Long-term monitoring reveals differing impacts of elephants on elements of a canopy shrub community
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Please do not remove this page Long-term monitoring reveals differing impacts of elephants on elements of a canopy shrub community Landman, Marietjie; Schoeman, David S; Hall-Martin, Anthony J; et.al. https://research.usc.edu.au/discovery/delivery/61USC_INST:ResearchRepository/12126086570002621?l#13127290600002621 Landman, M., Schoeman, D. S., Hall-Martin, A. J., & Kerley, G. I. H. (2014). Long-term monitoring reveals differing impacts of elephants on elements of a canopy shrub community. Ecological Applications, 24(8), 2002–2012. https://doi.org/10.1890/14-0080.1 Link to Published Version: https://dx.doi.org/10.1890/14-0080.1 Document Type: Published Version USC Research Bank: https://research.usc.edu.au research-repository@usc.edu.au Copyright © 2014 Ecological Society of America. Reproduced here in accordance with the publisher's copyright policy. Downloaded On 2021/02/22 08:18:26 +1000 Please do not remove this page
Ecological Applications, 24(8), 2014, pp. 2002–2012
Ó 2014 by the Ecological Society of America
Long-term monitoring reveals differing impacts of elephants
on elements of a canopy shrub community
MARIETJIE LANDMAN,1,4 DAVID S. SCHOEMAN,1,2 ANTHONY J. HALL-MARTIN,3,5 AND GRAHAM I. H. KERLEY1
1
Centre for African Conservation Ecology, Department of Zoology, Nelson Mandela Metropolitan University, Port Elizabeth,
South Africa
2
Faculty of Science, Health, Education and Engineering, University of the Sunshine Coast, Maroochydore, DC, Queensland, Australia
3
Hall-Martin Consulting CC, Somerset West, South Africa and Scientific Services, South African National Park, South Africa
Abstract. The conservation management of southern Africa’s elephants focuses on
identifying and mitigating the extent and intensity of impacts on biological diversity. However,
variation in the intensity of elephant effects between elements of biodiversity is seldom
explored, which limits our ability to interpret the scale of the impacts. Our study quantifies
.50 years of impacts in the succulent thickets of the Addo Elephant National Park, South
Africa, contrasting hypotheses for the resilience of the canopy shrubs (a key functional guild)
to elephants with those that argue the opposite. We also assess the impacts between elements
of the community, ranging from community composition and structure to the structure of
individual canopy species. We show the vulnerability of the canopy shrubs to transformation
as the accumulated influences of elephants alter community composition and structure. The
pattern of transformation is similar to that caused by domestic herbivores, which leads us to
predict that elephants will eventually bring about landscape-level degradation and a significant
loss of biodiversity. While we expected the canopy species to show similar declining trends in
structure, providing insight into the response of the community as a whole, we demonstrate an
uneven distribution of impacts between constituent elements; most of the canopy dominants
exhibited little change, resisting removal. This implies that these canopy dominants might not
be useful indicators of community change in thickets, a pattern that is likely repeated among
the canopy trees of savanna systems. Our findings suggest that predicting elephant impacts,
and finding solutions to the so-called ‘‘elephant problem,’’ require a broader and more
integrated understanding of the mechanisms driving the changes between elements of
biodiversity at various spatial and temporal scales.
Key words: Addo Elephant National Park, South Africa; conservation management; elephant impacts;
long-term studies; Loxodonta africana; monitoring; scale; succulent thickets.
INTRODUCTION et al. 2011), and are modified by other drivers of
Biological diversity has emerged as a central theme in ecosystem change (e.g., rainfall variability, fire frequen-
the conservation management of southern Africa’s cy, and the influences of coexisting large herbivores;
growing elephant (Loxodonta africana) populations. Mapaure and Moe 2009, Hayward and Zawadzka
Specifically, management focus has shifted from manip- 2010).
ulating the size of elephant populations to identifying Despite this general understanding, variation in the
and mitigating the extent and intensity of their effects on intensity of elephant effects between elements of
biodiversity (Owen-Smith et al. 2006). This change biodiversity is seldom explored. Identifying the distri-
brought with it the recognition that elephants influence a bution of impacts between components of biodiversity
range of ecological patterns and processes, from the may be particularly important, given that ecosystems
composition and structure of plant and animal commu- function across an integrated spatiotemporal hierarchy
nities, to soil resource availability, litter production, and of patterns and processes (Pickett et al. 1997). Such an
nutrient dispersal (see the most recent review by Kerley understanding may also provide key insights into the
et al. [2008]). The intensity and heterogeneity of the issues around regime shifts (i.e., extensive, often
effects typically vary in relation to the availability of key
irreversible, long-term changes), which are difficult to
resources, including surface water and the quantity and
predict and require indicators that provide advance
quality of food (Chamaillé-Jammes et al. 2007, Pretorius
warning (e.g., Carpenter et al. 2008, Hughes et al. 2013
and references therein). For example, the conversion of
Manuscript received 17 January 2014; revised 5 May 2014; tall woodlands to shrub coppice or treeless grasslands
accepted 12 May 2014. Corresponding Editor: B. P. Wilcox.
