Diversity and species turnover on an altitudinal gradient in Western Cape , South Africa : baseline data for monitoring range shifts in response to climate change

A temperature and moisture gradient on the equator-facing slope o f Jonaskop on the Riviersonderend Mountain. Westem Cape has been selected as an important gradient for monitoring the effects o f climate change on fynbos and the FynbosSucculent Karoo ecotone. This study provides a description o f plant diversity patterns, growth form composition and species turnover across the gradient and the results o f four years o f climate monitoring at selected points along the altitudinal gradient. The aim o f this study is to provide data for a focused monitoring strategy for the early detection o f climate change-related shifts in species’ ranges, as well as gaining a better understanding o f the role o f climate variability in shaping species growth responses, their distributions, and other ecosystem processes. INTRODUCTION The vegetation of mountain ranges of the Cape Floristic Region (CFR) is characterized by gradients of high species turnover. These gradients are associated with increasing precipitation and lower temperatures with increasing elevation, as well as edaphic diversity (Goldblatt 1997). The vegetation of Cape mountain ranges within the CFR is dominated by fynbos (Mucina & Rutherford 2006). However, in inland intermontane valleys in the Westem Cape interior, fynbos is replaced by Succulent Karoo, an arid shrubland, at lower eleva­ tions with an ecotonal zone in between, called renoster­ veld (Cowling & Holmes 1992). Ecotones, or areas of transition between distinct biomes or vegetation types, have been pointed out as important areas for monitoring the effects of global cli­ mate change (Kupfer & Caims 1996). Modelling of future climate scenarios for the Westem Cape indicates that large areas of fynbos will be subjected to warmer and drier climate conditions similar to what is currently associated with Succulent Karoo, and that vegetation shifts might take place resulting in a loss of distribu­ tion area of fynbos due to a southward shift of Succulent Karoo (Rutherford et al. 2000). Therefore vegetation gradients incorporating transitions between fynbos and Succulent Karoo are important areas to monitor for the early detection of climate change-induced biome shifts. The elevational gradient on the equator-facing slope of Jonaskop, one of the highest peaks in the Rivier* Department o f Conservation Ecology and Entomology, University o f Stellenbosch. Private Bag XI. Matieland 7602, South Africa. f Centre for Invasion Biology, University o f Stellenbosch. Private Bag XI. Matieland 7602, South Africa. ** Global Change and Biodiversity Programme, Ecology and Conser­ vation, South African National Biodiversity Institute. Private Bag X7, Claremont 7735, South Africa. X Department o f Botany and Zoology. University o f Stellenbosch. Private Bag XI. Matieland 7602, South Africa. + Corresponding author: e-mail address: KJEfasun ac.za. MS. received: 2008-01-18. sonderend Mountains. Westem Cape, has been identified as a key site for monitoring the effects of climate change on fynbos and the Fynbos-Succulent Karoo boundary. On the equator-facing foothills of Jonaskop. fynbos veg­ etation type FFsl3, North Sonderend Sandstone Fynbos, intersects with Succulent Karoo vegetation type SKv7, Robertson Karoo (Mucina & Rutherford 2006) through a renosterveld ecotone. The Jonaskop gradient has been described in terms of vegetation biomass turnover (Rutherford 1978). but no formal description of the plant communities or of species diversity turnover on the gra­ dient has been done to date. Other studies conducted at the site include investigations into Protea species and classic rodent pollinator interactions (Rourke & Wiens 1977; Wiens et al. 1983; Fleming & Nicolson 2002a, b), including rodent diets and metabolism (Johnson et al. 1999; 2004; 2006); a comparative study between nectar qualities of Cape and Australian Proteaceae (Nicolson & Van Wyk 1998); seed dispersal by small mammals (Midgley & Anderson 2005); lizard foraging behaviour (Cooper et al. 1999); leaf functional classification of a number of species in a Mediterranean context (Vile et al. 2005); xylem traits in relation to water stress tolerance (Jacobsen et al. 2007); plant phenophase responses to climate (Agenbag et al. 2004a); how biome boundaries might respond to climate change (Agenbag et al. 2004b, c); and sampling for vegetation structural changes as part of a biome-wide study (Campbell 1985). These studies mark Jonaskop as an important site for ecologi­ cal research that would benefit greatly from a formal description of the biotic and abiotic environment. The aim of this paper, therefore, is to describe the Jonaskop gradient in terms of its vegetation changes, species turnover, soil chemistry and climate and to serve as a source of reference for future monitoring. MATERIAL AND METHODS Study area The Riviersonderend Mountain is situated on the westem end of a chain of east-west trending moun162 Bothalia 38,2 (2008)


