IAWA Journal, Vol. 14 (2),1993: 173-185

BARK STRUCTURE AND PREFERENTIAL BARK UTILISATION BY THE AFRICAN ELEPHANT

by

J ohn W. Malan I and Abraham E. van Wyk 2

1 Mammal Research Institute, Department of Zoology, 2H.G.W.J. Schweickerdt Herbarium, Department of Botany, University of Pretoria, Pretoria, 0002 Republic of South Africa

Swnmary Bark fracture properties are thought to In any particular area, elephants show a influence the debarking of selected trees by preference for stripping and ingesting the the African elephant. This hypo thesis was bark of certain tree species; bark of Acacia tested for large riverine tree species in the tortilis and A. nigrescens are especially fa­ Northern Tuli Game Reserve, Botswana. An voured. Elephants loosen the bark with their index of bark breakage strength and pliability tusks and then strip it off. Even if only apart of secondary phloem tissue was compiled for of the bark is stripped off, debarked trees are 11 common riverine species, and the bark susceptible to further damage by fire, insect anatomy of these species was investigated to borers and fungi, and often succumb to their determine relative fibrosity. The majority of injuries (Thomson 1975). species preferred by elephants have strong Little is known about the effects of bark and pliable barks, associated with a high pro­ structure and strength on the extent of debark­ portion of fibres. However, not all preferred ing. In fact, the possible influence of plant species have these characteristics, which in­ fracture properties on animal behaviour is dicates that factors other than bark fracture rarely considered (Vincent 1990). Thomson properties affect species preference. Bark (1975) suggested that the ease with which structure influences the way pieces of bark bark is stripped from a tree may account, in are stripped from a tree trunk during debark­ part, for the differences in species preference ing. It is hoped that this paper will stimulate shown by elephants. Laws et al. (1975) re­ further studies on the effects of bark structure marked that the mechanical properties of bark, on the preferential feeding behaviour of the which either facilitate or hinder debarking, African elephant. may be involved in determining the incidence Key words: African elephant, bark anatomy, of debarking of a particular tree species. debarking, gelatinous fibres, mechanical The present study was initiated following properties, sclereids. a suggestion that the breakage strength and pliability of bark play an important role in the Introduction preference shown by elephants for certain The devastating effects that large numbers tree species in southern Africa (McKenzie, of elephants have on their habitat are evident pers. comm.). In this paper we explore the in many parts of eastern and southern Africa relations hip between the intensity of bark (Anderson & Walker 1974). The relatively damage caused by elephants and aspects of large size and high mean age of survival en­ bark anatomy. We test the hypothesis that able this species to have a major, often long­ fibrous barks tend to be tough, pliable, and term, impact on the ecosystem (Watson & hence easily stripped from extensive areas of Bell 1968), particularly in areas where the stern, whereas barks in which the sclerenchy­ movements of elephant herds are restricted. ma is completely lacking or is comprised of Elephants are second only to man in their ca­ mainly sclereids, are brittle and therefore less pacity for altering their environment (Skinner likely to be removed in relatively large pieces. & Smithers 1990, and references therein). This study forms part of a comprehensive

Downloaded from Brill.com09/24/2021 10:28:38PM via free access 174 IAWA Journal, Vol.14 (2), 1993

project on the association between African located along the main rivers within the Re­ e1ephants and the large riverine trees of the serve and inc1ude the Majale, Pitsani, Njaswe, Northern Tuli Game Reserve in Botswana. Matabole and Jwala rivers. Two hundred large trees, both living and dead, were sampled at each site. Only large trees were included in Materials 8lld Methods the sampie as these constitute the most im­ portant structural component of the riverine Studyarea forests. Large trees were defined as trees The Northern Tuli Game Reserve (NTGR) with a trunk circumference exceeding 1 m is situated in the eastern corner of Botswana, when measured just above the basal swelling. between 21° 55' Sand 22° 15' S, and 28° For each living tree sampled, the species 55' E and 29° 15' E, where Botswana, South and extent of bark damage inflicted by ele­ Africa and Zimbabwe meet. It is bounded phants were recorded. Bark damage was de­ in the north by the Tuli Circ1e, in the south fined as the width of bark removed at the by the Limpopo and Motloutse rivers, in the height of greatest width damage, expressed east by the Shashe River, and in the west by as a percentage of the stern circumference at the Tuli Block backline, a veterinary control that height: fence. The Reserve, which covers an area of Percentage bark removed (%BR) = 100' Ai / C approximately 60,000 ha, is neither gazetted nor proc1aimed as a private game reserve, but where: relies solelyon a shared interest in conserva­ Ai = the width of a debarked area at tree height x tion between the landowners of several pri­ C circumference of tree at height x vately owned farms. From aerial censuses x = the height at which the greatest area of bark conducted in 1986, 1987 and 1988 (Le Roux has been removed. 1989), it is evident that the density of ele­ phants within the NTGR is approximately Choice 0/ species one per square kilometre. The vegetation of the area is broadly c1assified into Colopho­ An objective of this study was to distin­ spermum mopane woodland and scrub wood­ guish, on the basis of bark damage, preferred land (White 1983). woody plants from non-preferred ones. A pre­ ferred species is defined as one that is utilised more frequently than indicated by its avail­ Choice 0/ trees ability in the environment (Petrides 1975). A stratified sampie of the riverine tree com­ For this purpose preference ratios were cal­ munities within the NTGR, comprising 16 culated using the following equations, adapt­ sites, was taken in May 1991. The sites were ed from Petrides (1975):

