Plant Ecology

, Volume 213, Issue 3, pp 483–491

Inter- and intra-plant variations in nitrogen, tannins and shoot growth of Sclerocarya birrea browsed by elephants


    • Department of AgricultureUniversity of Zululand
  • Robert W. Taylor
    • School of Biological and Conservation SciencesUniversity of KwaZulu-Natal
  • David Ward
    • School of Biological and Conservation SciencesUniversity of KwaZulu-Natal

DOI: 10.1007/s11258-011-9996-x

Cite this article as:
Scogings, P.F., Taylor, R.W. & Ward, D. Plant Ecol (2012) 213: 483. doi:10.1007/s11258-011-9996-x


Impacts of elephants (Loxodonta africana africana) on woody vegetation has attracted substantial attention for decades, but plant-level responses remain a gap in the understanding of savanna ecology. Marula (Sclerocarya birrea caffra) forms an important part of elephant diets. We investigated the relationships between browsing intensity and shoot/leaf size, nitrogen (N) and condensed tannin (CT) concentrations in upper and lower canopies of male and female marula individuals in Hluhluwe-iMfolozi Park, South Africa. Browsing intensity (54%) did not differ between sexes, suggesting no preference by elephants for either sex. Females had higher [CT] than males and tannin decreased with increasing browsing intensity in both sexes. In lightly or moderately browsed trees, [CT] was controlled by unmeasured factors such that within-tree impacts of browsing were more variable in lightly/moderately browsed than heavily browsed trees. There was little change in [N] up to ~60% browsing intensity, but [N] increased dramatically at higher intensity. Shoots and leaves on broken branches in the lower canopy were larger (2.5 and 1.2 times, respectively) than those on unbroken branches in either upper or lower canopies. Chemical responses were systemic and potentially influence browsing among trees, while growth responses were strongly localised and potentially influence browsing within trees. Although marula trees are able to compensate vigorously for browsing at the scale of individual organs, trees may become progressively carbon-deficient and have their lives shortened if total plant growth is negatively affected by chronic browsing, e.g. near permanent water.


Browsing lawnsCompensatory growthHerbivoryInduced responsesPlant defenceSpecific leaf area


Elephants (Loxodonta africana africana) are important non-ruminants that consume woody plants in African savannas. Elephant impacts have attracted attention for many decades (Kerley et al. 2008), but the focus of attention has been almost exclusively at scales of populations and communities of plants. Compared with the range of studies involving ruminant impacts, the paucity of studies on plant-level responses to elephants remains a critical gap in the body of fundamental knowledge of savanna ecology. Long-term severe browsing by elephants is reported to stimulate the production of nutritious material (Makhabu et al. 2006; Kerley et al. 2008). The phenomenon has been compared with grazing lawns in the Serengeti and the term “browsing lawn” has been used to describe it (McNaughton 1984; Fornara and du Toit 2007). The switch to a vegetative state characterised by N-rich/C-poor tissues puts heavily browsed plants in a positive feedback loop because their risk of being browsed again is increased (Fornara and du Toit 2007; Skarpe and Hester 2008).

One explanation given for positive feedback in savanna trees is preferential allocation of C to growth of new shoots rather than C-based secondary metabolites (CBSMs) such as condensed tannins (CTs), which are often assumed to function as chemical, anti-herbivore defences (Fornara and du Toit 2007; Skarpe and Hester 2008; Hrabar et al. 2009). Positive feedback can be achieved when the root:shoot ratio is altered such that shoot growth increases (Herms and Mattson 1992; Renton et al. 2007). A simultaneous increase in nutrients and photosynthesis may occur to meet the demands of increased growth (Herms and Mattson 1992; Glynn et al. 2007; Ågren 2008). However, very few studies have quantified the relationship between browsing by elephants (or simulation thereof) and either growth rate or nutritional value of individual trees, and the results are inconsistent. Increasing the severity of simulated elephant browsing of Colophospermum mopane trees once in the growth season led to increased investment in CBSMs, which remained elevated for 2 years (Wessels et al. 2007). In contrast, severe browsing (including branch breakage) by elephants had no effect on either N or CBSMs in C. mopane and led to increased shoot growth compared with no browsing (Hrabar et al. 2009). Increased shoot growth was observed in another four woody species in relation to severe browsing, compared with either no or low browsing intensity (Makhabu et al. 2006).

