Phytosociology of the farm Haribes in the Nama-Karoo biome of southern Namibia

Vegetation in Namibia has been described only on a few attempts in the past (Burke & Strohbach 2000; Strohbach & Jürgens 2010). To date, the only concise vegetation map of Namibia was produced in the 1970s, based on field observations rather than actual surveys (Giess 1971, 1998), and depicts broad categories at biome rather than landscape level. Later attempts to map the vegetation of Namibia had to rely on partial, often incomplete, regional updates used in conjunction with the Giess vegetation map to fill in gaps (Mendelsohn et al. 2002). Grazing capacity maps dating from the 1970s (Departement Landbou Tegniese Dienste 1979) are used for land-use planning, including the ongoing Land Reform Programme of the Government of Namibia. However, these maps are regarded as outdated and in urgent need of revision (Espach 2006; Espach, Lubbe & Ganzin 2006; Lubbe 2005). Current efforts to upgrade the grazing capacity estimation with the aid of satellite images require extensive ground surveys, including surveys of the structure and composition of vegetation (Espach et al. 2006; Espach, Lubbe & Ganzin 2010). With this in mind, but also as basis for a biodiversity inventory, the Vegetation Survey of Namibia project was initiated (Strohbach 2001).


Introduction
Vegetation in Namibia has been described only on a few attempts in the past (Burke & Strohbach 2000;Strohbach & Jürgens 2010).To date, the only concise vegetation map of Namibia was produced in the 1970s, based on field observations rather than actual surveys (Giess 1971(Giess , 1998)), and depicts broad categories at biome rather than landscape level.Later attempts to map the vegetation of Namibia had to rely on partial, often incomplete, regional updates used in conjunction with the Giess vegetation map to fill in gaps (Mendelsohn et al. 2002).Grazing capacity maps dating from the 1970s (Departement Landbou Tegniese Dienste 1979) are used for land-use planning, including the ongoing Land Reform Programme of the Government of Namibia.However, these maps are regarded as outdated and in urgent need of revision (Espach 2006;Espach, Lubbe & Ganzin 2006;Lubbe 2005).Current efforts to upgrade the grazing capacity estimation with the aid of satellite images require extensive ground surveys, including surveys of the structure and composition of vegetation (Espach et al. 2006;Espach, Lubbe & Ganzin 2010).With this in mind, but also as basis for a biodiversity inventory, the Vegetation Survey of Namibia project was initiated (Strohbach 2001).
Prof. O.H. Volk was one of only a few vegetation scientists active in Namibia before the 1980s.During 1956, he conducted extensive vegetation surveys on selected farms in the former South West Africa.Unfortunately, only one of these data sets had been analysed and formally published (Volk & Leippert 1971); the others that could be retrieved were all in the form of raw data sets with brief comments on the vegetation, generally held in farm archives.Nonetheless, these data sets have proved to be of immense value in reconstructing vegetation changes over the past five decades of land use, especially with regard to the effects of land use and/or global climate change.One of these data sets, surveyed in April 1956, was from the farm Haribes.
This study was initiated to provide data on the current vegetation, for later comparison with Volk's historic data.The aim of this paper, however, is to classify and describe the present vegetation of the farm Haribes as an attempt towards filling the immense gap in vegetation data in Namibia (Strohbach & Jürgens 2010).In a subsequent paper, we intend to compare the present vegetation with that documented in 1956.
Limited historic vegetation data (prior to the 1980s) are available for Namibia.Finding such historic data at Haribes prompted a follow-up survey of the vegetation.We present a classification of the recent data in this paper as a first step towards comparing the two data sets.Six new associations (three with two subassociations each) are formally described.The landscape at Haribes is dominated by a pan with surrounding hummock dunes.The pan supports the Lycio cinereum -Salsoletum, whilst on the hummock dunes the Salsolo -Tetragonietum schenckii can be found.The surrounding plains and escarpment can be divided into three landforms: the torras with the Monsonio umbellatae -Boscietum foetidae, the ranteveld with the Acacio senegal -Catophractetum alexandri (and two subassociations) and calcrete ridges with the Zygophylo pubescentis -Leucosphaeretum bainesii.Dry river beds on the farm support two subassociations of Anthephoro pubescentis -Ziziphodetum mucronatae.The area covered by each dominant landform has been calculated after being mapped.The composition and diversity of the associations are briefly compared to other known vegetation descriptions within the Nama-Karoo.Since November 2011, Haribes has been used as a resettlement farm.This may result in the overutilisation of the limited grazing resources, to the extent that the present, fairly dense Acacio senegal -Catophractetum alexandri of the ranteveld is feared to become degraded to resemble the Monsonio umbellatae -Boscietum foetidae of the torras.
Conservation implications: This paper describes six plant associations of the Nama-Karoo biome in arid southern Namibia.The information presented forms a baseline description, which can be used for future monitoring of the vegetation under altered land use.