Percentage utilisation (U i) Preference ratio (PR) Percentage availability (Ai)

where:

Number of debarked trees of species i in all sites sampled Ui = 100 • Number of debarked trees of all species within all sites sampled

Number of available trees of species i within all sites sampled Ai = 100 • Number of available trees of all species within all sites sampled

Downloaded from Brill.com09/24/2021 10:28:38PM via free access Ma1an & Van Wyk - Bark utilisation by the African elephant 175

For each tree, the time of debarking prior to Bark anatomy sampling (weeks, months, and/or years) was Fresh bark sampies, collected as described determined from the appearance of the expos­ above, were fixed in FAA (Johansen 1940). ed wood and/or stripped bark and/or signs Unembedded fixed material was softened of regenerative bark growth. From observa­ with steam and transverse and radial sections tions made on the response of trees from cut at 15 - 20 ~m on a sliding microtome. which we have removed bark (to test bark Sections were stained with safranin 0 and fast strength), the criteria were found to be a re­ green (Johansen 1940), and mounted in en­ liab1e measure of estimating the relative time tellan. The presence of lignin was determined of debarking. Preference ratios were calcu­ using phloroglucinol (Jensen 1962). Draw­ lated based on1y on bark damage that had ings of transverse sections were made using a occurred weeks and/or months prior to the projection light microscope. Unless other­ time of sampling (= recent damage) and only wise indicated, descriptive bark anatomy ter­ for species for which ten or more trees were minology follows Trockenbrodt (1990). sampled.

Bark strength Results The fracture properties of bark were tested Preference ratios for 11 NTGR riverine tree species. For each The number of large riverine trees sam­ species, bark from six randornly chosen trees, pled (both accessible and inaccessible to located along the Maja1e and Jwala rivers, elephants), and their percentage occurrence was collected in October 1991 (the end of the within the riverine tree communities of the dry season). The bark was removed from NTGR, are listed in Table 1. Based on bark each tree at breast height, and only from trees damage inflicted by elephants, preference with a trunk circumference of between 100 ratios calculated for species with ten or more and 200 cm when measured just above the trees sampled are given in Table 2. The me­ basal swelling. Bark sampled from the vari­ dian percentage bark removed from all trees ous species differed in thickness. For com­ of each species sampled (Table 3) was cal­ parative purposes, a standard 1 x 2 x 100 mm culated, and resuIts are plotted in Figure 1. axial strip of tissue, removed from the cen­ tral part of the living secondary phloem was Bark structure used in the following tests (n = 6 for each The predominant type of sclerenchyma in species). the secondary phloem of each investigated To measure breakage strength (stress at species is recorded in Tab1e 3. Sclerenchyma which the sampIe breaks), ahomemade wood­ arrangement, as seen in transverse section, is en clamp was attached to each end of a bark illustrated in Figures 2-13. Because the sec­ strip, leaving a 10 mm gap between each ondary sclereids that are associated with the clamp. The clamps were then pulled apart dilation zone are variable in occurrence and with a mini hand-wrench. The weight at degree of development, these cells were not which a bark strip snapped was read from a illustrated, nor taken into account for relating 100 kg weight scale and used to establish an sclerenchyma type to bark mechanical proper­ index of the breakage strength for the bark of ties, except in the case of Xanthocercis zam­ the species sampled. besiaca, where masses of secondary sclereids An index of pliability (toughness) for the form at an early stage. bark was based on the number of bends re­ Sclereids - Croton megalobotrys (Fig. quired to break a strip of bark, each strip of 13) is characterised by sclerenchyma consist­ bark being clamped as previously described. ing of scattered clusters (strands) of thick­ Abend is defined as a 900 change in the angle walled sclereids, each cluster wreathed by between the clamps (initially at an angle of crystalliferous cells. The two species of Com­ 1800 to each other), and the direction of suc­ bretum (Figs. 7,12) are also characterised by cessive bends is aItemated. thick-walled sclereids, though in this case the

Downloaded from Brill.com09/24/2021 10:28:38PM via free access 176 IAWA Journal, Vol. 14 (2),1993

Tab1e 1. Number of 1arge trees sampled (irrespective of their accessibility to elephants) and the percentage occurrence of each species within the riverine tree communities of the Northem Tuli Game Reserve.