4 E-mail: Marietjie.landman@nmmu.ac.za (e.g., Van de Vijver et al. 1999, Smallie and O’Connor
5 Deceased May 2014. 2000, Western 2006) may be a consequence of acceler-
2002December 2014 ELEPHANT EFFECTS VARY BY COMMUNITY TRAIT 2003
ated impacts on individual plant species that provide park comprises several fenced sections, with the majority
insight into the response of the community as a whole. of the elephant population confined to the Addo main
In the succulent thickets of the Addo Elephant camp section (AMC; covering 120 km2 during 2008).
National Park, South Africa, nearly 40 years of research AMC was originally fenced in 1954 (23.3 km2) to enclose
have demonstrated the consequences for biodiversity of the elephants of the region, before being incrementally
maintaining high levels of elephant utilization (reviewed expanded to support the steadily growing population
in Kerley and Landman 2006). The majority of this (from 22 individuals in 1954 to 384 in 2008; Kerley and
work quantified the impacts using snapshot natural Landman 2006). Three sites (exclosures; covering
experiments to contrast elephant-occupied areas with between 1.9 and 4.3 km2) that have excluded elephants
elephant exclosures. Results are particularly dramatic for .50 years, but are accessible to other large
for the canopy shrub community, and significant herbivores, e.g., kudu (Tragelaphus strepsiceros), bush-
declines in plant species richness, density, and biomass buck (Tragelaphus scriptus), and common duiker
have been recorded (Penzhorn et al. 1974, Barratt and (Sylvicapra grimmia), were established for monitoring
Hall-Martin 1991, Stuart-Hill 1992). However, these purposes.
simple contrasts limit our ability to predict the impacts. The region is semiarid, with 260–530 mm rainfall
Despite these weaknesses, Stuart-Hill (1992) and Kerley annually. In the absence of natural permanent surface
et al. (1999) argued that the top-down foraging of water, the number of provisioned water sources
elephants maintains the structure and ecological func- increased significantly through time: from six in 1954
tioning of thicket, particularly following 20 years of to 12 in 2008 (Fig. 1). The area comprises a series of low,
continuous use (Barratt and Hall-Martin 1991). Oppos- undulating hills (60–350 m in height) in the Sundays
ing this hypothesis is empirical evidence of the vulner- River valley, where nutrient-rich soils give rise to
ability of succulent thicket to transformation, as succulent thicket habitats (Vlok et al. 2003). These
prolonged utilization by domestic herbivores causes a thickets are typically evergreen, 2–4 m high, dense, and
gradual replacement of the canopy shrubs with ephem- characterized by a high diversity of growth forms. The
eral grasses (Kerley et al. 1995, Lechmere-Oertel et al. tree succulent Portulacaria afra is locally dominant and
2005a). Instead of reaching an equilibrium, however, occurs in a matrix of spinescent shrubs (e.g., Azima
transformed thicket continues along this trajectory of tetracantha, Capparis sepiaria, Carissa bispinosa, Searsia
decline, owing to the loss of key ecological processes spp.) and low trees (e.g., Euclea undulata, Schotia afra,
(Lechmere-Oertel et al. 2005b). Consequently, Gough Sideroxylon inerme). Grasses may be seasonally abun-
and Kerley (2006) predicted that similar patterns of dant where intensive utilization by elephants has
transformation might arise in the presence of elephants, removed the canopy shrubs (Landman et al. 2012).
and that thicket landscapes may be vulnerable to
degradation before any density-dependent population Experimental design and sampling
processes become apparent. Thus, despite the impor- Much of the history of AMC reflects an attempt to
tance of the canopy shrub community in the ecological reduce elephant densities in order to manage the impacts
functioning and resilience of succulent thicket (Kerley et (Kerley and Landman 2006). Thus, following Lombard
al. 1999, Lechmere-Oertel et al. 2005a, b), and evidence et al. (2001), we used the incremental expansion of AMC
of the impacts of elephants (Kerley and Landman 2006), between 1954 and 2008 (Fig. 1) to establish a gradient of
no clear understanding has emerged regarding its long- utilization, thereby quantifying elephant effects on the
term responses to elephants. canopy shrub community. The impacts were quantified
Using a unique synthetic design, our study quantifies at three levels: community composition (in terms of the
.50 years of elephant effects on the canopy shrubs of relative abundances of the canopy species contributing
the Addo Elephant National Park, contrasting hypoth- to the community), community structure (defined in
eses for its resilience to elephants (Stuart-Hill 1992, terms of the volume and density of all shrubs combined),
Kerley et al. 1999) with those that argue the opposite and the structure (volume only) of individual canopy
(Gough and Kerley 2006). We also assess the impacts species. Our approach assumed that areas utilized for an
between elements of this community, ranging from the extended period experienced relatively higher impacts,
composition and structure of the community as a whole, due to higher cumulative elephant densities, when
to the structure of individual canopy species. This long- compared to areas used for shorter periods; i.e., we
term and inclusive approach facilitates an understanding (initially) assumed an even distribution of elephants (but
of the scale of elephant impacts for monitoring and modified by other drivers of foraging intensity), and
management (Cumming et al. 2006, Lindenmayer and substituted space for time. We estimated elephant
Likens 2009). density for each site as the mean over 54 years, using
METHODS population numbers from K. Gough (unpublished data)
for every year. Twenty-nine experimental plots were
Study area located across three sites (6–15 plots per site) exposed to
Addo Elephant National Park (33831 0 S, 25845 0 E) is elephants since 1954 (site 1), 1977 (site 2), and 1984 (site
situated in the Eastern Cape, South Africa (Fig. 1). The 3), with an additional four plots located at the exclosuresEcological Applications
2004 MARIETJIE LANDMAN ET AL.
Vol. 24, No. 8
FIG. 1. Location and history of expansion of the Addo main camp (AMC) section (study area), Addo Elephant National Park,
South Africa. Experimental plots were located in succulent thicket habitats at sites exposed to elephants since 1954 (site 1), 1977
(site 2), and 1984 (site 3), and in three exclosures used as controls against which to measure elephant effects. Areas (covering 47.6
km2; 40% of AMC) included post-1984 were not surveyed.