Sclerocarya birrea (A. Rich) Hochst subsp. caffra (Sond.) Kokwara is a prominent southern African tree species that is favoured by elephants, although the leaves and bark are endowed with high tannin concentrations (Furstenburg and van Hoven 1994; van Wyk et al. 1997; Gadd 2002; Helm et al. 2009). S. birrea has compound leaves and is a deciduous, dioecious species, the fruits of which attract many animals (Pooley 1994; Coates Palgrave 2005). Elephants show varying degrees of preference for female trees, depending on elephant density (Gadd 2002; Hemborg and Bond 2007). How S. birrea individuals respond to browsing by elephants is important for the sustainable management of both of these iconic species and the savanna habitats in which they occur (Helm and Witkowski 2008). Elephants are known to negatively affect S. birrea populations by increasing mortality of mature trees (Jacobs and Biggs 2002a, b; Boundja and Midgley 2010), but there are no published accounts of browsing-induced responses of S. birrea individuals.

The general aim of this paper is to investigate the relationships between the intensity of browsing attributable to elephants, and shoot length, leaf size and leaf chemistry in different canopy positions on male and female S. birrea individuals. We assume that the measured variables are primarily consequences of browsing, rather than causes of browsing (Scogings et al. 2004; Rooke and Bergström 2007). On average, individuals of S. birrea were expected to display increased shoot/leaf size and increased [N]/reduced [CT] in relation to increased browsing intensity because severe browsing tends to reduce apical control and stimulates the growth of new shoots, which increases the demand for C (du Toit et al. 1990; Herms and Mattson 1992; Rooke and Bergström 2007; Skarpe and Hester 2008; but see Ward 2010). However, we were also interested in determining differences between leaves/shoots (i) on male or female trees, (ii) above or below the maximum reach of elephants (browse line), and (iii) on severely browsed (broken) or unbrowsed branches.

It was expected that shoots on female trees would be shorter and their leaves would be smaller and have lower [N]/higher [CT] than male trees because female trees are slower growing than males (Åhman 1997; Gayler et al. 2007). Differences between male and female shoot/leaf traits were expected to be influenced by browsing since slow-growing plants are expected to be more prone than fast-growing plants to C starvation (Gayler et al. 2007). Hence severe browsing of female individuals would be associated with a stronger expression of positive feedback (longer shoots, bigger leaves with higher [N]/lower [CT]) compared with male trees. Because plants are modular organisms and modules are semi-autonomous (Herrera 2009), we expected that responses to browsing would be (i) stronger below the browse line than above and (ii) strongest at sites of severe browsing such as broken branches. Shoots/leaves below the browse line would be expected to express positive feedback more strongly than shoots/leaves above the browse line because of the relative C shortage induced by shading below the browse line. We also expected that positive feedback would be most strongly expressed in shoots/leaves on broken compared with unbroken branches below the browse line because of the activation of lateral shoots near the break (du Toit et al. 1990; Herms and Mattson 1992; Rooke and Bergström 2007). If so, then there would be concomitant negative relationships between shoot/leaf size and [CT].

Methods and materials

Data collection

Seventy S. birrea trees were randomly sampled at a site in Huluhluwe-iMfolozi Park (28°06′S; 32º02′E), South Africa, during March 2010. The vegetation is Zululand Lowveld (Mucina and Rutherford 2006). Mean annual precipitation is 985 mm (Boundja and Midgley 2010). The park has a high density of both marula trees and elephants (Hemborg and Bond 2007; Boundja and Midgley 2010). Elephant density is ~4 km−2. Elephants prefer browsing at a height of 2–3 m, but they are known to browse to a maximum height of 5 m (Jachmann and Bell 1985). The mean height of the sampled trees was 7.5 m (SEM = 0.14). The trees were in a mature stand covering 10 ha (8.4 trees ha−1) on a hilltop. The site was chosen because it had sufficient trees of different browsing impact to allow us to collect the data we needed. The site was also readily accessible by road. Sampling was scheduled according to observations in similar environments elsewhere, which indicated that effects of browsing on chemistry of mature leaves is readily detectible in the latter half of the wet season (Scogings et al. 2011). Sampling only mature leaves at one time ensured that leaf phenology would not confound our results. The trees were selected by walking for random distances in random directions through the area and selecting the tree closest to each random point if it met the following criteria. Trees were sampled on condition that they were (1) taller than the maximum reach of elephants, (2) not obviously stressed by disease, insects, disturbance or neighbours, and (3) not obviously growing in nutrient-enriched patches. The fraction (%) of the productive canopy volume (including branches bearing leaves or shoots) below the browse line that was deemed to have been removed by elephants was visually estimated for each tree (Fig. 1). The method was based on one that we developed and tested across six other tree species in a similar environment (Scogings et al. 2011). The method was subjective, but all estimates were made by one observer to ensure repeatability. Most damage was presumed to be less than 3 years old, based on observations of the colour and texture of browsed points and age of regrowth (Page 2005). Toppled and pollarded trees were not sampled because direct comparisons between such growth forms and other trees are virtually meaningless because of the extreme differences in shoot demography and root:shoot ratio (Scogings and Macanda 2005). While evidence of browsing of shoot tips by giraffe was recorded, it comprised a negligible fraction of canopy removal and was ignored. Sex of each tree was determined by the presence or absence of fruit either on the tree or on the ground below.
Fig. 1