Species No. of trees Percentage Species No.oftrees Percentage sampled occurrence sampled occurrence

Colophospermum mopane 617 22.91 Albizia brevifolia 15 0.56 Combretum imberbe 543 20.16 Boscia foetida 12 0.45 Croton megalobotrys 489 18.l5 Spirostachys qfricana 8 0.30 Lonchocarpus capassa 257 9.54 Cassia abbreviata 6 0.22 Combretum hereroense 190 7.05 Markhamia acuminata 6 0.22 Acacia tortilis 149 5.53 Lannea schweinfurthii 5 0.19 Schotia brachypetala 102 3.78 Ficus sycamorus 4 0.15 Acacia nigrescens 83 3.08 Berchemia discolor 4 0.15 Boscia albitrunca 53 l.97 Sc/erocarya birrea 3 0.11 Combretum apiculatum 37 l.37 Commiphora glandulosa 3 0.11 Terminalia pruniodes 26 0.97 Salvadora angustifolia 2 0.07 Xanthocercis zambesiaca 23 0.85 Hyphaene natalensis 0.04 Ziziphus mucronata 21 0.78 Pappea capensis 0.04 Acacia albida 16 0.59 Kirkia acuminata 0.04 Sterculia rogersii 16 0.59

Table 2. Characteristics of debarked species sampled within the Northern Tuli Game Reserve riverine tree communities. Damaged = number of trees with recent bark damage; Total = number oftrees Iiving excluding trees inaccessible to elephants; Ui = percentage utilisation; Ai = percent­ age availability; PR = preference ratios; * = trees for which bark, strength and pliability, as well as bark anatomy, were recorded.

Species Damaged Total Vi Ai PR

Acacia albida * 7 16 5.98 0.61 9.73 Schotia brachypetala * 20 100 17.09 3.84 4.45 Acacia nigrescens· 8 79 6.84 3.l1 2.20 Xanthocercis zambesiaca * 17 21 1.71 0.81 2.l2 Acacia tortilis· 11 147 10.26 5.65 l.82 Combretum imberbe* 33 536 28.21 20.59 l.37 Colophospermum mopane* 20 617 17.09 23.70 0.72 Combretum hereroense* 6 189 4.27 7.26 0.59 Lonchocarpus capassa· 5 256 4.27 9.87 0.43 Croton megalobotrys· 5 488 4.27 18.75 0.23 Boscia albitrunca * 0 52 0.00 2.00 0.00 Ziziphus mucronata 0 20 0.00 1.15 0.00 Combretum apiculatum 0 30 0.00 1.15 0.00 Albizia brevifolia 0 II 0.00 0.42 0.00 Bosciafoetida 0 12 0.00 0.46 0.00 Terminalia pruniodes 0 26 0.00 l.00 0.00

PR < 1: species neglected; PR = 1: species neither neglected nor preferred; PR > 1: species preferred.

Downloaded from Brill.com09/24/2021 10:28:38PM via free access Malan & Van Wyk - Bark utilisation by the African elephant 177

Table 3. Extent of debarking compared to fracture properties, method of bark removal by ele­ phants, and the predorninant type of selerenchyma in the secondary phloem (extent of debarking and symbols explained below). Numbers preceding species names refer to numbers used in Figure 1.

Species Extentof Breakage Bending Methodof Typeof debarking strength (kg) strength debarking sclerenchyma Av±SD Av± SD

1. Acacia albida 5 1.14 ± 0.38 4.00 ± 2.50 SS F(G) 2. Xanlhocercis zambesiaca 5 < 1 1.00 ± 0.00 G F (C) + S 4. Schotia brachypetala 4 3.oo±I.41 3.00 ± 1.50 SS F(G) 3. Acacia nigrescens 4 19.25 ± 3.54 60+ LS F(G) 5. Acacia tortilis 3 27.64 ± 5.51 60+ LS F(G) 9. Boscia albitrunca 3 < 1 1.38 ± 0.52 G F 7 Colophospermum mopane 2 13.38 ± 1.92 60+ SS ces 8. Lonchocarpus capassa 2 6.00 ± 1.31 41.00 ± 5.50 SS F(G) 10. Combretum hereroense 2 1.13 ± 0.35 1.25 ± 0.46 B S 11. Croton megalobotrys < I 1.38 ± 0.52 G S 6. Combretum imberbe < 1 1.00 ± 1.50 B S

Extent of debarking: 1 =slight damage (1-15%); 2 = mild damage (16-25%); 3 = moderate damage (26-35%); 4 =extensive damage (36-45%); 5 =severe damaged (46-99%); where (x-y%) = mean percentage bark removed per species.

Bending strength: Number of90° bends before breaking of standard bark strip (I x 2 x 100 mm).

Method of bark rem oval: B = comes off in squares or blocks; G = tusk placed on bark and groove formed as tree is debarked; SS = removed as short strips; LS = removed as long strips.

