Simon Todd, Leslie Hill Institute for Plant Conservation, Botany Department, University of Cape Town, Rondebosch, 7701, South Africa (e-mail firstname.lastname@example.org).
1Gradients of animal impact known as piospheres tend to develop around artificial watering points, particularly in arid zones. Such grazing gradients represent a potential opportunity for differentiating the long-term effects of livestock activity from other environmental patterns. In this study, the impact of watering point provision on the plant cover, species richness and community structure of Karoo shrublands, South Africa, was investigated in the context of the evolutionary history and current grazing management practices of the region.
2The impacts of watering point provision were investigated by sampling plant cover and composition along transects placed at set distances, ranging from 10 m to 2200 m, from 11 watering points.
3Karoo vegetation cover and structure are relatively resilient to livestock grazing. Karoo plant diversity, as measured by species richness, evenness and dominance, was not as resilient. Twice as many species decreased as increased near watering points. The majority of species that decreased were regarded as being highly palatable to livestock. Heavy grazing, leading to death or repeated reproductive failure, is the most likely mechanism leading to the decline of such species.
4The highly disturbed area immediately adjacent to watering points was dominated by forbs and contained a large proportion of alien species. Adjacent to this was a zone dominated by widespread shrub species of medium to low palatability. Areas most distant from watering points contained a greater proportion of species known to be highly palatable to livestock. The ability of dominant Karoo shrubs to tolerate heavy grazing may have allowed rangeland managers to maintain stocking rates above that which can be tolerated by the majority of species but which are supported by a minority of grazing-tolerant species.
5Synthesis and applications. Highly palatable species are more abundant in areas distant from water points. Larger paddocks therefore provide a refuge for sensitive species that might otherwise be lost from the rangeland as a whole. Species that tend to occur away from watering points represent potentially useful indicators of grazing pressure. The use of these species as indicators of rangeland condition among landowners should be promoted.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Lange (1969) first coined the term piosphere to refer to the radial pattern of attenuating grazing and associated animal impact that develops around watering points. The ultimate cause of this pattern is the fact that, in arid zones, livestock are limited in how far they can move away from water sources because they have to drink regularly. Furthermore, the area available to graze increases with distance from the watering point, resulting in a reduction in the relative grazing intensity with distance (James, Landsberg & Morton 1999). The distribution of livestock impact is thus patterned and grades from many times higher than the overall average at the watering point to many times lower further away (Andrew 1988). Artificially created watering points, which are a common feature in semi-arid zones, are frequently not associated with or confounded by other environmental factors (Landsberg et al. 2003), as is usually the case with natural water sources such as springs and drainage lines. When appropriately selected, such grazing gradients represent an ideal opportunity for differentiating the long-term effects of livestock activity from other environmental patterns.
Within southern Africa, piosphere studies have generally been restricted to the arid and semi-arid savannas (Perkins & Thomas 1993; Van Rooyen et al. 1994; Jeltsch et al. 1997; Thrash 1998a,b) and a pair of unreplicated studies from the Succulent Karoo (Beukes & Ellis 2002; Riginos & Hoffman 2003). To date, no published studies have assessed the impact of watering point provision on the structure and species richness of Nama-Karoo shrublands in South Africa. This is surprising given that artificial watering points are an integral part of the extensive livestock industry that is the primary agricultural activity in the Karoo (Archer 2000) and that the Nama-Karoo occupies 607 235 km2 or almost 23% of the entire southern African subregion (Palmer & Hoffman 1997). Although there has been a long history of grazing research in the Nama-Karoo (Roux & Vorster 1983a; Hoffman 1988), the majority of this work has focused on increasing the production of the rangelands (Palmer & Hoffman 1997; Hoffman 1988). As a result, the potential of different grazing management systems and the responses of the dominant forage species to grazing are relatively well known (Vorster, Botha & Hobson 1983). However, the impact of livestock grazing on the less common species that constitute the bulk of the biodiversity of the Nama-Karoo remains poorly understood. In addition, there is considerable controversy surrounding the perceived negative impact of livestock on the vegetation of the Karoo. While some authors have claimed that the Karoo has become severely degraded since the arrival of European settlers (Acocks 1953; Roux & Vorster 1983b; Dean & Macdonald 1994) others maintain that the observed changes are simply short-term fluctuations that occur naturally in response to rainfall cycles (Milton & Hoffman 1994; O’Connor & Roux 1995). A better understanding of the impact of livestock watering points could contribute to this debate and potentially enhance sustainable rangeland management practices with attendant benefits for the biodiversity of the region.
