Short-term vegetation change on rehabilitated peatland on Rietvlei Nature Reserve

Natural peatlands occur on the Rietvlei Nature Reserve. Before the Pretoria City Council acquired the land, these peatlands were mined by private land-owners. Ditches were constructed to drain the area for mining and the peatlands became desiccated. Later the area was proclaimed as a nature reserve and has since then been managed as such. Rehabilitation of the drained peatland on Rietvlei Nature Reserve first started in 2000 as a Working for Water project. The aim of the rehabilitation was to close the ditches and rewet the peatland, to enable possible revival of the peatland. A baseline vegetation survey was undertaken during the summer (March to April) of 2001 to determine the nature of the pioneer communities that established on the rehabilitated area. This survey was repeated during the summer (March to April) of 2002 to detect changes in the vegetation. The same sample plots were used on both occasions. The initial pioneer vegetation was mostly composed of weedy annuals.


Introduction
Wetlands are very important in nature, because they form unique habitats for certain aquatic and hydrophilic plant species and also result in an increase in the number of waterfowl and other fauna typically associated with wetlands.Wetlands lessen the devastating effects of floods and are responsible for clearer and healthier surface water.They act as a filter for debris and certain toxic substances, such as heavy metals, and absorb large quantities of water that are eventually released over time into natural drainage channels (Hey & Philippi 1999).Wetlands in South Africa do not have high plant species diversity but some of the species that do occur in wetlands are restricted to these habitats (Bloem et al. 1993;Myburgh et al. 1995;Eckhardt et al. 1993;Smit et al. 1995).Additional to these functional and biological values, wetlands also have consumptive uses, e.g. they provide water to man and animals, they generate energy and can be used for fish farming.Non-consumptive uses include recreation, tourism and transport (Rand Water Scientific Services 1998).
The goal of a rehabilitation project may be to a particular species, or may include the rehabilitaion of a whole ecosystem or landscape, or to restore the ecosystem functions (Ehrenfeld 2000).South Africa ratified the Ramsar convention (Cowan & Marneweck 1996) and are therefore under an obligation to conserve wetlands.The rehabilitation of wetlands play an important role in wetland conservation, as mentioned in the Ramsar convention.
When rehabilitating wetlands it is important to understand wetland functions and to give the system time to react.Some rehabilitation projects may take up to 50 years before rehabilitation has been accomplished (Middleton 1999;Mitsch & Wilson 1996).A wetland ecosystem is not a stable system, because many disturbances such as flood pulsing, fire and herbivory occur naturally in many wetlands (Middleton 1999;Mitsch & Wilson 1996), especially in a country such as South Africa with its high frequency of thunder storms, fire-prone grasslands and large variety of large herbivores.A restored wetland should be able to deal with these kinds of disturbances.
Only a few peatlands have been rehabilitated in South Africa and information is limited.In 1995 it has been proposed that the Orange River Mouth salt marsh should be rehabilitated (Marneweck 1995), but in 1998 almost no progress has been made and many problems were experienced with the project (Dini 1998).Another similar project was the rehabilitation of Blesbokspruit (Dini 1998).In 1996 the Free State Department of Environmental Affairs and Tourism and Rand Water started the rehabilitation of the Seekoeivlei as well as other wetlands in the north-eastern Free State (Collins 1998).The rehabilitation of the Wilge River started as a Rand Water and Working for Water project in the same year (Rand Water & Working for Water Programme 1998).These projects take time and it is therefore difficult to determine if the projects have been successful.The Orange River Mouth, Blesbokspruit and Seekoeivlei are Ramsar sites (Cowan & Marneweck 1996).On the contrary, there is much infor-mation available on the rehabilitation of European peatlands, but those are completely different ecosystems, which developed under very different climatic conditions.
During peat mining operations in the past on Rietvlei Nature Reserve, ditches with sharp edges were constructed to drain the peatland.Some of the water was used for irrigation of farms in the area.There are four dolomite springs on the Reserve.The water from these springs is piped to the Rietvlei Water Works.A large amount of water was therefore drained from the peatland, which subsequently became dry and resulted in the vegetation being altered.
Rehabilitation of the Rietvlei peatland started as a "Working for Water" project.The aim of the rehabilitation project was mainly to rewet the peatland to allow hydrophytic vegetation to re-establish.The sharp edges of the main channel were sloped and gabions were placed in the main channel to stop the peat from rapidly drying out and to allow floodwater to easily overspill onto the adjacent peatland, thus enhancing the rewetting process.The peat recovered from the sloping process was used to fill some of the smaller ditches.Soil and straw gabions were placed in the ditches to stabilise them and to retain the water.Rock filled gabions were placed in the main channel to slow the flow, increase the flooding frequency and duration and to promote the deposition of sediment.
It was necessary to conduct a baseline vegetation survey to determine the pioneer plant communities that established naturally on the rehabilitated peatland.The communities closer to the climax, that occur on areas with less disturbance, were also surveyed, to gain knowledge on the potential climax vegetation.