Predominant type of sclerenchyma in secondary phloem (excluding dilatation tissue): ces =Iignified chambered crystalliferous strands; F = phloem fibres (Iignified); F (C) = phloem fibres. excIusively cellulosic; F (G) = phloem fibres, exclusively or predominantly gelatinous; S = scIereids.

clusters of sclerenchyma are markedly in­ Lonchocarpus capassa (Fig. 10) and the three terspersed by relatively large, thin-walled species of Acacia (Figs. 2,4, 6). The thickness crystalliferous cells. Xanthocercis zambesiaca and degree of lignification of the outer SI and (Fig. 3), a species with unusual, non-lignified S2 layers of these gelatinous fibres show con­ fibres in the secondary phloem, has abundant siderable interspecific variation. These latter secondary sclereids in its dilatation tissue. two layers are relatively thick and lignified in Because the selereids in X. zambesiaca form S. brachypetala, L. capassa, and Acacia albi­ relatively elose to the vascular cambium, and da, but are thin and inconspicuously lignified are abundant, these selereids may significant­ in A. nigrescens and A. tortilis. ly affect the fracture properties of the bark Fibres exclusively cellulosic - Slender (see Table 3). fibres with very thin, unlignified walls are Fibres with some lignification - Robust, present in the seeondary phloem of Xantho­ heavily lignified phloem fibres with excep­ cercis zambesiaca. These unusual fibres are tionally thick walls (resembling selereids in easily overlooked in stained seetions and, transverse seetion) are present in Boscia albi­ beeause of the strong birefringenee of their trunca (Fig. 11). Phloem fibres with an unlig­ walls, are best observed under polarised op­ nified innermost G-Iayer (gelatinous fibres) ties. They prove to be exeeptionally tough predorninate in Schotia brachypetala (Fig. 5), and elastie during seetioning.

Downloaded from Brill.com09/24/2021 10:28:38PM via free access 178 IAWA Journal, Vol. 14 (2), 1993

percentage most frequently damaged by elephants are compared to certain bark anatomical features. 70 Note that the barks with the highest breakage .-.- strengths also demonstrate the highest bending w.l ..... ;.j 1191\4 .01 60 ,..... strengths. Thus, the barks of Acacia nigres­ {velal.ll'lOV11 cens, A. tortilis, Colophospermum mopane

50 ~ FOrM and Lonchocarpus capassa proved to be both • lcesj strong and tough. In contrast, the bark is par­ g SdlmICl, .. 40 F,tJt., ' ~..a.cl ticularly weak and brittle in Boseia albitrunca, the two Combretum species, Croton megalo­ 30 botrys, Xanthocereis zambesiaca, and Acaeia albida.

20 Discussion . ~ In the NTGR, elephants show a preference ~ . 10 for the bark of certain tree species, and this • preference is not related to the relative avail­ 0 " 2 3 4 5 6 7 8 9 10 11 ability of each species. In particular, elephants species show a marked preference for the bark of Schotia brachypetala, Acacia albidn and Xan­ Fig. 1. The median percentage bark removed thocereis zambesiaca, with percentage occur­ from all trees of each species sampled, com­ rences of 3.78%, 0.59% and 0.85% respec­ pared to the type of sclerenchyma in the bark. tively, whereas Croton megalobotrys, a re­ Numbers refer to the species listed in Table 3. latively common species with percentage occurrence of 18.15%, remains almost un­ damaged. It is important to ask whether this Lignijied chambered crystalliferous strands preferential utilisation is due to intentional - The sc1erenchymatous elements in the bark selection by the elephant or is merely a con­ of Colophospermum mopane (Fig. 8) are ar­ sequence of the mechanical fracture/removal ranged in broad concentric zones, alternating properties of the barks. Alternatively, both with concentric secretory ducts which are as­ factors may playa role. The matter is compli­ sodated with zones ofparenchymatic second­ cated by observations suggesting that the ary phloem tissue, thus imparting to the bark preferential choice oftrees by elephants might a stratified arrangement in transverse sections. change over an extended penod (Anderson & Although probably derived from fibre initials, Walker 1974). the sclerenchyma of this species is comprised Studies attempting to explain the prefer­ of mainl y lignified, chambered crystalliferous ence shown by elephants for certain species strands. have usually concentrated on the chemical composition (taste, smell, nutrition) of plant Fracture properties material. Following the differential destruc­ In Table 3, the breakage strength and pli­ tion of Brachystegia boehmii in the Chizarira ability of bark from the eleven tree species Game Reserve, Zimbabwe, Thomson (1975)

Figs. 2-7. The arrangement of sclerenchyma as seen in transverse section of the inner bark (before conspicuous dilatation growth). Fibres, or elements considered to be homologous to fibres, shown in black (specific type mentioned under species); sclereids stippled. - 2: Acaeia albida, lignified ge1atinous fibres. - 3: Xanthocercis zambesiaca, sc1ereids and scattered cellu­ losie fibres. - 4: Acaeia nigrescens, gelatinous fibres. - 5: Schotia brachypetala, lignified gela­ tinous fibres. - 6: Acaeia tortilis, gelatinous fibres. - 7: Combretum imberbe, sclereids. - Scale bars = 100 ~m (Figs. 4-6), 200 ~m (Figs. 2, 3,7).

Downloaded from Brill.com09/24/2021 10:28:38PM via free access Malan & Van Wyk - Bark utilisation by the African elephant 179

Downloaded from Brill.com09/24/2021 10:28:38PM via free access 180 IAWA Journal, Vol.14 (2),1993

a .1..

