 130-150 cm, only phosphorus indicated positive relationship with *A. mellifera* distribution in the lunette dune-pan environment. In contrast, *G. flava* distribution had positive relationships with ECEC, sodium, calcium, potassium and




S30

S16 S17 S18

0

G. flav

S28

S7 S8

S3 S4

S6

ECEC

A. mell

S33 S34

P

%OC

A. mell

S31 S32

S35

S12

P

S23 S24

A. flec

S33

ECEC

Axis1

K

SD 130-150 cm were iron, sodium, calcium and ECEC.

A. flec

R.tenu

E. rigi

Axis2



Mokw

S10

Na

S24 S25

S22

Mg

Fe Ca

S23

Z. mucr S1 S5 S3S4

S11

S12 S16

S31 S32

S35

A. erio

B. albi

E. rigi

Ca

R. tric

Z. mucr

B. albi

S17 S18 S19S20 S21

S8 S9

G. flav

S28

S30

R. tric

S29

S6


0.0

S7

0.5

1.0

S36

1.5

2.0 2.5 iron in the soil. This therefore gave impetus to the observation that *G. flava* gained a competitive advantage over *A. mellifera* for soil nutrients as soil depth increased. Phosphorus, sodium, potassium, iron, calcium and ECEC were identified as significant soil variables in relation to plant species distribution in the lunette dune-pan environment.

Fig. 8. Relationships between soil properties (130-150cm depth) and plant species distribution.

Fig. 9. Relationships between soil properties (180-200cm depth) and plant species distribution.

Investigating Soils, Vegetation and Land Use in

erosion in the area.

pan which occurred gradually over the years.

within the lunette dune-pan environment.

pan as the main source of water for livestock.

to shortage of fodder and water in their areas.

**5.1 Pedo-geomorphology of the lunette dune complex** 

lunette dune-pan environment.

**5. Discussion** 

the perpetual development of some minor dunes in the area.

drivers linked with the continuous development of lunette dunes.

a Lunette Dune-Pan Environment: The Case of Sekoma Lunette Dune-Pan Complex, Botswana 245

They noticed an increase in dune size and height with simultaneous shrinkage of the

 They recognized that the lunette dunes were not a single hammock of sand, but a dune field of distinct sand dunes. Consequently, they identified the main lunette dunes as Tae, Kebuang, Boisi, Leremela and Tshube from east to west. They were also aware of

 Excessive wind erosion was identified as the main agent of soil transfer from the environment onto the lunette dunes. This coupled with the trees which had grown on the lunette dunes trapping the aeolian soil particles, were acknowledged as the main

 They noticed a rapid increase in the height and size of the lunette dunes between the years 1985-1987 which they attributed to the severe drought that occurred in the area during that period. They pointed out that due to the drought, vegetation was devastated leaving large areas of bare land, creating conducive conditions for rapid soil

They associated the drought with bush encroachment or thickening which was evident

 As evidence to the climatic changes that they observed in the area over a long period of time, they cited a decline in the amount of rainfall that the Sekoma area received over the years. Furthermore, they indicated that in the past, the annual rainfall was sufficient to fill the pan and that the pan was able to hold surface water for longer periods. They realized that this was no longer the case. They attributed the changes to loss of surface water holding capacity due to pan sedimentation. Pan sedimentation was associated with excessive soil erosion that continued to occur over the years causing the pan shrink. As a result, the community relied heavily on hand-dug wells located within the

 The community indicated that the lunette dunes did not contribute significantly to productivity in pastoral farming in their area due to shortage of fodder resources in the

 Finally, they pointed out that the lunette dune-pan environment was subjected to increasing land use pressure due to the increase in the population of the community and livestock in the area. However, they indicated that the situation had been aggravated by some farmers from other villages that had relocated close to Sekoma due

Soils in the Kalahari area are sandy grains constituted mainly by quartz and small amounts of zircon, garnet, feldspar, ilmenite and tourmaline (Wang et al., 2007; Leistner, 1967). Analysis of soil properties of the Sekoma lunette dune-pan environment did not indicate otherwise as the lunette dunes were more than 95% sandy up to the depth of 200 cm. A soil profile established in one of the lunette dunes indicated no signs of soil horizons up to the depth of 200cm. This showed dominance of sand fraction in the soil texture of the lunette dunes. Goudie and Wells (1995) and Lancaster (1978) pointed out that the deflation of sediment directly from the pan floor during dry climatic condition periods resulted in the


Table 4. Eigen values of the first four axes and the species-environment correlations.

The eigen values (λ) of the DCA and CCA were determined to further assess the degree to which the selected soil variables could explain plant species distribution in the lunette dunepan environment (Table 5). The eigen value is usually referred to as the "per centage variance accounted for" (ter Braak, 1988). It always ranges from one (1) to zero (0), and the higher the value the more important the ordination axis. Furthermore, eigen values of ca. 0.3 and higher are usually common in ecological applications (ter Braak, 1988). However, an ordination diagram that explains only a low per centage of the total variance in the species data may still be informative (ter Braak, 1988). Eigen values are usually in the form of a decreasing order with values for axes 1 and 2 being larger than those of axes 3 and 4 as is the case in Table 5 which shows the species-environment correlations (*r*) and the eigen values for the first four axes. It was observed that some eigen values were lower than 0.3 (Table 5). This suggested limitations on the use of data on selected soil variables to explain variation in plant species distribution. This was not out of the ordinary as it is widely acknowledged that plant species distribution in any ecosystem is a function of numerous environmental factors, and that it is practically impossible for any scientific research to exhaustively and concurrently incorporate all environmental factors of potential significance into a particular study. Therefore, the selected soil variables were considered sufficient to comprehensively shed light on the patterns of plant species distribution in the lunette dune-pan environment in Sekoma.

#### **4.4 Social survey**

It was established that the village of Sekoma did not originate where it was currently located. The village originated in the western side (Sekoma West) of the lunette dunes and a considerable portion of the community decided to migrate to the eastern side (Sekoma) of the lunette dunes between the years 1924-1927. However, some few members of the community decided to remain in Sekoma West and they still inhabited the area at the period of this research. They indicated that there was nothing major that caused the migration. However, observations indicated that some changes in their environment instigated the migration. For instance, observation of abandoned old hand-dug wells located in the western side of the pan suggested a possible exhaustion of underground water resources at that site. The migration implied a shift in land use pressure from one side of the lunette dune-pan complex to the other. During the discussions, it became apparent that over the years the local community had amassed a wealth of indigenous knowledge with regard to the changes in their environment. The following is an account of the perceptions of the local community pertaining to the lunette dune-pan environment:

 The community perceived the existence of the lunette dunes in their environs as a natural phenomenon.