Study area
Haribes, in the Mariental district of the Hardap region in southern Namibia (Figure 1), belonged to the Neue Haribes Company until 2010.In 1956 the farm covered 77 500 ha, but has been reduced to its present 50 100 ha through a process of selling off blocks of land.The present survey was conducted across blocks 019/1, 019/REM, 020, 672 and 673, which total 55 487 ha.Initially, Haribes was used for extensive commercial Karakul sheep farming.The most important land-use change took place in 1992, when all the sheep and goats were sold and replaced by cattle as the only domesticated commercial species.As a result of the selective utilisation of the grass layer, woody species, especially Catophractes alexandri, Rhigozum trichotomum and Acacia nebrownii, increased in abundance, resulting in a phenomenon commonly referred to as 'bush encroachment' (De Klerk 2004).A wide variety of game, often in large numbers, are found on the farm.Most numerous are springbok and oryx.There are also many bird species, including many ostriches.The farm was sold in May 2010 to the Government of Namibia as part of its land reform and resettlement scheme (Cloete 2010a;Pretorius 2010), with the first 13 families receiving their plots in November 2010 (Cloete 2010b;Kleinhans 2010).
Haribes is situated approximately 30 km west of Mariental and southwest of the Hardap Dam Nature Reserve within the Central Plateau 4-8 agro-ecological zone (De Pauw et al. 1998).This zone is described as a strongly dissected plain on Karoo rocks.The average rainfall is about 200 mm per year, with a coefficient of variation between 50% and 60% (Botha 1996).Temperatures rise to above 36 °C in summer, with minimum temperatures of between 2 °C and 4 °C in winter (Mendelsohn et al. 2002).
The geomorphology can be described as a plateau broken up by many dry washes and some bigger ephemeral rivers that are responsible for the drainage of the area [a dissected plain according to the terminology of the Digital Soils and Terrain Database (SOTER) (FAO 1995;Oldeman & Van Engelen 1993].The Riet River drains into a big pan surrounding the farmhouse complex.In and around the pan several welldeveloped dune hummocks have formed over time.The farm lies between 1120 m and 1320 m above sea level.The geological strata belong to the Fish River subgroup of the Nama sediments.The eastern part of the farm is situated on the Gross Aub Formation dominated by red shale and sandstone, whilst the larger, western part of the farm forms part of the older Nababis Formation with red sandstone and shales (Geological Survey 1980).The soils in the region (at a scale of 1:1 000 000) are classified as eutric leptosols, which means that the soils are fertile but generally very shallow (ICC, MAWRD & AECI 2000;Mendelsohn et al. 2002).
The farm falls within the Nama-Karoo biome (Irish 1994) and its vegetation is classified as Dwarf shrub savanna (Giess 1971(Giess , 1998)).This can broadly be described as open, short-tolow shrublands dominated by C. alexandri, R. trichotomum, Boscia foetida and several dwarf shrubs.The grass layer, which includes species such as Stipagrostis uniplumis, Eragrostis nindensis, Enneapogon cenchroides and Schmidtia kalahariensis, is relatively sparse.Trees and larger shrubs, including species such as A. nebrownii, Acacia karroo, Euclea pseudebenus and Ziziphus mucronata, are largely confined to dry washes and rivers.