5 J • \""" ','J ' • ...... \ J...... r..:... .:. ., , ,~ . ~ . '., •• '1:.1 ..., r:": ·:r::·t:- ...... : :[1.t-:- :.': .. :.. : ~ ~ ,-.;, '::. ":r:~i':' ::. :.:. :.:. :: ':.

IJ.. .. :t,: ...... ' ' 2~"'" ::::,) j ,./" .. , ...~ ,

~. •• • • . o.. ••• . 0. o • • r- . .. '.': .')

["1:':+:' .:.:' :' ..::, , '"::' "" " .. ~.". ". ~ ~;.;.,;;; .. '" ,~.~

, 7 \,I";": ~ :~':.' '.:. "'X'.:.:.: ' :"';':'.~:: ~jj"'"': ':. ~ '," .r:: :.L ..... ',,',r' . ~ , • , J 1 I I 'T:"" I -

Downloaded from Brill.com09/24/2021 10:28:38PM via free access Malan & Van Wyk - Bark utilisation by the African elephant 181 attempted to establish factors for the elephants' Thomson (1975) speculated that ease of preference for this species above four other bark removal may playa role in preferential species in the area, namely lulbernardia glo­ bark utilisation by elephants. Two factors are bif/ora, Brachystegia glaucescens, Colopho­ likely to affect the amount of damage: 1) the spennum mopane and Acacia albida. How­ activity of the vascular cambium at the time of ever, detailed chemical analysis for several debarking, and 2) the cellular structure of the elements, crude protein and ether extract con­ particular bark. Regardless of its structural tent, failed to indicate any significant differ­ characteristics, bark removal undoubtedly ences. In a study analysing only tree foliage, would require less effort during periods of Jachmann (1989) concluded that the mature increased cambium activity, for example, at foliage selected by elephants was character­ the beginning of tree growth following the ised by a high mineral and sugar content. dry season, or during the growth flushes that Neglected species were high in total phenols usually accompany good rains. On the other and steroidal saponins, and were often poor­ hand, Einspahr et al. (1984) found that dor­ ly digestible due to high concentrations of mant season wood/bark adhesion in several lignin. Bark is generally rich in phenolic and North American hardwood trees was posi­ varlous other secondary compounds (Srivas­ tively correlated with percent phloem fibres, tava 1964), and therefore these substances as weIl as secondary phloem toughness and are unlikely to playa significant role in bark strength. Wood/bark adhesion was negative­ selection. ly correlated with the percent sclereids in the In our study, crystals of calcium oxalate bark. However, Thomson (1975) noted that were abundant in most bark sampies. Fur­ elephant damage in the Chizarira Game Re­ thermore, secretory ducts/cells, sure to con­ serve in Zimbabwe mainly takes place from tain a diversity of secondary compounds, are August through November, aperiod likely to present in many species barks, e.g., Colo­ overlap with the start of active tree growth. phospermum mopane, Lonchocarpus capassa Guy (1976) suggested that the increase in and the Acacia species. However, the prepon­ debarking in the late dry season was due to derance of gelatinous fibres, which have a increased translocation from the roots to the low lignin content, in several of the heavily new leaves. Because the porcupine Hystrix utilised barks may improve their digestibility. africaeaustralis preferentially debarks certain Elephants rnight also benefit from the calcium trees during the spring, Yeaton (1988) con­ contained in the crystals of calcium oxalate cluded that this animal prefers tree species which are so abundant in most barks. In a capable of translocating sugars to their sterns number of tree species, Laws et al. (1975) prior to the beginning of the rainy season reported a significant correlation between the (e.g., Dombeya rotundifolia, one of the first extent of debarking and bark calcium content. savanna tree species to flower in spring). In It has also been suggested that bark removal contrast, Miquelle and Ballenberghe (1989) by elephants is a response to nutritional re­ suggested the seasonal preference for bark quirement for a substantial proportion of stripping by the moose Alces alces may be woody or fibrous material in their diet (Laws merely due to the low availability of grass et al. 1975). and browse during certain times of year. The

Figs. 8-13. The arrangement of sclerenchyma as seen in transverse section of the inner bark (before conspicuous dilatation growth). Fibres, or elements considered to be homologous to fibres, shown in black (specific type mentioned under species); sclereids stippled. - 8: Colo­ phospermum mopane, lignified chambered crystalliferous strands, alternating with concentric secretory ducts. - 9 & 10: Lonchocarpus capassa, lignified gelatinous fibres; high magnification (= 10) showing sclerenchyma interspersed with clusters of secretory ducts. - 11: Boscia albi­ trunca, lignified fibres. - 12: Combretum hereroense, sclereids. - 13: Croton megalobotrys, sclereids. - Scale bars = 10 I!m (Fig. 10), 100 I!m (Figs. 9, 11, 13), 200 I!m (Figs. 8, 12).