Examples of Sclerocary birrea trees sampled in the spectrum of browsing intensity at Hluhluwe-iMfolozi Park. The trees shown represent, from left to right: ~5, ~30, ~60 and ~90% of canopy removed below the browse line (broken line at 5 m). Note: The actual method used for assigning browsing intensity in the field involved an assessment on a continuous scale and was not based on these images, which only illustrate the range of browsing intensities included in the study

Seven current season shoots were randomly located on unbroken branches in each of two positions on each tree: (i) above the browse line and (ii) below the browse line. On each shoot, length of current season’s growth (mm) was measured and one leaf was collected for measuring the number of leaflets, total leaflet area (cm2) and total leaflet dry mass (g). Specific leaf area (SLA) was calculated (cmg−1). Where there was a branch broken below the browse line by elephants, a third sample of seven shoots and seven leaves was collected from regrowth on the branch and processed in the same way as listed above. Broken branches were sampled if they were big enough to ensure that no other animal could have broken them. Leaves not used for measuring leaf size parameters were pooled to form a composite sample per canopy position, which was oven-dried at 40°C and milled to 0.5 mm. The dried samples were analysed for nitrogen (N) concentration in a LECO FP2000 nitrogen analyser using the Dumas combustion method (AOAC 2000). CTs were quantified according to the acid–butanol assay (Hagerman and Butler 1989; Hagerman 2002) and expressed as quebracho equivalents (mgQE ml−1). We note that the use of quebracho as a standard may substantially over-estimate CT concentrations, compared with other standards (Li et al. 2010; Hattas and Julkunen-Tiitto, unpublished data), but the extent of disparity has not been established for S. birrea.

Data analysis

Because plants are modular organisms (Herrera 2009), it was assumed that samples of shoots and leaves from the different canopy positions could be regarded as independent data for the analysis. A completely randomised design was assumed for the analyses. Difference in browsing intensity between male and female trees was tested with a t test assuming unequal variances because of different numbers of male and female trees measured. Shoot and leaf size measurements were averaged per canopy position. For the analyses, [N] and [CT] were log10 transformed and shoot length and SLA were double log10 transformed to meet the assumptions of parametric statistical tests. The effects of position, sex and position*sex on each response variable were tested with browsing intensity (fraction of canopy removed below the browse line) as covariate in a two-factor ANOVA model with Type III sums of squares for unbalanced designs. Two subsets of the data were used: (i) samples from above or below the browse line excluding samples from broken branches, and (ii) samples from broken or unbroken branches below the browse line. When there was no effect of position or position*sex, then data were averaged across positions per tree and the effect of sex was tested with browsing intensity as a covariate after confirming that there was no significant interaction between sex and browsing intensity. Simple linear regression analysis was used to model the relationship between browsing intensity and [CT]. The relationship between browsing intensity and [N] was modelled as a piecewise linear regression analysis. All the above tests were done using SYSTAT 12 (SYSTAT Software 2007). Least absolute deviation (LAD) regression analysis (Cade and Richards 2005) was used to estimate median [CT] in relation to either shoot length or SLA because variation in [CT] was heterogeneous across the independent variables, which resulted in asymmetric distributions of data. Upper quantiles (90th, 95th and 99th percentiles) of [CT] were also estimated with LAD regression to explore the maximum rate of change (Cade and Richards 2005). Significance of all statistical tests was declared when P < 0.05. Only significant results were considered for presentation.


