• arid rangeland;
  • forage plants;
  • grazing history;
  • grazing productivity;
  • herbivore exclusion;
  • passive restoration;
  • rainfall;
  • resilience;
  • stocking rates


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

1. Up to 73% of the world’s rangelands are degraded, and increasing demand for meat in developing countries and a growing human population are likely to exert even greater pressures on rangelands in the next 20–50 years. Restoration of rangeland grazing potential and resilience is therefore important, particularly in the face of climate change.

2. We investigated the influence of past stocking rates (from 1910 to 1987), rainfall, and current grazing regimes (from 1988 to 2008) on plant assemblages, grazing potential, and diversity of palatable species in southern Karoo rangelands, South Africa.

3. We used herbivore exclusion experiments to test whether resting rangeland for 20 years enables recovery of plant assemblages (where seed sources are present within 50 m), regardless of previous grazing history. Mean annual rainfall over this period was 15% higher than the mean annual rainfall for the preceding 80 years and included two exceptionally wet years.

4. While rainfall was a primary driver of total vegetation cover, grazing history explained differences in plant species composition: plots with shared historical grazing intensity were more similar than plots with the same grazing regimes between 1988 and 2008.

5. In historically heavily-grazed exclusion plots, cover of the palatable species Tripteris sinuata (formerly Osteospermum sinuatum) returned to levels comparable to that in both exclusion and lightly-grazed plots with a moderate grazing history. Five palatable species (Pteronia empetrifolia, Tetragonia spicata, Berkheya spinosa, Hereroa latipetala and Ruschia spinosa) failed to re-establish, however, despite the presence of seed-producing plants nearby. Furthermore, only cover of P. empetrifolia increased significantly in historically moderately-grazed plots. Cover of unpalatable plants (e.g. Pteronia pallens) increased in all plots over time.

6.Synthesis and applications. These findings suggest that present species composition of arid shrublands reflects historical management at time scales greater than 20 years. Despite high rainfall enabling the return of grazing potential through recovery of a single forage species, rest alone did not ensure the return of all palatable species, with implications for rangeland resilience. Restoring the full suite of palatable species over management timeframes will require more complex interventions such as reseeding or selective clearing.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

As much as 40% of the world’s terrestrial area now functions as cropland, pasture or rangeland (Foley et al. 2005), and it is estimated that up to 73% of the world’s rangelands may be degraded (Lund 2007). In the context of a human population that is increasing at a rate of nearly 225 000 per day (Population Reference Bureau 2008), coupled with higher per capita consumption patterns in developed and emerging economies, human society needs rangelands to be highly productive.

The causes of degradation and decline of palatable cover in arid rangelands have been debated for over two decades (see review in Vetter 2005). Much of the argument over rangeland models has hinged on whether abiotic factors are more important than biotic factors as determinants of species abundance and composition (Ellis & Swift 1988; Illius & O’Connor 1999). Regardless of the underlying causes, however, the ability of rangelands to recover is still poorly understood. While there has been increasing global interest in the restoration of degraded rangelands (e.g. Aronson et al. 1993; Holmgren & Scheffer 2001; Curtin 2002), there are few long-term studies. Briske, Fuhlendorf & Smeins (2003) reported on 11 recently published (i.e. post 1995) investigations on the effects of grazing or grazing exclusion on rangelands, of which only six had datasets longer than 20 years, and only one of these (O’Connor & Roux 1995) had been conducted outside the USA. Since then, only three additional herbivore exclusion studies longer than two decades have been published, one in the USA (Valone & Sauter 2005), one in Mongolia (Sasaki et al. 2009), and one in the Sahel (Miehe et al. 2010).

A primary goal of rangeland restoration is the return of palatable cover, with a concomitant increase in grazing carrying capacity. A key question for restoration efforts concerns the role of palatable cover in restoring productivity: how complete a suite of (historically present) palatable species is necessary for restoration to be considered successful? In other words, what levels of functional redundancy (the ‘degree to which organisms have evolved to do similar things’; Rosenfeld 2002, p. 157) should be present in a restored system? An ecosystem is considered restored when it ‘contains sufficient biotic and abiotic resources to continue its development without further assistance or subsidy’ and ‘will demonstrate resilience to normal ranges of environmental stress and disturbance’ (Society for Ecological Restoration International Science & Policy Working Group 2004, p. 3), but as with many such definitions, the devil lies in the details.

A high level of functional redundancy can be expected to contribute to increased resilience (the amount of disturbance a system – including its fundamental structure, process and functions – can endure before changing to a qualitatively different state; Gunderson & Holling 2002). Different species respond differently to disturbance or environmental stress, so the presence of a number of functionally similar species enables communities to maintain ecosystem function in response to perturbation (Walker, Kinzig & Langridge 1999). As ecosystems change, and dominant species decline or are lost, functionally equivalent species can be expected to substitute for them, enabling ecosystem function to be maintained (Tilman & Downing 1994; Naeem & Li 1997; Walker, Kinzig & Langridge 1999; Petchey & Gaston 2002; Memmott et al. 2004). The original definition of ecosystem resilience (i.e. as return time to an equilibrium point following a perturbation) suggests that more resilient rangelands should be restored more rapidly, and that more diverse rangelands should be more resilient.