Prior to the colonization of the Karoo by European settlers 200 years ago, the abundance of ungulates in the Karoo was determined largely by the presence of surface water (Owen-Smith & Danckwerts 1997). Apart from sedentary ungulate populations near permanent water sources, migratory herds, particularly of springbok Antidorcas marsupialis Zimmermann, black wildebeest Connochaetes gnou Zimmermann, blesbok Damaliscus pygargus Pallas, quagga Equus quagga Boddaert and eland Taurotragus oryx Pallas, are known to have moved around the Karoo in response to drought and local food availability (Owen-Smith & Danckwerts 1997). The regularity and the impact of these migrations on the vegetation on the Karoo is, however, unknown. Populations of large herbivores had been substantially reduced or eliminated by the end of the nineteenth century (Hoffman 1997) and livestock were the dominant herbivores in the Karoo by the early twentieth century. Around the same time, the advent of windmills eliminated livestock dependency on naturally occurring surface water (Archer 2000).
As southern African ecosystems have had a long association with large indigenous herbivores (Owen-Smith & Danckwerts 1997), they should be relatively more resilient to livestock grazing impacts than other ecosystems that have not had such an association (Milchunas, Sala & Lauenroth 1988; Milchunas & Lauenroth 1993). This should confer several attributes to the vegetation of the region, including resistance to alien plant invasion and resilience in terms of vegetation structure, composition and diversity to a wide range of grazing intensities. In this study the validity of these assertions in the Nama-Karoo was tested by assessing the patterns of plant community structure and diversity around livestock watering points. The following predictions were tested, which should hold given the evolutionary history of the Nama-Karoo and the impacts of watering points that have been reported elsewhere.
1In common with studies from numerous ecosystems around the world, ephemeral species will be more common in the disturbed area near watering points.
2Alien species will be in the minority and are likely to be weedy ephemeral species, and will thus also be more abundant near watering points.
3The vegetation cover and composition of the Nama-Karoo should exhibit relative resilience to livestock grazing. This will manifest itself in the following. (i) The majority of species will recover rapidly away from watering points and will be reduced only in the highly impacted zone immediately adjacent to the watering point. (ii) Species that are abundant only in areas distant from watering points will, first, be in the minority and, secondly, be species known to be highly palatable to livestock. (iii) Vegetation structure and diversity will not be severely impacted by livestock grazing and will recover rapidly away from watering points.
Materials and methods
study site and sampling design
The vegetation of Nama-Karoo has been characterized as a grassy dwarf shrubland (Low & Rebelo 1998). Trees are infrequent but may occur along watercourses and in other moist habitats. The most broadly accepted descriptions of the vegetation of the Karoo include those of Acocks (1988), who described 16 vegetation types in the Nama-Karoo, and Low & Rebelo (1998), who recognized six vegetation types within the biome. In order to avoid the confounding effects of sampling across different vegetation types and soils, the current study was restricted to a single vegetation type known as Central Lower Karoo (Acocks 1988). This vegetation type covers an area in excess of 10 400 km2 and extends westward from the town of Pearston in the east to Beaufort West, on the extensive open plains of the area. The soils are typically very shallow and usually overlie a calcrete hardpan (Palmer & Hoffman 1997) that is frequently exposed. The annual rainfall in the area is between 150 mm and 250 mm and occurs mostly in autumn (March–May).
The primary agricultural activity in the Karoo is extensive livestock grazing with sheep and goats (Roux et al. 1981). In general, livestock graze in paddocks that range in size from less than 100 ha to more than 1000 ha. Each paddock is serviced by one or more watering points and there are few areas that are beyond 3 km from water. Paddocks are usually grazed on a rotational basis, with rest periods of several months to more than a year between grazing events (Hoffman 1988). The stocking rate in the region is between 2 and 4 ha per small stock unit [a sheep or goat equivalent to 1/6th of a large stock unit (LSU), which represents the metabolic equivalent of a 450-kg cow; Roux et al. 1981].
