**3. Methods**

Wetland areas occurring between the Tembe Elephant Park (TEP) and Kosi-Bay were identi‐ fied using Google Earth and 1:3000 Orthophotos, and verified with a field visit to the area. The wetlands were selected based on accessibility, safety, land owner consent, data availability, and land use.

Five wetland types were identified and investigated in this study (Figure 2):

**◦** Scattered depression type wetlands between vegetated coastal dunes,

**◦** Intense local utilisation of the fertile peaty soils for subsistence.

**◦** Clay lenses occur at 300 – 500 mm depth on the banks of the system,

**◦** A series of scattered seasonal pans occurring parallel to the MS system,

**◦** Outside the Park the pans are open and degraded (DP System),

**◦** Inside the Tembe Elephant Park the pans occur as open areas surrounded by closed

**◦** High clay content in the soil results in a perched water table for several months per year

**◦** Located on the upland flat area between the Tembe Elephant Park and Manguzi,

**◦** Slightly undulating Lala Palm veld with interspersed spaces of open, moist grassland,

**◦** These wetlands are seasonal and water table fluctuation plays a prominent rolep [17].

Wetland areas occurring between the Tembe Elephant Park (TEP) and Kosi-Bay were identi‐ fied using Google Earth and 1:3000 Orthophotos, and verified with a field visit to the area. The wetlands were selected based on accessibility, safety, land owner consent, data availability,

**•** Interdune-depression (IDD) System –

278 Biodiversity - The Dynamic Balance of the Planet

**•** Muzi North Swamp System (MS) – **◦** A linear valley-bottom system,

woodland (PP System),

**•** Upland Wetland (PL) System –

[31],

**3. Methods**

and land use.

**◦** Linked with the regional water table,

**◦** Peaty soil in the pristine wetlands [33],

**◦** Linked with the regional water table,

**◦** The permanently wet areas of the system are peaty,

**•** Perched Pan (PP) and Depression (DP) Systems –

**◦** The pans are clay-rich, calcareous duplex soils.

**◦** Depressions occur in large patches in the Palm Veld.

**Figure 2.** Wetland systems occurring in the northern parts of the Maputaland Coastal Plain.

Between three to five wetlands in each system were selected to be surveyed. These wetlands were first stratified into their various vegetation zones. Between three and five vegetation zones were identified in each wetland. For the purpose of the data collection in the field the different zones sampled were based on vegetation communities observed and not on hydro‐ logical regime. Therefore these zones were not termed 'permanent', 'seasonal', 'temporary', or 'terrestrial', but rather as Zone 1, Zone 2, Zone 3, etc (Figures 3 and 4). However after the data analysis the different zones were grouped into the different wetness zones as listed above and discussed accordingly under the discussion section.

x 2 m) were placed within the different zones in a stratified random manner. Plant species were identified in the field, while the unknown plant species were collected and sent to the

The Ecology and Species Richness of the Different Plant Communities Within Selected…

http://dx.doi.org/10.5772/58219

281

The vegetation relevés were captured into TURBOVEG for Windows 1.97 [23] and exported to JUICE 6.5 [24]. A modified TWINSPAN was performed in JUICE using the Whittaker's betadiversity, with the following pseudospecies cut-levels: 0, 1, 5, 25, 50, and 75. The Fisher Exact Fidelity Test at P<0.001 were used. The final classification was manually refined according to the Braun-Blanquet procedure [25]. No re-arrangements of clusters or relevés were done, but

Six different ordination methods were applied to the plant community data in PCOrd [26]-the Bray-Curtis ordination, Canonical Correspondence Analysis, Weighted Averaging, Reciprocal Averaging, Detrended Correspondence Analysis (DCA), and Nonmetric Multidimensional Scaling (NMS). The DCA and NMS analyses gave the best results. The DCA ordination results are presented in this study, as it emphasized the variation and combination of the plant communities better than the NMS results. Various environmental factors thought to influence the distribution of the vegetation communities were superimposed on the ordination results. The overlay of the floristic communities identified by [27], the five wetland systems, and

The Chi-Square Test [28] was performed on the data to determine whether significant differ‐

The modified TWINSPAN analysis [29] resulted in the identification of 11 plant communities that can be grouped into eight major communities and six sub communities. The results of the DCA ordination for all the plant communities are contained in Figure 5. From the DCA ordination axes 1 and 2 were selected as it was the most interpretable ordination. An Eigen‐

The clay communities (communities 1 – 3) are positioned distinctly to the right of the ordination diagram. The communities which are located on predominantly sandy substrates (e.g. Community 4) are found on the extreme opposite end from the clay communities. The close proximity between Community 2 and sub community 6.2 is because both originate from the PP system. Sub communities 7.1, 7.2, and 7.3, all from the MS system, are affiliated with each other despite hydrological differences between the different zones. Sub community 5.1 and most of Community 8 originate from the PL System, explaining this association. The significant distance between sub community 5.1 (PL System) and 5.2 (IDD System) is as a result of the fact that they occur in different systems, despite similar environmental settings. Sub community 7.3 has a wide distribution, as some of its dominant species occur in other communities as well. Of these the graminoids *Stenotaphrum secundatum* and *Cynodon dactylon* are known to be

ences exist between the species richness of the different plant communities.

value of 0.933 and 0.828 were obtained for Axis 1 and Axis 2 respectively.

variable in their habitat preference, and are not limited to a certain environment.

South African National Biodiversity Institute for identification.

only species groups were manually re-arranged.

substrate type are included in the final results.

**4. Results**

**Figure 3.** The Interdune-depression (IDD) System with an example of the zone delineation.

**Figure 4.** The Upland Wetland (PL) System, with an example of zone delineation.

Vegetation surveys were conducted in March 2010 following the Zurich-Montpellier (Braun-Blanquet) School of total floristic compositions approach [22]. A total of 72 sample plots (2 m

x 2 m) were placed within the different zones in a stratified random manner. Plant species were identified in the field, while the unknown plant species were collected and sent to the South African National Biodiversity Institute for identification.

The vegetation relevés were captured into TURBOVEG for Windows 1.97 [23] and exported to JUICE 6.5 [24]. A modified TWINSPAN was performed in JUICE using the Whittaker's betadiversity, with the following pseudospecies cut-levels: 0, 1, 5, 25, 50, and 75. The Fisher Exact Fidelity Test at P<0.001 were used. The final classification was manually refined according to the Braun-Blanquet procedure [25]. No re-arrangements of clusters or relevés were done, but only species groups were manually re-arranged.

Six different ordination methods were applied to the plant community data in PCOrd [26]-the Bray-Curtis ordination, Canonical Correspondence Analysis, Weighted Averaging, Reciprocal Averaging, Detrended Correspondence Analysis (DCA), and Nonmetric Multidimensional Scaling (NMS). The DCA and NMS analyses gave the best results. The DCA ordination results are presented in this study, as it emphasized the variation and combination of the plant communities better than the NMS results. Various environmental factors thought to influence the distribution of the vegetation communities were superimposed on the ordination results. The overlay of the floristic communities identified by [27], the five wetland systems, and substrate type are included in the final results.

The Chi-Square Test [28] was performed on the data to determine whether significant differ‐ ences exist between the species richness of the different plant communities.