Continuous selective grazing in natural rangelands typically affects palatable species, leading to declines (or even losses) of this functional group (O’Connor 1991; Milton 1995a). Two questions then arise – can rangelands regain their grazing productivity with rest, and is rest alone sufficient to enable the return of the full suite of palatable species. Yayneshet, Eik & Moe (2009) found rest was sufficient to restore plant species composition, biomass and cover to heavily-grazed rangelands, but a review by Curtin (2002) concluded that in general, rangelands do not recover with rest alone, and that climate is an overwhelming factor in determining restoration of rangelands. Holmgren & Scheffer (2001) argued that favourable weather conditions should be used opportunistically, along with grazing control, to bring about effective restoration of degraded rangeland. Regeneration of the desired components of rangelands has been shown to be triggered by good rainfall events in many arid and semi-arid systems (Gutiérrez et al. 1997; Holmgren et al. 2001). These ideas are consistent with the consensus that restoration attempts are most effective when managers can restore and sustain natural processes, using appropriate climatic conditions and disturbance factors that interact with grazing to elicit desired ecological effects such as shifts in vegetation composition (Westoby, Walker & Noy-Meir 1989; Curtin 2002).

In the arid rangelands of South Africa, historical livestock census data show that stock numbers peaked in the country in the 1920s and have subsequently fallen to half the peak density value (Milton 1995a). It has been argued that these livestock declines were driven by a reduction in forage availability (Dean et al. 1995), rather than by a decline in the market for wool and meat or by stock reduction policies. Despite scientific advances over the past couple of decades, our knowledge of long-term rangeland recovery in southern Africa remains limited. In 1988, a set of exclusion plots was established at our study site in the southern Karoo. Our 20 year data set incorporates plant transects from areas that experienced different grazing intensities and histories, dating back a century. Between 1988 and 2008, mean annual rainfall was 15% higher than the mean for the preceding 80 years, and included two exceptionally high rainfall years, allowing the effects of above-average rainfall on rangeland recovery to be compared with and without grazing exclusion. Although most species in our study plots are ‘medium’ (5–20 years) to ‘very long-lived’ (>80 years) (Wiegand et al. 2000), we anticipated some changes in plant species composition over this time frame.

In this study we asked, using two decades of grazing exclusion:

  • 1
     How have plant assemblages in areas with historically different grazing intensities changed in response to grazing exclusion?
  • 2
     Does rest alone lead to return of palatable plant cover (grazing productivity), regardless of whether grazing in the past was heavy or moderate?
  • 3
     Given above-average rainfall and a seed source within 50 m, do all species depleted by grazing return to rested rangelands within twenty years, i.e. do the rangelands exhibit resilience in timeframes of management or ownership?

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study area

The study was carried out in the Sandrivier valley (33°10′S, 22°17′E, 800 m altitude) of the southern Great Karoo on the adjacent sheep ranches of ‘Argentina’ and ‘Tierberg’ (more fully described in Milton 1992; Milton, Dean & Kerley 1992). Rainfall is low (range 50–400 mm year−1; mean 175 mm), but mean annual rainfall during the 20 year observation period (1987–2006) exceeded the mean annual rainfall in any of the four preceding 20-year periods by 15% (Table 1) and included 2 years with roughly double the mean annual rainfall – 1996 (346 mm) and 1998 (363 mm). Although there were two relatively dry years, where rainfall was marginally below 60% of the long-term mean (1990 and 2003), there were no droughts during the two decades over which this experiment was run.

Table 1.   Mean annual rainfall for five 20-year periods, based on rainfall data from the South African Weather Bureau (South African Weather Service 2007), supplemented by records kept by SJM and WRJD for 2008–2009
20-year periodMean annual rainfall (mm)Standard deviation (mm)

Peak monthly rainfall averages are recorded in autumn (March–May). Annual variability in both seasonality and quantity is high. Temperatures range from −5 °C (winter) to 43 °C (summer). In about 1910 a fence was erected, separating Tierberg from Argentina, and an additional fence constructed in 1987 within Tierberg led to the establishment and separation of the Tierberg Karoo Research Centre (TKRC) from the greater Tierberg ranch. The vegetation is characterized by low succulent and non-succulent shrubs of under 0·5 m in height.

Grazing history differs between sites. Tierberg (including TKRC) underwent moderate grazing by merino sheep (stocked at 6 ha SAU−1; SAU = small animal unit approximately equal to one sheep) until 1987. In 1987, TKRC and Tierberg were separated by fencing that allowed wild herbivores such as hares Lepus capensis and steenbok Raphicerus campestris to move easily between the two sites. In that year, shrubs palatable to sheep totalled almost 70% of ground cover, and stood at densities of 21 700 plants ha−1 (TKRC) and 18 600 ha−1 (Tierberg) (Milton 1995b). The third site, Argentina, has been more degraded (Dean et al. 1993), and experienced overstocking prior to 1960 (2·7 ha SAU−1, with Dorper sheep and domesticated ostriches Struthio camelus, after which stocking density was reduced to 6 ha SAU−1 (Milton 1992). Palatable plant density on Argentina was only 4200 ha−1 (19% of vegetation cover) in 1987 (Milton 1995b). Currently, TKRC has no sheep, and Argentina and Tierberg hold about 6 ha per sheep-equivalent per year.

