Utilization by elephants of the Brachystegia woodlands of the Kasungu National Park, Malawi

Utilization by elephants of the trees of the Brachystegia woodlands of Kasungu National Park, Malawi, was investigated. Of forty-one common species thirtyfive species were eaten, of which thirteen species were selected by elephants. The chemical composition of the leaf material was analysed and a significant correlation was found between the utilization of certain species and the protein and sodium content, whereas the crude fibre content showed no significant correlation but in general appeared to be relatively low in highly favoured species. The pushing over and uprooting of trees by elephants appear to be part of a feeding strategy which improves the availability of food for elephants during the dry season. The number of trees browsed increased with an increasing tree density up to 300 trees ha^-1, where browsing intensity remained constant for both selected and non-selected species. The reasons why the species composition of Brachystegia woodlands is hardly affected by elephant use are briefly discussed.     

Elephant herbivory in Majete Wildlife Reserve,Malawi

Elephants have a major influence on vegetation structure, composition and ecosystem processes, and are primary agents of habitat change in Africa. At moderate‐to‐high population densities, elephants can damage vegetation, especially when enclosed in protected areas. This study examines the effects of elephant browsing on woody trees in Majete Wildlife Reserve (WR), south‐western Malawi. Regression analysis is used to assess the associations of six factors known to drive elephant browsing in other areas and determine which ones have the most influence on browsing at Majete WR. Twenty‐four per cent of tagged trees had been subject to elephant browsing. The model with vegetation type, stem diameter and distance from permanent water correctly predicted browsing for 80% of the observations. Elephants mostly favoured riparian woodlands, followed by Acacia‐dominated woodland and Brachystegia‐dominated woodland. Browsing occurrence was negatively related to distance from permanent water and diameter at breast height(DBH). A larger number of trees, sampled at random and covering a larger portion of the reserve would provide more reliable estimates of browsing and related factors. Knowledge of time‐ and site‐specific factors affecting elephant browsing can be used to forecast future habitat transformations and manipulate the range of the elephants within the reserve.     

Effects of large herbivores and fire on the regeneration of Acacia erioloba woodlands in Chobe National Park, Botswana

Acacia erioloba woodlands provide important forage and shade for wildlife in northern Botswana. Mortality of mature trees caused by browsing elephants has been well documented but the lack of regeneration of new trees has received little attention. Annual growth of new shoots and changes in height were measured to determine the influence of elephants and small ungulate browsers, rainfall and fire on the growth and survival of established A. erioloba seedlings from 1995 to 1997 in the Savuti area of Chobe National Park. All above‐ground vegetation was removed from 40% of established seedlings in 1995 and 28% in 1997 by browsing elephants, and the mean height of remaining seedlings decreased from >550 mm to <300 mm. When seedlings browsed by kudu, impala and steenbok but not elephants are considered, mean seedling height increased <50 mm per year, even though mean new shoot growth remaining at the end of the dry season was 100–200 mm. Fires burned portions of the study area in 1993 and 1997, killing above‐ground vegetation, but most established A. erioloba seedlings survived, producing coppice growth from roots. While elephants and fire caused the greatest reduction in established seedling height and number, small browsers suppressed growth, keeping seedlings vulnerable to fire and delaying growth to reproductive maturity.     

Effect of low rainfall and browsing by large herbivores on an enclosed savannah habitat in Kenya

Savannah ecosystems in East Africa are rarely stable and can experience rapid local changes from dense woodlands to open plains. In this 3‐year study there was a reduction of 16.3% in a height‐stratified sample of nearly 1000 individually marked Acacia drepanolobium trees. The study was carried out in an enclosed fire‐free wooded grassland habitat in the Laikipia region of Kenya. The trees were monitored from 1998 to 2001, a period that included 12 months when rainfall was 60% below average. Elephants were responsible for the loss of 40% of the trees, black rhinos 33% and 27% died from the effects of the drought. Low rainfall was correlated with increased damage as elephants switched diet from grass to trees. Heavy browsing by giraffes reduced tree growth rates and increased their susceptibility to drought. Hence the combination of low rainfall and heavy browsing by elephants, black rhinos and giraffes led to the rapid tree loss. These findings have implications for research into the causes of instability in savannah ecosystems and the management of enclosed reserves.     

