Emergent trees in Colophospermum mopane woodland: influence of elephant density on persistence versus attrition

Study area

The study was conducted in south-eastern Zimbabwe within Gonarezhou National Park (GNP) and the adjacent Malilangwe Wildlife Reserve (MWR) (Fig. 1). MWR is 394 km² in area, and is fenced. GNP has an area of 5053 km² and is partly fenced. The environment and vegetation of MWR has been described in Clegg & O’Connor (2012), and of GNP in Cunliffe, Muller & Mapaura (2012). Each study area experiences a similar seasonal climate of hot, wet summers and warm dry winters. Mean annual rainfall is approximately 550 mm, and daily maximum temperatures may exceed 30 °C during every month of the year. The Chiredzi River, joined by the Nyamasikana River, flows through MWR to join the Runde River, which flows west to east across northern GNP until it enters Moçambique. Woodlands in which mopane is dominant or conspicuous cover large parts of both MWR (Clegg & O’Connor, 2012) and GNP (Cunliffe, Muller & Mapaura, 2012). This study focussed on mopane woodland on alluvium, described as the Mopane Woodland on Alluvium (vegetation type 4.9) in GNP (Cunliffe, Muller & Mapaura, 2012) and as Colophospermum mopane-Courbonia glauca shrub open tall woodland in MWR (Clegg & O’Connor, 2012). This woodland type has low species richness, is dominated by mopane, and usually supports tall, emergent trees of this species.

Data collection

The Gonarezhou Conservation Trust granted permission for this study. The study was structured as two components. The first component addressed the effects of elephant density using data collected in MWR and GNP during 2001 and 2014. The second component examined the effects of elephants at high density in GNP from 2014 to 2022.

Data analysis

All analyses were conducted using R version 4.2.2 (R Core Team, 2022), with data manipulation and plotting done, respectively, using the ‘tidyverse’ (version 1.3.2; Wickham et al., 2019) and ‘dplr’ (version 1.0.10; Wickham et al., 2022) packages. The average of a plot is the unit of analysis unless otherwise stated.

Effects of elephant density (2001 and 2014)

Vegetation structure.

Elephant density had a marked influence on vegetation structure (Table 1). Under a sustained high elephant density in GNP, compared with low or intermediate density in MWR, tree canopy volume and total canopy volume were approximately halved; shrub canopy volume was unaffected; mean tree height was 3 m lower; the density of trees >10 m in height was more than halved, with a corresponding approximately six-fold increase in the density of pollarded stems, although the density of trees <10 m in height, and of live standing trees, were unaffected (live prostrate trees were almost absent); dead trees were two thirds less; and four and two species of shrubs and trees, respectively, had been lost. Effects of an increase from low to intermediate elephant density on Malilangwe Reserve over 14 years were apparent as a 21% loss of live standing trees, a 38% loss of trees <10 m in height, a corresponding four-fold increase in the density of pollarded main stems, a 29% reduction in shrub density, and a marginal loss of tree and shrub species.

Table 1:

Structure and extent of use of Colophospermum mopane woodland under three elephant densities.