(Fig. 1); the latter were used as a control against which example, plots at site 1 experienced between 23 and 54
to measure elephant effects. Plots were permanently years of utilization over the sample period (1977–2008),
marked in 1977, when they were first surveyed, with while impacts at site 3 were initiated only following the
further monitoring in 1981, 1989, and 2008 (providing 1981 survey (Table 1).
temporal coverage of 31 years). This meant that during Experimental plots were 5 m wide, while plot length
our final survey, sample sites represented 0, 24, 31, and (13–45 m) scaled inversely with the abundance of the
54 years of elephant utilization, respectively, with mean dominant shrub taxa. We estimated the volume (m3/m2)
densities of 0–2.4 elephants/km2 (Table 1). However, of all canopy shrubs (34 spp.: seven succulents, 27
intensity of use also varied with each survey. For woody shrubs) encountered by measuring the maximum
TABLE 1. Characteristics of sample sites incrementally exposed to elephants and surveyed between 1977 and 2008.
Site 1 Site 2 Site 3 Exclosure
Area (km2) 23.3 14.4 33.5 10.4
Total time (yr) utilized by elephants 54 [23, 27, 35, 54] 31 [0, 4, 12, 31] 24 [0, 0, 5, 24] 0 [0]
Mean elephant density (no. elephants/km2) 2.4 [0.9–4.0] 1.5 [1.8–3.2] 1.2 [1.8–3.2] 0 [0]
Density of permanent water points in 1977 and 2008 0.26, 0.13 0.07, 0.14 0.03, 0.09 0
(no./km2)
Notes: Area for site 3 includes a 10.6 km2 area exposed to elephants since 1982. However, due to small sample sizes (n ¼ 2),
results for this site were combined with those for site 3. Total time utilized by elephants refers to the sample period: values
presented in brackets for total time refer to how many years the site had been utilized by elephants at the time of each sampling
(1977, 1981, 1989, and 2008). Elephant density is shown with the range in brackets (estimated as the mean over 54 years [1954–
2008] using population numbers from K. Gough [unpublished data] for every year). Note that, because densities were standardized
to 54 years, these are generally smaller than the range estimated according to the time the site was utilized by elephants.December 2014 ELEPHANT EFFECTS VARY BY COMMUNITY TRAIT 2005
height and canopy diameters of individual plants. Since At a landscape scale, elephant foraging intensity may
most shrubs are multi-stemmed re-sprouters, stems vary with proximity to water, topography, and the
within 50 cm of each other at ground level were availability and quality of food (e.g., Chamaillé-Jammes
considered to be of the same individual. Individuals et al. 2007, Pretorius et al. 2011). However, because our
were measured if at least half the rooted area occurred experimental plots were generally located on even
within the plot. We calculated shrub density as the terrain with similar soils (a proxy for food quality;
number of individuals per unit area. Pretorius et al. 2011), we expected surface water
Our approach assumed that herbivory by elephants availability to be the primary determinant of elephant
was the primary determinant of vegetation structure in effects at this scale. Thus, for each sample period, we
AMC, dominating the effects of other herbivores (e.g., determined the distance between each experimental plot
kudu, bushbuck, common duiker) and other drivers of and the nearest permanent water point (range: 0–4422
ecosystem change (particularly rainfall; Kerley et al. m; see Table 1 for trends in water provision at each site),
1995, Hayward and Zawadzka 2010). Although this and included this as a covariate (log-transformed to
assumption should be treated with caution (e.g., Land- reduce the effects of extreme values) in our models. Plots
man et al. 2008), it reflected the fact that elephants located at sites that excluded elephants (i.e., the
comprise roughly 80% of large herbivore biomass in exclosure, but also those with no elephants at the time
AMC (South African National Parks, unpublished data), of sampling) were assigned distances equaling 4500 m;
and have been managed at densities that exceed (two- to i.e., slightly further than the most extreme observed
eightfold) recommended levels (0.3–0.5 elephants/km2) distance to water, which we took to be the distance
for 50 years (Kerley and Landman 2006). beyond which the impacts are most likely to be
asymptotic (Landman et al. 2012). Robustness tests of
Data analysis this assumption confirmed that model estimates were
We described elephant effects for each site, recogniz- relatively insensitive to the value assigned to this point.
