Changes in population biology of two succulent shrubs along a grazing gradient

Authors


*Present address and correspondence: Corinna Riginos, Graduate Group in Ecology, c/o Department of Environmental Horticulture, University of California, Davis, CA 95616, USA. E-mail: corinna.riginos@alumni.brown.edu

Summary

  • 1Heavy livestock grazing in Namaqualand, South Africa, is threatening the region's unique diversity of succulent shrubs. This is especially true in the communally managed lands, where grazing is centred around fixed enclosures (stockposts) in which animals stay overnight. In this study we set out to determine the effects of a semi-permanent stockpost on the composition of the surrounding vegetation and the mechanisms by which grazing limits the persistence of leaf-succulent shrub populations.
  • 2We used the grazing gradient created by a stockpost to examine the impacts of grazing on vegetation composition and changes in mortality, reproductive output and seedling establishment for the leaf-succulent species Ruschia robusta and Cheiridopsis denticulata.
  • 3Vegetation composition was found to change from a community dominated by the unpalatable shrub Galenia africana at high grazing intensities to a community dominated by the palatable leaf-succulent shrub R. robusta at lower grazing intensities.
  • 4Mortality of the leaf-succulents R. robusta and C. denticulata was high at the sites closest to the stockpost, while fruit production and seedling germination were substantially reduced over distances of 800 m and 2 km for the two species, respectively. Seedling establishment was not limited by either grazing or microsite availability. Thus reduction in reproductive output is the greatest impact of heavy grazing on these two species.
  • 5Synthesis and applications. This study demonstrates that marked zonation in vegetation composition and population biology can develop around a fixed stockpost and that the greatest impact of grazing on the two leaf-succulent species studied is the suppression of flower and fruit production. Consistent suppression of reproductive output could have long-term consequences for the persistence of succulent shrub populations in the heavily grazed communal lands of Namaqualand. We recommend that (i) herders should be encouraged to relocate their stockposts regularly to prevent the development of centres of degradation, and (ii) areas should be relieved periodically of all grazing pressure to allow for successful seed set of native shrubs.

Introduction

Heavy grazing by livestock is considered to be among the main threats to the floristic diversity of arid and semi-arid regions world-wide (Dregne 1983; Middleton & Thomas 1997). The Succulent Karoo biome of South Africa, especially the Namaqualand region, is an area of exceptional plant diversity and endemism (Cowling & Hilton-Taylor 1999; Desmet & Cowling 1999), particularly within the family Mesembryanthemaceae (Cowling, Esler & Rundel 1999), most of which are shrubs of the leaf-succulent life form (short-statured shrubs with woody stems and succulent leaves). However, overgrazing in the communally managed areas, which constitute 25% of the land area of Namaqualand, is considered to pose a substantial threat to the floristic diversity of the region (Cowling & Pierce 1999).

Management practices in the communal areas, where stocking rates have, on average, been twice the National Department of Agriculture's recommended rate over the last 30 years (Hoffman et al. 1999), have been shown to bring about decreased perennial cover and altered community composition relative to privately owned farms, which are stocked at or below the recommended rate (Todd & Hoffman 1999). In general, sustained heavy grazing in the Succulent Karoo tends to result in an increased cover of unpalatable perennials, with concomitant decreases in succulent and palatable non-succulent perennial cover and diversity (Milton & Dean 1995). Although stem- and leaf-succulents appear to be negatively affected by heavy grazing, little is known about the mechanisms by which grazing impacts the population dynamics of perennial succulent shrubs (Milton & Dean 1995; Milton, Davies & Kerley 1999).

Grazing may have a negative impact on perennial shrub populations via several mechanisms: (i) by directly increasing mortality among established adults (Hunt 2001); (ii) by reducing the competitive ability of grazed shrubs relative to ungrazed, less palatable shrubs (Milton, Gourlay & Dean 1997; Assaeed & Al-Doss 2001); (iii) by reducing the reproductive output of grazed individuals (Milton & Dean 1990; Milton 1994a; Stokes 1994); or (iv) by reducing the success of seedling establishment (Milton 1994b, 1995; Hunt 2001). Although all of these mechanisms are thought to be at work in the Succulent Karoo (Milton & Dean 1995), it has not yet been established which mechanism most limits succulent shrub persistence. An understanding of the limiting process is critical for informing management decisions that balance conservation with resource utilization, particularly in the communal areas where high population pressure and demand for access to rangelands preclude measures such as complete livestock exclusion (Hoffman et al. 1999).

A grazing gradient analysis provides the ideal system by which to examine plant responses to a range of grazing intensities (Andrew 1988). Gradients of grazing intensity and impact have been shown to develop around artificial features of the landscape such as watering points (Andrew 1988; Pickup & Chewings 1994; Fusco et al. 1995; James, Landsberg & Morton 1999) and human settlements (Turner 1998a,b).

In the communal areas of Namaqualand, individual herds (comprising sheep and goats) follow daily grazing orbits around ‘stockposts’ that they leave in the morning and return to in the evening. Stockposts are semi-permanent camps typically comprising a fenced enclosure where the livestock over-night and living quarters for the herder responsible for that particular herd. Stockposts are part of the traditional method of livestock husbandry and afford the herds greater protection from predation by jackals Canis mesomelas and other carnivores. Stockposts occur at a high density (approximately one per 700 ha) and are typically located 1–3 km from one another. Stockposts have the potential to create centres of grazing intensity and impact (H. Hendricks, unpublished data), which has caused the stockpost system to be identified as one of the major contributors to landscape degradation in the communal areas.