Elephant density	Low (L) (n= 8)	Intermediate (I) (n= 8)	High (H) (n= 7)	L v I	L v H	I v H
Total shrub canopy volume (m3 ha−1)	4716 ±2402.7	2420 ±380.7	3715 ±1025.0	t =1.0263; P = 0.3389	t =0.3793; P = 0.7126	t =1.1162; P = 0.2991
Total tree canopy volume (m3 ha−1)	23194 ±3942.0	25838 ±5765.0	11404 ±1619.0	t =0.5366; P = 0.6082	t =2.7386; P = 0.0217	t =2.3980; P = 0.0424
Total canopy volume (m3 ha−1)	27911 ±4286.5	28258 ±5451.7	15119 ±1918.7	t =0.0563; P = 0.9567	t =2.6919; P = 0.0227	t =2.2558; P = 0.0508
Shrub density (ha−1)	1406 ±246.0	1004 ±269.1	2100 ±811.3	t =3.3788; P = 0.0118	t =0.7699; P = 0.4667	t =1.2404; P = 0.2580
Density of dead trees (ha−1)	32 ±6.0	24 ±3.6	9 ±3.5	t =1.3517; P = 0.2185	t =3.3899; P = 0.0057	t =2.9285; P = 0.0119
Density live trees (ha−1)	259 ±59.3	205 ±48.0	193 ±24.1	na	na	na
Density of standing live trees (LS) (ha−1)	258.8 ±10	205 ±48.0	193 ±24.1	t =2.3526; P = 0.0509	t =1.0194; P = 0.3333	t =0.229; P = 0.8235
Density of live prostrate trees (ha−1)	0.5 ±0.5	0	0	na	na	na
Density of trees <10 m height (ha−1)	220 ±67.9	136 ±46.7	178 ±23.9	t =2.7931; P = 0.0268	t =0.5734; P = 0.5805	t =0.7898; P = 0.4468
Density of trees >10 m height (ha−1)	40 ±10.8	69 ±13.4	14 ±2.3	t =1.7408; P = 0.1320	t =2.2929; P = 0.0522	t =3.9998; P = 0.0046
Density of pollarded main stems (ha−1)	28 ±8.6	120 ±23.9	191 ±27.6	t =5.5654; P = 0.0008	t =5.3064; P = 0.0011	t =1.8751; P = 0.0853
Mean tree height (m)	7.9 ±0.81	9.4 ±1.47	5.1 ±0.28	t =1.8235; P = 0.1110	t =3.1826; P = 0.0113	t =2.8499; P = 0.0227
Species richness of shrubs	11.5 ±3.2	9.3 ±3.5	7.4 ±2.1	t =2.1223; P = 0.0715	t =2.7357; P = 0.0224	t =2.2280; P = 0.0448
Species richness of trees	3.4 ±1.3	2.4 ±0.5	1.1 ±0.4	t =1.9296; P = 0.0950	t =5.5624; P = 0.0004	t =4.0909; P = 0.0027
Trees: Old damage by elephants (%)	5.3 ±1.0	21.3 ±3.0	50.1 ±4.3	t =6.4378; P = 0.0004	t =9.4453; P = 4.7e−05	t =5.2335; P = 0.0003
Trees: Old damage by unknown agents (%)	9.7 ±1.7	14.3 ±2.2	5.2 ±2.3	t =1.5402; P = 0.1674	t =1.4833; P = 0.1658	t =2.7430; P = 0.0171
Trees: New damage by elephants (%)	0.1 ±0.0	2.4 ±1.1	11.3 ±0.9	t =2.1572; P = 0.0679	t =11.921; P = 2.06e−05	t =6.3255; P = 2.65e−05
Trees: New damage by unknown agents	0	0	0	na	na	na
Shrubs: Old damage by elephants (%)	4.2 ±2.1	36 ±4.7	28.8 ±3.6	t =7.3440; P = 0.0002	t =5.5755; P = 0.0003	t =1.1864; P = 0.2570
Shrubs: Old damage by unknown agents (%)	3.6 ±1.1	4.8 ±1.8	1.3 ±0.5	t =0,5591; P = 0.5935	t =1.8806; P = 0.08917	t =1.8841; P = 0.0951
Shrubs: New damage by elephants (%)	0.1 ±0.0	6.1 ±1.7	6.6 ±0.9	t =3.6176; P = 0.0085	t =6.5194; P = 0.0006	t =0.2681; P = 0.7935
Shrubs: New damage by unknown agents (%)	0	1.6 ±1.4	0	na	na	na

DOI: 10.7717/peerj.16961/table-1

Notes:

Differences in vegetation structure and woody species richness across Colophospermum mopane woodland areas in south-east Zimbabwe experiencing low, intermediate and high elephant density. Cell values denote mean ± SE (na denotes not applicable).

Elephant density further influenced canopy volume per height layer (Fig. 2; Table S1). The least canopy volume occurred in the 0–1.0 m layer (P < 0.05), whilst the other three height layers did not differ among themselves (P > 0.05). The smallest canopy volume under the highest elephant density shown in Table 1 was apparent only for the two height layers >3.0 m, but not for those <3.0 m in height (i.e., interaction effect). Elephant density further influenced the height structure of the tree population (Fig. 3; χ² = 336.3; df = 6; P = 2.2e−16). Smaller trees (3.1–6.0 m in height) were well represented under all three levels of elephant density; trees from 7.1 to 11.0 m in height had been, respectively, markedly reduced or completely eliminated under an intermediate or high elephant density. There was a lower density of trees 6.1 to 14.0 m in height under a high elephant density, although a proportion of trees >14.1 m in height had persisted.