Selective disturbance by elephants (<Emphasis Type="Italic">Loxodonta africana</Emphasis>) on eight tree species in a West African savannah

African elephants (Loxodonta africana) are ecosystem engineers in African savannahs, but their role in the modifications of the populations of trees by means of their browsing activities has been poorly studied so far in West Africa. We studied the disturbance of elephants to eight selected species of trees in the Nazinga Game Ranch (Burkina Faso), in 54 transects at the end of dry season 2008. We fitted simple models describing the number of dead individuals for each tree species in relation to their initial population density, and in relation to dung-pile density, in the assumption that there should be a positive relationship between elephant density and dung-pile density. Generalized Linear Model analyses and regression analyses showed that the more dense the woodland cover, the least the percentage of damaged plants by elephants. For each plant species, the initial density and density of elephant dung-piles explained a high proportion of the variance in the density of dead individuals. Stochastic models, generated by a purposely created simple computer program written in GW-BASIC programming language, predicted changes in tree and shrub abundance under different assumptions about elephant numbers. The models suggested that elephant browsing may cause considerable change in the selected plant populations, especially with regard to such species as Acacia gourmaensis, Vitellaria paradoxa and Maytenus senegalensis. These changes may possibly increase the meat harvest from controlled hunting activities, thus improving the income for surrounding communities.     

Woody vegetation damage by the African elephant during severe drought at Pongola Game Reserve,South Africa

Elephants (Loxodonta africana) significantly alter ecosystem structure and composition through browsing (e.g. pollarding, debarking and toppling). Such browsing is predicted to intensify during severe drought which may become more common with climate change. Here, we make use of an elephant impact survey from 2012 to 2015 and during the El Nino drought of 2015–2016 at Pongola Game Reserve (107 km²), KwaZulu-Natal, to investigate how severe drought influenced damage severity of different tree heights and species by elephants in this small reserve. Contrary to expectations, damage to common species did not change with severe drought. Crown damage had the highest predicted probability across heights (29%–90%) and species (46%–75%) regardless of drought. However, we found severe drought increased the predicted probabilities of crown damage to smaller trees <4 m, mortality >6 m and severe damage at 4–6 m. Consequently, elephant damage during severe drought may alter vegetation structure by severely damaging or killing large trees (>4 m) and extensively damaging the crowns of trees <4 m. Long-term monitoring of elephant effects on woody vegetation is essential to enable science-based management in response to future drought and elephant damage (e.g. range expansion, beehive deterrents) to protect elephants and conserve woody vegetation.     

Effects of natural and simulated herbivory on spine lengths of Acacia drepanolobium in Kenya

We present experimental evidence supporting the hypothesis that increased spine length in acacia species is a defense induced by herbivory. Acacia drepanolobium is the dominant tree over large areas of East Africa. Each individual tree is occupied by one of four ant species at our study site. Using two types of electric fences, we have effectively controlled herbivory by megaherbivores (elephants and giraffes) and other large mammalian herbivores at a field site in Laikipia, Kenya since 1995. Mean spine lengths of new spines on trees occupied by the most abundant ant species (presumed to be a defensive mutualist) have shown a slow and steady decline over the first five years of the experiment on branches protected from these herbivores. This reduction has been 35–40%, or approximately half of the reduction in spine length that we anticipate will eventually occur, based on trees that have been protected from herbivory for many years. In contrast, trees occupied by a resident ant species that systematically prunes shoots have shown no reduction in spine length associated with herbivore exclusion treatments. Experimental pruning of shoots similar to that carried out by this ant species resulted in longer spines on seedlings in a greenhouse setting. Simulated large mammal browsing in the field rapidly (re-)induced greater spine lengths on trees that had been protected from large mammals for five years. The slow relaxation of spine length in the absence of herbivory, contrasted with its rapid induction after simulated browsing, suggests that there is a difference in the reliability of these two signals. Spine length responses to herbivory were extremely local (limited to individual branches). These branch-specific responses are consistent with the hypothesis that induced defense in this system evolved in the context of within-tree spatial variation in herbivore pressure, in particular variation in branch height.     