INTRODUCTION
The vegetation of mountain ranges of the Cape Floristic Region (CFR) is characterized by gradients of high species turnover.These gradients are associated with increasing precipitation and lower temperatures with increasing elevation, as well as edaphic diversity (Goldblatt 1997).The vegetation of Cape mountain ranges within the CFR is dominated by fynbos (Mucina & Rutherford 2006).However, in inland intermontane valleys in the Westem Cape interior, fynbos is replaced by Succulent Karoo, an arid shrubland, at lower eleva tions with an ecotonal zone in between, called renoster veld (Cowling & Holmes 1992).
Ecotones, or areas of transition between distinct biomes or vegetation types, have been pointed out as important areas for monitoring the effects of global cli mate change (Kupfer & Caims 1996).Modelling of future climate scenarios for the Westem Cape indicates that large areas of fynbos will be subjected to warmer and drier climate conditions similar to what is currently associated with Succulent Karoo, and that vegetation shifts might take place resulting in a loss of distribu tion area of fynbos due to a southward shift of Succulent Karoo (Rutherford et al. 2000).Therefore vegetation gradients incorporating transitions between fynbos and Succulent Karoo are important areas to monitor for the early detection of climate change-induced biome shifts.
The elevational gradient on the equator-facing slope of Jonaskop, one of the highest peaks in the Rivier-sonderend Mountains.Westem Cape, has been identified as a key site for monitoring the effects of climate change on fynbos and the Fynbos-Succulent Karoo boundary.On the equator-facing foothills of Jonaskop.fynbos veg etation type FFsl3, North Sonderend Sandstone Fynbos, intersects with Succulent Karoo vegetation type SKv7, Robertson Karoo (Mucina & Rutherford 2006) through a renosterveld ecotone.The Jonaskop gradient has been described in terms o f vegetation biomass turnover (Rutherford 1978).but no formal description of the plant communities or of species diversity turnover on the gra dient has been done to date.Other studies conducted at the site include investigations into Protea species and classic rodent pollinator interactions (Rourke & Wiens 1977;Wiens et al. 1983;Fleming & Nicolson 2002a, b), including rodent diets and metabolism (Johnson et al. 1999;2004;2006); a comparative study between nectar qualities of Cape and Australian Proteaceae (Nicolson & Van Wyk 1998); seed dispersal by small mammals (Midgley & Anderson 2005); lizard foraging behaviour (Cooper et al. 1999); leaf functional classification of a number of species in a Mediterranean context (Vile et al. 2005); xylem traits in relation to water stress tolerance (Jacobsen et al. 2007); plant phenophase responses to climate (Agenbag et al. 2004a); how biome boundaries might respond to climate change (Agenbag et al. 2004b, c); and sampling for vegetation structural changes as part of a biome-wide study (Campbell 1985).These studies mark Jonaskop as an important site for ecologi cal research that would benefit greatly from a formal description of the biotic and abiotic environment.
The aim of this paper, therefore, is to describe the Jonaskop gradient in terms of its vegetation changes, species turnover, soil chemistry and climate and to serve as a source of reference for future monitoring.

Study area
The Riviersonderend Mountain is situated on the westem end of a chain of east-west trending moun- tain ranges at the southern edge of the Cape Fold Belt.Jonaskop, 33°58'00 S 19°30'00 E, altitude 1 646 m (Figure 1) is located within the Riviersonderend Mountain Catchment-a protected area that is man aged by CapeNature as an important water source to the Riviersonderend River below the southern slopes and the Breede River towards the north.The protected area is bordered by privately owned agricultural land.A service road leading up to a Sentech radio mast on the moun tain summit provides access to the equator-facing slope of Jonaskop.
The Jonaskop gradient spans a change of roughly 1 200 m in elevation between foothills in the Breede River Valley and Jonaskop Peak.Whereas the southfacing slope of Jonaskop is extremely steep, the equa tor-facing slope rises somewhat less rapidly, except for a flattened 'step' in the middle of the gradient, at ± 900 m.
Several drainage lines leading down from the northern slope converge in the Sand River, which at ± 400 m is the lowest point of the gradient.This small stream joins the Doom River, a tributary of the Breede River, further down the valley.
Soils on the mountain are shallow and very rocky.The coarse, grey, sandy soils of the mountain slope are derived from quartzitic sandstone of the Table Mountain Group.At the foot of the mountain (below 600 m) soils are finer grained and less rocky compared to those of the Table Mountain Group.Here the geology is alternating bands of arenaceous shale and argillaceous sandstone from the Bokkeveld Group (Besaans 1966).
Vegetation changes from Succulent Karoo at the very lowest elevations (± 500-600 m) through an ecotonal area of renosterveld with succulent as well as fynbos ele ments (roughly around 600-800 m), to fynbos from 800 m upwards.The boundary between the Succulent Karoo and ecotone coincides with the transition from shalederived to sandstone-derived soils.
Ecotonal and fynbos vegetation on either side of the road leading to the top of the mountain differs in fire his tory: with few exceptions, one side burned fairly recently (2000), whereas on the other side, mature vegetation has established after the last fire in 1992.