Research method
A total of 177 plots were sampled between March and May 2005.Below average rain (c.150 mm, D. Poggenpoel, pers.comm., May 2005) had fallen that season, with the first substantial rains falling only in March.
To compare the present composition with that documented by Volk in 1956, we made exhaustive attempts to relocate his original sampling sites according to route descriptions, odometer distance readings (in miles) and old farm maps.Although this information greatly eased the task of relocation, the exact position of those plots could not be determined.Additional vegetation types were sampled using a stratified sampling approach to achieve a good average for the main vegetation types.For the purpose of stratification, a satellite map of the farm (based on Landsat 7 ETM scene path 177 row 077, 26 March 2001) was used as base map.Relatively homogenous areas were delineated as derived from segmentations calculated by Definiens 5 (Definiens 2006).Occasional azonal vegetation types (especially small vleis and related features) were also sampled.
Surveying followed the Braun-Blanquet concept as applied during the Vegetation Survey of Namibia project (Strohbach 2001).For each 20 m × 50 m plot, all species were recorded, as well as their typical growth form and average height.Abundance was estimated according to the modified Braun-Blanquet scale (Mueller-Dombois & Ellenberg 1974).In addition, the position of the plot and environmental parameters such as altitude, aspect, slope, topography, stone cover, erosion and disturbances were noted.Owing to time constraints, no soil samples were taken.Instead, available soil data (ICC et al. 2000;Mendelsohn et al. 2002) were interpolated for the various vegetation associations described.Unknown plant species were collected and identified at the National Herbarium of Namibia.The relevé data were captured in the TurboVeg database (Hennekens & Schaminée 2001).
The data were classified with JUICE (Tichý 2002), using the Modified TWINSPAN classification procedure (Roleček et al. 2009).The pseudospecies cut level was set to presence/ absence, and total inertia was selected as heterogeneity measure.The classification process concluded after 14 divisions.At this stage, two communities (hummock dunes and the ranteveld) were already subdivided owing to high internal heterogeneity and had to be recombined manually (Figure 2).The classification result was ordered as a table using the phi coefficient fidelity measure (Chytrý et al. 2002) as sorting criterion.
The classification results were confirmed with an ordination using Detrended Correspondence Analysis (DCA) (Hill & Gauch 1980), prepared with PC-Ord 5 (McCune, Grace & Urban 2002).Two iterations were performed, with the first excluding only two wetland relevés (as outliers) whilst the second also excluded the pans and associated hummock dunes.Although DCA has been criticised in recent years for 'detrending' the arch effect, this very effect proved useful in illustrating the relationships between the various vegetation associations, as Reciprocal Averaging (RA) depicted only a severe arch effect.The DCA ordination diagram was also used to depict habitat relations, which could often only be overlaid as categorical data.
Mapping was performed using a Landsat 7 ETM satellite image (path 177, row 077;26 March 2001).In addition, a digital terrain model derived from Shuttle Radar Topography Mission (SRTM) data (available from http://srtm.csi.cgiar.org/) (Jarvis et al. 2008) was used to elucidate the topography.The satellite image was clipped to an area slightly larger than the study area before further processing.This clip was imported into the Definiens software package (Definiens 2006) and segmented into relatively homogenous areas.The segments were classified using the classified sample sites as ground truth data.In a final step, the resulting shape file was clipped according to the farm boundaries and areas for each landscape calculated using the IDRISI Andes software package (Eastman 2006).