We examined the effects of a grazing gradient created by a long-term stockpost on the surrounding vegetation, focusing on two species of leaf-succulents of the family Mesembryanthemaceae, the shrub species Ruschia robusta Schwantes and the dwarf shrub species Cheiridopsis denticulata N.E. Brown. Both are common in lightly grazed areas but appear to be sensitive to heavy grazing, particularly C. denticulata. In this study we addressed the following questions. (i) Does the sustained utilization of a stockpost affect changes in vegetation composition in response to grazing intensity? (ii) Is the persistence of succulent shrub populations in the face of heavy grazing most threatened by adult mortality, reduced reproductive output or diminished establishment success? (iii) Over what distance does a stockpost have significant effects on plant population dynamics?

Materials and methods

study site

The study area was located in the Paulshoek area (30°24′S; 18°08′E) of the Leliefontein communal reserve, Namaqualand. Paulshoek covers approximately 20 000 ha and is located on the eastern slopes of the Kamiesberg, an escarpment 900–1500 m above sea level characterized by 150–250 mm annual rainfall. The landscape is composed of rocky granite hills interspersed with sandy plains. Livestock, a mixture of sheep and goats, has been kept at twice the recommended rate over the past several decades, resulting in a reduction in perennial plant cover relative to moderately grazed neighbouring farms that are privately owned (Todd & Hoffman 1999; Carrick 2001).

Most stockposts in the Paulshoek landscape are relocated with a frequency ranging from several months to a decade or more. In this study, we focused on one stockpost known to have been stationary for 15 years and the surroundings of which have not been ploughed for cropping in the past. The average number of animals kept at this stockpost between 1998 and 2001 was 101, comparable to the mean herd size of 104 animals for the 28 herds in the Paulshoek area. There were two neighbouring stockposts, one 1·2 km to the north and one 2·5 km to the west of this stockpost. The fenced boundaries of the communal area lie 1 km south and 2·5 km east of the stockpost. Eight focal sites were chosen at increasing distances (50, 100, 200, 400, 800, 1200, 1600 and 2000 m) along a transect running south-east from the stockpost. An attempt to standardize for habitat was made by locating all focal sites in sandy lowland areas.

grazing intensity

Grazing intensity along the gradient was quantified by tracking herd routes and by dung collection. Daily herd routes for a period of 6 months (July–December 2000) were tracked by the herder and drawn onto transparencies overlying an aerial photograph of the area. The routes were then digitized and a grazing intensity index determined based on the frequency of herd passages through an area (Combrinck 2003).

Accumulation of livestock dung provides a good index of the amount of time that livestock spends grazing in a particular area (Morton & Baird 1990) and can thus be used as a measure of grazing intensity. Dung was collected from four 2 × 4-m plots at each of the eight focal sites, dried for 1 week at 70 °C, and weighed. Each plot was initially cleared and dung was subsequently collected at 2-month intervals for 10 months. Accumulated dung data for all plots and collection dates were pooled for each focal site.

soil parameters

Soil samples were collected from four 4 × 4-m plots at each of the eight focal sites. In each plot, four 100-mm deep soil samples were collected and compiled into one composite plot sample. Resistance of a saturated soil paste was determined in a United States Department of Agriculture soil cup, and electrical conductivity was calculated as 1000/resistance. Soil pH was determined in 1 m KCl, and exchangeable cations were determined in a 1-m ammonium acetate extract. Phosphorus was measured by the Bray-2 method (Bray & Kurtz 1945). Total nitrogen was established by digestion in a LECO FP-528 nitrogen analyser (LECO Corp., Michigan, USA) and organic carbon was determined by means of the Walkley–Black method (Nelson & Sommers 1982).

Infiltration rate was established by placing an open-ended cylinder of diameter 75 mm into the soil to a depth of 5 mm and measuring the time needed for 50 ml of water to be absorbed into the undisturbed soil surface (Dean 1992). Four measurements of infiltration rate were taken in each plot. Soil depth was measured at the south-west corner of each plot by hammering a sharpened metal stake into the ground until it hit bedrock.

species composition and demography

To examine changes in species composition and population demography for the leaf-succulents R. robusta and C. denticulata as well as the unpalatable shrub Galenia africana L. Aizoaceae, four 4 × 4-m plots were established at each focal site. Species, height and canopy diameter in two axes were recorded for all perennial plants in each plot. Dead plants were also counted and, where possible, identified to species level. Curves of cumulative species counts for all focal sites approached their asymptote after two or three plots, indicating that sampling was sufficient to capture local species composition. Canopy cover of individual plants was calculated assuming elliptical canopies (Phillips & MacMahon 1981).

The number of mature fruits was counted for all reproducing individuals of R. robusta and C. denticulata. Because flowers and fruits of Mesembryanthemaceae may be eaten by livestock before they have matured into woody, unpalatable capsules, all fruit counts were made in the summer months (November 2000–February 2001), when capsules had reached the woody stage but could still be differentiated from previous years’ fruits.