西双版纳野象谷景区亚洲象取食作用对五种植物的影响

摘要：本研究在西双版纳利用样线调查法调查亚洲象的取食作用对野象谷景区常见分布的5种植物即酸苔树、象鼻藤、粗糠柴、翅果麻和黄牛木的影响,调查中记录样线上每一目标植株的基径、高度、影响类型、影响程度等信息。结果表明：1）植物受影响的数量与其植株总数之间的相关性不大（Pearson correlation, r=0.608, P=0.277）;2）亚洲象对5种植物的偏爱程度不一样,酸苔树、象鼻藤、粗糠柴和翅果麻都是大象喜爱的食物,尤其是象鼻藤（偏爱指数=1.4855）,而黄牛木不受大象喜爱（偏爱指数=0.5855）;3）亚洲象对5种植物有不同的影响方式和影响程度,酸苔树以主干折断和枝叶损失为主,受大象的取食影响最严重,象鼻藤虽然是大象最喜爱的食物,但受到的取食影响最小,粗糠柴以枝叶损失和主干折断为主,翅果麻以枝叶损失为主,同时也是唯一被亚洲象剥皮的植物,黄牛木以枝叶损失为主,被推倒的情况最普遍。树干基部直径为3-8cm的植株最受大象偏爱。亚洲象对植物的选择性以及植株自身的特性是导致上述差异的主要原因。建议后续开展长期的调查以了解亚洲象对这些植物种群数量及群落结构的影响,同时应采取有效保护措施以避免象群在当地过于集中的现象。     

How does a re-colonizing population of Asian elephants affect the forest habitat?

The Asian elephant Elephas maximus is currently re-colonizing the Bardia National Park in lowland Nepal. We studied their impact on woody vegetation in the nutrient-rich floodplain and in the relatively nutrient-poor sal forest. The types and extent of tree impact were recorded along fixed-width transects (335 km). Species composition, density and size classes ≥8 cm diameter breast height (dbh) were recorded in 15-m radius random plots ( n =95). Impact was higher in the floodplain complex than in the sal-dominated forest. Our hypothesis that elephants were more selective on species in the nutrient-poor sal forest was only partly supported; the niche breadth of impacted trees was slightly higher in the floodplain complex. Pushed-over trees accounted for the highest proportion of impact (55%), followed by killed trees (39%). Of the pushed trees, 10% were not used for food. Among food trees, elephants selectively impacted size class 12–16 cm dbh, whereas non-food trees were impacted independently of size. A large proportion of the freshly browsed trees had been felled previously, indicating that most felled trees survived, enabling elephants to feed on them again. This may reflect an evolutionary adaptation among long-lived species with high site fidelity. Owing to preferential use but low abundance, two species in sal forest, Grewia spp. and Desmodium oojeinense , were found to be particularly vulnerable to local extinction due to elephants. Although the elephants had impacted a large number of species (62, 73% of all), 56.4% of the impacted trees consisted of Mallotus phillippinensis . A recently observed increase in the density of M. phillippinensis and the concurrent reduction of the hardly utilized Shorea robusta indicates that the rapidly growing elephant population may modify the composition of the forest by increasing its preferred food species.     

Long‐term response of temperate canopy trees to removal of browsing from an invasive arboreal herbivore in New Zealand

Defoliation of forest tree canopies by herbivores and other agents, leading to tree mortality and reduced productivity, threatens the ecological stability of forests globally. This study shows that long‐term control of a mammalian arboreal folivore (brushtail possums; Trichosurus vulpecula Phalangeridae) reduces crown dieback and increases foliage cover in browsing‐damaged canopy trees. We monitored indices of possum density, possum browsing, tree foliage cover and crown dieback for 20 years following initiation of possum control in 1994 that repeatedly reduced possum densities to near zero every 5–6 years and kept the population below 35% of pre‐control levels over the entire period. Observable possum browsing was recorded on 20–49% of individuals of three palatable tree species at the time of first control. Those percentages fell to zero after control and never exceeded 2–10% for individual species over the next 19 years. We recorded significant increases in foliage cover attributable to recovery from defoliation by possums for all three species during the first 10 years. Large increases in foliage cover occurred on individuals that were heavily browsed in 1994 (mean increases: 36–89%), but mean population increases were modest (3–19%) because only 10–19% of trees were initially heavily browsed. Twenty‐year mortality rates were similar for plants with, or without, initial possum browsing, indicating no residual impact of pre‐control browsing on tree mortality. Times for full recovery of crown foliage cover varied from 10 years for the youngest trees and faster growing species to more than 20 years for mature individuals of the slowest growing species.     