Utilisation.

Consistent with an increase in elephant density was an approximately ten- and hundred-fold increase in old and new elephant damage, respectively, for trees (Table 1). Trees had lost about 25% or 60% of canopy volume to old and new elephant utilisation combined under intermediate or high elephant density, respectively, compared with 5% for the area under low elephant density. By contrast, old and new elephant damage on shrubs was considerably higher for intermediate or high elephant density than for low elephant density. Damage by unknown agents was minor, accounting for five to 14% of canopy volume, and fire damage was so slight it could be disregarded.

Areas under different elephant densities differed in terms of old or new bark damage inflicted by elephants, and for old bark damage by unknown agents, that depended on tree height (Fig. 4; Table S2). Only trees > 7 m in height were debarked to any meaningful degree. Areas under high elephant density showed a 30-fold greater level of old debarking by elephants than areas under low elephant density, but that of areas under intermediate density did not differ (P > 0.05) from either (Fig. 4A). Areas under different elephant densities differed only marginally in terms of new bark damage by elephants (Table S2), attributed to the near absence of new debarking in areas under low density (Fig. 4B). Old debarking by unknown agents was higher under an intermediate than under a low or high elephant density (Fig. 4C; Table S2).

Influence of chronic elephant utilisation in GNP: 2014–2022

Mopane contributed more than the other 26 shrub species combined to either average shrub density or average shrub canopy volume (Fig. S2), and almost all trees were mopane. The combined value of all species was therefore used for analyses. For trees and shrubs combined, no change was evident between 2014 and 2022 for either total density (t = 0.51; df = 11.81; P = 0.6208) or for total canopy volume (t = 0.8632; df = 9.8475; P = 0.4086). For the analysis of variance, there was no main effect of year (2014 to 2022) on changes in density by height class, or changes in canopy volume by height class or by height layer (Table S3). Nor were there any changes in density or canopy volume of individual height classes, or canopy volume of individual height layers (P > 0.05 for all paired t-tests) between 2014 and 2022 (Fig. 5). Although plants <1 m in height constituted most, and plants >6 m in height constituted very little of population number in either year (Fig. 5A), the bulk of canopy volume was provided by plants between 3.1 and 6.0 m in height (Fig. 5B), whereas the 1.1 to 3.0 m height layer contained the greatest amount of canopy volume (Fig. 5C). The abundance of mopane in the smallest size class (Fig. 5A) indicates this species has recruited well. Other than some dead emergent trees, dead stems and trees were conspicuously almost absent, presumably having long since been knocked over by elephants.

Mopane trees had been severely utilised by 2022 (Table 2). In terms of old elephant damage, two thirds of the tree stem population had been pollarded and about 11% had escaped attention, whereas removal of branches was the main recent impact because very few stems remained to be pollarded or pushed over. The height at which a stem was pollarded was from close to ground level up to 3 m. Shrubs received less attention; only 23.4% of shrubs <1 m in height, and 80.2% of shrubs 1.1-3 m in height, had been used. The main stem of a tree was unlikely to be pollarded by elephants if it was approximately >45 cm in diameter (Fig. 6), which is the size of the emergent canopy trees >10 m in height (Fig. S3).

In summary, canopy volume was effectively maintained over an eight-year period despite chronic utilisation by elephants.

Emergent trees (n = 89) experienced attrition between 2014 and 2022, with 73% surviving intact, 15.8% having lost one or more of the main stems (>90 cm circumference), and 11.2% of trees having been lost completely. Twelve of the 15 trees with two main stems had lost the smaller stem, and two of the six trees with more than two stems had lost at least one stem. Expressing this as a stem population (n = 127) of these trees in 2022, 71.7% of the stems remained standing, and 28.3% had been lost as large stems owing to being pollarded (18.1%), falling over (8.7%), or other causes (1.6%). Percent of circumference debarked of an emergent standing tree increased on average by 25% between 2014 and 2022 (Fig. S4). The proportion of canopy volume of an emergent tree lost to dieback in relation to the extent of debarking was described by a logistic relationship (Fig. 7). Taken together, collapse of tall stems was primarily attributed to windthrow of a stem preconditioned by advanced canopy dieback once debarking exceeded 50% of stem circumference (Fig. 7). One exception was probable collapse directly from windthrow. Thus elephants indirectly caused the toppling of emergent trees and stems through canopy dieback resulting from debarking; complete ringbarking was not necessarily required.