Plants were divided into two life-stage categories (seedlings and adults) based on size. Adults were classified as all individuals of size greater than the minimum necessary to reproduce. For G. africana, which can reproduce after only 1 year of growth, adults were classified as all individuals of height greater than 15 cm, the maximum observed height for 1-year-old plants. For R. robusta and C. denticulata, the respective relationships between volume and reproductive ability were established by measuring height and width in two dimensions and counting the number of fruits for 35 individuals of each species growing in a fenced grazing exclosure in the vicinity of the focal sites. Volumes were calculated assuming plants were half-ellipsoid in shape. The volume cut-off for R. robusta was 100 cm3 and for C. denticulata 30 cm3.

seedling germination, survival and growth

In order to examine the effects of grazing on reproductive success, we conducted a survey of post-rain germination under field conditions. Seeds of the Mesembryanthemaceae tend not to persist in soil seed banks (Esler 1999), so that post-rain germination represents recent reproductive efforts. The survey was conducted after two significant rain events in the autumn of 2001 had deposited a combined 65 mm of rainfall on the study area. Namaqualand perennial shrubs typically germinate after the first or second major rainfall of the autumn season (Cowling, Esler & Rundel 1999; Esler 1999). Seedlings were sampled along two perpendicular 15-m transects at each focal site. A circular ring with a diameter of 20 cm was placed every 50 cm along each transect, and all seedlings within the ring were identified to species and counted.

Seedling survival under different grazing intensities was studied by transplanting established seedlings of R. robusta and C. denticulata to each of the focal sites and monitoring their subsequent survival in the field over a period of 10 months. Transplants were used rather than in vivo germinants due to the scarcity of seedlings at the field site. Capsules were collected from multiple plants in the Paulshoek area in the summer of 1999–2000 and germinated in February 2000. Seedlings were transplanted in September 2000 into individual 3 × 3 × 10-cm pots containing soil collected from the field site and drought-hardened for 1 month.

Seedlings were randomized and transplanted to the field in October 2000, at which time they were equivalent in size to those of a similar age in the field. In order to determine whether seedling establishment under a heavy grazing regime is limited by the availability of preferred microsites, seedlings were planted in two different microsites: in the open or under the canopy of an adult R. robusta, the dominant perennial shrub in lightly and moderately grazed areas. We transplanted 10 seedlings per species per treatment into each of the eight focal sites and into a fenced livestock exclosure proximate to the 1200- and 1600-m focal sites. Seedlings were monitored for survival on a monthly basis for 10 months, until August 2001. Where possible, cause of death was recorded.

statistical analyses

Statistical analyses of changes in all variables, with the exception of survival of transplanted seedlings, along the grazing gradient were carried out using linear regression. The independent variable, distance from stockpost, was log-transformed for all models, and for each dependent variable the best-fit model (linear or log-transformed) was used. All models were examined for normality of residuals and homoscedasticity and discarded if these assumptions were violated. In the analysis of soil parameters, we conducted a sequential Bonferroni test on the regression results to minimize type I error in rejecting the null hypothesis of soil homogeneity among sites (Rice 1989).

Differences in seedling mortality across sites and treatments were examined using Pearson's χ2 tests. As the effect of microsite on mortality showed no interaction with the effect of increasing distance from the stockpost, the effect of distance on mortality was examined by pooling data for each focal site across microsites, and the effect of microsite was examined by pooling survival data across the eight focal sites.

Causes of death for the seedlings were divided into two categories: abiotic (water stress or burial by soil redistribution during rain events) and due to livestock presence (grazing or trampling). The number of seedling deaths in each of these categories was counted (deaths due to unknown or other causes were not included) and the likelihood of any pattern in seedling deaths was tested using Pearson's χ2.

All statistical analyses were performed with JMP (version 3.1; SAS Institute 1994).

Results

grazing intensity

The frequency of animal visitation decreased with distance from the stockpost. Both indices of grazing intensity, the number of times the herd passed through an area and total dung accumulation, decreased exponentially over the 2-km transect from the stockpost (Fig. 1). This suggested that animals spent far more time grazing at sites closer to the stockpost than at sites further away.

Figure 1.

Plots of grazing intensity at each of the eight focal sites against distance from the stockpost. Grazing intensity is quantified by (a) total dung accumulation and (b) frequency of herd passage through each focal site (after Combrinck 2003).

soil parameters

There was considerable variance in soil variables among sites (Table 1), with individual soil variables showing a significant relationship with distance from the stockpost. However, none of the soil parameters was significantly related to distance at the table-wide level after a sequential Bonferroni test at α = 0·05 (Table 2). Therefore, there were no consistent trends to indicate that the soil profile as a whole varied with distance from the stockpost.