Climate monitoring
Climate change monitoring and experimental stations have thus far been placed at six sites along the length of the gradient (Figure 1).Weather stations (Watchdog Model 600 Weather Station 3325WD), recording air temperature, relative humidity, rainfall, soil moisture and wind speed and direction, are situated at the top (1 303 m), middle (953 m) and lowest (545 m) end of the gradient, whereas data loggers (Watchdog 450 Relative Humidity/Temperature Data Logger) record rel ative humidity, air temperature and soil moisture at the intermediate sites (1 196 m, 1 044 m, 744 m).Weather stations have been recording climate continuously since February 2002 (with some breaks at all stations from time to time due to equipment malfunction and damage, e.g. from rodent gnawing on exposed cables.Where such breaks occurred, data were interpolated from adjacent or nearby weather stations assuming average air tempera ture lapse rates or average annual altitude-related rain fall trends).Readings are taken automatically every 30 minutes.Climate data were analysed to show patterns in temperature, rainfall and wind conditions across the gra dient.

Soil analysis
Soil samples were taken at each of the monitoring sites along the gradient and analysed for P, Ca, Mg, K, Na, N and pH according to standard methods: to analyse for P, a 5 g sample was added to 50 ml 1 % citric acid solution.To an aliquot of the clear and colourless extract was added an acidified ammonium molybdate solu tion.The phosphomolybdate complex was then reduced with stannous chloride and the absorbance of the result ing blue colour measured with a spectrophotometer and compared with the absorbances obtained from standard phosphorus.To analyse for N, a known mass of soil was digested with sulphuric acid using selenium as a catalyst.The resultant ammonia was distilled into a satu rated boric acid solution and titrated with standard acid.Macro elements (Ca, Mg, Na, K) were determined by measuring a 1 mol dm-3 ammonium acetate extraction by atomic absorption against known standards.

Field sampling methods
For the vegetation analysis, two more sites were added to the existing climate monitoring stations: one on the lower sandstone slopes of the ecotonal zone (690 m), and another one near the mountain summit (1 576 m).Vegetation sampling was done in October and November 2003.Releves of 10 x 10 m were used, with two releves located in mature vegetation and one in recovering veg etation at each site.Species were scored following the Braun-Blanquet cover-abundance scale (Braun-Blanquet 1928).At the Succulent Karoo site, at the lowest end of the gradient (545 m), a releve was added to include sam pling of vegetation on nutrient-enriched heuweltjies or mima-like mounds usually found in association with termitaria in this region (Esler & Cowling 1995).
Environmental characteristics noted at each releve included slope, aspect, percentage rock cover, soil type and soil depth.Soil depth was determined up to 0.5 m deep at 10 points at regular intervals across one diago nal of each releve.Soil depth for each releve was then expressed as a range from shallow to deep, with depths of more than 0.5 m indicated by 0.5 m+.

Data analysis
Species data were assembled into a phytosociological table and sorted according to constancy and affinity to determine plant communities and their characteris tic species.Braun-Blanquet scores were then converted to percentage cover values for the calculation of spe cies diversity (See Table 1 for conversion values).The diversity of each releve was calculated according to the Shannon-Wiener Index (Kent & Coker 1994) with the formula:

1=1
where s is the number of species, and pi is the proportion of the i-th species to the total vegetation cover.Diversity values of the releves of each site were then averaged to arrive at a diversity estimate for each site.Within-site comparisons of diversity between recently burned and mature vegetation were also done for all sites except the karoo site (545 m), which did not bum.
All sampled species were classified according to growth form.Growth forms were assigned according to a scheme adapted from Cowling et al. (1994) (Table 2).The contribution of each growth form to total vegetation cover in each releve was calculated by summing con verted percentage cover values of all species belonging to each growth form, and expressing it as a proportion of total vegetation cover.Site specific growth form com position was then calculated by averaging cover values of the releves of each site.Comparisons of growth form composition between recently burnt and mature vegeta tion were done for all sites except the lowest karoo site (545 m).Releves of mature vegetation of all sites were lumped, and the average growth form composition was  Similarity of species composition of different sites were compared using Jaccard's coefficient (Kent & Coker 1994).The formula is: where S is the Jaccard similarity coefficient, a is the number of species common to both sites compared, b is the remaining number of species present at the first site, and c is the remaining number of species present at the second site.S was multiplied by 100 to arrive at a per centage similarity.The average turnover rate across the gradient was determined according to the methods of Itow (1991).Log percentage similarity between every pair of sites (on y-axis) was plotted against their differ ence in altitude (on x-axis).The slope of the regression line fitted is taken as the average turnover rate of species per 1 m altitudinal difference.