Elephant-mopane relations

Conversion of mopane woodland to hedged woodland by elephant utilisation has been widely reported, including in northern Botswana (Ben-Shahar, 1996), Tuli Block, Botswana (Styles & Skinner, 1997; Styles & Skinner, 2001), Luangwa Valley, Zambia (Caughley, 1976; Lewis, 1991), Kruger National Park and Limpopo Valley, South Africa (Trollope et al., 1998; Smallie & O’Connor, 2000), and, in Zimbabwe, in Sengwa Wildlife Research Area (Anderson & Walker, 1974), Hwange National Park (Boughey, 1963), and Gonarezhou National Park (Guy, 1981; Tafangenyasha, 1997; this study). In this study, conversion of riverine mopane woodland to hedged woodland depended on elephant density. In MWR, conversion of riverine mopane woodland had been initiated between 2001 and 2014 when elephant density rose from 0.23 to 0.59 elephants km⁻², as evidenced by an increase in the density of pollarded stems and a loss of canopy volume (Table 1). However, the tallest height classes were maintained (Fig. 3). By contrast, riverine mopane woodlands in GNP had been converted to hedged woodland by 2014 through the loss of all trees between 6.1 m to 10 m in height, with a low density of relict tall trees too large to be pollarded (Fig. 6) remaining. Prior to the 1991/92 drought, elephant density in GNP was mostly maintained below 1.0 individual km⁻² through population reduction, and was reduced to 0.8 individuals km⁻² by this drought, but thereafter the population grew exponentially at 6.2% per annum to attain a density of >2 elephants km⁻² in 2013 (Dunham et al., 2013). Riverine mopane woodland apparently was converted during this 23-year period. The responses observed in this study are consistent with woodland conversion becoming apparent at an elephant density of about 0.5 individuals km⁻² (Cumming et al., 1997). Ground observations emphasised the role bulls play in woodland conversion, with the changes on MWR between 2001 and 2014 strongly influenced by an influx of about 70 bulls from GNP before 2014.

Conversion of tall woodland to a hedged woodland in GNP by 2014 had not decreased the amount of canopy volume under 3 m in height available to elephants (Fig. 2), nor did overall canopy volume decrease under eight years of chronic use by elephants (Figs. 5B and 5C). Furthermore, coppicing of mopane increases the availability and palatability of foliage (leaf, twig, or twig bark) (Smallie & O’Connor, 2000; Smit & Rethman, 1998; Styles & Skinner, 1997; Styles & Skinner, 2001; Hrabar, Hattas & Du Toit, 2009a) that should improve the foraging efficiency of elephants. Mopane leaf is the staple foodstuff of female elephants in south-eastern Zimbabwe (Clegg, 2010). We therefore propose that the increased availability of this foodstuff through hedging is an important influence on maintaining a high density (∼2 km⁻²) of elephants in a semi-arid environment.

Conversion of mopane woodland to hedged woodland has not threatened persistence of the mopane population. (A caveat is the dearth of knowledge about the effect of chronic utilisation on below-ground growth of a plant.) Elephants have transformed the growth form of an individual plant and, thereby collectively, of the vegetation structure of a woodland, but complete mortality (i.e., no coppicing) of mopane recorded in this study was relatively slight. Mortality of large trees as a result of ringbarking observed in this study is an expected result (Lewis, 1991). Reports of apparently high mortality of mopane in GNP as a consequence of toppling or pollarding by elephants (Tafangenyasha, 1997) are equivocal because trees may appear dead based on a once-off visual assessment of tree loss, but subsequent monitoring of toppled mopane trees in GNP has revealed that most affected trees coppice from the base months after impact (Bromwich, 1972). Lewis (1991) found that complete mortality of mopane trees toppled or pollarded by elephants in the Luangwa Valley, Zambia, depended on edaphic characteristics, and was precipitated by drought. Elsewhere in semi-arid, non-riparian mopane woodlands, a single drought event has caused the loss of between 4.5% and 6.9% of mopane individuals, usually smaller plants (Scholes, 1985; O’Connor, 1999), and the loss of patches of adult mopane trees on degraded habitats where water retention had been compromised (MacGregor & O’Connor, 2002). Fire does not cause conspicuous mortality of mopane because of its coppicing ability (Timberlake, 1995) and, in any event, riverine mopane woodland in Gonarezhou National Park rarely carries sufficient fuel for a burn.