Table 1.  Means (± SD) for soil variables among four plots at each of the eight focal sites
Dependent variableDistance from stockpost (m)
50100200400800120016002000
pH 6·1 5·8 4·9 6·4 5·8 5·9 5·1 6·2
Electrical conductivity (dS m−1) 0·73 (0·45)  0·21 (0·09)  1·02 (1·04) 0·49 (0·25) 0·35 (0·16) 0·67 (0·20) 4·71 (2·32) 0·59 (0·19)
Phosphorus (mg kg−1)20·5 (8·35)18·3 (5·50)17·0 (4·83)29·8 (19·4)44·8 (24·5)79·5 (50·8)32·0 (7·26)56·8 (39·5)
Sodium (cmol(+) kg−1)  0·20 (0·14)  0·08 (0·07)  0·32 (0·20) 0·12 (0·02) 0·14 (0·11) 0·19 (0·12) 1·29 (0·60) 0·25 (0·14)
Potassium (cmol(+) kg−1)  0·43 (0·15)  0·30 (0·10)  0·28 (0·07) 0·59 (0·14) 0·35 (0·10) 0·42 (0·07) 0·51 (0·11) 0·36 (0·08)
Calcium (cmol(+) kg−1) 2·9 (2·5) 1·7 (0·6) 1·7 (0·8) 3·5 (1·8) 3·5 (1·6) 3·6 (1·4) 3·0 (1·9) 5·1 (4·0)
Magnesium (cmol(+) kg−1) 1·2 (0·5) 0·8 (0·3) 1·4 (0·5) 1·8 (0·4) 1·8 (0·8) 1·6 (0·4) 2·4 (1·0) 1·7 (0·3)
Nitrogen (%) 0·02 (0·003) 0·02 (0·013) 0·01 (0·004) 0·01 (0·001) 0·03 (0·016) 0·04 (0·011) 0·04 (0·008) 0·03 (0·005)
Carbon (%) 0·42 (0·06)  0·28 (0·13)  0·54 (0·12) 0·30 (0·06) 0·51 (0·11) 0·59 (0·28) 0·62 (0·12) 0·45 (0·10)
Depth (cm)44 (11)47 (14)29 (6)30 (11)26 (3)27 (6)26 (11)35 (16)
Infiltration rate (ml s−1) 5·55 (3·48) 5·71 (3·26) 2·82 (2·37) 1·00 (0·79) 3·41 (3·46) 2·71 (2·41) 1·31 (1·12) 1·54 (0·90)
Table 2.  Regression parameters relating soil variables to distance (n= 8 distances) from the stockpost. Site means of the dependent variables were used for the regression. None of the relations between the variables and distance were significant at the table-wide α level of 0·05
Dependent variableRegression parameters
Slope*Constantr2P
  • *

    Log-transformed.

pH−0·02 5·800·000·956
Electrical conductivity (dS m−1)−0·03 0·660·000·303
Phosphorus (mg kg−1)* 0·33 0·640·650·016
Sodium (cmol(+) kg−1) 0·28−1·450·210·311
Potassium (cmol(+) kg−1) 0·04 0·300·050·596
Calcium (cmol(+) kg−1) 1·30−0·300·480·059
Magnesium (cmol(+) kg−1) 0·65−0·120·650·016
Nitrogen (%)* 0·25−2·290·340·128
Carbon (%) 0·12 0·150·300·157
Depth (cm)*−0·12 1·820·510·048
Infiltration rate (ml s−1)−2·44 9·450·630·045

species composition and demography

Total cover increased significantly with distance from the stockpost. Species richness of perennial shrubs did not change significantly over the grazing gradient (Tables 3 and 4), although the abundance of individual species did show considerable differences among sites. At the two most heavily grazed sites (50 m and 100 m), G. africana contributed 68% and 74% to the total perennial shrub cover, respectively. The dominant shrub in terms of cover at the two sites furthest from the stockpost (1600 m and 2000 m) was R. robusta, with 75% and 50% of the total cover, respectively.

Table 3.  Means (± SD) for species richness, total percentage cover and plant demography of Ruschia robusta, Cheiridopsis denticulata and Galenia africana among four plots at each of the eight focal sites. Fruit production data were not obtained for G. africana. A dash indicates that no shrubs were present in that category
VariableDistance (m)
50100200400800120016002000
  • *

    Mean is calculated for all shrubs in all plots.