Results
A total of 218 species, representing 42 families, were observed on the 177 relevés.Poaceae was the most abundant familiy, represented by the most species (54), followed by the Fabaceae (27 species), Asteraceae (16 species) and the Aizoaceae (13 species).
The classification of these relevés revealed eight communities, of which three were divided into subcommunities.The phytosociological table is presented in an online appendix 1, whereas online appendix 2 is a synoptic table.Figure 2 depicts a dendrogram of the classification result.The ordination diagrams of the DCA are shown in Figure 3a (all data except for two outlier relevés) and Figure 3b (also excluding the relevés representing pans and associated hummock dunes).In Figure 3c, the lithology (i.e. the rock types which gave rise to the soils) has been imposed onto the DCA ordination diagram of Figure 3b to illustrate the habitat relationships between vegetation communities.
The pan and hummock dune communities form a welldefined continuum to the right of the ordination diagram in Figure 3a, indicating most likely a combined salinity and soil moisture gradient from the pan centre (community 1.2; upper right corner) to the surrounding hummock dunes (community 3; lower right corner).In contrast, the remaining communities (plains, ranteveld and rivers) are compressed on the left of the main axis.Splitting out the relevés representing the pan and hummock dune communities resulted in a continuum spread between the stony ranteveld, plain and torra communities as well as the rivers (Figure 3b).The limited habitat data available indicate that this is a gradient of increased stoniness towards the left of the main gradient (Figure 3c), representing the ranteveld.Together with the lithology (i.e. the rock type from which the substrate is derived), the stoniness forms the main ecological gradient in this environment.
The numbers at the bottom represent the community numbers, as used in the description of the vegetation types.Note that communities 3 (hummock dunes) and 8.2 (ranteveld) were divided owing to high internal heterogeneity and had to be recombined manually.

Description of the plant associations
We formally describe six of the observed communities as associations, three with subassociations, following the International Code of Phytosociological Nomenclature (Weber, Moravec & Theurillat 2000).The two wetland communities, however, are described only informally, as the number of samples taken was too small for proper description.

Lycio cinereum -Salsoletum ass. nov.
The typicum for this association is relevé 2514, sampled at 24°35'36"S, 17°32'58"E on 09 March 2005 (Figure 4b).The association is represented by 21 relevés and characterised by the presence of Salsola spp.dwarf shrubs, which also dominate the vegetation.The structure is a short, open shrubland (sensu Edwards 1983), with an average shrub cover of 6%.Total vegetation cover barely reaches 10%.In total, 58 species have been observed on these pans, with an average species density of 12 per 1000 m 2 .
The Lycio cinereum -Salsoletum is typical of the large Haribes pan on the farm.The soils originate from weathered sandstones and some calcretes consisting of fine material which is very prone to capping and surface sealing (Figure 4c).The cation levels are probably elevated to form a halophitic environment, but not sufficiently so to render the pan a salt pan (M.Coetzee, pers.comm., March 2010).
Two subassociations can be distinguished, as described below.
Along the outer parts of the pan and on the floodplains to the pan the vegetation is better developed, forming a tall, moderately closed shrubland (Figure 5).Few perennial herbaceous species occur here; this layer is dominated by ephemerals.The species density is on average 14 species per 1000 m 2 and in total 39 species have been observed here.

The Lycio cinereum -Salsoletosum subass. nov.
This subassociation forms the typicum for the Lycio cinereum -Salsoletum.It is represented by 16 relevés and characterised by near-monospecific stands of Salsola shrubs.Other species that occur here are mostly of ephemeral nature.The shrubs are on average only 0.5 m -1 m high and reach a cover of only about 6%, thus forming a short, open shrubland (Figure 5).On average, only five species occur per 1000 m 2 , with a total of 31 species being observed on these pans.