Caughley (1976) proposed, based on his studies of mopane woodland and baobabs in the Luangwa Valley, Zambia, that elephant-woodland relations may follow a stable limit cycle, with a periodicity of about 200 years in the Luangwa Valley. His proposal questioned the prevailing assumption of management of that period that a stable equilibrium point exists between elephants and woodland. Putting aside theoretical and empirical challenges to his proposal (e.g., Cumming, 1982; Lewis, 1986; Lewis, 1991; Duffy et al., 1999; Baxter & Getz, 2005), we do not consider a stable-limit cycle to be an appropriate conceptual model for elephant-mopane woodland relations in south-eastern Zimbabwe. This model requires a close coupling between elephants and mopane, which apparently exists in south-eastern Zimbabwe where mopane provides the bulk of the food intake of cows (Clegg, 2010). However, this study showed that the abundance of mopane did not materially diminish even after a few decades of chronic utilisation by elephants. Furthermore, elephants show a catholic use of food species using more than 100 species in south-eastern Zimbabwe (Clegg, 2010) and comparable numbers in other systems where mopane does (Williamson, 1975; Guy, 1976) or does not occur (De Boer et al., 2000). Elephants therefore have many options for foodstuffs should mopane decline; there is no convincing evidence that elephants have a close coupling with any individual plant species.

Can relict emergent trees survive?

An hypothesis put forward for this study was that an increased availability of forage as a result of hedging would decrease use of remaining emergent trees, provided that elephant density did not continue to rise. The condition of elephant density remaining approximately stable was met (Dunham, Van der Westhuizen & Madinyenya, 2021). The elephant population remained at approximately 9,000-10,000 individuals (∼2 individuals km⁻²) from 2014 to 2021. The conditions were also met that hedging would not result in a decreased availability of mopane for elephants (Fig. 2), and that the amount of mopane canopy volume would be maintained under chronic elephant utilisation (Figs. 5B and 5C). However, the hypothesis was rejected on the basis of ongoing attrition of emergent trees through debarking, at a rate of 1.4% per annum lost over eight years. On MWR, elephants make greater use of debarking toward the end of the dry season when the availability of other foodstuffs has declined (Clegg, 2010). Debarking is interpreted as a foraging action of necessity rather than a preferred action because elephants require about 18 h a day to meet their foraging needs, with an adult male consuming 1–1.2% of its body weight per day (Owen-Smith, 1992), and debarking of large stems is an energetically costly process, taking about five times longer to harvest and chew a mouthful of bark from the main stem than a mouthful of leaves (Clegg & O’Connor, 2016). The time required for debarking mopane trees may be less because large sections of bark can be relatively quickly stripped off a large mopane stem once an incision has been made, compared with some tree species (e.g., Sclerocarya birrea) for which small fragments have to be tediously chiselled off (Clegg, 2010). However, bark of large mopane stems is conspicuously fibrous. Bark therefore seems less than ideal for meeting foraging needs, unless it perhaps offers an essential constituent (cf. Anderson & Walker, 1974), which has not yet been identified.