ing that the intensity of utilization varied with sample Analyses were initiated with all fixed effects in the
period (Table 1). Data for the 1977, 1981, and 1989 model. We then optimized random effects (constant or
surveys were available from Barratt and Hall-Martin random slopes for the covariate sample period, and
(1991). allowing for heteroscedastic variances) for this full
Nonmetric multidimensional scaling (NMDS) ordi- model on the basis of restricted maximum likelihood
nations, based on Bray-Curtis resemblance matrices of fits, primarily using standard likelihood-ratio tests (a ¼
shrub density data (Clarke 1993, Clarke and Gorley 0.05), supplemented with Akaike’s information criterion
2006), were used to visualize differences in community (AIC) where significance was marginal (Burnham and
composition between sites. Six plots located 300 m Anderson 2002, Zuur et al. 2007). Once the random
from permanent water that showed extensive changes in effects structure was ascertained, we used the same
shrub composition due to the effects of elephants (i.e., approach to optimize fixed effects, but using maximum
the near-complete replacement of the shrub community likelihood fits. Final inspection of the model was used to
with grasses; Landman et al. 2012) were excluded from decide whether to review random effects again (where
these analyses because they dominated the ordinations little or no variance was explained by a random effect).
across sites. Data were square-root transformed to Data were examined for linearity prior to analyses, and
reduce the influence of extremely dominant species, standard diagnostic plots were inspected for deviations
and ordinations were corroborated with hierarchical from the model assumptions. Estimates for the coeffi-
agglomerative cluster analyses (Clarke 1993). Analyses cients from optimal models were generated using
of similarity (ANOSIM; 5000 Monte Carlo permuta- reduced maximum likelihood fits (Zuur et al. 2007).
tions) were used to test the null hypothesis of no RESULTS
difference in shrub composition between sample periods
for each site. Multivariate analyses were performed with Community composition
Primer version 6 (Clarke and Gorley 2006). The NMDS ordinations showed a clear change in
We modeled trends in shrub volume and density (i.e., shrub composition over the sample period for all
community structure) using linear mixed-effects models elephant-occupied sites and the exclosure (Fig. 2).
(package nlme in R 2.12.1; R Development Core Team However, inspection of ANOSIM R values indicated
2010) as described by Zuur et al. (2007). Analyses were that the magnitude and trajectory of these changes
repeated for the volumes of five canopy dominants for varied between sites (Table 2). With the exception of
which we had sufficient data: P. afra, E. undulata, S. site 2, the combined differences in shrub composition
afra, A. tetracantha, and C. sepiaria. In these models, we between 1977 and 2008 were comparable between sites
specified that the factors sample period (0–31 years from (global R ¼ 0.34 –0.39), despite substantial variations in
1977 to 2008), site (four levels: 1–3 and exclosure), and the intensity of utilization (Table 1). Importantly, these
their interaction were fixed, and that plots nested within differences were also similar between sites with and
site were random. The random effects fulfilled the role of without elephants, although it is likely that the
assigning repeated measures (Zuur et al. 2007). trajectory of change varied. Using the 1977 survey asEcological Applications
2006 MARIETJIE LANDMAN ET AL.
Vol. 24, No. 8
TABLE 2. Analyses of similarity (ANOSIM) results of the change in shrub composition between 1977 and 2008.
Pairwise comparisons
Global test 1977 to 1981 1977 to 1989 1977 to 2008
Site R P R P R P R P
1 (since 1954) 0.34 ,0.001 0.14 0.022 0.33 ,0.001 0.51 ,0.001
2 (since 1977) 0.26 0.001 0.13 0.786 0.51 0.008 0.35 0.008
3 (since 1984) 0.39 ,0.001 0.06 0.510 0.72 ,0.001 0.59 ,0.001
Exclosure 0.36 0.003 0.17 0.743 0.34 0.114 0.58 0.029
Notes: The 1977 survey was used as the base case for comparison. R values indicate the degree of change in composition between
sample periods, with values approaching unity indicating a clear separation. The global R reflects the combined differences between
all surveys.
the base case for comparison, shrub communities at site with the introduction of elephants, but a degree of
1 (intensive utilization) and the exclosure followed a stabilization thereafter.
trend of increasing dissimilarity with sample period
(Table 2). For site 1, these dissimilarities were Community structure
statistically significant throughout, while only the Results from the mixed-effects models showed a clear
2008 survey was different for the exclosure (P ¼ linear relationship between shrub volume and sample
0.029). In contrast, shrub communities at sites 2 and period (Fig. 3), and model fit deteriorated when we
3 initially showed increased dissimilarities associated removed the sample period 3 site interaction or any of
FIG. 2. Nonmetric multidimensional scaling ordinations of the change in shrub composition between 1977 and 2008. Each
point on a bi-plot represents the data from a single experimental plot. Stress indicates the goodness of fit of the three-dimensional
plot.December 2014 ELEPHANT EFFECTS VARY BY COMMUNITY TRAIT 2007 FIG. 3. Best-fit linear mixed-effects models of total shrub volume (V; m3/m2) and shrub density (D; number of individuals/m2) as a function of sample period (Sp; 0–31 years from 1977 to 2008) and distance to water (Dw; m). See Table 1 for intensity of elephant use per site. the individual factors (Appendix). Using this parame- the intensity of utilization; Appendix). After controlling terization, shrub volume declined significantly with for distance to water, model estimates indicated that sample period at all elephant-occupied sites (Fig. 3), volume declined weakly with sample period at sites 1 but with variation in slope among sites (and thus with (14.2%) and 3 (16.0%), while decreasing strongly at site 2
Ecological Applications
2008 MARIETJIE LANDMAN ET AL.
Vol. 24, No. 8
FIG. 4. Best-fit linear mixed-effects models of shrub volume as a function of sample period and distance to water for individual
canopy species. Abbreviations are as in Fig. 3. Surface plots for Azima tetracantha are not shown because best models did not
include site or distance to water; distance to water was also not important in the best-fit model for Schotia afra.