Temperature
Temperatures decrease with increasing altitude (Barry 1992); however, the rate at which temperatures decrease can vary spatially as well as seasonally (Rolland 2003).On Jonaskop, there is an average difference of 4.4°C between the highest (1 303 m) and lowest site (545 m) on the gradient (Table 3).Temperatures along this gra dient generally decrease by 0.58°C with every 100 m altitude gained, as indicated by the slope of a regression line fitted to mean annual temperatures recorded at each of the monitoring sites (not shown).However, diurnal and seasonal temperature lapse rates on the mountain slope vary between -0.40°C.100m 1 for winter minimum temperatures (April to September) and -0.77°C.100m 1 for summer maximum temperatures (October to March) (Figure 2).This is consistent with patterns found in mountainous regions elsewhere (Rolland 2003 and ref erences therein).Complex factors contribute to seasonal and diurnal variation in lapse rates.These include wind regime, cloud cover, amount of incoming solar radiation and the moisture content of the air (Barry 1992) and it is therefore difficult to explain the variations observed on Jonaskop without detailed analysis of other meteorologi cal patterns.
It should be noted that the middle site (953 m) is somewhat colder than the site directly above it (1 044 m) (Table 3).This effect is particularly strong at night during winter (Figure 2), when mean minimum temper atures recorded at the site are on average 1.6°C colder than the expected trend.Such local temperature inver sions are generally the result of cold air drainage, and are often observed in valley bottoms (O' Hare et al. 2005).The fact that the middle site is located on a plateau prob ably results in cold air flowing downwards from the steeper slopes above and collecting at this site during the long winter nights.
Monthly mean temperature summaries indicate that August is the coldest month at the site (Figure 3), with mean minimum temperatures ranging from 3.3-6.9°Cbetween the highest and lowest monitoring points.Highest temperatures were recorded during February, when daily maximum temperatures are on average around 30°C at the karoo site (545 m) at the lowest end of the gradient.

Rainfall
During three years of climate recording (2002)(2003)(2004), the highest site (1 303 m) received an average of 719.6 mm annual precipitation, the middle site (953 m) 411.3 mm and the bottom karoo site (545 m) 315.4 mm.The highest site receives on average a slightly lower pro portion of its annual rainfall in winter (Table 3, Figure 4).Higher summer rainfall at the highest site is pos sibly due to orographic effects, with southerly winds pushing clouds over the mountaintop and bringing rain to the highest site but not to lower sites.According to Aschmann (1973), Mediterranean climates are defined by winter rainfall constituting at least 65 % of the annual rainfall.The proportions of winter rainfall recorded across the Jonaskop gradient, which ranges from 57-66 %, therefore indicates that the study site does not fall within the strictly winter rainfall zone of the CFR.Long term rainfall data for the Riviersonderend Mountain con firm that many large rainfall events, associated with postfrontal cut-off lows occur in summer, resulting in the  area not being subjected to as severe summer droughts as elsewhere in the CFR (R.M. Cowling pers.comm.).
Monthly rainfall patterns recorded during three years of this study were very variable (Figure 5).The year 2002 had good winter rains and a relatively dry summer.The high monthly total precipitation of March 2003 was due to a single extreme rainfall event on 24 March when 174.5 mm was recorded at the highest site (1 303 m) and 103.9 mm at the lowest site (545 m).Year 2003 was an extremely dry year throughout the whole Westem Cape, and the low rainfall, especially during the winter months, is reflected in the data from Jonaskop.Although 2004 had, in total, a much higher annual precipitation, it also had a very dry winter, with most of the annual precipita tion recorded during October.Cowling et al. (2005) highlighted the importance of rainfall reliability in terms of interannual variation in rainfall, as well as the size and structure of rain fall events, as a driver of plant traits in Mediterranean ecosystems.Rainfall data recorded at Jonaskop were, therefore, further analysed to explore seasonal pat terns in the size, duration and frequency of rainfall events.Whenever rain was recorded on a number of consecutive days, rainfall recorded over the period was added together as a single rainfall event.Cowling et al. (2005) also mentioned the importance of the regularity of rainfall events in Mediterranean climates, especially with regard to the germination of seedlings.Mustart & Cowling (1993) showed that the duration of dry periods between rainfall events is an important factor determin ing the successful germination of Proteaceae seeds, and that rainfall patterns during germination stages have a large impact on the distribution patterns of Proteaceae populations observed in the field.Therefore a frequency analysis of the number of days between rainfall events was also done.At all sites, rainfall events occur most often as small (< 5 mm) events lasting only one day (Figure 6).This trend was observed during winter and summer.However, larger rainfall events (> 10 mm), and events lasting lon ger than two days occur much more frequently in winter than summer at all sites.In terms o f the duration of dry periods between rainfall events, patterns are more diver- gent among sites and seasons.At the highest site (1 303 m), rainfall events were most often separated by less than four days during winter and summer.At the middle site (953 m), the winter rainfall events in contrast, are far more frequently within four days of each other than in the summer.Summer rainfall events show a larger frequency distribution towards longer dry intervals.At the lowest site (545 m), the winters tend to have longer dry intervals at low altitudes, indicating that even during winter, rainfall at this site can be very sporadic.At this site, dry intervals of longer than 14 days are also more common during winter and summer than at the higher sites.