Do browsing elephants damage female trees more?

Herbivory in unisexual plants has mainly been assessed in studies of invertebrates, mostly insects. At a different scale of disturbance, browsing by large animals has rarely been studied. Elephants can seriously damage or kill trees by removing parts of the canopy, debark, or break the stem. We compared browsing by African elephants (Loxodonta africana) in male and female Marula trees (Sclerocarya birrea ssp. caffra). We expected females to incur more damage because of their sweet and fleshy fruits, known to attract elephants from far. Following this, we expected females to give more resistance through a denser branch wood that would break off at a larger diameter than in males. We tested this by pulling branches down to simulate elephant browsing as to compare branch resistance and design. All trees showed signs of browsing and females were generally more injured. However, there was no sexual difference in wood density and diameter at breaking point. Female branches were shorter, less ramified and carried fewer reproductive shoots than male branches. This sexual dimorphism could result from costly reproductive allocation in females and sexual selection, promoting intensive male flowering. Indirectly, elephants could reinforce this difference by engaging females more in tissue repair after browsing.     

Megafaunal influences on tree recruitment in African equatorial forests

The forests of central Africa are distinct from counterpart forests in Amazonia by having fewer trees ≥ 10 cm dbh ha^?1, especially small trees < 20 cm dbh, and in having sapling cohorts with less diversity than canopy trees. We tested four hypotheses to investigate whether herbivory, in particular, browsing by forest elephants, could be a factor in these differences. We found that the density of small saplings and diversity of large saplings were inversely associated with local density of elephants. We then tested the hypothesis that steep slopes might serve as refugia from elephant foraging, but found that elephants routinely forage on slopes with an inclination of less than ca 30 deg. Nevertheless, the diversity of small trees (≥ 10 cm, < 20 cm dbh) was higher on slopes than on paired level‐ground sites. The incidence of break scars on saplings ≥ 2 cm dbh and < 6 cm dbh was greater (107/100 stems) on level ground than on slopes (77/100 stems), although high variability precluded statistical significance. After correcting for background breakage not caused by elephants, an estimated 71% of breaks on level ground and 43% of breaks on slopes were attributable to elephants. Liana loads borne by trees at different sites were highly variable and unrelated to slope. Apparently, disturbances are more critical to liana development than herbivory. Elephants, along with other large mammals such as gorillas, duikers, red river hogs and rodents, appear to act as powerful filters on the tree recruitment process in African forests that still retain intact megafaunal communities.     

Inter- and intra-plant variations in nitrogen,tannins and shoot growth of <Emphasis Type="Italic">Sclerocarya birrea</Emphasis> browsed by elephants

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

Defensive traits of savanna trees – the role of shoot exposure to browsers

To investigate patterns of plant defence, a study was conducted on 13 tree species in Botswana. We tested the hypotheses that (1) shoots on small, young trees are more defended than shoots at the same height on tall, mature individuals and (2) shoots within browsing height are more defended than shoots above the reach of browsers, on mature trees. Plant traits studied were total phenolics, tannin activity, nitrogen (N), acid-detergent fibre (ADF), and degree of spinescence (in three species). We found some support for the first hypothesis. Overall young trees had a significantly higher concentration of total phenolics than mature trees and on species level these differences were significant for three of the 13 species. However, we found no differences in tannin activity, spinescence or ADF concentration between young and mature trees, and four species had higher levels of N in young trees, suggesting higher overall palatability. Contrary to our second hypothesis, shoots within browsing height were less chemically defended (total phenolics and tannin activity) than shoots above browsing height in six of 13 species. Two species had less ADF in lower shoots, and no significant differences were found in N or spinescence for any species. This study included a broad range of tree species at two sites and the results indicate that general theories of plant defence, originating from boreal and northern temperate forests, cannot be simply applied to savanna ecosystems.     