(42.5%). At the exclosure, shrub volume nearly doubled most species, mixed-model fits improved when we
(93.1% increase) over the same period (Fig. 3). Plots included sample period, site, and distance to water as
located near water (,300 m) had severely reduced factors (Appendix). However, neither site nor distance
volumes (Appendix), especially at sites 1 and 2, which to water was important in the best model for A.
were exposed to elephants for the longest period. tetracantha, while distance to water was also not
Similar to the best model for shrub volume, model fit important for S. afra, although this latter result is
for shrub density was best (lowest AIC) when we marginal. Where distance to water featured as an
included all factors in the model (Appendix). Using this important factor, shrub volume declined significantly
parameterization, shrub densities declined strongly near water for all species (Fig. 4). After controlling for
adjacent to water and with sample period at Site 1 distance to water, model estimates showed that P. afra,
(18.6%; Fig. 3). After controlling for distance to water, E. undulata, and C. sepiaria volumes generally varied
all other elephant-occupied sites and the exclosure little with sample period and site (Fig. 4; Table 3):
showed increasing densities (site 2, 13%; site 3 and exceptions were site 2 for P. afra (significant decline),
exclosure, 23.4%). site 3 for E. undulata (significant increase), and the
exclosure for C. sepiaria (significant increase). Only S.
Structure of the canopy species afra declined significantly at all elephant-occupied sites
As expected, the effects of elephants on the structure over the survey period, albeit that this decline was
of the canopy dominants varied between species. For significantly lower following intensive utilization at siteDecember 2014 ELEPHANT EFFECTS VARY BY COMMUNITY TRAIT 2009
TABLE 3. Percentage of change and significance levels of shrub volume over the sample period
(1977–2008) as predicted by linear mixed-effects models (see Fig. 3).
Change (%)
Canopy species Site 1 Site 2 Site 3 Exclosure
a b a
Portulacaria afra 4.0 61.4 * 6.3 55.3a
Euclea undulata 30.6a 39.7a 65.5b** 34.1a
Schotia afra 36.0a*** 64.3b** 64.3b*** 3.3c
Capparis sepiaria 14.6a 16.1a 37.7a 152.0b**
Notes: Positive values show an increase with sample period, while negative values show a decline.
Estimates were standardized using median distance to water. Different superscripted letters denote
significant (P , 0.05) between-site effects.
* P , 0.05; ** P , 0.01; *** P , 0.001.
1; we observed no change in the canopy volume of S. vores) are the rightful conservators of succulent thicket
afra at the exclosure (Table 3). Shrub volumes for A. (Stuart-Hill 1991). While this notion is supported by
tetracantha varied significantly with sample period evidence of the role of elephants in various ecologically
(Appendix), increasing by 163% (coefficient ¼ 0.01; SE important processes (Kerley and Landman 2006), our
¼ 0.01) between 1977 and 2008. results challenge these ideas by demonstrating the effects
of intensive utilization. Specifically, we show that the
DISCUSSION accumulated influences of elephants change the compo-
Despite nearly 60 years of scientific research on the sition and reduce the structure of the canopy shrub
subject (reviewed in Kerley et al. 2008), our ability to community. This pattern of transformation is no
predict the consequences of elephants on ecological different than that caused by domestic herbivores, and
systems is limited by a paucity of long-term quantitative is characterized by a gradual replacement of vulnerable
studies (e.g., Barnes 1983, Trollope et al. 1998, Van de species (e.g., those that recruit or regenerate poorly or
Vijver et al. 1999, Western 2006, Mapaure and Moe are susceptible to uprooting; O’Connor et al. 2007) with
2009). Our study expands on nearly 40 years of research ephemeral grasses, and the loss of ecological functioning
in the Addo Elephant National Park (Penzhorn et al. (Kerley et al. 1995, 1999, Lechmere-Oertel et al.
1974, Barratt and Hall-Martin 1991, Stuart-Hill 1992) to 2005a, b, Landman et al. 2012). Contrary to predictions
develop a detailed overview of elephant impacts on the that an equilibrium might be reached (Barratt and Hall-
thicket canopy shrub community. With this, we also Martin 1991), the decline continues even after 50 years
expand on other elephant studies that cover a wide of intensive use and despite the incremental expansion of
temporal range (from six to 60 years, e.g., Barnes [1983], the area to reduce the impacts. Thus, while the
Trollope et al. [1998], Van de Vijver et al. [1999]), but are equilibrium hypothesis probably emerged as AMC was
typically limited by poor temporal replication (reducing expanded and the intensity of the impacts declined (but
the power for detecting trends; Lindenmayer and Likens only by spreading impacts to novel areas; see for
2009). Thus, we provide the first spatially explicit models example sites 2 and 3 during the 1989 survey; Table 2),
(using empirical data; see the simulation model of Baxter we predict that the effects of elephants will eventually
and Getz [2005]) of the long-term effects of elephants on bring about landscape-level degradation (cf. Gough and
any plant community that may be used as a tool to Kerley 2006) and a significant loss of biodiversity. These
monitor and manage the impacts. results suggest that attempts to use range expansion as a
Ideally, the impacts of elephants on ecosystems should management tool to reduce elephant impacts may fail if
be understood in relation to the resilience of the system implemented without limiting population numbers and
to irreversible changes (Carpenter et al. 2008, Hughes et controlling local densities (e.g., by reducing surface
al. 2013). In succulent thicket, the canopy shrub water availability).