Wind
An analysis of wind patterns at the top (1 303 m), middle (953 m) and lowest (545 m) sites (Figure 7) shows altitudinal as well as seasonal differences.All sites experience predominantly southerly winds dur ing summer months (October to March).During winter (April-September) wind patterns shift to predominantly northwesterly at the top site and westerly at the middle site, whereas winter wind patterns at the lowest site are essentially the same as during summer conditions, with a slight shift towards more north and northwesterly winds and less easterly winds.
Wind speeds tend to increase with increasing altitude (Barry 1992), and can have an impact on plant growth.High wind speeds in combination with very low tem peratures on exposed mountain peaks can cause stunted growth in plants, for example, in the prostrate cushion forms of pine trees found in the Krummholz zone of alpine regions (Komer 1999).However, mean seasonal wind speeds on Jonaskop do not indicate an increase of wind speed with increasing altitude (Figure 8).
When wind speeds are compared across the gradient, lowest mean wind speeds are recorded during both win ter and summer at the middle site (953 m).The southerly winds of summer reach highest speeds (mean 11.8 km/h, maximum 28 km/h) at the lowest site.At the middle and lowest sites, winds are stronger during summer than in winter, but the predominantly northeasterly winds asso ciated with advancing rain-bearing cold fronts are much stronger during winter at the highest site.It is difficult to place wind speed data recorded on Jonaskop into con text, as windspeed data for mountain slopes elsewhere in the CFR are not readily available.According to Barry (1992), mean wind speeds of around 25 km/h are typi cal for mountain peaks in the mid-latitudes of the north ern hemisphere, which is much higher than wind speeds recorded on Jonaskop.Whether wind on Jonaskop is likely to affect the vegetation is also not certain.Controlled experimental studies have indicated that wind speeds higher than 10 km/h negatively affects the growth of herbaceous annuals and grasses (Whitehead 1962;Woodward 1993).On Jonaskop, mean wind speeds during the growth season (summer) are higher than 10 km/h only at the lowest site, but it is unlikely to affect growth in the sturdy perennial shrubs of the site, except when associated with high temperatures and low relative humidity, as under berg wind conditions.

Soil characteristics
Soils on the gradient are generally very shallow and rocky, the rockiness increasing with increasing altitude (Table 4).This has an effect on soil moisture, as soils at  Interval between events (days) the highest site (1 303 m) are generally drier than the site just below (1 196 m) (data available on request).At the 1 303 m site, slopes are steeper, and the soil is shallower and rockier than at the 1 196 m site (Table 4).These factors imply that faster drainage and runoff of rainfall occurs at the highest site, resulting in soils being drier.It is important to note that the soil moisture sensors did not record values drier than 100 kPa.and thus indicate when water was freely available, and then only in shallow sur face layers of the soil.The calibrated range of the sen sors was not wide enough to record the onset of stressful soil moisture conditions, which are generally associated with soil water tensions lower than -1500 kPa (Miller & Gardiner 1998).Soil moisture values do not give an indication of water availability to deep-rooted species.However, our data (available on request) does illus trate that there is a trend toward different soil moisture dynamics at different altitudes that is related to rainfall.
The main differences in soil chemistry between shalederived soils at the karoo site (545 m) and the sandstonederived soils of the rest of the mountain was in terms of P. Ca and Mg. which were much higher in the shalederived soils.Soil on the mounds (heuweltjies) was very' different from that in the sandstone-derived soils, as well as from off-mound shale-derived soils (see also Ellis 2002).Heuweltjie soils have much higher Ca, K and N content than either sandstone or off-mound shale.Offmound shale has roughly double the Na content of either heuweltjie or sandstone-derived soils.

Vegetation
A total of 286 species were recorded at sampling sites along the gradient.The highest number of species recorded in a single 10 x 10 m releve was 50 species at the middle site (953 m) (Appendix 1 & 2).Species accu mulation curves, calculated for each site, revealed that vegetation sampling did not approach complete sam pling of communities (data not shown).Typically, the vegetation of the Fynbos Biome has three strata and is a mid-tall to tall shrubland (Campbell 1985).The veg etation recorded in this study differed only at the sum mit where extreme subalpine conditions have resulted in dwarf shrubland vegetation, less than 0.5 m tall.