There was no difference in browsing intensity (overall mean = 54%) between male and female trees (t = 0.16; df = 61.6; P = 0.875), suggesting no preference by elephants for females. Females had higher [CT] than males (F1,66 = 7.73; P = 0.007) and [CT] decreased with increasing browsing intensity in both sexes (F1,66 = 9.59; P = 0.003), although the relationship was weak (Fig. 2). Mean [CT] was 13.4 mgQE ml−1 (SEM = 1.66; n = 27) in females and 8.6 mgQE ml−1 (SEM = 1.05; n = 42) in males. Nitrogen concentration increased with increasing browsing intensity (Fig. 2). There was a slight decline in [N] up to ~60% browsing intensity, but above ~60% browsing intensity, [N] increased dramatically.
Fig. 2

Concentrations of condensed tannin (mg quebracho tannin ml−1) and nitrogen (%) in leaves of Sclerocarya birrea trees in relation to intensity of browsing by elephants expressed as the fraction of canopy removed below the browse line (%) at Hluhluwe-iMfolozi Park (March 2010). Concentrations are presented on log scales. Linear regression models for condensed tannin: y = 1.2881 − 0.0049x (r2 = 0.15; F1,25 = 4.485; P = 0.044) in female trees (upper regression line) and y = 1.025 − 0.004x (r2 = 0.11; F1,40 = 5.204; P = 0.028) in male trees (lower regression line). Piecewise linear regression model for nitrogen: yx ≤ 56.905 = [1.915 × (56.905 − x) + 1.759 × (x − xmin)]/(56.905 − xmin); yx > 56.905 = [1.759 × (xmax − x) + 2.279 × (x − 56.905)]/(xmax − 56.905) (r2 = 0.34; F3,66 = 11.308; P < 0.001)

Current season’s shoots on broken branches were longer than those on unbroken branches below the browse line (F1,97 = 26.91; P < 0.001) (Fig. 3). Mean shoot length was 67.7 mm (SEM = 8.34; n = 52) on broken branches and 25.0 mm (SEM = 1.74; n = 50) on unbroken branches. Leaflets were also more abundant per leaf on broken branches than on unbroken branches (F1,97 = 4.41; P = 0.038) (Fig. 3). Mean number of leaflets per leaf was 10.2 (SEM = 0.23; n = 52) on broken branches and 9.6 (SEM = 0.19; n = 50) on unbroken branches. Specific leaf area was higher (more area per unit dry mass) below the browse line than above (F1,114 = 7.44; P = 0.007) and was also higher on broken branches than unbroken branches below the browse line (F1,114 = 4.21; P = 0.043) (Fig. 3). Specific leaf area was 81.0 cm2 g−1 (SEM = 1.94; n = 70) above the browse line, 87.4 cm2 g−1 (SEM = 1.83; n = 49) below the browse line and 99.9 cm2 g−1 (SEM = 4.02; n = 52) on broken branches. Scatter plots of relationships between [CT] and either shoot length or SLA revealed triangular distributions (Fig. 4). Only the median regression models were significant (P < 0.05), although weak; hence maximum rates of change could not be successfully modelled.
Fig. 3

Box plots illustrating the effects of canopy position and proximity to branch breakage on length (mm) of current season’s shoots, leaf size (leaflets leaf−1) and specific leaf area (cm2 g−1) on Sclerocarya birrea trees browsed by elephants at Hluhluwe-iMfolozi Park (March 2010). The lower and upper sides of boxes indicate 25th and 75th percentiles. Lines within boxes mark the medians. Whiskers indicate the 10th and 90th percentiles. Dots indicate the 95th and 5th percentiles
Fig. 4

Scatter plots showing relationships between condensed tannin concentration (mg quebracho equivalents ml−1) and either current season’s shoot length (mm) or specific leaf area (cm2 g−1) on Sclerocarya birrea trees browsed by elephants at Hluhluwe-iMfolozi Park (March 2010). Significant quantile regression models are shown for the 50th percentiles (solid lines). Shoot length: y = 11.811 − 0.069x (Coefficient of determination, R1 = 0.008; Rank Score test statistic = 0.078; n = 70; P = 0.021). Specific leaf area: y = 20.815 − 0.137x (R1 = 0.087; Rank Score test statistic = 0.172; n = 70; P = 0.0006). The broken line represents a marginally significant quantile regression model for the 90th percentile: y = 45.833 − 0.254x (R1 = 0.058; Rank Score test statistic = 0.046; n = 70; P = 0.074)