Experimental layout

We used 40 of the plots established in 1988 on the study site, more fully described in Milton (1992). On the ranch with a history of heavy grazing (Argentina), we used eight plots that were currently (i.e. between 1988 and 2008) grazed by sheep, and six plots from which all grazing had been excluded over the 20-year period. On Tierberg and TKRC, which share a moderate grazing history, we used six plots that are now grazed by sheep, 10 plots that are now grazed by wildlife, and 10 plots from which all grazing, including small mammals and reptiles, had been excluded over the last two decades.

Sampling trends in vegetation cover and composition

Over 20 years, vegetation cover and species composition were recorded for each plot on eight separate occasions (December 1988, July 1989, April 1990, April 1991, December 1998, May 2005, June 2006 and November 2008). Percentage cover was measured using the line-intercept method (Mueller-Dombois & Ellenberg 1974). Four line-intercept transects were established at marked points in each plot, measured to an accuracy of one decimeter (0·1 m), and divided over the length of the sampling line. Plant nomenclature follows Germishuizen & Meyer (2003).

Data analysis

We used a two-way crossed analysis of similarities (anosim) (Clarke & Warwick 1994) to ascertain if plant assemblages differed based on grazing history or grazing regime (treatment) over the period 1988–2008. We then explored species composition between different histories and treatments using cluster analysis, based on untransformed percentage plant cover in each plot, sorting the data using group averaging. We used a simper analysis to ascertain which species were responsible for the patterns observed. These multivariate analyses were conducted using primer v. 6 (Clarke & Gorley 2001).

We then selected five species of differing palatability and longevity and tracked their responses to the different grazing treatments over the period 1988–2008. The species selected were Pteronia pallens (very long-lived, unpalatable), Tripteris sinuata (long-lived, palatable), Pteronia empetrifolia (long-lived, palatable), Drosanthemum montaguense (medium-lived, slightly palatable), and Brownanthus ciliatus (medium-lived, unpalatable). We first ascertained if there was any relationship between cover and time for each of the species, and sought a best-fit regression model for each species. The best-fit model was that which, if significant overall, explained the greatest variation. Mathematically, R2 values are higher for models that include more variables, so we determined whether higher variance explained by non-linear models was statistically significant, using F tests. If variance explained was not significantly greater, we elected to use the simpler linear models. We corrected for false detection rates using sequential Bonferroni tests (Rice 1989). Finally, for those species showing significant linear changes in cover over time from the beginning of the experiment, we used linear-mixed effects models with maximum likelihoods. We assessed model fit using Akaike Information Criteria (AIC), comparing a null model, grazing treatment alone, grazing history alone and grazing treatment nested within grazing history as models for assessing changes in percentage cover with time. Grazing history, grazing treatment and time were held as fixed effects, with plot number as a random effect. We also hypothesized that exclusion plots that were heavily grazed in the past might see a greater rate of increase in cover than plots that have been moderately grazed in the past, owing to relatively little competition compared to moderately-grazed plots, along with release from grazing. We also hypothesized that historically heavily-grazed plots that continued to be grazed by sheep would show a slower rate of recovery than all other plots. We therefore also grouped all moderate treatments, historically heavily-grazed exclusion plots, and historically heavily grazed but continually grazed by sheep, as three different groups, and assessed the fit of this model (the ‘hypothesized model’) against the other models. These analyses were carried out in R (R Development Core Team 2005).

To ascertain the recovery of palatable species diversity overall, we compared average percentage cover of four species considered highly palatable (Hereroa latipetala, Ruschia spinosa, Berkheya spinosa and Tetragonia spicata, along with Tripteris sinuata, in exclusion plots on both the historically heavily and moderately-grazed farms. We used Mann–Whitney U-tests to compare percentage cover in 1988 with cover in 2008, carried out in Unistat (Unistat Ltd 2009).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

A total of 31 perennial plant species was surveyed in the permanent plots used in this study. Long-term grazing history (prior to 1987, i.e. between 1910 and 1987; hereafter referred to as ‘grazing history’) emerged as a significant factor driving the observed similarities in species composition across all grazing treatments (‘grazing treatments’ = grazing between 1987 and 2007; ANOSIM: Global R = 0·31, < 0·01). Across all grazing history groups, however, treatments were not significantly different from each other (anosim: Global R = −0·028, > 0·05). Inspection of the dendrogram produced from the cluster analysis confirmed these results (Fig. 1).


Figure 1.  Plots that were heavily grazed in the past were more similar in species composition, regardless of grazing regime between 1988 and 2008. Heavily-grazed plots differed from those that had been moderately grazed in the past by 59·65%. Species that explained 66% of this dissimilarity are listed in Table 2.

Download figure to PowerPoint

Five species explained almost two-thirds of the differences in grazing history (Table 2). As a rule, palatable species had higher percentage cover in areas that had histories of light grazing. The four species responsible for almost 60% of the dissimilarity were all medium to long-lived or long-lived. However, only T. sinuata and B. ciliatus could be considered good discriminating species, with average percentage cover of T. sinuata being nearly 40% greater in plots of moderate grazing history compared with those of heavy grazing history (5·4% vs. 7·4%, respectively). Similarly, B. ciliatus had an average percentage cover seven times higher in plots of heavy grazing history compared to those with a history of moderate grazing (3·5% vs. 0·5%, respectively).