Study area
The Rietvlei Dam Nature Reserve is located at about 25º50'S and 28º20'E.The reserve became established as a result of the Rietvlei Water Scheme to provide drinking water to the people of Pretoria.The Rietvlei Dam is located in the northern portion of the reserve and the peatland is located upstream from the dam in the southern portion of the reserve.
The City Council of Pretoria acquired the farm in 1929.The dam was built in the Ses Myl Spruit during the Great Depression and it was completed in 1934.The catchment area of the Ses Myl Spruit is to the south of Pretoria, towards Kempton Park.The area was declared the Rietvlei Game Reserve in 1937.In 1948 it became the Rietvlei Reserve for Game and indigenous Flora.In 1954 the area became the Van Riebeeck Nature Reserve and in 1992 the name was changed again to the Rietvlei Nature Reserve (Culture and Recreation Department 1997).
To drain the peatland, a deep ditch was dug through it, and this now forms part of the Ses Myl Spruit main channel.The main channel has been severely eroded and the water flows very fast, with high energy levels.The main channel and the peatland on the western side of the main channel have been rehabilitated.Certain portions of the peatland, on the eastern side of the main channel, were extensively mined and almost all the peat was removed.This area has not been rehabilitated.The underlying clay soil was exposed.On the western side of the main channel deep deposits of peat still remained intact, but they were severely disturbed by drainage ditches.The vegetation was disturbed during the construction of the ditches and the ditches altered the hydrological regime of the peatland.

Methods
The peatland area was stratified into the following units: ditches filled with peat; ditches filled with soil; sloped main channel edges; reed areas on old peatdeposits; and the mined area east of the main channel.
During the summer of 2001, 22 sample plots of 4 m x 4 m were placed randomly in these different units.The exact locality of each plot was recorded using a Global Positioning System (GPS).In each plot all plant species present were listed and each assigned the approriate Braun-Blanquet cover-abundance value (Werger 1974).Some aspects of the habitat were also noted such as soil type (peat or clay), degree of disturbance and rehabilitation measures.
The floristic data were entered into the TURBOVEG data basis (Hennekens 1996a) and the data were subjected to classification and ordination procedures to identify plant communities.TWINSPAN (Hill 1979b) was used for the initial classification of the vegetation, and the result was refined by using MEGATAB (Hennekens 1996b).The final classification is presented in a phytosociological table.The procedure of Detrended Correspondence Analysis was performed by applying the DECORANA software (Hill 1979a).This ordination was used to indentify environmental gradients responsible for plant species distribution, and hence the occurrence of plant communities.
To determine a probable change in vegetation (path of succession) after one year, the floristic survey was repeated during the summer of 2002 on the same sample plots as 2001.The data of the two surveys were pooled and to determine the direction of change in vegetation, the ordination was re-applied to the pooled data set.

Classification
The result of the classification is presented in the phytosociological table (Table 1).
The classification of the plant communities found in the baseline survey (2001 data set) is as follows:
A hierarchical and habitat interpretation of the recognised plant communities is given in Fig. 1.
The plant community occurs on the ditches that were filled with peat or soil during the rehabilitation of the peatland.The sub-community has 10 species and an average of seven species per relevé.