Number of species 6·8 (2·2) 5·0 (1·4) 7·8 (0·5)  6·0 (1·8)  8·8 (3·9)  8·0 (2·2)   8·5 (1·7)  9·0 (3·2)
Cover (%)10·4 (3·9)15·3 (3·1)13·9 (2·4) 18·1 (2·8) 23·1 (8·7) 36·8 (9·7)  34·9 (5·4) 34·9 (0.029)
Number of live adult shrubs
R. robusta 3·3 (2·1) 4·3 (3·2)16·5 (11·4) 28·0 (14·9) 17·5 (9·7) 15 (9·1)  42·8 (13·6) 21·8 (11·4)
C. denticulata 0·3 (0·5) 0 (0) 2·5 (1·0)  3·3 (1·5)  7·00 (7·4) 12·3 (9·4)  22·0 (11·4) 17·0 (10·4)
G. africana12·5 (4·7)11·8 (5·4) 9·0 (6·5)  0 (0)  4·3 (4·3)  7·3 (7·2)    0 (0)  2·0 (2·7)
Number of live seedlings
R. robusta 0·3 (0·5) 0 (0) 1·8 (0·5)  1·8 (1·5)  0·5 (1·0)  1·3 (0·9)   4·8 (2·9)  1·5 (1·0)
C. denticulata 0·3 (0·5) 0 (0) 0·3 (0·5)  0·8 (1·5)  0·5 (1·0)  0·5 (1·0)   3·3 (2·2)  2·0 (1·4)
G. africana 6·8 (3·8) 5·5 (5·8) 3·5 (2·1)  0 (0)  0·5 (1·0)  2·0 (1·8)   0 (0)  0·3 (0·5)
Dead adult shrubs as percentage of total number of adult shrubs
R. robusta52·8 (9·6)54·6 (11·0)20·7 (8·1) 15·3 (8·4) 22·1 (11·8) 12·0 (7·8)  19·9 (7·4) 18·2 (10·6)
C. denticulata    –    –66·5 (11·9) 57·3 (27·7) 22·5 (9·8) 26·0 (16·9)   7·7 (2·1) 19·2 (12·4)
G. africana14·4 (6·2)13·3 (12·4)    –     –     – 23·3 (20·1)      –100·0 (0·0)
Number of fruits
R. robusta 0 (0) 1·0 (1·4)83 (97)175 (206)890 (397)514 (441)1248 (474)761 (542)
C. denticulata 0 (0) 0 (0) 0 (0)  0 (0)  0·5 (1)  0·5 (0·6)   3·8 (2·4)  0·8 (0·9)
Number of fruits per adult shrub*
R. robusta    – 0·6 (1·3)14·0 (22·0) 29·0 (59·0) 99·6 (117·1) 67·5 (120·0)  60·9 (96·2) 64·6 (82·2)
C. denticulata    –    – 0 (0)  0 (0)  0·1 (0·4)  0·1 (0·2)   0·3 (0·5)  0·1 (0·2)
Table 4.  Regression parameters relating species richness, total percentage cover and plant demography of Ruschia robusta, Cheiridopsis denticulata and Galenia africana to distance from the stockpost (n= 8 distances except where otherwise noted). Data from the four plots at each focal site were pooled for the regressions
Dependent variableRegression parameters
Slope*Constantr2P
  • *

    Log-transformed.

  • n= 6 distances.

  • n= 4 distances.

Number of species  3·90  5·800·27  0·189
Cover (%) 16·8−20·80·86  0·001
Total number of individuals
R. robusta*  0·53  0·360·69  0·010
C. denticulata*  1·15 −1·850·90< 0·001
G. africana−27·5 96·50·63  0·019
Number of adult shrubs
R. robusta*  0·53  0·350·70  0·009
C. denticulata*  0·91 −1·150·95< 0·001
G. africana−18·2 67·70·49  0·055
Number of seedlings
R. robusta*  0·47 −0·560·47  0·059
C. denticulata*  0·53 −0·850·71  0·008
G. africana−16·7 54·20·79  0·003
Dead adult shrubs as percentage of total number of adult shrubs
R. robusta* −0·32  2·220·64  0·016
C. denticulata* −0·76  3·630·73  0·030
G. africana*  0·42  0·340·68  0·176
Number of fruits per adult shrub
R. robusta 52·68−96·850·73  0·007
C. denticulata −0·37  0·150·38  0·190

Sites close to the stockpost were found to have a high abundance of G. africana, some individuals of R. robusta, and few or no individuals of C. denticulata, while this pattern was reversed at the further sites (Table 3). Changes in the abundance of adult plants over the gradient were significant or marginally significant for all three species (Table 4). Similar trends in the abundance of established seedlings of the three species were observed (Tables 3 and 4). Both adult plant and seedling abundance for R. robusta and C. denticulata were best described by an exponential increase with distance from the stockpost, while G. africana abundance was best described by a linear decrease away from the stockpost.

The number of dead adult shrubs relative to all adult shrubs (dead and alive) decreased exponentially over the gradient for both R. robusta and C. denticulata (Tables 3 and 4). Galenia africana mortality increased with distance but this trend was not significant.

Fruit production for the two leaf-succulent species was also found to vary under different grazing intensities (Table 3). All individuals of R. robusta close to the stockpost produced few or no fruits. At sites 800 m or further from the stockpost, fruit production among adult R. robusta shrubs was variable but overall much higher, with certain individuals producing hundreds of fruits. Cheiridopsis denticulata adults, in contrast, produced few fruits even at the furthest distances along the gradient. The number of mature fruits per adult plant increased significantly over the grazing gradient for R. robusta, while there was no significant relationship between fruit production and distance from the stockpost for C. denticulata (Table 4) due to the extremely low fruit production in this species at all sites.

seedling germination and survival

Post-rain germination of leaf-succulent seedlings at the eight focal sites followed the fruit production trends for these species. Almost no seedlings of R. robusta were found at the 50- and 100-m sites, while seedlings were present in increasing abundance with distance at the other sites (Fig. 2a). In concordance with the very low fruiting success rate for C. denticulata, no seedlings of this species were encountered in the post-rain germination survey. The abundance of G. africana seedlings decreased exponentially with distance from the stockpost (Fig. 2b).

Figure 2.

Plots of post-rain seedling emergence at each of the eight focal sites against distance from the stockpost for (a) Ruschia robusta and (b) Galenia africana. The total number of seedlings of each species at each focal site is used for the regression.

Although seedling abundance changed across the grazing gradient, survival of transplanted seedlings did not vary greatly (Fig. 3). For both R. robusta and C. denticulata, seedling survival did not differ significantly among focal sites along the gradient (R. robusta: χ2 = 3·36, n= 180, d.f. = 7, P= 0·850; C. denticulata: χ2 = 2·80, n= 180, d.f. = 7, P= 0·903). Overall seedling survival was no better for R. robusta when grown in a livestock exclosure than when exposed to possible herbivory (χ2 = 0·30, n= 200, d.f. = 1, P= 0·587) but C. denticulata did have higher survival in the exclosure than at the focal sites (χ2 = 10·59, n= 200, d.f. = 1, P= 0·001).