Table 2.   Results from simper analysis, showing species that explained up to two-thirds of the dissimilarities in plant assemblages with different grazing histories. Species with high Dissimilarity/SD ratios (‘Ratio’ in this table) contributed consistently across replicates. Species for which this ratio was greater than 1·3 can be considered as good discriminating species (Clarke & Warwick 1994); these taxa are underlined in the table. ‘Short-lived’ species are those that live for less than 5 years; ‘medium-lived’ for 5–20 years; ‘long-lived’ for 20–25 years and ‘very long-lived’ for greater than 50 years
SpeciesHeavy Grazing historyModerate grazing historyAverage DissimilarityRatioContribution %Cumulative %
Average % coverAverage % cover
Tripteris sinuata(very palatable, medium to long-lived)5·47·411·821·3619·819·8
Pteronia pallens (poisonous, very long-lived)10·96·210·241·1417·237·0
Ruschia spinosa (palatable, long-lived)0·34·37·121·2811·948·9
Drosanthemum montaguense (slightly palatable, long-lived)3·01·35·541·129·358·2
Brownanthus ciliatus(unpalatable, medium-lived)3·50·54·971·488·366·5

Plots with a history of heavy grazing had an average similarity of 45·5% (Table 3). Pteronia pallens accounted for over half (53·3%) of the similarity between historically heavily-grazed plots, regardless of whether grazing had been excluded or not over the last 20 years. Plots with a moderate grazing history had an average similarity of 49·9%, and Tripteris sinuata was the only species that could be considered a good discriminating species, explaining 34·3% of the similarity, regardless of grazing treatment (Table 3).

Table 3.   Species that explained 90% of the similarity between plots with a history of heavy grazing, regardless of grazing treatment, or moderate grazing, regardless of grazing treatment (simper analysis). Species with high Similarity/SD ratios (‘Sim/SD’ in this table) greater than 1·3 can be considered good discriminating species (Clarke & Warwick 1994), and are underlined in the table
Group: Heavy grazing history (Average similarity: 45·48)
SpeciesAverage abundanceAverage similaritySim/SDContribution %Cumulative %
Pteronia pallens(poisonous, very long-lived)0·1124·221·7953·2653·26
Brownanthus ciliatus (unpalatable, medium-lived)0·036·681·1414·6967·95
Tripteris sinuata (very palatable, medium to long-lived)0·065·170·7511·3679·31
Galenia fruticosa (slightly palatable, medium-lived)0·032·850·666·2685·58
Malephora lutea (unpalatable, short-lived)0·022·750·646·0491·62
Group: Moderate grazing history (Average similarity: 49·91)
SpeciesAverage abundanceAverage similaritySim/SDContribution %Cumulative %
Tripteris sinuata(very palatable, medium to long-lived)0·0717·141·634·3434·34
Pteronia pallens (poisonous, very long-lived)0·0614·191·0628·4462·78
Ruschia spinosa (palatable, long-lived)0·047·630·8915·2978·08
Galenia fruticosa (slightly palatable, medium-lived)0·023·810·997·6485·72
Pteronia empetrifolia (palatable, long-lived)0·0330·536·0291·74

Species responses to changing stocking rates over time

Although all species in all treatments showed best-fit relationships with time (percentage cover vs. months) that were linear, only some of these were statistically significant after application of sequential Bonferroni tests to correct for false detection rates. Brownanthus ciliatus cover appeared to decline over time, but the decline was not statistically significant. Drosanthemum montaguense cover increased with time (Spearman Rank R = 0·378, = 0·0017) in plots that had a history of heavy grazing in which sheep grazing was ongoing. Cover of Pteronia pallens was weakly but positively correlated with time (Spearman Rank R = 0·254; = 0·006) in plots with a history of moderate grazing from which all grazers had been excluded for the last 20 years.

For all treatments and all grazing histories, cover of the palatable T. sinuata increased linearly with time (Fig. 2). We therefore used a linear-mixed effects model to assess changes in percentage cover of this species with time. Model selection using AIC selected our hypothesis (‘hypothesized model’) that moderately-grazed treatments would have a slower rate of increase than historically heavily-grazed exclusion plots, but that cover in historically heavily-grazed plots that continued to be grazed by sheep had increased slower than all other treatments, as the best model (Null model AIC = 1717·9; Percentage cover of T. sinuata increasing at the same rate within grazing histories: AIC = 1713·9; Percentage cover of T. sinuata over time increasing at the same rate within grazing treatments: AIC = 1708·3; Percentage cover of T. sinuata increasing at the same rate within grazing histories and treatments: AIC = 1672·2; Hypothesized model: AIC = 1644·1).


Figure 2.  Relationship between percentage cover of the palatable shrub T. sinuata with time over nearly 20 years. The relationship between time and cover revealed a significantly greater increase in cover in plots that had a history of heavy grazing, from which grazing was excluded, compared to all other plots. Plots that had a history of heavy grazing and continued grazing increased significantly slower than all other plots. The rates of increase of T. sinuata cover in plots with a history of moderate grazing were not significantly different to each other.