Broadly, we hypothesised that individuals of S. birrea would display (i) increased shoot/leaf size, or (ii) increased [N]/reduced [CT] in relation to increased browsing intensity; we hypothesised further that these relationship would be most evident (iii) on female compared with male trees, (iv) among shoots below compared with above the browse line, and (v) among shoots on broken compared with unbroken branches. Our hypotheses were only partially supported. Chemical variables were related to browsing intensity and sex, but the relationships were weak, while shoot/leaf size was affected by canopy position. The strongest chemical response to browsing was among the most severely browsed trees, which had highest [N] and lowest [CT], while female trees had consistently higher [CT] than males. That female trees were not browsed at different intensity than males suggests that elephants are not affected by the difference in [CT], which may be attributed to their large size and hind-gut fermentation that allows them to utilise feeds of relatively low quality (Codron et al. 2011). The strongest growth response was on broken branches, which had longer shoots, more leaflets per leaf and higher SLA than unbroken branches, regardless of the intensity of browsing or sex of the tree. Hence, the chemical response, although weak, was systemic and potentially influences browsing among individual trees, while the growth response was strongly localised and potentially influences browsing within individual trees.

Both the absence of a relationship between shoot/leaf size and browsing intensity among trees, and increased shoot/leaf size on broken branches reflect increased shoot/leaf growth rates (Renton et al. 2007; Mopipi et al. 2009). Thus, whichever way shoot growth is viewed, either among trees or within trees, the rate of shoot growth was elevated in relation to browsing intensity and canopy position. Therefore, S. birrea trees are able to vigorously compensate (or even over-compensate) for browsing by elephants at the scale of individual organs, which is consistent with observations of other tree species browsed by elephants (Makhabu et al. 2006; Hrabar et al. 2009). We recognise the possibility that a non-significant result consistent with the claim of compensation could be due to a Type II statistical error (Ward 2010), but we deem this unlikely, considering the sample size of 70. In support of elevated shoot/leaf growth rates, increased SLA suggests photosynthetic rates of leaves on broken branches were also elevated (Reich et al. 1998), but the contribution of this to the C budget of the total canopy is unclear and requires further research. The role of increased SLA is presumably contingent upon the proportion of canopy comprising broken branches. Clearly, however, total plant growth (in terms of metrics such as plant height or canopy diameter) is negatively affected by chronic browsing by elephants, e.g. near permanent water (Fig. 1; Makhabu et al. 2006). A consequence of simultaneous within-plant growth compensation and whole-plant under-compensation is that trees become progressively C deficient, which shortens their expected life spans (Teague 1988; Hester et al. 2006). Therefore, while shoot growth may increase under frequent browsing and be maintained for several years, long-term browsing may be neither beneficial to a plant nor sustainable (Stuart-Hill and Tainton 1988; Teague 1989; Rooke and Bergström 2007). Woody plants are seldom monitored through to mortality, so the extent to which long-term browsing reduces life expectancy is virtually unknown. However, heavy defoliation of Acacia karroo by goats every 2 weeks during one growth season did increase mortality rate in the following year (Teague 1989).

Responses to browsing intensity similar to those we observed have been described previously (Hester et al. 2006), although mechanistic explanations remain unclear. Assuming that increasing the intensity of browsing reduces apical dominance and stimulates the growth of new, N-rich shoots, which increases the demand for C, we expected to observe a concomitant decrease in concentrations of CTs in leaves of S. birrea (limited supply of C relative to demand) (du Toit et al. 1990; Herms and Mattson 1992; Skarpe and Hester 2008). The positive relationship we observed between browsing intensity and [N] supports our hypothesis, as does the negative relationship between browsing intensity and [CT], which corroborates observations of other deciduous, spineless species that are well endowed with CTs, e.g. Combretum apiculatum (Rooke and Bergström 2007; Scogings et al. 2011). However, our results contradict observations of C. mopane, which is another deciduous, spineless, CT-rich species heavily used by elephants (Wessels et al. 2007; Hrabar et al. 2009). The contrasting responses of species such as S. birrea and C. apiculatum, compared with C. mopane, may be related to differences in nutrient acquisition strategies. For example, C. mopane is found more often than either S. birrea or C. apiculatum in arid environments and therefore may be expected to have evolved a stronger defense response than S. birrea or C. apiculatum according to the growth–differentiation balance hypothesis (Herms and Mattson 1992). In addition, C. mopane is a legume. N2-fixation may facilitate compensatory photosynthesis leading to excess C being available for secondary metabolism, but the N and C dynamics of woody plants browsed at different intensities in savannas will remain unclear without further research. Further explanations are suggested by Ward and Young (2002). Essentially, because plants have a fixed position in space (except as seeds), if there are large differences between plants in their abilities to take up nutrients (acquisition) and relatively small differences in their defence strategies (allocation), then a positive correlation (i.e. not a trade-off) may exist between one type of defence and another regardless of the level of browsing. Alternatively, contrasting results may simply be caused by fundamental differences among research methods, such as manual simulation of browsing versus long-term real browsing.