community shapes both the structural and functional Although our results show the vulnerability of the
complexity of the landscape. Thus, it is recognized that canopy shrubs, it will be important to develop a greater
this complexity declines following prolonged utilization understanding of the ecological thresholds in thickets.
by domestic herbivores, which causes the system to lose Landman et al. (2012) predicted that such a threshold is
resilience as it tends toward a degraded grassland state exceeded near water where the impacts intensify. Our
(Kerley et al. 1995, 1999, Lechmere-Oertel et al. results corroborate these ideas, as the accelerated decline
2005a, b). Elephants, however, are thought to maintain in shrub volume at site 2 (and perhaps even the
the structure and ecological functioning of the canopy continued decline at site 1) is likely a consequence of
shrubs, since their top-down foraging strategy promotes abundant water provisioning (Fig. 1; Table 1). Of
vegetative reproduction and resource trapping at ground significance is the fact that shrub densities declined only
level (Stuart-Hill 1992, Kerley et al. 1999). Herein lies in the vicinity of water and following intensive
the notion that elephants (rather than domestic herbi- utilization (site 1) where the generally persistentEcological Applications
2010 MARIETJIE LANDMAN ET AL.
Vol. 24, No. 8
rootstocks were completely removed. This implies that trees are also long-lived and slow-growing, and they may
some recovery of the canopy shrubs might be possible be manipulated by elephants in a variety of ways, from
with a release in the intensity of utilization elsewhere breaking branches and stems, to toppling and uprooting
(with unknown consequences for ecological functioning (O’Connor et al. 2007). The ecological consequences of
and resilience), but that such a recovery is unlikely near such effects differ considerably and determine the rate
water (Landman et al. 2012). Thus, while our study was and trajectory of change, and therefore the utility (or
undertaken at only a single site, and can therefore not be strength) of trees as indicators. This suggests that
generalized, we argue that dense networks of water predicting and monitoring the impacts requires a
points, as in AMC, compromise both biodiversity and broader and more integrated understanding of the
conservation objectives as the utilization gradients that mechanisms driving the changes between elements of
develop around water coalesce (Chamaillé-Jammes et al. biodiversity at various spatial and temporal scales
2007, Landman et al. 2012). (Kerley et al. 2008, Landman et al. 2008).
Predicting the impacts of elephants for management An important species’ response that emerged from our
requires a detailed understanding of their spatial and study requires further exploration. The tree succulent P.
temporal extent (Kerley et al. 2008), and this should be afra is widely accepted to be particularly tolerant of
based on robust insights into the distribution of the elephant impacts (e.g., Barratt and Hall-Martin 1991,
effects between elements of biodiversity that typically Stuart-Hill 1992), which we confirmed in our study. A
differ in their vulnerability. Understanding such distri- notable exception, however, was the significant decline
butions may further contribute toward establishing in canopy volume at site 2 (61.4%), where the impacts
appropriate indicators for monitoring, particularly since accelerated, possibly owing to abundant water provi-
monitoring programs (and by implication management) sioning. Given the vulnerability of succulent thickets to
often fail, owing to poor planning and limited evidence transformation (Kerley et al. 1995, 1999, Lechmere-
for the utility (or strength) of the indicators (Cumming Oertel et al. 2005a, b, Landman et al. 2012), we speculate
et al. 2006, Lindenmayer and Likens 2009). However, that such novel responses should generate significant
with the exception of the work of Levick and Rogers concerns in light of potential system shifts.
(2008) on woody species and patch responses to large
browsers, these relationships are rarely established. CONCLUSION
Because we observed extensive changes to the thicket The issues around elephant management are complex
shrub community at all levels explored, we expected the and should be approached with caution where uncer-
influences on the canopy species to provide insight into tainties exist (Biggs et al. 2008). By way of a
the response of the community as a whole. Specifically, precautionary approach, these uncertainties may be
we thought that canopy volume would decline steadily dealt with either by keeping elephant numbers low in
as the intensity of utilization increased across sites. the hope that this prevents the loss of biodiversity, or by
Instead, and with the exception of near-consistent allowing densities to increase until the levels of
declines around water, the majority of the canopy utilization that reduce diversity have been established
dominants exhibited little change and resisted removal, (Owen-Smith et al. 2006). The decision will largely be
whereas A. tetracantha might have benefitted from being driven by society through the values attached to
utilized. Given the generally poor regeneration dynamics elephants and biodiversity in general. The long-term
of most thicket plants (Vlok et al. 2003), this probably perspectives generated in this study and the information
reflects the top-down foraging strategy of elephants that reviewed by Kerley et al. (2008), however, clearly show
promotes vegetative reproduction at ground level the deleterious consequences for succulent thickets and
(Stuart-Hill 1992). These so-called hedging-effects are other elephant habitats maintaining high elephant
not novel (see, for example, effects on Colophospermum densities and abundant water provisioning. For the
mopane; Smallie and O’Connor 2000), and elephants are Addo Elephant National Park, this is ahead of any
often thought to select previously hedged plants due to evidence that its elephant population might stabilize at
increased browse availability and quality; hence their some resource-limited level (Gough and Kerley 2006).