Download figure to PowerPoint

A comparison of average percentage cover of Tripteris sinuata in 2008 showed that only the historically heavily-grazed site that had continued to be grazed by sheep between 1988 and 2008 had a significantly lower percentage cover (Kruskal–Wallis One-way anova Chi-square = 9·67, = 0·046, = 5; Multiple comparisons with t-distributions), whilst the exclusion treatment plot on the historically heavily-grazed site was not significantly different to the historically moderately-grazed sites for all current grazing treatments.

Pteronia empetrifolia, another palatable species, showed a very different response. In areas that had been heavily grazed, even in exclusion plots, P. empetrifolia failed to re-establish. However, in areas that had been moderately grazed, cover increased significantly between 1988 and 2008 (Table 4). A comparison of average percentage cover in exclusion plots between 1988 and 2008 of four other palatable species (Tetragonia spicata, Berkheya spinosa, Hereroa latipetala, Ruschia spinosa) on the farm with a history of heavy grazing showed that 20 years of rest did not result in any significant change in percentage cover, which remained close to zero for all four species. Although these palatable species did not return to plots with a history of heavy grazing, most of these species also did not exhibit change in cover between 1988 and 2008 in plots with historically moderate grazing (Table 4).

Table 4.   Changes in median percentage cover of five palatable species between 1988 and 2008 in exclusion plots that had been moderately grazed in the past. Only Pteronia empetrifolia showed a significant increase in cover. In plots that had been heavily grazed in the past, mean percentage cover remained close to zero for all five species
Species1988: Median cover (Lower and upper quartiles)2008: Median cover (Lower and upper quartiles)Mann–Whitney UP
Pteronia empetrifolia0·5% (0%; 2%)3·5% (0%; 7%)111< 0·05; (n = 38)
Hereroa latipetala1% (0·5%; 2%)1% (0%; 1·5%)188·5> 0·05; (n = 40)
Tetragonia spicata0·05% (0%; 0·1%)0·01% (0%; 0·2%)212> 0·05; (n = 40)
Ruschia spinosa6% (1%; 8%)3·5% (1%; 6%)242·5> 0·05; (n = 40)
Berkheya spinosa0% (0%; 0%)0% (0%; 0%)240·5> 0·05; (n = 40)


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Drivers of plant assemblage composition

Despite two decades of differing grazing treatments, historical grazing intensity emerged as the single significant driver of plant species composition in this arid shrubland. This finding shows that a considerable period of rest, including two above-average rainfall events, may still be insufficient to allow some plant assemblages to recover. Although the combination of favourable climatic events and the removal of grazing has seen some change (average percentage of palatable cover increased), the time period has been insufficient to see a change in cover of long-lived unpalatable species, like Pteronia pallens, which explained more than 50% of the similarity between sites with a history of heavy grazing, regardless of grazing treatment.Wiegand et al. (2000) estimate that 90% of P. pallens individuals reach ages of between 80 and 100 years, which may help explain why P. pallens is the primary species driving similarities between historically heavily-grazed sites. In addition, amongst shorter-lived species, which would perhaps be expected to exhibit changes in percentage cover over the time scale of this experiment, few species (regardless of palatability) showed significant changes in percentage cover. Selective grazing in heavily-stocked areas leads to an increase in densities of unpalatable species. The resultant disproportional representation of unpalatable species leads to a cycle of increased seed input, and higher seedling survival of unpalatable species, relative to moderately-grazed areas (where competition from established plants could be expected to increase seedling mortality; Milton 1995a,b). The expected endpoint is an incremental decrease in carrying capacity (Todd & Hoffman 1999).

Sites with a moderate grazing history owed their similarity to the presence of the palatable T. sinuata (which accounted for more than a third of the observed similarity between these sites). Tripteris sinuata, although poorly dispersed in space and time, is relatively fast growing and able to take advantage of good rainfall events (Wiegand et al. 2000), enabling it to increase its representation in areas where there were both parent plants and restricted grazing. Furthermore, T. sinuata has been shown to resprout after damage, which may also help explain its recovery (Rahlao et al. 2009).

Return of rangeland grazing potential

We found that T. sinuata returned to heavily-grazed rangeland over the 20 year study period if the site was within 50 m of rangeland in good condition, and where the species was plentiful. The average percentage cover of the palatable T. sinuata in plots with a history of heavy grazing regardless of grazing treatment was not significantly different to those with a moderate grazing history, indicating recovery. Our hypothesis that cover of T. sinuata would show a greater rate of increase in exclusion plots was supported by the data, probably because of reduced intra- (and perhaps inter-) specific competition in previously heavily-grazed plots, along with escape from herbivory.

The recovery of T. sinuata in protected plots suggests that rest can enable carrying capacity to return provided that seed is available. Most of the increase in cover can be attributed to a single recruitment event following above-average summer rainfall in 1996 (Milton & Wiegand 2001). Thus, with rest, the expected endpoint need not be a rangeland with reduced carrying capacity for livestock, although recovery potential may vary with species, depending upon dispersal ability or current population size and the climatic events suitable for establishment.