Ordination
In the ordination on the 2001 data set (Fig. 2), the separation of communities 1 and 2 on the drained, rehabilitated side of the channel and community 3 on the mined side of the channel is clear.Most of the relevès in a sub-community are close together but there are a few exeptions where the relevès are in a state of transition between two communities.Although relevè 2 is still in Sub-community 1.2 it is also close to Sub-community 2.2.It is therefore closer to the climax (Subcommunity 2.2) than the other relevès in Sub-community 1.2.The environmental gradient from dry to wet can be seen in the ordination as well as the gradients from the pioneer community to the two different close-to-climax communities.These communities occur in the areas with less recent disturbance and are more similar to climax vegetation than the pioneer communities.The succession of community 1 will be to either one of them.The one close-to-climax community should not change to become like the other close-to-climax community.

Monitoring vegetation change
The following is the result of the ordination as performed on the pooled data set for the specific purpose to monitor vegetation change from 2001 to 2002.The ordination (Fig. 3) indicates that the vegetation of subcommunity 1.1 moved away (to the top left in the diagram) from the other communities.This is probably because it is situated at the edge of the wetland and has more terrestrial species than wetland species and is becoming more terrestrial, like the adjacent grassland community.
The vegetation of sub-community 1.2 changed (moved downwards in the diagram) in the direction of the of the reed-dominated community 2.2.This is probably an indication of succession form a pioneer community towards the climax Phragmites australis dominated vegetation.
The vegetation of Sub-community 2.1 stayed almost the same and although reeds are present it is dominated by the tall weed Phytolacca octandra.It does however share a lot of species with the reed-dominated sub-community 2.2.Relevé 11 moved closer to 2.1 and it therefore seems that it has degraded, due to later disturbance by the floods.
Sub-community 2.2 represents the close-toclimax vegetation dominated by Phragmites australis.This vegetation stayed almost the same, however, the vegetation of sample plots 14 and 15 seems to have degraded, and moved towards sub-community 1.2, which is an earlier successional stage.
Community 3 stayed almost the same, but the vegetation at relevé 22 moved towards the more reed-dominated group.This may be because it is the only relevé in community 3 with Phragmites australis.Relevé 20 in community 3 was probably more disturbed

Discussion
The baseline survey The first division of TWINSPAN on the 2001 baseline data separated the un-rehabilitated, mined areas and areas disturbed by the rehabilitation.The rehabilitation disturbed the sites because of trampling, the construction of gabions and the movement of peat from one site to another.The mined area is on the eastern side of the main channel and the disturbed rehabilitated area is on the western side.These communities differ largely and have few species in common.These species are in species groups G and J and include Persicaria lapathifolia, Tagetus minuta and Cyperus fastigiatus, mostly pioneer, annual species.The DECORANA ordination diagram also shows significant separation between these plant communities.There are two communities on the western, rehabilitated, side of the main channel and each of these have two sub-communities.There is only one community on the eastern side, with two sub-communities.Each community and subcommunity occurs on a particular habitat.
Community 1 is on the filled ditches that were rehabilitated recently and the weedy pioneer plant species of the two sub-communities are mostly the same.Most of the pioneer species are exotic weeds.The vegetation of sub-community 1.1 differs from 1.2, due to the dominance of the perennial stoloniferous grass Cynodon dactylon, while the annual weedy Solanum sisymbrifolium is very conspicuous in 1.1.Persicaria species and Amaranthus hibridus are more dominant in sub-community 1.2 and the annual grass Urochloa panicoides occurs in this sub-community.Phragmites australis is becoming established in sub-community 1.2, but not in 1.1.The differences in the vegetation of the two sub-communities are probably a result of the difference in soil.Sub-community 1.1 is on relatively dry clay and sub-community 1.2 on moist peat.There are very few general species.Communities 1 and 2 share some species, but they do not share many species with community 3. Community 1 has the highest species richness and community 3 has the lowest species richness.The pioneer communities therefore contribute more to species diversity than the close-to-climax communities, which are dominated by tall groups of reeds or sedges, with few other species present.Community 3 is already stable and there is probably more competition between dominant species, while community 1 is a pioneer community with many weeds and open spaces for plants to establish.Few individuals of these weeds occur in the established communities and they should disappear in time.In contrast, the flood regime of the wetland may cause sufficient disturbance for the weedy species to persist in the pioneer communities for some time.Wetland species are adapted to these disturbances, but they need time to establish in a disturbed wetland and the flooding regime still need to stabilise.