The paddocks used in the study were chosen using several criteria aimed at reducing, as far as possible, the number of possible extraneous factors influencing vegetation patterns. To be eligible for sampling, a paddock had to be serviced by a single watering point; paddocks containing multiple watering points would contain multiple grazing gradients that would be difficult to interpret. Watering points associated with livestock handling facilities or where there was evidence of supplementary feeding were avoided. Paddocks containing large drainage lines and other conspicuous vegetation changes were also avoided as these have been shown to influence livestock activity (Lange 1985). Finally, the paddock had to have been established in its current configuration, regarding its size, shape and the location of the watering point, for at least 15 years. Although time since last grazing is likely to have some influence on plant cover in particular, this effect could not easily be controlled for without greatly reducing the potential number of watering points suitable for sampling. While some paddocks had been rested for longer than others had, no paddocks that were currently being grazed or had been grazed within the previous month were sampled. A total of 11 watering points was found to meet these criteria adequately and sampled. It was assumed that these criteria were adequate in eliminating extraneous influences as far as possible and that consistent trends in the vegetation observed across different piospheres would represent a common response to the piosphere-related gradient rather than a response to other unknown factors. The attributes of the paddocks used in the study are described more fully in Appendix S1 in the supplementary material.
Vegetation composition and diversity were sampled using transects placed at set distances away from each watering point. As it was anticipated that vegetation change would be more rapid in the zone immediately adjacent to the watering point than at a distance from it, the sampling resolution was consequently increased towards the watering point. Transects were placed every 10 m from 10 to 50 m, at 70 m and 100 m, every 100 m thereafter until 1000 m and then every 200 m thereafter as far as the paddock allowed. Along each transect, the vegetation was sampled by identifying and estimating the cover of all species present within 10 1-m2 (0·5 × 2-m) quadrats. Except for the first one or sometimes two transects immediately adjacent to the watering point, which were necessarily shorter, transects were 50 m wide. The presence and cover of all species that occurred in the 1-m2 plots was recorded over each transect for all watering points. All fieldwork took place between August 2002 and the end of December 2002. Taxonomic nomenclature follows Germishuizen & Meyer (2003).
Because of differences in the size of the paddocks involved in the study, not all the watering points studied could be sampled to the same distance. Two paddocks each were sampled to 1000 m and 1400 m, one paddock each to 1200 m and 1800 m and five paddocks to 2200 m. Although some of the paddocks were sampled to greater distances, these data were excluded because of the low number of replicates that were available at these distances. The final data matrix used in the analysis consisted of a total of 215 transects, sampled at 22 sampling distances around 11 watering points.
patterns in cover and species richness within plant growth forms
The species recorded in the study were classified into one of the following growth forms: shrub, forb, grass, dwarf succulent and geophyte (Palmer & Hoffman 1997). Plants such as Asparagus, which have perennial woody above-ground parts and fleshy tubers or rhizomes, were classified as shrubs and were not considered to be geophytes. Annual and perennial forbs were combined into a single category because few of the forbs in the area are truly perennial and the majority do not persist through the dry season. Although succulent shrubs are known to respond to grazing differently from woody shrubs (Todd & Hoffman 1999), all shrubs, including those with succulent leaves, were classified into a single category. The motivation for this included the fact that only six succulent shrub species were recorded in the study and the responses of the succulent species were mixed and could not be distinguished from that of woody shrubs.
For each transect around each watering point sampled, the species richness and mean cover of all species and within each growth form was calculated. Regression analysis was then used to investigate the relationship of these grazing response variables to distance from water. Although many studies have shown that response patterns tend to follow a logistic curve, other regression models were also used where they were able to better fit the observed data (Thrash & Derry 1999).
individual species’ responses
To determine how individual species responded to the grazing gradient associated with watering points, a best-fit regression model was sought for each species (Landsberg et al. 2003). The best-fit model was the simplest model that explained the greatest amount of variation. A non-linear model was considered to be the best-fit model only when it explained significantly more of the variation than the linear model. Species that occurred fewer than five times were considered too rare to detect a significant pattern, but were combined and tested overall to assess whether or not rare species tended to be more abundant away from watering points, as might be expected if such species are rare as a result of grazing impact. This resulted in all species being classified among nine possible patterns, seven of which could be considered responses to the grazing gradient. These responses were classified among four basic patterns: increasing abundance away from watering points; decreasing abundance away from watering points; peaked abundance at intermediate distances from watering points; consistent abundance along the grazing gradient. The results were then summarized according to the number and growth form of species responding according to each pattern. In order to assess the relative importance of each group in terms of community dominance, the contribution of each group to the overall cover of the rangeland was calculated.