Karoo site (545 m)
Out of a total o f 56 species recorded at this site, 45 (80 %) occurred at no other sites on the gradient.A num ber of species were specific to heuweltjies alone (Group II, Appendix 1), namely Galenia africana, Schismus barbatus and Galium tomentosum.The off-mound com munities are defined by the presence of Dicerothamnus rhinocerotis, Oedera squarrosa and Pteronia panicu lata (Group III, Appendix 1).General species typify ing the vegetation at this site are Euphorbia burmannii, Pteronia incana and Ruschia lineolata.These species are found in high abundances on heuweltjies, and to a lesser extent elsewhere at the site (Group I, Appendix 1).Four species found frequently at the karoo site, that are able to successfully cross the soil barrier between the karoo and ecotonal site are Anthospermum aethiopicum, Drosanthemum speciosum, Montinia caryophyllacea and Passerina obtusifolia (Group VII, Appendix 1).Eight other species were found on both sandstone and shalederived soils, but these were rare and had low cover val ues.
Dicerothamnus rhinocerotis (renosterbos) and Oedera genistifolia, which both have fairly high cover values in releves 6B & 6C (Appendix 1), are generally associ ated with renosterveld (Mustart et al. 1997;Goldblatt & Manning 2000), a vegetation type of the Fynbos Biome which is found on more nutrient-rich soils (Rebelo 1998).According to Mucina & Rutherford (2006), the vegetation on the lower eastern slope of Jonaskop, which is quite close to the karoo site, is classified as Breede Shale Renosterveld (FRs8).This is probably a function of mapping scale as Holmes (2002), in an environmen tal impact study on the vegetation of the Breede River Valley northeast of Jonaskop, found that in this area, Succulent Karoo is found on the foothills of equator-fac ing slopes, wheareas renosterveld is found on pole-fac ing slopes.The karoo site is located between the foot of the equator-facing slope of Jonaskop and the southfacing slope of a low hill, with releves 6B & 6C (plot OFM3 in Figure 1) located on the south-facing hillslope.The species composition of this site, considered in com bination with the topography, points either to an altitu dinal banding of the vegetation types, or to transitional vegetation, or to a mosaic of adjacent Succulent Karoo and renosterveld, which has resulted in an inevitable mix of species from both vegetation types found in the sam pling plots.The substrate, in conjunction with the local climate, appears to be the main driving force determin ing the vegetation found here (Boucher & Moll 1981), but it is still unclear whether this is due to soil nutritional or soil water retention constraints.
The site also has a relatively high incidence of unpal atable asteraceous shrubs such as Chrysocoma ciliata (bitterbos), Galenia africana (kraalbos) and species of Pteronia, which may indicate that the vegetation at this In all, 105 species were recorded at the two sites located at low altitude sandstone sites.Fifty five species were found nowhere else on the gradient.Species typical of the low altitude sandstones are Cannomois scirpoides, Ficinia oligantha, Protea humiflora and Ruschia sp.(L.Agenbag 5A09, Group IV, Appendix 1).The lower site (690 m) has a number of highly abundant species occur ring only at this site, which defines it as supporting a subcommunity within the ecotonal Low-altitude Inland Sandstone Fynbos (Group V, Appendix 1).They are Cliffortia crenata, Ischyrolepis sieberi, Lachnospermum fasciculatum and Polygala fruticosa.A number of spe cies were also found unique to the higher ecotonal site (744 m), namely, Hermannia rudis, Othonna ramulosa and Phvlica sp.(L.Agenbag 5A03) (Group VI, Appendix 1).
The vegetation of the ecotonal Low-altitude Inland Sandstone Fynbos recorded here is very similar in spe cies composition and physical aspects to the Cannomois parviflora-Passerina obtusifolia Shrublands described by McDonald (1993), which occurs on the lower north ern slopes of the nearby Langeberg.These arid Sandstone Fynbos shrublands are also found near the transition between Table Mountain Sandstone and Bokkeveld Shales, which support Little Karoo vegetation.
Site 4: Protea repens-Tetraria flexuosa Closed Tall Mid-altitude Sandstone Fynbos, which is located in the middle of the gradient (953 m), has by far the highest species number on the gradient, with 85 species recorded at this site alone.It has a high number of rare species which were recorded in low abundances and low fre quencies.Tetraria flexuosa and Wahlenbergia neorigida (Group IX, Appendix I), typify this community.
Sites 1-3: Protea repens-Tetraria ustulata Closed Tall Mid-altitude Sandstone Fynbos, is typified by Tetraria ustulata and the other less frequent species listed in Group X (Appendix 1).The sites at 1 196 m and 1 303 m, however, do have characteristic species, but in both cases, these species were not found in the recently burned vegetation.These latter communities are there fore given lower ranking.
Ehrharta ramosa-Restio triticeus High Altitude Sand stone Dwarf Fynbos (1 576 m): Structurally, vegetation at the mountain summit is very low compared to the rest of the gradient (less than 0.25 m tall).Some species found at this site, that also occur elsewhere on the gradient, such as Helichrysum zwartbergense, are found near the summit as dwarfed growth forms, compared to taller individuals lower down.Soils are extremely shallow (surface rock to pockets of soil not often more than 10 cm deep) and very rocky, with an average of 75 % rock cover.Species numbers at this site are very low, with only 26 species recorded in the plots.The vegetation at this site is domi nated by grasses, sedges and restios, although dwarf shrubs are present throughout.Species characteristic of this high-altitude summit vegetation include Ehrharta ramosa, Metalasia sp.(L.Agenbag 0B02), Restio tri ticeus and Stoebe sp.(L.Agenbag 0C04) (Group XV, Appendix 1).