Vegetation change
The study was conducted over a short period with only two data sets one year apart.This is important to take into consideration, since wetlands take a long time to rehabilitate.The study is at the start of rehabilitation and it is difficult to predict what the vegetation will look like after a few years.This study will hopefully give us some indication.
After the flooding disturbance at the end of the year, that destroyed most of the pioneer vegetation, the peat was wetter.This flooding was caused by the rehabilitation measures.It seems that the rehabilitation was successful and that succession is taking place towards the climax vegetation, even after only a single year.Sub-community 2.1 may be a problem, because it does not seem to advance towards a reed-dominated community, but stays the same.This is a problem because this community are dominated by weeds.It would be interesting to see if this community would change over a longer period of time.Further monitoring would confirm whether the changes in vegetation is of a permanent nature, or whether the vegetation is merely fluctuating due to yearly changes in the moisture regime.The moisture regime is changing due to the rehabili-taion, but yearly fluctuations in the amount of rainfall also plays a role.Although it seems as if the peatland is recovering the time frame of the study is to short to confirm this as a fact.
Other disturbances in the peatland are grazing, by the animals occuring in the reserve as well as trampling by the buffalo.Any of the disturbances could have caused the degeneration of some of the relevés, for example relevé 15.

Fig. 2 .
Fig. 2. Scatter diagram of a DECORANA ordination illustrating the difference between the communities along an environmental gradient using the 2001 dataset.Axis 1 was plotted against Axis 2.

Fig. 3 .
Fig. 3. Scatter diagram of a DECORANA ordination illustrating the change of the vegetation of the communities after one year using the pooled dataset of the 2001 and 2002 data.Axis 1 was plotted against Axis 2.

Table 1
Phytosociological table of pioneer vegetation on the rehabilitated peatland at Rietvlei Nature ReserveThe following species were encountered in the peatland, but were not of any significant value to the phytosociological table Oenothera rosea and Cirsium vulgare are conspicuous in the vegetation.Paspalum dilatatum occurs mostly on clay and loam soil and in wet areas and is often a weed in disturbed areas (Van Oudtshoorn 1999).
(Van Wyk & Malan 1988)nts a pioneer stage, which established directly after the new disturbance caused by species is Persicaria.Furthermore Amaranthus hibridus, Persicaria lapathifolia and Cyperus fastigiatus are also conspicuous.Phragmites australis is starting to establish within this sub-community.Persicaria lap-Phytolacca octandra is a weed restricted to disturbed areas and Persicaria lapathifolia is a weed that often occurs in wet areas(Van Wyk & Malan 1988).The sub-community with 13 species is relatively poor in species Fig. 1.Dendrogram showing the hierarchy of TWINSPAN divisions for the vegetation data of the rehabilitated peatlands on Rietvlei Nature Reserve on the 2001 dataset.This sub-community is drier than sub-community 3.2, because is is on slightly higher ground, and the vegetation is shorter, mostly lower than 1 m.The sub-community is characterised by the absence of species group I and it is dominated by Cyperus species.community and the dominant species is the perennial Typha capensis.Epilobium hirsutum, Persicaria lapathifolia, Cyperus sp., Cyperus fastigiatus and Oenothera rosea are conspicuously present.Typha capensis occurs in wet areas and Epilobium hirsutum occur in moist grassland and are often found in vlei areas (Van Wyk & Malan 1988).These species are indicators of the wet soil.