dominance in the diet (sensu Cromsigt and Kuijper Thus, in the absence of a clearer understanding of the
2011). Nevertheless, the disparities in the responses ecological thresholds in thickets, this suggests that
between communities and species limit our understand- limiting elephant numbers should be a conservation
ing of the scale of the impacts, and imply that the management priority. Expanding the area available to
canopy dominants might not be useful indicators of elephants in Addo did not achieve this goal, which
community change in succulent thicket. further calls for a revision of the approaches to manage
Similarly, large trees are iconic elements of savanna the impacts. Elsewhere, we expect similar trends to
landscapes that play an important role in community develop where elephants are confined to small fenced
structure and ecological functioning (e.g., Van de Vijver areas, with the exception that other drivers of ecosystem
et al. 1999, Manning et al. 2006). As a consequence, they change may accelerate and limit our interpretation of
are considered obvious and suitable indicators for the impacts. Thus, a predictive understanding of the
monitoring (e.g., Druce et al. 2009). However, these spatial and temporal variations of elephant impactsDecember 2014 ELEPHANT EFFECTS VARY BY COMMUNITY TRAIT 2011
between elements of biodiversity and the mechanisms during slow, unrecognized regime shifts. Trends in Ecology
driving these changes are key to their management, and and Evolution 28:149–155.
Kerley, G. I. H., M. H. Knight, and M. De Kock. 1995.
central to finding solutions to the so-called elephant Desertification of subtropical thicket in the Eastern Cape,
problem (Caughley 1976). South Africa: are there alternatives? Environmental Moni-
toring and Assessment 37:211–230.
ACKNOWLEDGMENTS Kerley, G. I. H., and M. Landman. 2006. The impacts of
We thank South African National Parks for funding and elephants on biodiversity in the Eastern Cape subtropical
permission to conduct the study in the Addo Elephant National thickets. South African Journal of Science 102:395–402.
Park. Mazda Wildlife Fund provided field transport. Com- Kerley, G. I. H., D. Tongway, and J. A. Ludwig. 1999. Effects
ments from Norman Owen-Smith and two anonymous of elephant browsing on soil resources in succulent thicket,
reviewers greatly improved the manuscript. Eastern Cape, South Africa. VI International Rangeland
Congress 1, 116–117.
LITERATURE CITED Kerley, G. I. H., et al. 2008. Effects of elephant on ecosystems
Barnes, R. F. W. 1983. Effects of elephant browsing on and biodiversity. Pages 146–204 in R. J. Scholes and K. G.
woodlands in a Tanzanian National Park: measurements, Mennell, editors. Elephant management: a scientific assess-
models and management. Journal of Applied Ecology ment for South Africa. Wits University Press, Johannesburg,
20:521–539. South Africa.
Barratt, D. G., and A. J. Hall-Martin. 1991. The effects of Landman, M., G. I. H. Kerley, and D. Schoeman. 2008.
indigenous herbivores on Valley Bushveld in the Addo Relevance of elephant herbivory as a threat to important
Elephant National Park. Pages 14–16 in P. J. K. Zacharias plants in the Addo Elephant National Park, South Africa.
and G. C. Stuart-Hill, editors. Proceedings of the first Valley Journal of Zoology, London 274:51–58.
Bushveld/subtropical thicket symposium, Grahamstown, Landman, M., D. S. Schoeman, A. J. Hall-Martin, and G. I. H.
South Africa, 1990. Grassland Society of Southern Africa, Kerley. 2012. Understanding long-term variations in an
Howick, South Africa. elephant piosphere effect to manage impacts. PLoS ONE
7:e45334.
Baxter, P. W. J., and W. M. Getz. 2005. A model-framed
Lechmere-Oertel, R. G., R. M. Cowling, and G. I. H. Kerley.
evaluation of elephant effects on tree and fire dynamics in
2005a. Patterns and implications of transformation in semi-
African savannas. Ecological Applications 15:1331–1341.
arid succulent thicket, South Africa. Journal of Arid
Biggs, H. C., et al. 2008. Towards an integrated decision
Environments 62:459–474.
making for elephant management. Pages 146–204 in R. J.
Lechmere-Oertel, R. G., R. M. Cowling, and G. I. H. Kerley.
Scholes and K. G. Mennell, editors. Elephant management: a
2005b. Landscape dysfunction and reduced spatial heteroge-
scientific assessment for South Africa. Wits University Press,
neity in soil resources and fertility in semi-arid succulent
Johannesburg, South Africa.
thicket, South Africa. Austral Ecology 30:615–624.
Burnham, K. P., and D. R. Anderson. 2002. Model selection
Levick, S., and K. Rogers. 2008. Patch and species specific
and multimodel inference: a practical information-theoretic responses of savanna woody vegetation to browser exclusion.
approach. Springer, New York, New York, USA. Biological Conservation 141:489–498.