The Eragrostis rotifer -Leptochloa fusca community
This community was found once in relevé 2599 from a sample on a deeper vlei, which is regularly inundated.This The codes used correspond to those used in the description of the vegetation types.is indicated by the presence of the geophytic hydrophyte Aponogeton cf.desertorum and Eragrostis rotifer, also generally associated with water-logged soils (Figure 4d).These uncommon vleis form in relatively small depressions, mostly on top of the plateau area in pans that range from 20 m to 50 m in diameter.The pans are surrounded by sandstone stones varying in size between 5 cm and 80 cm.The reddish colour indicates that the soil is most probably relatively high in nutrient content (M.Coetzee, pers.comm., March 2010).
The association is dominated by T. schenckii.The typical structure is a tall, moderately closed shrubland (Figure 5), with the shrubs reaching a height of between 1 m and 1.5 m.On average 11 species are found per 1000 m 2 , with a total diversity of 56 species.
The Salsolo -Tetragonietum schenckii is found on hummock dunes probably originating from a deflation process during the formation of the Haribes pan, similar to that described for the Koës pan and other depressions in the Kalahari (Bullard & Nash 2000;Lancaster 1986;Shaw 1988;Shaw & Thomas 1997).The dunes vary in length, generally between 30 m and 50 m.The slopes of the dunes are often quite steep.The occurence of Salsola spp. on these dunes indicate a high base status, which, in turn, supports the hypothesis that the dunes are the result of the deflation process of the pan.

The Cullen obtusifolia -Leptochloa fusca community
This community was identified in three relevés within the larger Haribes pan ecosystem.The presence of Leptochloa fusca and Indigofera alternans indicates water-logging at these vleis, whilst the absence of E. rotifer and Aponogeton sp. and the presence of the flat-growing therophytes Cullen obtusifolia and Platycarpha carlinoides indicate a drier environment than occurs in the deep vleis of the E. rotifer -Leptochloa fusca community.Typically for these vleis, A. nebrownii is dominant along the fringes (Figure 4f).These pans are small (50 m -80 m in diameter) and very shallow.The high clay content of the soil allows the water to remain in the pan for approximately 2-3 weeks.In most cases, these pans are surrounded by reddish gravel, but no larger stones.
There are very few of these small pans that differ greatly from the big pans on the farm.Some are relatively small depressions, mostly on top of the plateau area, but there are also small pans below the plateau, especially near the old road along the tar road between Mariental and Maltahöhe.These depressions allow water to accumulate, resulting in microhabitats with a wetland character.

Anthephoro pubescentis -Ziziphodetum mucronatae ass. nov.
The The Anthephoro pubescentis -Ziziphodetum mucronatae represents the riverine woodlands (sensu Giess 1998).There are two types of drainage system on the farm, with dry washes, which contain water only for very short periods during rain storms, being most common.Geomorphologically they can be classified as very young river systems.In many cases the slope is relatively steep and the run-off very fast.They occur mostly on the plateau or at the edge of the plateau.Towards the centre of the farm, more mature rivers, like the Riet, Holz and Lewer, are responsible for drainage.There is a big dam near the farmhouse, where the water of these rivers is captured.These two different habitats support two different vegetation subassociations (5.1 and 5.2).
Although the vegetation of the riverbed and the riverbank differ, they were sampled as one for the purposes of this study and are thus included in the same plant association.