Changes in vegetation structure were assessed in terms of the relative abundance of the different species in the community using Simpson's D and the evenness index Evar (Smith & Wilson 1996), which were calculated for each transect around each watering point. To assess the response in plant community composition to distance from watering points, the data were also subjected to ordination. As time since grazing could not reliably be controlled for and is likely to affect plant cover more than presence, plant frequency data were used in preference to cover for this analysis. The data set was first reduced by calculating the mean frequency of each species at each distance, across all the watering points (i.e. the mean frequency of species x in all transects at distance y). The data were then subjected to ordination by non-metric multidimensional scaling (NMDS) using the Bray–Curtis distance coefficient (Bray & Curtis 1957). The NMDS ordination was produced using multiple runs and following the stress and stability criteria described in detail by McCune & Grace (2002).
A total of 105 species was recorded in the study, of which 22 were forbs, 12 were grasses and 57 were shrubs. There were also eight geophytes and six dwarf succulents that together accounted for less than 0·2% of the plant cover and that were not analysed further in terms of growth forms because of their low abundance.
Perennial plant cover increased rapidly away from watering points and reached an asymptote within 300 m (Fig. 1a). Even in the transects most distant from the watering points, plant cover did not exceed 40%, suggesting that this was the maximum perennial plant cover possible under the prevailing climate conditions in the area. The increase in cover away from watering points was largely determined by the abundance of shrubs (Fig. 1b) rather than that of grasses (Fig. 1c), which did not appear to respond as a group to watering points.
The cover of forbs varied by an order of magnitude between watering points, apparently as a consequence of the interaction between recent grazing history and rainfall events. Despite there being a clear gradient in forb cover within each piosphere, the high degree of variability between watering points tended to obscure the overall response pattern when the data were combined. As it was the response pattern rather than the absolute abundance that was of primary interest, the cover abundance of forbs around each watering point was standardized by the maximum value for that piosphere. The abundance of forbs around the different watering points could thus all be compared on the same relative scale. This analysis revealed that forb abundance followed the opposite pattern to shrub cover and that forbs contributed significantly to cover only within 100 m of the watering point, and decreased sharply thereafter (Fig. 1d). Beyond 200 m forbs tended to remain at a fairly consistent but low level of cover abundance.
Plant species richness increased with distance from watering points (Fig. 2a). A logistic regression provided a poor fit to the data, which were much better explained in terms of a logarithmic curve. This suggested that species richness did not reach an asymptote, at least within the distances available in this study. Shrub species richness followed a similar trend and was clearly responsible for the overall pattern (Fig. 2b). Grass species richness also increased with distance, but the fit of the regression was relatively poor, with more scatter about the line, particularly at sites closer to the watering point (Fig. 2c). This may partly be explained by the relatively low overall diversity of grasses in the study area, but also by contrasting responses by different grass species. Forb species richness decreased away from watering points, but the rate of decrease was not nearly as marked as that for cover (Fig. 2d).
Simpson's D and Evar (Fig. 3) illustrated a similar pattern that was best explained in terms of the different processes that operate near and further away from watering points. The area near watering points was dominated by annuals and forbs that respond to disturbance patterns in this zone and whose dynamics are essentially independent from the long-lived woody shrubs that dominate the rest of the rangeland. Dominance and evenness were high close to watering points because this area was dominated by a narrow suite of species tolerant of the heavy trampling and grazing pressure. Dominance reached a second peak 100–200 m away from the watering points at the same time that evenness reaches its lowest levels. At this point, the forbs that dominated near watering points had declined and the rangeland was highly dominated by a few grazing-tolerant shrub species. Beyond this point, dominance declined and evenness increased as the abundance of less grazing-tolerant species increased in the community. Linear regression fitted to the data from transects beyond the highly disturbed zone within 100 m of the watering points indicated that Simpson's D declined significantly (R2 = 0·77, P < 0·0001) and that Evar increased (R2 = 0·64, P < 0·001) across this zone.
A single NMDS ordination axis was sufficient to explain the majority of the variance in the data set. The stress of the final 1-D solution was 10·21 and the instability 10−5, indicating acceptable levels of stress and stability (McCune & Grace 2002). The R2 of the correlation between distance in the ordination space and distance and the original n-dimensional space was 0·989. These results strongly suggested that the data set was highly structured and that only one major gradient was present in the data set. The single NMDS axis corresponded to the grazing gradient associated with the watering points, as illustrated by the high correlation between the axis scores and the distance of transects from watering points (Fig. 4). The relatively large dispersion of the points representing transects in the first 100 m of the piosphere indicated that a large amount of change in community composition occurred over this zone compared with over the rest of the piosphere.