Rates of recovery were affected by both current and past grazing regimes. Recovery of T. sinuata was significantly slower in plots with a history of heavy grazing that endured ongoing grazing by sheep, compared with plots with a history of moderate grazing and the equivalent treatment. This is probably because once the vegetation is in a state of low biomass, the grazing pressure needed to subsequently suppress vegetation re-growth is far lower than that needed to cause the collapse in the first place (Noy-Meir 1975). The implications are that lower densities of palatable species (and consequently lower seed availability), combined with continued grazing, and perhaps also increased competition from unpalatable species, hamper rates of recovery.

Rangeland resilience

Despite the contribution of T. sinuata in restoring the grazing potential of the rangeland to previous levels, certain palatable species (Pteronia empetrifolia, Tetragonia spicata, Berkheya spinosa, Hereroa latipetala and Ruschia spinosa) did not increase or establish at all in historically heavily-grazed plots, and showed no appreciable increase in cover in moderately-grazed rangelands, despite rainfall over the two decades of the experiment exceeding the long-term mean by 15%. The study period included two exceptionally high rainfall years, which would have been expected to trigger recruitment events (since seed sources were within 50 m of the exclusion plots of even the heavily-grazed rangeland) and increases in cover. Milton (1995b) found that if there were good post-germination rains in winter, spring and summer, after flowering and seed set, seedlings could recruit successfully. In response to such rains in 1996 and 1998, Tripteris sinuata showed a marked increase in cover and recruitment (Milton & Wiegand 2001). Within the heavily-grazed rangeland, dominance by unpalatable species like P. pallens is ongoing probably because these plants, avoided by herbivores, continue to grow and because many of them have long life-expectancies. Mature (and large) unpalatable plants can be expected to produce more seed than the palatable species that are kept small by browsing. Previous studies have found persistent seed banks to be relatively unimportant to perennial shrubs in the southern Karoo (Esler, Cowling & Ivey 1992; Milton 1995b), and it seems as if population renewal in these species depends on annual production of seed. The low density of palatable species also implies a lower representation in both the annual seed rain and the seed bank, and so lower likelihood of recruitment of these species.

Palatable species are believed to be superior competitors to their unpalatable counterparts (Moretto & Distel 1997; Olff & Ritchie 1998), an advantage that falls away in the face of heavy herbivory and the concomitant impacts on seed production. It is known that Pteronia empetrifolia usually outcompetes its congeneric P. pallens (Yeaton & Esler 1990). Pteronia empetrifolia was absent in 1987 on sites with a heavy grazing history, but still present on those with a moderate grazing history. The implication of this absence is that grazing tips the balance in favour of the unpalatable P. pallens. Milton (1995a) showed that when grazed by sheep during bud development, P. empetrifolia inflorescence density fell by almost 90%. In the absence of sheep, however, P. empetrifolia produced more flowers and seeds per bush than did the inedible P. pallens and increased significantly in rested areas with a history of moderate grazing (Milton 1995a).

Grazing potential can be restored by rest in highly stocked arid rangelands, as illustrated in this study. However, the problem with the return of only a single palatable species is that even though it may restore productivity, it leaves the system less resilient to future perturbations. Continuous heavy grazing in the past has led to a lower diversity of palatable species; this loss may not affect grazing potential in times of low stress, but if external perturbations (e.g. climate change) cause a shift in plant species composition, the palatable species that could respond by becoming more abundant may not be present to do so. The loss of palatable species diversity at ranch scale does not mean that these species will never return. What is apparent, however, is that the return interval is greater than time frames of management and ownership (approximately two decades). Furthermore, the potential for alternate stable states may mean that degraded rangeland never returns to its former state (Suding, Gross & Houseman 2004).

The findings of this 20-year data set suggest that future plant composition of arid rangeland can be dictated by current stocking practices, as is evident from this example in the Karoo. Rest alone, even when combined with auspicious climatic events, cannot be expected to reverse the effects of management practices today. Rangeland with a history of continuous heavy grazing, with two decades of rest, will still have more in common with other heavily-grazed land than with rangeland consistently exposed to moderate grazing (even though such rangeland never experiences any rest). The shift from plant assemblages diverse in palatable species to those dominated by poisonous and/or inedible species, has potentially serious implications for the current and future resilience and productivity of rangeland. A return of resilient productivity to rangelands that have lost diversity of palatable species is unlikely without active restoration efforts, and farmer support programmes will need to consider this. These interventions will probably include reduction in density of long-lived unpalatable plants (Milton 1994), combined with other effective restoration methods such as reseeding, mulching, application of gypsum, use of nurse plants, brushpacks, and tilling, in appropriate seasons (Beukes & Cowling 2003; Jones & Esler 2004; Visser, Botha & Hardy 2004).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The continued existence of the Tierberg Research site is sponsored by a grant administered jointly by the South African Department of Science and Technology and BIOTA southern Africa. We thank Dr. Res Altwegg for statistical advice, two anonymous reviewers and the associate editor for comments which improved the manuscript. Data for the Tierberg study site are available at