Changes in Elephant Abundance Affect Forest Composition or Regeneration?

While overall numbers of African elephant have declined dramatically in recent times, some populations are now confined to protected areas and are locally overabundant—an undesirable situation for both biodiversity conservation and elephants. In forested protected areas, options to manage elephants are limited because it is difficult to safely approach animals, yet it is vital that these populations are managed because browsing by elephants can dramatically alter forest ecosystems. Using data collected over 50 yr in Kibale National Park, Uganda, we examine the prediction that increasing elephant numbers and associated changes in their foraging behavior have caused a shift in tree community composition. Although the relative abundance of elephants increased significantly between 1996 and 2010, the population structure of their preferred tree food species did not change, nor did tree community composition change in favor of species able to re‐sprout after elephant damage. Furthermore, over the last 50 yr Kibale elephants have not become more selective foragers, as would be expected if more nutritious tree species were declining. However, elephants are more abundant in disturbed areas dominated by shrubs and grasses and appear to have arrested forest succession in these areas. At their current abundance, elephants have not selectively altered the composition of intact old growth forest, but they do inhibit the regeneration of disturbed areas.     

Observations of tree growth, woodland structure and elephant damage on Colophospermum mopane in Luangwa Valley, Zambia

Six samples of tagged Colophospermum mopane were monitored for five years in locations with varying soil characteristics but with similar elephant densities. Physiognomic variation among the samples was related to soil differences, which also correlated with different browsing habits by elephants. The impact of elephant browsing further influenced both the physiognomy and demography of C. mopane. Results from this study suggest that the influence of soils and elephants on C. mopane alter successional transitions from grassland to woodland. Soils that promote coppicing of C. mopane yield less stable woodlands when associated with elephants than soils promoting woodlands with large bolus, non-coppicing trees. The dynamics of the latter are determined more by tree recruitment as influenced by such agents as other browsers or frequency and seasonality of bush fires. Implications for forest/elephant management are discussed.     

Livestock reduces juvenile tree growth of alien invasive species with a minimal effect on natives: a field experiment using exclosures

Many alien invasive tree species were originally introduced to their non-native ranges for use in forestry and as urban trees. These alien species were selected for their fast growth and not necessarily for possessing mechanisms which deter browsing. Instead, many tree species native to semiarid areas of the world evolved mechanisms which deter browsing, presumably at the cost of slower growth. In a semiarid rangeland we observed that livestock exclusion greatly promoted the growth of juveniles of several alien species but not of native species, and we hypothesized that this increase in growth of aliens was due to livestock preference for alien and not native trees. With the objective of quantifying our observations and understanding the mechanism underlying the increased growth rates of alien juvenile trees under livestock exclusion, we assessed growth and browsing levels in juveniles of two alien invasive and four abundant native tree species within three parcels where livestock was excluded and three parcels with livestock at 0.20 cattle equivalents.ha^?1. Alien species grew around four-fold faster under livestock exclusion than with livestock and, as predicted, received five times more browsing than natives. Instead, native species did not significantly increase their growth rate with livestock exclusion. The results support our hypothesis and the implications for management would be that stocking paddocks with livestock to browse existing alien juveniles and re-growth of felled adults should be effective in delaying invasions of trees used for forestry without significantly affecting the growth of the most abundant native trees.     

Disentangling the causes of damage variation by deer browsing on young Thuja plicata