Figure 3.

Seedling survival by focal site (eight distances from stockpost, given in metres, and grazing exclosure) and microsite for (a) Ruschia robusta and (b) Cheiridopsis denticulata, n= 200 seedlings per species.

Because there was no effect of distance on survival, survival data were pooled across all focal sites to examine the effect of microsite on seedling survival. For R. robusta, survival was not significantly different between the two microsite treatments (Fig. 3a; χ2 = 0·08, n= 200, d.f. = 1, P= 0·776). Mortality in this species was mostly due to drought stress (Fig. 4a), both in the open (χ2 = 9·33, n= 67, d.f. = 1, P= 0·002) and under the canopy of an adult (χ2 = 13·00, n = 52, d.f. = 1, P= 0·003). For C. denticulata, survival was not significantly different between seedlings planted under the canopy of adult R. robusta individuals and those growing in the open (Fig. 3b; χ2 = 3·60, n = 200, d.f. = 1, P= 0·058). Mortality for C. denticulata was more likely to be due to abiotic factors than due to livestock presence in both microsites (open: χ2 = 7·68, n= 47, d.f. = 1, P = 0·006; under adult R. robusta: χ2 = 6·53, n= 30, d.f. = 1, P= 0·011; Fig. 4b).

Figure 4.

Causes of seedling mortality by microsite for (a) Ruschia robusta and (b) Cheiridopsis denticulata.

Discussion

effects on community

Orbits of grazing intensity and vegetation degradation around artificial features of the landscape such as waterholes and human settlements have been well-documented in the semi-arid rangelands of Australia (Andrew & Lange 1986; Pickup & Chewings 1994; James, Landsberg & Morton 1999), North America (Fusco et al. 1995) and Africa (Tolsma, Ernst & Verwey 1987; Perkins & Thomas 1993; Van Rooyen et al. 1994; Moleele & Perkins 1998; Turner 1998b). These grazing orbits provide useful systems in which to study the responses of vegetation and other variables to a range of grazing intensities (Andrew 1988).

In this study, we used the grazing gradient created around a stockpost to examine the impacts of heavy grazing on the population biology of two leaf-succulent species and an unpalatable non-succulent shrub. The presence of a grazing gradient was first established by analysis of soil parameters and grazing intensity at increasing distances from the stockpost.

Soil analysis revealed no underlying gradient of soil properties present. Previous work has demonstrated that organic matter, nitrogen and calcium are the soil parameters correlated with the abundance of leaf-succulent shrubs in the Paulshoek environment (Carrick 2001). These parameters, however, did not vary significantly with distance from the stockpost, further evidence that soil alone cannot explain differences in species composition and population dynamics along the gradient.

Indicators of grazing intensity, conversely, showed strong patterns of change with distance from the stockpost. Grazing intensity was found to be much higher in the area closer to the stockpost than in the area further from the stockpost. Thus the differences in succulent shrub population dynamics observed across the eight focal sites in this study provide a basis for understanding the impacts of different grazing intensities on species composition, percentage cover and shrub mortality, reproduction and establishment.

Grazing orbits around fixed points have been described as exhibiting three distinct zones of response (Andrew 1988). The first zone, in the immediate vicinity of the central point, is a ‘sacrifice zone’ of high disturbance intensity and is typically characterized by bare soil or ephemeral cover of annual plants (James, Landsberg & Morton 1999). The second zone is one of intermediate and progressively decreasing influence of grazing (Andrew 1988), characterized by a shift in dominance from unpalatable to palatable perennial plants with distance from the centre (James, Landsberg & Morton 1999). The third zone lies beyond the range of grazing activity. In the Paulshoek communal area, this last zone is rarely present, as livestock are kept at high densities throughout the area (Todd & Hoffman 1999). Daily grazing routes for the stockpost used in this study showed that the animals’ grazing orbit is bound by the presence of neighbouring stockposts and by the boundary of the communal area (Combrinck 2003). Thus the impacts of grazing on vegetation reported here do not include the third zone of the orbit model, and even the second zone may be somewhat truncated.

The zone of sacrifice observed at this stockpost was less than 50 m in radius. The grazing gradient from 50 m to 2 km exhibited a significant decrease in abundance of G. africana and an increase in abundance of the palatable R. robusta and C. denticulata. Galenia africana and R. robusta were the most abundant species and contributed more than half of the canopy cover of all perennial shrubs at sites close and far from the stockpost, respectively. The switch in dominance from a palatable leaf-succulent to an unpalatable woody shrub under heavy grazing follows the established pattern for the Paulshoek area (Todd & Hoffman 1999) and for the Succulent Karoo in general (Milton & Dean 1995).