Downloaded from Brill.com09/24/2021 10:28:38PM via free access 182 IAWA Journal, Vol. 14 (2),1993

size of the area of bark stripped from individ­ Our results on the mechanical properties ual trees by the grey squirrel, Sciurus caroli­ of fibrous barks show that there are marked nensis has been claimed to be strongly related differences in strength and toughness be­ to phloem volume, but not to sugar concen­ tween species. Generally, barks with gelati­ tration (Kenward & Parish 1986). nous fibres and slight walllignification, as in Buechner and Dawkins (1961) suggest Acacia nigrescens and A. tortilis, proved to feeding on bark by elephants is concentrated be very strong and tough. The bark of Colo­ during the period of regrowth when the bark phospermum mopane, with concentric zones will peel easily. The frequency with which of abundant crystalliferous strands, but with grey squirrels remove bark from trees is very thin, lignified secondary cell walls, largely dictated by the ease with which they showed similar mechanical qualities. The can remove it (Hampshire 1985, as quoted in gelatinous phloem fibres in the bark of the Vincent 1990). In a study on a number of tem­ investigated species are not associated with perate Northern Hemisphere trees, the tough­ leaning sterns. Tension wood, which occurs ness of the cambium has been found to be re­ in many angiosperm trees at the upper side of duced by a factor of five or more when the cells leaning sterns and branches, is characterised are actively dividing (Vincent 1990). How­ by the presence of gelatinous fibres. Some ever, as this will probably affect all species of authors suggest the development of gelati­ trees to more or less the same extent, increased nous fibres is initiated by the large tensile cambium activity on its own is unlikely to ex­ stresses generated in the xylem by gravity plain preferential bark stripping by elephants. (Cote et al. 1969; Boyd 1977 and references Structurally, bark can be broadly inter­ therein, but see also Höster & Liese 1966; preted as consisting of various proportions of Chalk 1983). Bark tissue found in straight soft (parenchymatous) and hard (sclerenchy­ trunks, however, also generates longitudinal matous) elements. The basis for the particular tensile stresses as a result of the growth of mechanical properties of bark is therefore xylem and phloem at the vascular cambium. complex, undoubtedly involving engineering Because of the sporadic occurrence of gela­ principles associated with composite materials tinous fibres in bark, growth tensile stresses (Harris 1980). The most obvious structural are probably not the primary cause of their variations between mature bark of different initiation. Our results suggest that, unlike species usually involve aspects such as the wood, the presence of gelatinous fibres in the composition, distribution and amount of scle­ bark of a particular species is consistent, and renchyma tissue. The most common scleren­ of potential taxonomie significance (recorded chyma cell types are fibres and sclereids (for during this study in members of the Mimosa­ definitions see Trockenbrodt 1990). ceae, Caesalpiniaceae and Fabaceae only). The mechanical properties of bark are vir­ In spite of the presence of exclusively cel­ tually unknown (Diener et al. 1968; Murphy lulosic fibres, the bark of Xanthocercis zam­ & RisheI1977; Einspahr et al. 1984). In one besiaca turned out to be weak and brittle. of the few studies on bark strength, Einspahr During the preparation of the bark sections, et al. (1984) presented evidence suggesting these fibres proved to be tougher and more that bark strength is increased by the pres­ resilient than any of the other species inves­ ence of fibres, but decreased by sclereids. tigated, making it impossible to obtain intact Murphy and Rishel (1977) also reported that transverse sections. The abundant secondary fibrous barks are much stronger in compres­ sclereids in this species may affect the mechan­ sion than the conglomerate (sclereidal) types. ical properties of this species. Despite the paucity of quantitative evidence, In the remaining fibrous barks, strength the toughness and strength of certain fibrous and toughness appear to decline with an in­ barks are taken advantage of by man, e.g., crease in the thickness of the lignified portion for binding material (cord) - examples from of the fibre walls. This may explain the rela­ southern Africa include the bark of several tively brittle bark in Acacia albida and Sclwtia species of Acacia, Brachystegia and Coloplw­ brachypetala; both have gelatinous fibres, but spermum mopane (Palmer & Pitman 1972). with the SI and S 2 layers conspicuously lig-