Emergent mopane trees in riverine mopane woodland, GNP, appear set to experience an ongoing slow decline in density if conditions remain similar to those recorded over the eight years of study—a high elephant density and a lack of foraging alternatives at the height of the dry season. Bulls are primarily responsible for debarking mopane, but they are preferentially grazers of green grass (Clegg, 2010, and references therein). Clegg (2010) proposed that the impact of bulls on woody vegetation would be considerably less if they had access to suitable grasslands, especially floodplain or riverine grasslands (e.g., Lewis, 1986), wetlands and reedbeds, during the dry season, which, he proposed, was the historical norm before unbridled human expansion. The far-reaching ranging patterns of elephant bulls, for which travel in the order of 50-100 km is not uncommon (summary in Dolmia et al., 2007) should enable them to access winter foraging grounds of such a nature both within and outside of GNP. Options within GNP are limited. Extensive reedbeds have occurred along portions of the Runde River within GNP that have been stripped by large floods and then regrown. For example, the flood resulting from Cyclone Eline in 2000 denuded a four kilometre stretch of reedbeds, which then recovered within about a decade, but was again stripped by Cyclone Dineo’s flood in 2017, and has not yet re-established (Appendix S1). Options outside of GNP should be substantial with the creation of the 89,000 km² Greater Limpopo Transfrontier Conservation Area in 2002 (Ferreira, 2004) that potentially offers extensive wetlands or riverine grasslands within the Banhine and Zinave National Parks, and along the Save River, in Moçambique. To date, a current study using satellite-tracked elephants has revealed that bull elephants have begun to explore the habitat available in Moçambique, but not yet for a duration that would make a material difference to their impact within GNP (B Madinyenya, pers. comm., 2023). Their subdued use of Moçambique is attributed, in part, to the hunting pressure they encounter along the border between GNP and Moçambique (Dunham, Van der Westhuizen & Madinyenya, 2021), compounded by an increase in human settlement along the Save River and its tributaries. If these constraints change, then Moçambique may become an important foraging area for, at least, bull elephants that should diminish impact on emergent mopane trees.

Implications of structural transformation for supported biodiversity

Although an abundant mopane population has persisted in GNP in the face of chronic elephant utilisation, the potential consequences of dramatic transformation of vegetation structure for supported biota need to be considered because an aim for protected areas is to conserve all elements of indigenous biodiversity. Herremans (1995) showed an effect of elephant-transformed vegetation on avifaunal composition of comparable mopane woodland in northern Botswana, but this topic is essentially unstudied for mopane woodland. By contrast, structural change of miombo woodland in Zimbabwe by elephants had a pronounced negative effect on the richness of other plant species, birds, and some invertebrate taxa (Cumming et al., 1997). The purpose of this section is to collate selected facets of indirect evidence to propose that such biodiversity impacts are also considerable in mopane woodland, in support of further study.

Tall mopane trees possess stem cavities as a consequence of their heartwood disappearing with age (heart rot; Timberlake, 1995) which are used by a large number of vertebrate species for nesting or as a home. Many species of birds use these cavities for nesting, including hornbills, barbets, and chats (Hockey, Dean & Ryan, 2005). Tree squirrels Paraxerus cepapi are sometimes termed mopane squirrels because of their penchant for using this species as a home and place for breeding; in addition, the large seeds of mopane are an important constituent of their diet in mopane woodland (Skinner & Chimamba, 2005). On account of seed production of mopane being related to tree height (Timberlake, 1995), availability of mopane seed in hedged riverine mopane woodlands in GNP was low because only the relict emergent mopane trees, which occurred at a low density (Fig. 5), produced seed, pollarded trees did not (T O’Conner, pers. obs., 2022), although hedged mopane woodland produced seeds, albeit at a reduced amount, in Tuli Block, Botswana (Styles & Skinner, 2001). Emergent trees were also the only source of cavities, but there was a slow attrition of emergent trees primarily through debarking. The potential value of the mopane woodlands we studied for squirrels appears to have been compromised by a dramatic reduction in mopane seeds and a loss of breeding sites that is expected to impact on their numbers and on the numbers of the many predators which prey upon them. Use of tree cavities by reptiles is not as conspicuous as by birds, but some snake, skink, and agama species, as well as the rock monitor Varanus albigularis, do so (Alexander & Marais, 2007).

Caterpillars of the lepidopteran Imbrasia belina (Saturniidae; mopane worm) may exceed the biomass of elephants in semi-arid savannas during periods of irruption that affects ecosystem functioning and vegetation structure (Duffy, O’Connor & Collins, 2018; De Swardt, Wigley-Coetsee & O’Connor, 2018). However, chronic utilisation of mopane by elephants, among other factors, has been implicated in reducing their abundance or even their disappearance (Styles & Skinner, 1996; Hrabar & Du Toit, 2014), in part because of the disappearance of tall trees which are their preferred sites for laying eggs (Hrabar, Hattas & Du Toit, 2009b). Mopane worms are consumed by a suite of bird species that changes in composition as instars develop (Gaston, Chown & Styles, 1997)—reduction in the availability of mopane worms therefore has obvious ramifications for trophic patterns.

In summary, conversion of mopane woodland to hedged woodland is expected to have had far-reaching effects on the biodiversity supported by these woodlands through disruption of trophic and non-trophic linkages of mopane with faunal elements.