In addition to changes in palatable shrub cover, the second zone of a typical grazing orbit exhibits an increase in species richness with distance from the centre of the sphere (James, Landsberg & Morton 1999). Our results, however, show no change in perennial species richness over the gradient. This may be a consequence of the scale of the grazing gradient studied; in a review of the impacts of grazing around water points on the ecologically similar chenopod shrublands of Australia, James, Landsberg & Morton (1999) report no consistent change in species richness over grazing gradients of 2–3 km. Moreover, other studies have shown that the relationship between grazing intensity and plant species richness in arid systems can vary greatly in space (Landsberg et al. 2002) and time (Oba, Vetaas & Stenseth 2001).

effects on population biology

In addition to the zones of grazing impact on community composition, zones of grazing impact on population dynamics can be found within a grazing orbit. Hunt (2001) describes three zones for the palatable shrub Atriplex vesicaria in Australia's chenopod shrubland. In the first zone, close to the water point, adult shrubs are heavily grazed and eventually die due to the combined effects of grazing and drought. In the second zone, as well as in the first zone, recruitment of shrub seedlings is generally suppressed, whereas in the third zone grazing has little impact on population processes.

We found similar zones of grazing impact for the palatable shrubs R. robusta and C. denticulata, and inverted zones for the unpalatable shrub G. africana. Adult shrub abundance for R. robusta and C. denticulata increased exponentially with distance from the stockpost, closely following the exponential decrease in grazing intensity. The patterns of abundance indicate that C. denticulata is more sensitive to the effects of heavy grazing than R. robusta; R. robusta was present at all sites, whereas live adults of C. denticulata were not found at the two focal sites closest to the stockpost.

Patterns of relative adult mortality suggest that grazing causes high adult shrub mortality only where grazing intensity is very high. Similar studies of shrub population dynamics in Australian rangelands have also shown that adult shrub mortality due to grazing only limits population persistence in areas subject to very heavy grazing (Watson, Westoby & Holm 1997; Hunt 2001).

Recruitment within the two leaf-succulent species studied was reduced in a wider zone around the stockpost than the zone of high adult shrub mortality. Ruschia robusta suffered reduced fruit production and post-rain seedling emergence at all sites less than 800 m from the stockpost, with especially low fruiting success at sites less than 400 m from the stockpost, but had high fruit production at the less heavily grazed sites. Fruit production was more severely impacted by grazing in C. denticulata, which produced very few fruits even at the least heavily grazed sites. We observed adult plants of this species on the adjacent lightly grazed privately owned farm and in permanent grazing exclosures within the communal area to produce numerous fruits per shrub during the same reproductive season. The general failure of this species to set seed in the heavily grazed communal area was confirmed by the absence of any seedlings at any sites in the post-rain seedling emergence survey. Thus, for C. denticulata, the zone of recruitment suppression around the stockpost extends for at least 2 km.

The pronounced effects of grazing on the number of fruits and seedlings of R. robusta and C. denticulata indicate that recruitment of these species is seed-limited even where adult plants can be found in relative abundance. The greater sensitivity of C. denticulata to grazing in terms of reproductive output may be explained by the fact that this species produces few large, showy flowers, whereas R. robusta produces many small flowers. Furthermore, flowers of C. denticulata appear to be actively sought-out as forage by livestock. Thus trade-offs in flower number and size, as well as floral palatability and nutrition, may play important roles in determining the impact of grazing on reproductive success.

Despite these differences, however, both species are significantly impacted in terms of seed production where grazing pressure is high. Similar trends have been observed in the Succulent Karoo for palatable perennial shrubs (Milton & Dean 1990; Milton 1994a; Todd 2000) and some leaf-succulent shrubs (Stokes 1994), and in Australian shrublands (Andrew & Lange 1986).

Seed production has been identified as one of the processes most sensitive to herbivory among long-lived perennial plants (Fenner 1985; O’Connor & Pickett 1992). Continued low levels of reproductive output under a heavy grazing regime can have negative ramifications for population persistence (Hunt 2001). Given that stockposts in the communal areas of Namaqualand are typically located 1–3 km apart, a significant reduction in seed set of leaf-succulent species in spheres 800 m to 2 km from the stockposts could pose a serious threat to the persistence of these shrubs in the landscape. This is especially the case for species such as C. denticulata that produce very few fruits even at the least heavily grazed sites. As succulent shrubs of the Mesembryanthemaceae lack long-distance dispersal mechanisms (Esler 1999; Parolin 2001) and do not persist in soil seed banks (Esler 1999), recolonization of an area where local extinction has occurred is highly unlikely.

Once germinated, however, seedlings of succulent shrubs may not be greatly impacted by heavy grazing. Survival of R. robusta and C. denticulata seedlings in this study did not vary with grazing intensity. Seedling mortality was much more often due to abiotic causes than to grazing or trampling. Studies of woody perennial seedling mortality under varying grazing intensities, both in the Succulent Karoo (Milton 1994a, 1995) and in Australia's chenopod shrubland (Lange, Coleman & Cowley 1992), have also found seedling mortality to be independent of grazing, but this result has not been documented previously for succulent shrubs of the Karoo. The primary cause of seedling mortality in semi-arid shrublands is generally thought to be drought-stress (Milton 1995; Watson, Westoby & Holm 1997) and competition from neighbouring adult plants (Milton 1994a,b, 1995). Succulent shrubs, although palatable, are usually not the most preferred livestock forage, and the cost of searching out inconspicuous seedlings of these species probably outweighs the benefit in terms of palatability (Lange, Coleman & Cowley 1992).