Downloaded from Brill.com09/24/2021 10:28:38PM via free access Malan & Van Wyk - Bark utilisation by the African elephant 183 nified. Nanko and Cöte (1980) have suggest­ new prospects for interdisciplinary co-oper­ ed that the thin secondary walls and large ation. Such research will not only be of sig­ lumina in phloem fibres are features that will nificance in the field of animal behaviour, but generally enable them to collapse and become may also help in elucidating some of the flexible. The very thick and lignified second­ evolutionary pressures to which plants are ary walls of the fibres in Boscia albitrunca exposed. In the process even engineers may (lumina almost absent) probably account for benefit - biological designs have already been the relatively brittle bark in this species. In successfully transferred to engineering appli­ the case of wood, however, mechanical cations (e.g., Mattheck 1989). strength (hardness) largely depends on the proportion of fibres and the thickness of cell Conclusions walls, while bending and toughness qualities Species with bark that is relatively strong cannot, at least in some species, be specific­ and pliable are preferred by elephants, but the ally referred to features of anatomical struc­ reverse is not necessarily true. For example, ture (Wilson & White 1986). Acacia albida, Xanthocercis zambesiaca and All the investigated barks in which scler­ Schotia brachypetala, although markedly pre­ eids predominate are, as expected, weak and ferred by elephants, have barks that demon­ brittle. For these species (notably species of strate a relative low pliability and small break­ Combretwn), elephants tend to remove small age strength. It appears that, in addition to pieces of bark from around the circumference bark strength, factors such as chemical com­ of the trunk, so that ring-barking is the prin­ position, moisture content and nutritive value cipal form of damage. Species often ring­ of bark affect species choice. There is a ten­ barked in this manner include C. imberbe, dency for fibrous barks to predorninate among Acacia albida and Trichilia emetica (Anderson the tree species with the highest preference & Walker 1974). For tree species with fibrous ratios. However, fracture properties (breakage barks, bark stripping by the elephant often and bending strength) of these barks differ takes the form of ripping long strips of bark and therefore do not show a dear and consis­ from the trunk. In the Sengwa Wildlife Re­ tent pattern. This variation is probably due to search Area, Anderson and Walker (1974) specific characteristics of the phloem fibres, found this type of debarking to be most pro­ notably thickness and degree of lignification nounced in Acacia tortilis (which was marked­ of the secondary cell walls, and the presence ly preferred by elephants), Acacia nigrescens or absence of an inner gelatinous walllayer. and Brachystegia boehmii. Bark structure has adefinite influence on In contrast to wood (e.g., leronimidis the way in which pieces of bark break away 1976,1980; Boyd 1980; Vincent 1990), there from the trunk during debarking. Fibrous have been few attempts to relate the mechani­ barks with high breakage and bending strength cal properties of mature bark tissue to the mor­ are removed in relatively long axial strips, re­ phology of the cell elements. The approach sulting in relative easy removal of large quan­ followed in our study allows only generalisa­ tities of bark from a trunk. Brittle barks, and tions. Vincent (1990) stressed the necessity in partieular those where sclereids predorni­ to use 'proper' engineering and materials nate in the bark sderenchyma, break away in science methods to measure the fracture prop­ small pieces, resuIting in the damage being erties of plant tissues. The method used here localised on the stern. These trees are prone was, however, chosen for its simplicity and to be ring-barked by the elephant. served mainly as an index to be used for Comparative evidence on bark structure is comparative purposes. lacking for the vast majority of African trees. There is a need for more studies on the It is hoped that this paper will serve as a basis relationship between fracture properties and for further studies on the possible influence anatomieal structure in plants. The appliea­ of bark structure on the preferential feeding tion and refinement of methodologies previ­ behaviour of the African elephant, and its ef­ ously mainly developed and employed by fects on the dynamics of specific plant com­ materials scientists, should open exciting munities.

Downloaded from Brill.com09/24/2021 10:28:38PM via free access 184 IAWA Journal, Vo!. 14 (2),1993