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Aronson, J., Floret, C., Le Floch, E., Ovalle, C. & Pontanier, R. (1993) Restortation and rehabilitation of degraded ecosystems in arid and semi-arid lands. 1. A view from the south. Restoration Ecology, 1, 817.
  • Beukes, P.C. & Cowling, R.M. (2003) Evaluation of restoration techniques for the succulent Karoo, South Africa. Restoration Ecology, 11, 308316.
  • Briske, D.D., Fuhlendorf, S.D. & Smeins, F.E. (2003) Vegetation dynamics on rangelands: a critique of the current paradigms. Journal of Applied Ecology, 40, 601614.
  • Clarke, K.R. & Gorley, R.N. (2001) Plymouth Routines In Multivariate Ecological Research (PRIMER). Primer-E Ltd, Plymouth.
  • Clarke, K.R. & Warwick, R.M. (1994) Change in marine communities: an approach to statistical analysis and interpretation. Plymouth Marine Laboratory, Plymouth, MA.
  • Curtin, C.G. (2002) Livestock grazing, rest, and restoration in arid landscapes. Conservation Biology, 16, 840842.
  • Dean, W.R.J., Milton, S.J., Du Plessis, M.A. & Siegfried, W.R. (1993) Dryland degradation: symptoms, stages and hypothetical cures. Wildland shrub and arid land restoration symposium (eds B.A.Roundy, E.D.McArthur, J.S.Haley & D.K.Mann), pp. 178182. US Department of Agriculture, Forest Service, Las Vegas, NV.
  • Dean, W.R.J., Hoffman, M.T., Meadows, M.E. & Milton, S.J. (1995) Desertification in the semi-arid Karoo, South Africa: review and reassessment. Journal of Arid Environments, 30, 247264.
  • Ellis, J.E. & Swift, D.M. (1988) Stability of African pastoral ecosystems: alternate paradigms and implications for development. Journal of Range Management, 41, 450459.
  • Esler, K.J., Cowling, R.M. & Ivey, P. (1992) Seed biology of three species of Mesembryanthema in the southern succulent Karoo. South African Journal of Botany, 58, 343348.
  • Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., Chapin, F.S., Coe, M.T., Daily, G.C., Gibbs, H.K., Helkowski, J.H., Holloway, T., Howard, E.A., Kucharik, C.J., Monfreda, C., Patz, J.A., Prentice, I.C., Ramankutty, N. & Snyder, P.K. (2005) Global consequences of land use. Science, 309, 570574.
  • Germishuizen, G. & Meyer, N.L. (2003) Plants of southern Africa: an annotated checklist. Strelitzia, 14, 11231.
  • Gunderson, L.H. & Holling, C.S. (2002) Panarchy: Understanding transformations in human and natural systems. Island Press, Washington, DC.
  • Gutiérrez, J.R., Meserve, P.L., Herrera, S., Contreras, L.C. & Jaksic, F.M. (1997) Effects of small mammals and vertebrate predators on vegetation in the Chilean semiarid zone. Oecologia, 109, 398406.
  • Holmgren, M. & Scheffer, M. (2001) El Niño as a window of opportunity for the restoration of degraded arid ecosystems. Ecosystems, 4, 151159.
  • Holmgren, M., Scheffer, M., Ezcurra, E., Gutiérrez, J.R. & Mohren, G.M.J. (2001) El Niño effects on the dynamics of terrestrial ecosystems. Trends in Ecology & Evolution, 16, 8994.
  • Illius, A.W. & O’Connor, T.G. (1999) On the relevance of nonequilibrium concepts to arid and semiarid grazing systems. Ecological Applications, 9, 798813.
  • Jones, F.E. & Esler, K.J. (2004) Relationship between soil-stored seed banks and degradation in eastern Nama Karoo rangelands (South Africa). Biodiversity and Conservation, 13, 20272053.
  • Lund, H.G. (2007) Accounting for the world’s rangelands. Rangelands, 29, 310.
  • Memmott, J., Waser, N.M. & Price, M.V. (2004) Tolerance of pollination networks to species extinctions. Proceedings of the Royal Society B: Biological Sciences, 271, 26052611.
  • Miehe, S., Kluge, J., Von Wehrden, H. & Retzer, V. (2010) Long-term degradation of Sahelian rangeland detected by 27 years of field study in Senegal. Journal of Applied Ecology, 47, 692700.
  • Milton, S.J. (1992) Effects of rainfall, competition and grazing on flowering of Osteospermum sinuatum (Asteraceae) in arid Karoo rangeland. Journal of the Grassland Society of South Africa, 9, 158164.
  • Milton, S.J. (1994) Small-scale reseeding trials in arid rangeland: effects of rainfall, clearing and grazing on seedling survival. African Journal of Range and Forage Science, 11, 5457.
  • Milton, S.J. (1995a) Effects of rain, sheep and tephritid flies on seed production of two Karoo shrubs in South Africa. Journal of Applied Ecology, 32, 137144.
  • Milton, S.J. (1995b) Spatial and temporal patterns in the emergence and survival of seedlings in arid Karoo Shrubland. Journal of Applied Ecology, 32, 145156.
  • Milton, S.J., Dean, W.R.J. & Kerley, G.I.H. (1992) Tierberg Karoo Research Centre: history, physical environment, flora, and fauna. Transactions of the Royal Society of Southern Africa, 48, 1546.