More than 40% of the species that occurred more than five times showed an increasing trend in abundance away from watering points (Table 1). Of these, only four species showed an exponential asymptotic increase away from watering points and could be considered be highly resilient to grazing pressure. These four species however, constituted the bulk of the plant cover of the rangeland (see Table S1 in the supplementary material). Sixteen species showed a linear increase away from watering points. All of these species were rare or did not occur close to watering points and represented the species that were most affected by grazing. These species nevertheless constituted more than 10% of the available plant cover, suggesting that they remained an important component of the rangeland. Of the 31 species that increased away from watering points, the majority were shrubs, four were grasses and two were dwarf succulents. Five species were most abundant at intermediate grazing intensities, and an additional two did not appear to respond to the grazing gradient and showed consistent abundance levels across all transects. Fifteen species decreased in abundance away from watering points. Of these, the majority were forbs while two were shrubs and two grasses. Although these species were dominant near watering points, they were restricted to the highly disturbed zone around the watering points and did not constitute an important part of the overall rangeland cover. Five of the six alien species that were encountered in the study belonged to this group and decreased in abundance away from watering points. The prediction that ephemeral and alien plants would be more common near watering points was clearly supported by the results.
ephemeral and alien species are associated with disturbed areas around livestock watering points
the resilience of the nama-karoo to livestock grazing
The patterns in plant cover and species richness away from watering points indicate that changes in species richness continue to occur well beyond the zone in which plant cover is impacted. This suggests, first, that plant species richness is more sensitive to grazing impact than plant cover and, secondly, that the major impact of livestock beyond the highly disturbed zone immediately adjacent to the watering point is on plant community composition rather than structure. That the Karoo should exhibit relative resilience to livestock grazing holds true in the sense that livestock grazing did not result in a large change in plant cover or vegetation structure (Milchunas, Sala & Lauenroth 1988).
The dominant species in study area were the shrubs Pentzia incana (Thunb.) Kuntze, Rosenia humilis (Less.) K.Bremer and Zygophyllum lichtensteinianum Cham. & Schltdl., which are all widespread species of moderate to low palatability (Shearing 1994). The abundance of these species was reduced only in the highly disturbed zone around the watering point, indicating that these species are tolerant of heavy grazing pressure. It is, however, likely that the abundance of these species has increased as a result of livestock grazing at the expense of more palatable and less grazing-tolerant species (Milchunas, Sala & Lauenroth 1988). Indeed, a large proportion of species, the majority of which are regarded as highly palatable to livestock (Shearing 1994), increased away from watering points, suggesting that current grazing intensities have a negative impact on vegetation composition. Livestock grazing, even at moderate stocking rates, has been shown to negatively affect seed production and recruitment of palatable Karoo shrubs (Milton 1992, 1994, 1995). Heavy grazing of the above-ground plant parts, leading to death or repeated reproductive failure, is the most likely mechanism leading to the decline of these species (Andrew & Lange 1986b; O’Connor & Pickett 1992; Milton 1994; Todd 1999; Riginos & Hoffman 2003). These results support the contention that when herbivores are limited by forage availability rather than water availability, the majority of the rangeland is easily accessible to herbivores and may be prone to severe impact at high herbivore densities (Owen-Smith 1996). The relative resilience of dominant Karoo shrubs to livestock grazing may confer these rangelands with an ability to maintain relatively high stocking rates without obvious impact. As the dominant species are tolerant of or favoured by heavy grazing, the impacts seen may be an inevitable consequence of rangeland managers maintaining stocking rates above that which can be tolerated by the majority of species but which are supported by a minority of grazing-tolerant species.
The lack of response in grass cover with distance appears to be the result of contrasting responses by different grass species. Grazing- and trampling-resistant species such as Eragrostis lehmanniana Nees and Cynodon incompletus Nees (Gibbs Russell et al. 1990) tend to increase near watering points, while grazing-sensitive species such as Stipagrostis ciliata (Desf.) De Winter and S. obtusa (Delile) Nees tend to decrease.
the majority of nama-karoo species do not recover rapidly away from watering points
Landsberg et al. (2002) examined the effects of grazing away from water points at multiple scales and found that, at the local level, more species showed a pattern of increasing than decreasing abundance with proximity to watering points. They suggest that grazing increases species richness at a local scale by providing opportunities for more species to establish. In this study, the number of species that responded positively (15) is much less than the number of species that declined (31) in response to proximity to watering points. The difference between the study site and that of Landsberg et al. (2002) may be the abundance of short-lived and ephemeral species in the latter site, which tended to increase near watering points but which are relatively uncommon on the calcrete plains habitat of the Nama-Karoo. The majority of species that decreased in abundance with proximity to watering points in the study of Landsberg et al. (2002) were, however, palatable perennial shrubs, which is consistent with the trends described in this study.