Acknowledgments Einspahr, D.W., R.H. van Eperen & M.L. We would like to thank His Excellency, Fiscus. 1984. Morphological and bark Dr. Quett D. J. Masire, President of the strength characteristics important to wood/ Republic of Botswana and the landowners of bark adhesion in hardwoods. Wood & the Northern Tuli Game Reserve for extend­ Fiber Sci. 16: 339-348. ing their permission to conduct the project. Guy, P.R. 1976. The feeding behaviour of Our sincere thanks to Mr. M. J. Potgieter, elephant (Loxodonta africana) in the Seng­ who prepared the microscope slides used for wa Wildlife Research Area, Rhodesia. S. this study, and to Miss G.L. Day, who criti­ Afr. J. Wildl. Res. 6: 55-63. cally read and improved the manuscript. The Hampshire, R.J. 1985. A study in the social senior author would like to express his grati­ and reproductive behaviour of captive grey tude towards Prof. J.D. Skinner for financial squirrels (Sciurus carolinensis). PhD The­ assistance, Dr. AA McKenzie for scholarly sis, University of Reading. advice and for suggesting this study, and the Harris, B. 1980. The mechanical behaviour McKenzie and Steyn families for their gener­ of composite materials. In: F.V. Vincent ous hospitality. Financial support from the & J.D. Currey (eds.), The mechanical University of Pretoria and the Foundation for properties of biological materials: 38-74. Research Development is gratefully acknowl­ 34th Symp. Soc. Exper Bio!. Cambridge edged. University Press, Cambridge. Höster, H.R. & W. Liese. 1966. Über das References Vorkommen von Reaktionsgewebe in Anderson, G.D. & B.H. Walker. 1974. Wurzeln und Ästen der Dikotyledonen. Vegetation composition and elephant dam­ Holzforschung 20: 80-90. age in the Sengwa wildlife research area, Jachmann, H.. 1989. Food selection by ele­ Rhodesia. J. Sth. Afr. Wild!. Mgmt Assoc. phants in the 'Miombo' biome, in relation 4: 1-14. to leaf chemistry. Biochem. Syst. Ecol. 17: Boyd, J. D. 1977. Basic cause of differen­ 15-24. tiation of tension wood and compression Jensen, W.A 1962. Botanical histochemis­ wood. Austral. For. Res. 7: 121-143. try. W. H. Freeman & Co., San Francisco. Boyd, J.D. 1980. Relationship between fibre Jeronimidis, G. 1976. The fracture of wood morphology, growth strains and physical in relation to its structure. In: P. Baas, properties of wood. Austral. For. Res. 10: A.J. Bolton & D.M. Catling (eds.), Wood 337-360. structure in biological and technological Buechner, H.K. & H.C. Dawkins. 1961. research. Leiden Bot. Series 3: 253-265. Vegetation changes induced by elephants Leiden University Press, Leiden. and fire in Murchison Falls National Park, Jeronimidis, G. 1980. Wood, one ofnature's Uganda. Ecology 42: 742-766. challenging composites. In: F.V. Vincent Chalk, L. 1983. Fibres. In: C.R. Metcalfe & & J.D. Currey (eds.), The mechanical L. Chalk (eds.), Anatomy of the dicotyle­ properties of biological materials: 169- dons 2: 28-38. Clarendon Press, Oxford. 182. 34th Symp. Soc. Exper Bio!. Cam­ Cote, W.A, A.C. Day & T.E. Timei!. 1969. bridge University Press, Cambridge. A contribution to the ultrastructure of ten­ Johansen, D.E. 1940. Plant microtechnique. sion wood fibers. Wood Sci. Technol. 3: McGraw-Hill Book Co. Ine., New York. 257-271. Kenward, R.E. & T. Parish. 1986. Bark­ Diener, R.G., J.H. Levin & B.R. Tennes. stripping by grey squirrels (Seiurus earo­ 1968. Directional strength properties of linensis). J. Zoo!., Lond. (A) 210: 473- cherry, apple, and peach bark and the in­ 481. fluence of limb mass and diameter on bark Laws, R.M., LS.C. Parker & R.C.B. damage. Trans. Am. Soc. Agric. Engrs. Johnstone. 1975. Elephants and their habi­ 11: 788-791. tats. Clarendon Press, Oxford.

Downloaded from Brill.com09/24/2021 10:28:38PM via free access Malan & Van Wyk -- Bark utilisation by the African elephant 185 ~------~------

Le Roux, P.G. 1989. Report on the 1989 Srivastava, L.M. 1964. Anatomy, chemistry, aerial game census. Unpublished report, and physiology of bark. In: J.A. Rom­ Mashatu Game Reserve. berger & P. Mikola (eds.), International Mattheck, C. 1989. Engineering components review of forestry research: 203-277. grow like trees. Report no. 4648, Karls­ Academic Press, New York. ruhe Nuclear Research Centre, Karisruhe. Thomson, P.J. 1975. The role of elephants, Miquelle, D.G. & V. van Ballenberghe. 1989. fire and other agents in the decline of Impact of bark stripping by moose on as­ Brachystegia boehmii woodland. J. Sth. pen-spruce communities. J. Wildl. Mgmt. Afr~ Wildl. Mgmt Ass. 5: 11-18. 53: 577-586. Trockenbrodt, M. 1990. Survey and discus­ Murphy, W.K. & L.E. Rishel. 1977. Prop­ sion of the terminology used in bark anat­ erties and potentials of bark as a raw ma­ omy. IAWA Bull. n.s. 11: 141-166. terial. Progress Report 363, Agricultural Vincent, J.F.V. 1990. Fracture properties of Experimental Station, Pennsylvania State plants. Adv. Bot. Res. 17: 235-287. University, Pennsylvania. Watson, R.M. & R.H.V. Bell. 1968. The Nanko, H. & W.A. Cöte. 1980. Bark struc­ distribution, abundance and status of ele­ ture of hardwoods grown on southern pine phant in the Serengeti region of northern sites. Syracuse Univ. Press, Syracuse. Tanzania. J. Appl. Eco!. 6: 115-132. Palmer, E. & N. Pitman. 1972. Trees of White, F. 1983. The vegetation of Africa Southern Africa, vo!. 1. A. A. Balkema, (and accompanying Unesco/ AETFAT / Cape Town. UNSO vegetation map of Africa). Unes­ Petrides, G.A. 1975. Principal foods versus co, Paris. preferred foods and their relations to stock­ Wilson, K. & D.J.B. White. 1986. The ing rate and range condition. Bio!. Con­ anatomy of wood: its diversity and varia­ servo 7: 161-169. bility. Stobart & Son Ltd., London. Skinner, J.D. & R.H.N. Smithers. 1990. Yeaton, R.I. 1988. Porcupine, fires and the The mammals of the southem African sub­ dynamics of the tree layer of the Burkea region, ed. 2. Univ. of Pretoria, Pretoria. africana savanna. J. Ecol. 76: 1017-1029.

Downloaded from Brill.com09/24/2021 10:28:38PM via free access