Long‐lived trees experience different levels of damage due to mammalian herbivores. To untangle the mechanisms that underlie this variation, we combined chemical with dendrochronological analyses to study variation in browsing on Western redcedars (Thuja plicata) on Haida Gwaii (British Columbia, Canada). Since the last glaciation, Haida Gwaii forests had lacked large herbivore browser until Sitka black‐tailed deer (Odocoileus hemionus sitkensis) were introduced at the beginning of the 20th century. Dendrochronology yielded information on radial growth and plant annual responses to environmental stresses including herbivory. Secondary metabolite content and plant nutritional quality provided insights into proximate causes of food choices made by herbivores. We sampled lightly‐ and heavily‐browsed young trees at four sites: three clear‐cut sites with high browsing pressure and one old‐growth forest site where browsing pressure had, until recently, been lower. Heavily‐browsed young trees had lower concentrations of secondary metabolites and were of lower nutritive value than lightly‐browsed trees at all sites. Under high browsing pressure, tree growth patterns suggested that all young trees were initially severely browsed until some trees, currently scored as lightly‐browsed, started to escape deer. At the old‐growth site, both lightly‐ and heavily‐browsed trees tended to have lower overall average secondary metabolite concentrations than those of all other sites, a trend possibly related to greater canopy closure. Lightly‐browsed trees were older than heavily‐browsed ones which resulted, during the period of lower browsing pressure, in higher growth rate and a same pattern of change in growth from one year to the next year. This suggests that, under low browsing pressure, selection of young trees related to chemical defense was weak and that growth differences due to other factors than browsing could be expressed. Under strong browsing pressure, however, all young trees had equally low growth rates until trees with better genetic potential to produce effective defenses were able to escape deer. This suggests that selection by deer could occur on a long‐lived tree.     

Effects of simulated winter browsing on mountain birch foliar chemistry and on the performance of insect herbivores

Winter browsing by mammalian herbivores is known to induce a variety of morphological and physiological changes in plants. Browsing has been suggested to decrease the carbohydrate reserves in woody plants, which might lead to reduced tannin production in leaves during the following summer, and consequently, to increased herbivore damage on leaves. We conducted a clipping experiment with mature mountain birch trees and measured the effects of clipping on birch growth, leaf chemistry and toughness, as well as on the performance of insect herbivores. Leaves grew larger and heavier per unit area in the clipped ramets and had a higher content of proteins than leaves in the control trees. Clipping treatment did not affect the total content of sugars in the leaves (mg g^?1), suggesting that a moderate level of clipping did not significantly reduce the carbohydrate pools of fully‐grown mountain birch trees. Furthermore, the contents of proanthocyanidins (condensed tannins) and gallotannins were slightly higher in the leaves of clipped ramets, contrary to the hypothesis of reduced tannin production. The effects of clipping treatment on leaf and shoot growth and on foliar chemistry were mainly restricted to the clipped ramets, without spreading to untreated ramets within the same tree individual. The effects of clipping on leaf characters varied during the growing season; for instance, leaf toughness in clipped ramets was higher than toughness in control trees and ramets only when leaves were mature. Accordingly, clipping had inconsistent effects on insect herbivores feeding at different times of the growing season. The generally small impact of clipping on herbivore performance suggests that the low intensity of natural browsing at the study area, simulated by our clipping treatment, does not have strong consequences for the population dynamics of insect herbivores on mountain birch via enhanced population growth caused by browsing‐induced changes in food quality.     

Nutrient and secondary metabolite concentrations in a savanna are independently affected by large herbivores and shoot growth rate

Carbon-based secondary metabolites (CBSMs) such as tannins are assumed to function as plant defences against herbivores. CBSMs are thought to be inversely related to growth rate and nutrient concentrations because a physiological trade-off exists between cellular growth and differentiation, but CBSM concentrations can be altered by herbivory-induced changes in the trade-off. We predicted that a significant interaction exists between herbivory and growth phase, such that the effects of large herbivores (or their exclusion) on nutrient or CBSM concentrations are greatest during phases of rapid shoot or leaf growth. Leaf samples were collected during phases of different growth rate from six woody species 4 years after establishment of a large-scale long-term herbivore exclusion experiment in Kruger National Park, South Africa. Samples were analysed for N, P, condensed tannins and total phenolics. Interactions between growth phase and herbivores were rare. However, the assumption that elevated nutrients and reduced CBSMs occurs during fast phases of growth was supported by four species (consistent with the growth-differentiation balance hypothesis), but not the other two. Large herbivores generally did not affect nutrients, but CBSMs in four species were reduced by large herbivores other than elephants, while CBSMs in two species were reduced by elephants. Carbon limitation ultimately prevailed among woody plants taller than 2 m under long-term browsing. Large herbivores and plant growth phase are independent and important determinants of nutrients or CBSMs in African savannas, but the effects depend on the interacting assemblages of species, which poses challenges to the application of current general hypotheses of plant defence.