Ehretrio albae -Ziziphodetosum mucronatae subass. nov.
The  5), with the trees only 4 m -5 m tall and shrubs reaching height of 1 m -2 m.In this subassociation, 99 species have been recorded; on average 21 species occur per 1000 m 2 .
The Ehretrio albae -Ziziphodetosum mucronatae occurs along the dry washes.Their slopes vary from steep to almost level (15°-20° and < 3°, respectively).Erosion is a common occurrence on the steeper slopes.Most of these dry washes start on top of the plateau and generally run eastward or southward to join the Riet, Holz and Lewer rivers.They are all typically ephemeral and are active only during and shortly after thunderstorms.Varied soil character is observed: although most soils are very shallow, occasionally with big rocks, gravel and small stones also represent a common soil cover.At the base of the plateau some of the washes have deeper soils, which are also higher in nutrient content.
Owing to the higher moisture content in the soils, the trees grow up to 7 m tall, thus forming a short, open woodland with a tree cover of 2% (Figure 5).Species density is on average 28 species per 1000 m 2 , whilst a total of 109 species have been observed in this subassociation.
The Heteropogo contorti -Ziziphodetosum mucronatae occurs along the more mature rivers (Riet, Holz and Lewer).Their slopes are normally less than 3° and water remains standing in river pools for long periods.The soils in the riverbeds are normally rough gravels, which can be very stony with big rocks.The nutrient level in these gravels is low, but water content is usually relatively high.The soils of the river banks are sandy loams with a relatively high nutrient content, whilst the moisture content varies depending on the slope of the river.
Zygophylo pubescentis -Leucosphaeretum bainesii is associated with shallow calcrete soils, either along the northern Haribes pan edge, as part of the flood plain system of larger rivers or on dolerite ridges.The soils are shallow (20 cm -30 cm deep), whitish in colour and have a high clay content.The soils are expected to be alkaline with a pH of more than 7 (M.Coetzee, pers. comm., March 2010).Stone cover in the form of calcrete gravel and pebbles is often found mixed with medium to large dolerite blocks and can reach up to 80%.
Steep terraces, with slopes of between 5° and 10°, are formed along the pan edge.

Monsonio umbellatae -Boscietum foetidae ass. nov.
The typicum for this association is relevé 2632, sampled at 24°37'10"S, 17°37'03"E on 15 May 2005 (Figure 4j).This association was found on 17 plots in the study area.5).The shrubs are between 0.5 m and 2 m high and represent less than 5% of the ground cover.Density is on average 17 species per 1000 m 2 , with a total of 67 species being observed in this association.
The Monsonio umbellatae -Boscietum foetidae typically occurs on the torras on the farm.Torras are defined as flat plains, with little or no internal relief, covered with reddish brown gravel on the surface (often more than 80%).The soils are loamy, with a crust, and with very low water infiltration capacity.This low water infiltration capacity results in patchy, nearbanded vegetation, with run-off from the open patches feeding the vegetated patches (run-on).Similar vegetation patterns have been described as brousse tigree in the Sahel (Boaler & Hodge 1962;Bromley et al. 1997;Couteron et al. 2000;Thiéry, D'Herbès & Valentin 1995) and banded mulga in arid Australia (Dunkerley 1997a(Dunkerley , 1997b)).

Acacio senegal -Catophractetum alexandri ass. nov.
The typicum for this association is relevé 2606, sampled at 24°32'31"S, 17°28'59"E on 13 May 2005 (Figure 4l).5), with a shrub cover of less than 5%.Shrub height ranges between 0.5 m and 1.8 m.The species density is on average 18 species per 1000 m 2 , whilst a total of 87 species have been found in this subassociation.
The Schmidtio kalahariensis -Boscietosum foetidae occurs on flat plains and footslopes as well as on ranteveld proper and forms a transition between the torras and the ranteveld.These plains are underlain by sandstones of the Nababis formation.
Fragments of these form the stone cover on the soil surface, ranging from gravel to medium-sized stones and covering up to 80% of the soil surface.The low vegetative cover indicates a tendency to degrade to a similar vegetation as the Monsonio umbellatae -Boscietum foetidae on the present torras (Figure 4j).
Occasional trees stand 4 m -5 m tall.On average, 19 species have been found per 1000 m 2 , whilst the total species diversity is 105.
The Acacio senegal -Catophractetosum alexandri forms the ranteveld and occurs mainly in areas on top of the plateau.The plateau is dominated by very stony outcrops and small escarps (rante) of Nababis sandstone.The soils are leptosols, red in colour and very shallow, of moderately coarse to medium texture, with many coarse fragments and well drained.The surface layer is typically 10 cm -20 cm thick.These soils have a moderate hydraulic conductivity and present a moderate infiltration rate with a low water holding capacity (ICC et al. 2000).The landscape is relatively flat on top of the plateau, but forms occasional low hills and an escarpment area to the west of the farm, with steep-sloped ravines in between.