Diversity
Shannon diversity indices showed highest diversity at the low to middle altitudes (690-953 m) and sites on a sandstone substrate (Figure 9A).When diversity was compared within sites between recently burned and mature vegetation, burned releves had higher diver sity in all fynbos sites (953, 1 044, 1 196 and 1 303 m) as well as at the summit (1 576 m. Figure 9B).Lowaltitude inland sandstone fynbos sites (690 and 744 m) have lower diversity in recently burned releves.Other studies in fynbos have also found that richness is highest in the first few years after fires, due to the presence of ephemeral species (Bond & Van Wilgen 1996;Holmes & Cowling 1997).

Growth form composition
A comparison of the relative contribution of various growth forms to vegetation cover shows clear shifts in dominant growth forms across the gradient (Figure 10A).From the karoo site (545 m) up to the lowest fynbos site (953 m), vegetation is dominated by low shrubs.At the karoo site, succulents are also a significant contributor to vegetation cover, but not at any of the other sites.Above 953 m, the dominant growth form shifts to graminoids, which includes grasses, sedges and restios.The low veg etation at the summit site (1 576 m) is reflected in a high proportion of vegetation cover represented by dwarf shrubs, with heights lower than 0.25 m.At this site, herbs also represent a larger proportion of the vegetation cover than at other sites.
When growth form composition between recently burn ed and mature vegetation is compared, ecotonal low-alti- tude inland sandstone sites show very similar composition (Figure 10B).In mid to high-altitude fynbos sites, how ever, recently burned vegetation has higher cover in herbs, succulents and graminoids, wheareas mature vegetation consists largely of long-lived low and mid-high shrubs, as well as graminoids.
Taking a closer look at the relative contribution of the families Cyperaceae, Poaceae and Restionaceae to graminoid cover across the gradient, more interesting patterns are revealed (Figure 11 A, Appendix 3).Grasses are more common at the karoo site (545 m), and are found only in very low proportions elsewhere on the gradient, except again for the summit (1 576 m), where grasses, sedges and restios make roughly equal contributions to total graminoid cover.Restionaceae show increas ing dominance from the low-altitude sandstone sites upwards to 953 m, after which sedges become slightly more important.Comparing graminoid cover between recently burned and mature vegetation reveals that in fynbos, Cyperaceae dominate recently burned vegeta tion, whereas Restionaceae are dominant in mature veg etation (Figure 11B).In the low-altitude sandstone sites, grasses are slightly more abundant in recently burned than mature vegetation.V

Similarity and turnover rates
Similarity between all sites is relatively low, suggest ing high species turnover even between sites with similar vegetation.Similarity between the karoo site and the rest of the gradient, as well as between the summit and the rest of the gradient is the lowest (Table 5), suggesting that strong climatic and/or soil factors are limiting spe cies distributions between these and other areas on the gradient.Highest similarities were found among the ecotonal low-altitude sandstone sites (25.6 %) and among fynbos sites (15.7-25.5 %).Similarities of 11.3 %, 17.1 % and 16.2 % between the (low-altitude sandstone) ecotone and lower mid-altitude fynbos sites (953 and 1 044 m) suggests that the low-altitude sandstone ecotonal vegetation is more closely related to fynbos than karoo, as similarities between the latter two sites are only 7.5 % (690 m) and 2.7 % (744 m).Mean turnover rate for the gradient was determined as -0.0014, with the equa tion fitted to the plot of difference in altitude against similarity being Log % similarity 0.0014 x difference in altitude + 1.41, r2 = 0.52 (Figure 12).Similarity val ues between the karoo site and other sites were gener ally much lower than predicted by the regression line for similar changes in altitude elsewhere along the gradient, again suggesting that the area between 545 and 690 m is governed by strong environmental factors affecting spe cies distributions.