Carpenter, S. R., et al. 2008. Leading indicators of trophic Lindenmayer, D. B., and G. E. Likens. 2009. Adaptive
cascades. Ecology Letters 11:128–138. monitoring: a new paradigm for long-term research and
Caughley, G. 1976. The elephant problem—an alternative monitoring. Trends in Ecology and Evolution 24:482–486.
hypothesis. East African Wildlife Journal 26:323–327. Lombard, A. T., C. F. Johnson, R. M. Cowling, and R. L.
Chamaillé-Jammes, S., M. Valeix, and H. Fritz. 2007. Pressey. 2001. Protecting plants from elephants: botanical
Managing heterogeneity in elephant distribution: interactions reserve scenarios within the Addo Elephant National Park,
between elephant population density and surface-water South Africa. Biological Conservation 102:191–203.
availability. Journal of Applied Ecology 44:625–633. Manning, A. D., J. Fischer, and D. B. Lindenmayer. 2006.
Clarke, K. R. 1993. Non-parametric multivariate analyses of Scattered trees are keystone structures—implications for
changes in community structure. Australian Journal of conservation. Biological Conservation 132:311–321.
Ecology 18:117–143. Mapaure, I., and S. R. Moe. 2009. Changes in the structure and
Clarke, K. R., and R. N. Gorley. 2006. Primer V6: user manual/ composition of miombo woodlands mediated by elephants
tutorial. Primer-e, Plymouth, UK. (Loxodonta africana) and fire over a 26-year period in north-
Cromsigt, J. P. G. M., and D. P. J. Kuijper. 2011. Revisiting the western Zimbabwe. Africa Journal of Ecology 47:175–183.
browsing lawn concept: evolutionary interactions or pruning O’Connor, T. G., P. S. Goodman, and B. Clegg. 2007. A
herbivores? Perspectives in Plant Ecology, Evolution and functional hypothesis of the threat of local extirpation of
Systematics 13:207–215. woody plant species by elephant in Africa. Biological
Cumming, G. S., D. H. M. Cumming, and C. L. Redman. 2006. Conservation 136:329–345.
Scale mismatches in social-ecological systems: causes, conse- Owen-Smith, N., et al. 2006. A scientific perspective on the
quences, and solutions. Ecology and Society 11:14. management of elephants in the Kruger National Park and
Druce, D. J., et al. 2009. Ecological thresholds in the savanna elsewhere. South African Journal of Science 102:389–394.
landscape: developing a protocol for monitoring the change Penzhorn, B. L., P. J. Robbertse, and M. C. Olivier. 1974. The
in the composition and utilisation of large trees. PLoS ONE influence of the African elephant on the vegetation of the
3:e3979. Addo Elephant National Park. Koedoe 17:137–158.
Gough, K. F., and G. I. H. Kerley. 2006. Demography and Pickett, S. T. A., R. S. Ostfeld, M. Shachak, and G. E. Likens.
population dynamics in the elephants Loxodonta africana of 1997. Enhancing the ecological basis of conservation:
Addo Elephant National Park, South Africa: is there heterogeneity, ecosystem function and biodiversity. Chap-
evidence of density dependent regulation? Oryx 40:434–441. man and Hall, New York, New York, USA.
Hayward, M. W., and B. Zawadzka. 2010. Increasing elephant Pretorius, Y., et al. 2011. Soil nutrient status determines how
Loxodonta africana density is a more important driver of elephant utilize trees and shape environments. Journal of
change in vegetation condition than rainfall. Acta Therio- Animal Ecology 80:875–883.
logica 55:289–299. R Development Core Team. 2010. R: A language and
Hughes, T. P., C. Linares, V. Dakos, I. A. van de Leemput, and environment for statistical computing. R Foundation for
E. H. van Nes. 2013. Living dangerously on borrowed time Statistical Computing, Vienna, Austria. www.r-project.orgEcological Applications
2012 MARIETJIE LANDMAN ET AL.
Vol. 24, No. 8
Smallie, J. J., and T. G. O’Connor. 2000. Elephant utilization of savanna during 25 years. Journal of Tropical Ecology
Colophospermum mopane: possible benefits of hedging. 15:545–564.
African Journal of Ecology 38:352–359. Vlok, J. H. J., D. I. W. Euston-Brown, and R. M. Cowling.
Stuart-Hill, G. C. 1991. Elephant, the rightful conservators of 2003. Acocks’ Valley Bushveld 50 years on: a new perspective
the Valley Bushveld. Veld and Flora 77:9–11. on the delimitation, characterization, and origin of subtrop-
Stuart-Hill, G. C. 1992. Effects of elephants and goats on the ical thicket vegetation. South African Journal of Botany
Kaffrarian succulent thicket of the Eastern Cape, South
69:27–51.
Africa. Journal of Applied Ecology 29:699–710.
Trollope, W. S. W., et al. 1998. Long-term changes in the Western, D. 2006. A half century of habitat change in Amboseli
woody vegetation of the Kruger National Park, with special National Park, Kenya. African Journal of Ecology 45:302–
reference to the effects of elephants and fire. Koedoe 41:103– 310.
112. Zuur, A. F., E. N. Ieno, and G. M. Smith. 2007. Analyzing
Van de Vijver, C. A. D. M., C. A. Foley, and H. Olff. 1999. ecological data. Springer Science and Business Media, New
Changes in the woody component of an East African York, New York, USA.
SUPPLEMENTAL MATERIAL
Ecological Archives
The appendix is available online: http://dx.doi.org/10.1890/14-0080.1.smYou can also read