Vegetation mapping
It was not possible to map all vegetation types, for several reasons.Firstly, many units are just too small to map with the 30-m ground resolution of Landsat 7 ETM satellite images.This holds true for the two vlei communities, E. rotifer -L.fusca and C. obtusifolia -L.fusca, as well as the Zygophylo pubescentis -Leucosphaeretum bainesii, which occurs on ridges with very limited spatial extent.
Differentiating between the Monsonio umbellatae -Boscietum foetidae and the Schmidtio kalahariensis -Boscietosum foetidae proved extremely difficult owing to their similar structures and especially similar low vegetative cover.For this reason, a landscape map was produced (Figure 6) to differentiate between plains and torras, and ranteveld.Table 1 lists the mapped landscapes, their spatial extent and their included vegetation associations or subassociations.

Discussion
Four distinct higher syntaxonomic groupings can be recognised, namely the saline pans and associated hummock dunes (as represented by the Lycio cinereum -Salsoletum and the Salsolo -Tetragonietum schenkii), the wetlands (as represented by the two vlei communities), the river systems (as represented by the Anthephoro pubescentis -Ziziphodetum mucronatae) and the surrounding plateaus (as represented by the Zygophylo pubescentis -Leucosphaeretum bainesii, the Monsonio umbellatae -Boscietum foetidae and the Acacio senegal -Catophractetum alexandri).Unfortunately the data set is too limited to define and describe such higher syntaxonomic levels at this stage.Volk and Leippert (1971)  This description is illustrated with three plates.Yet the Acacio senegal -Catophractetum alexandri has a too tall structure (2 m -3 m) to be regarded as true dwarf shrub savanna.Within the broader Nama-Karoo (of which Giess' dwarf shrub savanna forms the northern extent), the Acacio senegal -Catophractetum alexandri can structurally be compared to the Acacia mellifera open shrubland of the Blouputs karroid thornveld as described for the Augrabies Falls National Park (Bezuidenhout 1996;Mucina & Rutherford 2006).However, the shrub component here is dominated by A. mellifera subsp.detinens, a species rarely found in the Acacio senegal -Catophractetum alexandri.In this respect, the Acacio senegal -Catophractetum alexandri is to be regarded as a xeric outlier of the highland savanna (sensu Giess 1998).The rocky habitat with shallow soils, and often steep slopes, is similar in both vegetation types.
The Salsolo -Tetragonietum schenckii resembles the Saline desert and dwarf shrub savanna fringe (sensu Giess 1998) described for the Etosha pan.Here the structure of the hummock dune vegetation can specifically be compared to the Okondeka duneveld (Le Roux 1980).However, the composition varies widely, with T. schenkii not occurring in northern Namibia.For the Lycio cinereum -Salsoletum no comparable vegetation type is known.It is likely that these, in a future synopsis of syntaxa, will form a class on their own.
The very close relationship between the Monsonio umbellatae -Boscietum foetidae and the two subassociations of the Acacio senegal -Catophractetum alexandri is well illustrated in the ordination diagram in Figure 3b.This leads to the assumption that these three units form degradation stages of one another: Acacio senegal -Catophractetosum alexandri degrading to Schmidtio kalahariensis -Boscietosum foetidae, and in very extreme cases (possibly also owing to soil degradation