  • Milton, S.J. & Wiegand, T. (2001) How grazing turns rare seedling recruitment events to non-events in arid environments. Sustainable land-use in deserts (eds S.W.Breckle, M.Veste & W.Wucherer), pp. 197207. Springer, Heidelberg.
  • Moretto, A.S. & Distel, R.A. (1997) Competitive interactions between palatable and unpalatable grasses native to a temperate semi-arid grassland of Argentina. Plant Ecology, 130, 155161.
  • Mueller-Dombois, D. & Ellenberg, H. (1974) Aims and methods of vegetation ecology. Wiley, New York, NY.
  • Naeem, S. & Li, S. (1997) Biodiversity enhances ecosystem reliability. Nature, 390, 507.
  • Noy-Meir, I. (1975) Stability of grazing systems: an application of predator-prey graphs. Journal of Ecology, 63, 459481.
  • O’Connor, T.G. (1991) Local extinction in perennial grasslands: a life-history approach. The American Naturalist, 137, 753773.
  • O’Connor, T.G. & Roux, P.W. (1995) Vegetation changes (1949-71) in a semi-arid, grassy dwarf shrubland in the Karoo, South Africa: influence of rainfall variability and grazing by sheep. Journal of Applied Ecology, 32, 612.
  • Olff, H. & Ritchie, M.E. (1998) Effects of herbivores on grassland plant diversity. Trends in Ecology & Evolution, 13, 261265.
  • Petchey, O.L. & Gaston, K.J. (2002) Extinction and the Loss of Functional Diversity. Proceedings Biological Sciences, 269, 17211727.
  • Population Reference Bureau (2008) World Population Clock, 2008, Available at: Acc-essed on 5 August 2009.
  • R Development Core Team (2005) R: A language and environment for statistical computing, reference index version 2.2.1. R Foundation for Statistical Computing, Vienna. ISBN 3-900051-07-0, Available at:,accessed24/3/10.
  • Rahlao, S.J., Milton, S.J., Esler, K.J., Van Wilgen, B.W. & Barnard, P. (2009) Effects of invasion of fire-free arid shrublands by a fire-promoting invasive alien grass (Pennisetum setaceum) in South Africa. Austral Ecology, 34, 920928.
  • Rice, W.R. (1989) Analyzing tables of statistical tests. Evolution, 43, 223225.
  • Rosenfeld, J.S. (2002) Functional redundancy in ecology and conservation. Oikos, 98, 156162.
  • Sasaki, T., Okayasu, T., Ohkuro, T., Shirato, Y., Jamsran, U. & Takeuchi, K. (2009) Rainfall variability may modify the effects of long-term exclosure on vegetation in Mandalgobi, Mongolia. Journal of Arid Environments, 73, 949954.
  • Society for Ecological Restoration International Science & Policy Working Group (2004) The SER International Primer on Ecological Restoration.,accessed5/9/09 . Society for Ecological Restoration International, Tucson.
  • South African Weather Service (2007) Rainfall records for Prince Albert (1877-2007). SAWS, Pretoria.
  • Suding, K.N., Gross, K.L. & Houseman, G.R. (2004) Alternative states and positive feedbacks in restoration ecology. Trends in Ecology and Evolution, 19, 4653.
  • Tilman, D. & Downing, J.A. (1994) Biodiversity and stability in grasslands. Nature, 367, 363365.
  • Todd, S.W. & Hoffman, M.T. (1999) A fence-line contrast reveals effects of heavy grazing on plant diversity and community composition in Namaqualand, South Africa. Plant Ecology, 142, 169178.
  • Unistat Ltd. (2009) Unistat 5.6 for Excel,,accessed8/8/09 .
  • Valone, T.J. & Sauter, P. (2005) Effects of long-term cattle exclosure on vegetation and rodents at a desertified arid grassland site. Journal of Arid Environments, 61, 161170.
  • Vetter, S. (2005) Rangelands at equilibrium and non-equilibrium: recent developments in the debate. Journal of Arid Environments, 62, 321341.
  • Visser, N., Botha, J.C. & Hardy, M.B. (2004) Re-establishing vegetation on bare patches in the Nama Karoo, South Africa. Journal of Arid Environments, 57, 155177.
  • Walker, B., Kinzig, A. & Langridge, J. (1999) Plant attribute diversity, resilience, and ecosystem function: the nature and significance of dominant and minor species. Ecosystems, 2, 95113.
  • Westoby, M., Walker, B. & Noy-Meir, I. (1989) Opportunistic management for rangelands not at equilibrium. Journal of Range Management, 42, 266274.
  • Wiegand, T., Milton, S.J., Esler, K.J. & Midgley, G.F. (2000) Live fast, die young: estimating size-age relations and mortality pattern of shrubs species in the semi-arid Karoo, South Africa. Plant Ecology, 150, 115131.
  • Yayneshet, T., Eik, L.O. & Moe, S.R. (2009) The effects of exclosures in restoring degraded semi-arid vegetation in communal grazing lands in northern Ethiopia. Journal of Arid Environments, 73, 542549.
  • Yeaton, R.I. & Esler, K.J. (1990) The dynamics of a succulent Karoo vegetation: a study of species association and recruitment. Vegetation, 88, 103113.