Seedlings of many Karoo shrubs have been shown to exhibit preferences in terms of microsites for germination and recruitment (Yeaton & Esler 1990; Esler 1999). In semi-arid shrublands, seedlings growing under the canopy of adult plants are thought to benefit from thermal stress alleviation (McAuliffe 1984; Franco & Nobel 1989) or protection from herbivory (Turner, Alcorn & Olin 1969; Todd 2000), benefits that may outweigh the effects of competition from the proximate adult plant under certain conditions. Thus heavy grazing, by changing species composition and reducing total plant cover, may negatively affect seedling establishment by reducing the availability of favourable microsites.

In this study, however, seedlings were not found to be microsite-limited. Seedlings of R. robusta and C. denticulata survived equally well in the open and under the canopy of R. robusta, demonstrating that adult plants do not provide essential protection from herbivory. Moreover, seedling mortality for both species was much more often caused by abiotic factors in both microsites.

While R. robusta and C. denticulata abundance, mortality and reproduction exhibited strong patterns of grazing influence around the stockpost, the unpalatable shrub G. africana was less responsive to the changes in grazing intensity. The abundance of adult and seedling G. africana shrubs decreased linearly with distance from the stockpost, in contrast to the exponential increase in abundance among the succulent shrubs. Relative mortality of G. africana did not change significantly over the gradient and was low at all sites. In the post-rain germination survey, seedlings of this species were present in high abundance at sites close to the stockpost and, although abundance dropped off at sites further than 400 m from the stockpost, seedlings were present even at the least heavily grazed sites. This contrasts strongly with the trends for R. robusta and C. denticulata, for which no seedlings were found at the most heavily grazed sites.

In general, G. africana populations at the least heavily grazed sites did not show strong signs of decline, whereas population processes within both species of leaf-succulent species were strongly impacted at the more heavily grazed sites. The difference in response between G. africana and the palatable leaf-succulents provides insight into the process by which vegetation degradation around a stockpost (or comparable central point of grazing intensity) occurs.

With reduced recruitment, populations of palatable shrubs are unable to replace themselves and eventually become locally extinct when adult shrubs die off in areas of highest grazing intensity. Without a local source of seed, recolonization by these species is highly unlikely. In contrast, the unpalatable G. africana survives and recruits successfully even in less heavily grazed areas, and, because it appears to disperse seeds more widely than the Mesembryanthemaceae, it can quickly colonize an area where palatable shrub populations are in decline. With continued heavy grazing, the radius of unpalatable forage around a stockpost increases, shifting the focus of grazing further and further away from the stockpost, thus creating an expanding centre of degradation in the landscape that is limited only by the distance over which animals travel (Jeltsch et al. 1997).

management implications

The stockpost system can have significant impacts on the vegetation of the surrounding areas (H. Hendricks, unpublished data). The presence of a stockpost, however, will have the greatest effect on shrub populations and communities if the stockpost remains stationary. In this study we found that marked zonation has developed around a stockpost that has been stationary over a period of 15 years. Although caution should be exercised in generalizing results based on only one stockpost, our informal observations at other semi-permanent stockposts in the Namaqualand communal areas suggest that they exhibit similar zonation. Continued and concentrated grazing impacts on plant population dynamics could ultimately lead to the local extinction of grazing-sensitive succulent species in an increasing area around a stockpost. Stockposts that relocate regularly, on the other hand, are not likely to suppress recruitment and cause adult shrub mortality in distinct zones. Frequent movement (every 1–3 years) of stockposts would allow different patches of the landscape to set seed in different years. We therefore recommend that herders be encouraged to relocate their stockposts at frequent intervals.

The results of this study support the theory that reduced seed set is the main mechanism responsible for the decrease in abundance of Karoo shrubs in areas subject to heavy livestock grazing (Milton 1995; Todd 2000; Carrick 2001), although further research is needed to confirm this conclusion for the guild of leaf-succulents in general. In light of the apparent sensitivity of fruit production to grazing, it is advisable that grazing pressure be reduced or removed from particular areas periodically. Succulent shrub populations in areas that are grazed heavily and continuously may never have adequate opportunity to replace themselves. The alleviation of grazing pressure, especially preceding and during the months of peak flower and fruit maturation and during high rainfall years when fruit production is most likely to be successful, would allow seed set to occur and thus increase the chances of successful recruitment. Further research is needed to determine whether shrubs that have been heavily grazed can set seed successfully in the first growing season after the removal of grazing pressure, or whether rest periods of several years are necessary. In general, however, it is of great importance that grazing pressure be relieved periodically, rather than concentrated around fixed points, in order to ensure the persistence of Namaqualand's unique succulent shrub species.

Acknowledgements

We are indebted to the Paulshoek community for permission to work on their land and support for our research. We especially thank Sors Cloete, the herder at the stockpost used in this study. Sue Milton provided helpful input at many stages of the study. We are grateful to Natasha Gabriels for providing seedlings and David Ward for his statistical advice. This manuscript benefited from the comments of Peter Carrick, Sue Milton, Nicky Allsopp and two anonymous referees. The Mazda Wildlife Fund provided a courtesy vehicle for field use. This research was funded by a Fulbright Student Scholarship to C. Riginos and by the European Commission under INCO-DC: International Cooperation with Developing Countries (1994–98), Contract No. ERBIC18CT970162. The European Commission cannot accept responsibility for any information provided or views expressed.

Ancillary