Conclusion
Haribes is unlikely ever to become a conservation area.On the contrary, the recent change in ownership also introduces an entirely different land use, with the farm being targeted for resettlement.Some 13 families have already been resettled on small parcels of the farm, each ranging between 1700 ha and 7500 ha (Cloete 2010b;Kleinhans 2010).These units, meant to support the livelihoods of these families (Werner & Odendaal 2010), have been calculated according to grazing capacity maps from the 1970s (Department Landbou   Tegniese Dienste 1979).However, the general indications are that the grazing capacity has drastically reduced throughout the entire country over the past decades (Espach et al. 2010).This is due to a multitude of factors generally associated with land degradation [e.g.bush encroachment (De Klerk 2004)].It is thus unlikely that these small farming units will be sustainable and severe degradation of grazing resources may result (Werner & Odendaal 2010).
Within the Nama-Karoo biome in Namibia, livestock farming depends mainly on browsing of dwarf shrub species (Espach et al. 2006), with grazing playing a secondary role.This does not hold true for Haribes.A comparison of the shrub, dwarf shrub, grass and total vegetation cover of the various associations (Figure 5) shows that farming on Haribes is more dependent on grazing than browsing.The pans and surrounding dunes are mostly covered by shrubs dominated by Salsola spp.and T. schenckii, which are both known to be poisonous (Mannheimer, Marais & Schubert 2008).This makes roughly 13% of the farm unsuitable for grazing.
The Zygophylo pubescentis -Leucosphaeretum bainsii on the calcrete ridges provide some dwarf shrub browse, but is very limited in extent.The Monsonio umbellatae -Boscietum foetidae (torras) and the Acacio senegal -Catophractetum alexandri (ranteveld) provide the main grazing resources on the farm.However, as pointed out in the discussion, these seem to form a degradation gradient from each other; thus, extensive overgrazing will lead to desert-like torra plains.
Global climate change has been identified as a reality in recent years.Temperature increases coupled with less precipitation, in more erratic events, are predicted for southern Africa (Parry et al. 2007).Based on these forecasts, it is predicted that large parts of southern Namibia will become unsuitable for livestock farming, as it will turn into a virtual desert (Midgley et al. 2005).Similar scenarios have been developed for the Kalahari basin, with forecasts of remobilisation of sand dunes as far north as southern Angola (Thomas, Knight & Wiggs 2005).In addition, a human-induced drought, as described for other Nama-Karoo regions (Booysen & Rowswell 1983), threatens.This is aggravated by the fact that farming on Haribes depends largely on grazing, meaning that protection of the perennial grass cover is essential for sustainable use of the farm in the long term.Perennial grasses have been shown to be able to withstand prolonged droughts (i.e.periods of below-normal precipitation), but are unable to withstand heavy grazing together with arid conditions (O'Connor 1991(O'Connor , 1994)).It is thus likely that overgrazing in the ranteveld may lead to a composition (and vegetation cover) similar to that of the present torras.When planning the redistribution of the farm Haribes, it should be kept in mind that only 75% of the farm (the ranteveld) provides adequate grazing and that even this grazing needs to be utilised judiciously to prevent severe degradation to desert-like conditions.

FIGURE 1 :
FIGURE 1: Location of the farm Haribes in the Hardap region of southern Namibia.Also indicated on this map are the long-term average annual rainfall isohyets (in grey), with their values (in mm) indicated to the right.

FIGURE 2 :
FIGURE 2: Dendrogram based on the Modified TWINSPAN classification of the vegetation types in the study area.

FIGURE 3 :
FIGURE 3: Detrended Correspondence Analysis ordination diagrams compiled from (a) all data, excluding two outlying wetland relevés and (b) data excluding wetland, hummock dune and pan relevés; (c) indicates the lithology of each plot, with a biplot indicating the stoniness of the substrate.

FIGURE 4 :
FIGURE 4: Typical vegetation at different sampling sites on the farm Haribes.

FIGURE 5 :
FIGURE 5: Box-and-whisker plots of the various structural components for each of the described associations or subassociations.

FIGURE 6 :
FIGURE 6: Map of the major landscapes on the farm Haribes.

TABLE 1 :
Landscapes and associated vegetation associations or subassociations on the farm Haribes.