The pattern of animal impact identified in this study corresponds well with the generalized pattern identified by James, Landsberg & Morton (1999). Compared with many other studies, the highly impacted ‘sacrifice zone’ detected in this study was relatively small (Nash et al. 1999; Thrash & Derry 1999). Although this may be partly the result of the grazing tolerance of the dominant shrub species, the environment and livestock present also probably play an important role. Trampling effects associated with small hoofed animals such as sheep and goats, which predominate in the study area, are likely to be less than those associated with cattle and other larger herbivores. The poorly developed soils may also contribute to limiting the negative effects of animal activity on soil properties, such as compaction, infiltration and the development of footpaths around water points. The extent of the sacrifice zone is also affected by the number of animals using the watering point and, because paddocks in the Nama-Karoo are relatively small, the actual number of animals using a watering point will be relatively low even at high stocking rates.
In Australian studies, a zone in which the cover and species richness of palatable species reached stable values was frequently identified, usually beyond the range of domestic livestock (James, Landsberg & Morton 1999). Such a zone cannot be identified in this study, most probably because paddocks in Australia are usually an order of magnitude larger than those in the Karoo. In the study area, which is typical of the Karoo in general, the maximum distance from a watering point rarely exceeds 3 km. This coincides with the maximum foraging distance of sheep under hot conditions (Lange 1969), suggesting that most areas in the Karoo are accessible to, and hence potentially impacted by, livestock (Owen-Smith 1996). This conclusion lends potential support to the somewhat controversial contention that grazing-sensitive species that now tend to be restricted to those areas distant from watering points may, in pre-colonial times, have been the dominant species in terms of plant cover (Acocks 1979; Hoffman & Cowling 1990).
The results of this study clearly illustrate that grazing gradients associated with watering points are a primary determinant of the vegetation structure and composition of the Central Lower Karoo. Areas close to watering points tend to be overutilized and associated with a reduction in plant species richness. From a rangeland management perspective, reducing the distance livestock need to move in order to obtain water would facilitate a more even utilization of the rangeland (Landsberg et al. 2003). Under correct grazing management this would have positive consequences for overall rangeland condition as the tendency to overutilize part of the paddock would be reduced. However, under poor management or overstocking, reducing the distance to water would have a negative impact because a larger proportion of the rangeland would be easily accessible to livestock and would tend to be overutilized (Owen-Smith 1996). Large paddocks with areas distant from watering points provide a refuge for grazing-sensitive species that might otherwise be lost from the rangeland as a whole. Furthermore, as some species may be very sensitive to grazing and may be eliminated even under moderate stocking rates (Landsberg et al. 2002), larger paddocks with areas distant from water may, from a conservation perspective, provide the most secure long-term protection for these species, irrespective of stocking rate.
While the denuded zone around watering points is a conspicuous feature associated with watering point provision in arid regions, the impact of watering points on rangeland condition extends well beyond this zone. The less perceptible impact of watering point provision on plant species richness implies that plant cover alone does not provide a good indicator of overall rangeland condition. The majority of species that are abundant immediately adjacent to the watering point are often restricted to such highly disturbed sites and do not provide a useful index of general rangeland condition, being indicative only of severely impacted rangeland. Grazing-sensitive species that tend to occur away from the watering points represent potentially good indicators of rangeland condition and grazing impact. Furthermore, as these species are likely to be the first to respond to changes in grazing pressure, they represent the best candidates for monitoring and predicting future changes in rangeland condition. A limiting factor to the broad applicability of these species is, however, the fact that palatability tends to vary according to soil type, season and vegetation composition (Palmer & Hoffman 1997), and so not all the grazing-sensitive species identified in this study will be useful as indicators in other environments. Promoting the use of these species as indicators of rangeland condition within the appropriate environments will nevertheless lead to an enhanced understanding by landowners of the impacts of livestock on the vegetation of the Karoo, and ultimately contribute to improving the state of grazing management in the region.
This work forms part of the GEF Conservation Farming Project, co-ordinated by the National Botanical Institute. The Mazda Wildlife fund is gratefully acknowledged for the use of a courtesy vehicle. Two anonymous referees made useful comments on earlier drafts. Christy Bragg is thanked for field assistance and comments on earlier drafts.