The triangular distributions of data illustrate the relationships between [CT] and shoot length or SLA. Our results suggest that [CT] is limited by the production of biomass in heavily browsed trees (>65%), which have larger SLA, and longer, more N-rich shoots than moderately or lightly browsed trees. We have postulated that this is a result of browsing causing N to become more available as root:shoot increases and C becomes limiting as growth increases, reflecting the classical pattern in which growth compromises CBSM production (Herms and Mattson 1992). However, our results also indicate that [CT] in lightly or moderately browsed trees, which have reduced SLA and shoot length, is not limited only by growth, but may be controlled by other extrinsic or intrinsic factors (e.g. sex, light, soil moisture or nutrients) leading to [CT] being substantially reduced below the limits imposed by growth. For example, we postulate that it is feasible to expect most shoots on heavily browsed trees to be situated within resource-rich environments (a reduced canopy means more light, water and nutrients are available per shoot), but shoots on lightly or moderately browsed trees are in environments of variable resource availability (especially with regards to light). Hence, within-tree impacts of browsing would be more noticeable in lightly or moderately browsed trees than in heavily browsed ones.

Plants experience substantial temporal variations in resource availability, which may alter the relationship between browsing intensity and leaf chemistry (Gayler et al. 2007; Glynn et al. 2007). For example, baseline concentrations of CBSMs are typically low and inducible when resource availability is high (Bryant et al. 1991; Scogings and Mopipi 2008). Therefore, pending further research, we postulate that the slope of the relationship between browsing intensity and [CT] for S. birrea and other deciduous, spineless species with high [CT] is different early in the wet season, compared with later in the wet season. In particular, the slope may be positive early in the wet season because water and nutrients are most available then (Owen-Smith 2002; Scholes et al. 2003). This would imply that very high browsing intensity is needed early in the wet season to induce C shortage relative to N and thus increase potential palatability of S. birrea. However, this is unlikely to happen because elephants prefer to eat grass in the wet season (Codron et al. 2011, but see Shrader et al. 2011). In contrast, browsing intensity may not need to be very high later in the wet season to induce increased potential palatability of S. birrea. Even if elephants prefer grass in the wet season, they may still browse at an intensity that is sufficient to induce C shortage later in the wet season. Hence, if conservation of both elephants and S. birrea is an important objective in a protected area such as Hluhluwe-iMfolozi Park, then further research is needed in such areas to validate an approach that would promote managing elephant density such that reproductive S. birrea trees are not severely browsed, e.g. not more than 65% of a mature tree’s canopy is removed below 5 m.

Although our study was on one tree species, there remains a paucity of studies on comparable species in African savannas from which to draw broad generalisations. Clearly, further research is required for advancing the understanding of tree responses to browsing in African savannas. Studies are needed on browsed species representing various life histories in different soil and rainfall conditions, including both field-based studies and controlled manipulations in nursery/greenhouse experiments (see Scogings and Mopipi 2008 for further details). Additional research is also needed to elucidate long-term influences of annual and seasonal variations in climatic and browsing conditions on woody species in African savannas.


The National Research Foundation, University of Zululand and University of KwaZulu-Natal provided financial support. Bruce Page contributed to designing the methods. Dave Druce provided logistical support. Douglas Makin, Megan Welsford, Tiffany Pillay, Vanessa Stuart and Desale Okubamichael helped in the field and lab. The comments from three anonymous reviewers improved the manuscript.

Copyright information

© Springer Science+Business Media B.V. 2011