CONCLUSION
Climate change studies report increasing evidence that species' ranges shift higher on mountain slopes as a result of climate warming, and that high altitude species are therefore particularly vulnerable to extinction as they reach the limits of mountain summits (Grabherr et al. 1994;Parmesan & Yohe 2003).Therefore the summit com munity on the Jonaskop gradient, with its very restricted range, is almost certainly vulnerable to the warmer and drier conditions predicted for the Western Cape.
However, monitoring species for population mortal ity, extinctions and shifts in ranges across the entire gra dient will provide valuable insights into the responses of fynbos, renosterveld and Succulent Karoo vegetation dynamics to climate change.Anecdotal evidence sug gests that Protea species (mainly genus Leucadendron) may be vulnerable to drought stress on Jonaskop (Hannah et al. 2007).Jacobsen et al. (2007) studied 19 species from nine angiosperm families along the Jonaskop gradient, and found significant variation in their water stress tolerance.Species studied were not as tolerant of water stress as chaparral shrubs occurring in climatically similar California, USA.They suggested that the measurement of xylem density may be a useful tool to assess drought tolerance of large numbers of spe cies.The area around the lower ecotonal site (550-700 m) is the most important to monitor for the first signs of shifts in species and growth form composition as indication of an upward shift of Succulent Karoo into renosterveld-fynbos territory.Since karoo vegetation is known to be distinct from the vegetation recorded in this study through analysis of species similarity, diversity and growth form composi tion, changes should be detected relatively easily.

-
Monitoring efforts should not only focus on upward shifts in species ranges, but also on contraction of ranges at the lower elevational limits of species (Hampe et al. 2005).Data collected in this study revealed clearly defined upper and lower altitudinal limits on the gradient for many species sampled, and thus, focusing monitoring on changes in abundance of these species at their upper and lower limits, along with continued climate data recording, could reveal much about the climatic controls of species' ranges.
It is of course necessary to investigate the importance of the change in soil type between 545 m and 690 m as a potential barrier to climate change-induced shifts in spe cies' ranges, and this is possibly best achieved through experimental studies.As Dunne et al. (2004) have indi cated, a combination of gradient monitoring and experi mental investigation strategies provide the best insights into complex ecosystem responses to climate change.We submit that this study will provide useful baseline data for a future focused and directed monitoring effort lead ing to a better understanding of the potential effects of climate change on fynbos and the fynbos-renosterveldsucculent karoo boundary.

A C K N O W L E D G E M E N T S
The Du Plessis family is acknowledged for access to the study site and helpful ongoing support.This study was financially supported through a Franco-RSA (CNRS/NRF) grant (GUN 2065318 to G.F. Midgley), NRF (GUN 2053516 to K.J. Esler) and the Centre for Invasion Biology (to K.J. Esler).We acknowledge the support from SANBI and especially field assistance pro vided by Deryck de Wit.' F-
FIGURE 4.-Total annual rainfall (April to March) recorded at top (1 303 m), middle (953 m) and lower (545 m) monitoring points on Jonaskop gradient.Total annual rainfall is split into winter rainfall season (April to September) and summer growth rainfall season (October to March).
FIGURE 5.-Monthly rainfall recorded from April 2002 to March 2(K)4 at the top (1 303 m), middle l?53 m) and lower (545 m) monitor ing points on the Jonaskop gradient.

FIGURE 6 .
FIGURE 6.-Amount o f rain per rainfall event, duration o f rainfall events, and duration o f dry intervals between rainfall events summarized as mean frequencies per season.Altitude where rainfall was recorded on gradient is indicated above each column.Error bars indicate 1 standard error.
FIGURE 7.-Seasonal wind patterns recorded at top, middle and lower monitoring points on Jonaskop gradient.Lengths o f bars indicate no.days.

FIGURE 8 .
FIGURE 8.-Mean seasonal wind speeds recorded at top (1 303 m), middle (953 m) and lowest (545 m) end o f Jonaskop gradient.Wind speeds measured at 1 m above soil surface.Error bars indicate 1 standard errort

FIGURE
FIGURE 9.-A, Shannon-Wiener diversity indices (H') recorded across gradient.H' values for each altitude is mean o f three 10 x 10 m releves; B, com parative diversity indices (H ') between recently burned and mature vegetation at selected altitudes along Jonaskop gradi ent.Diversity values for mature vegetation is mean o f two 10 x 10 m releves, whereas only one releve was sampled in recently burned vegetation at each alti tude.Error bars in A indicate 1 standard error.

FIGURE
FIGURE 10.-A, comparative con tributions o f various growth forms to total vegetation cover at selected altitudes along gra dient; B, comparative growth form composition between recently burned and mature vegetation in fynbos sites.FUB, fynbos mature vegetation; FB. recently burned fynbos; EUB, ecotone mature vegetation; EB. recently burned ecotone vegetation.
S. f *i N ' K "T >5 for species occurring on Jonaskop altitudinal gradient in Riviersonderend Mountain Catchment o f Western Cape, South Africa.Growth form codes (GFC) are provided in

Structured releve table for Jonaskop altitudinal gradient i n Riviersonderend Mountain Catchment of Western Cape, South Africa. Data were collected in October 2003. See Appendix 2 for species with low frequency occurrences (cont.) F Bothalia 38,2 (2008) 177
T © Q III APPENDIX I