Symbiotic Ants Research Articles

Automated detection of an insect‐induced keystone vegetation phenotype using airborne LiDAR

Abstract Ecologists, foresters and conservation practitioners need ‘biodiversity scanners’ to effectively inventory biodiversity, audit conservation progress and track changes in ecosystem function. Quantifying biological diversity using remote sensing methods remains challenging, especially for small invertebrates. However, insect aggregations can drastically alter landscapes and vegetation, and these ‘extended phenotypes’ could serve as environmental landmarks of insect presence in remotely sensed data. To test the feasibility of this approach, we studied symbiotic ants that alter the canopy shape of whistling thorn acacias (Acacia [syn. Vachellia] drepanolobium), a keystone tree species of the black cotton soils of east African savannas. We demonstrate a protocol for using light detection and ranging (LiDAR) data to collect, prepare (including a customizable tree‐segmentation algorithm) and apply a convolutional neural network‐based classification for the detection of ant‐inhabited acacia tree phenotypic variations. Applying this protocol enabled us to effectively detect intra‐specific tree phenotypic variation induced by insects. Surveying ant occupancy across 16 ha and 9680 acacia trees took 1000 work hours, whereas surveyed patterns of ant distribution were replicated by our trained classifier using only an hour‐long airborne LiDAR collection time. We suggest that large‐scale surveys of insect occupancy (including insect‐vectored disease) can be automated through a combination of airborne LiDAR and machine learning.

Morphological divergence of domatia in ant-free populations of the widespread Neotropical myrmecophyte Miconia tococa (Melastomataceae)

Abstract Co-evolving organisms experience multiple selection pressures that may lead to trait mismatches among different populations and sites. In defensive ant–plant mutualisms, host plants (myrmecophytes) produce specialized shelters (domatia) to harbour specialized ant-partners in exchange for protection against enemies. Although populations of myrmecophytes without ants occur in some locations, there are no records of changes in domatia morphology—at the population level—due to the absence of symbiotic ants. We conducted broad-scale samplings of Miconia tococa (Melastomataceae) populations across the Brazilian Cerrado and a 2-year transplant experiment to test whether domatia morphology changes when symbiotic ants are naturally absent. Domatia were 33.9% smaller in ant-free populations than in ant-inhabited populations. Transplants revealed that host plants from ant-inhabited sites still developed larger domatia than those from ant-free sites, even in the absence of ant-partners. These findings point to a change of M. tococa traits associated with biotic defences where symbiotic ants are absent. What may have begun as a plastic adjustment to ant-free environments appears to have been transformed into fixed (genetic) interpopulation differences over time, indicating a potential local destabilization of the mutualism or a mechanism to stabilize the interaction at the landscape scale.

Trophobiosis between a new species of Acropyga (Hymenoptera, Formicidae) and new Neochavesia (Hemiptera, Xenococcidae) from Peru, and establishment of the Acropygasmithii species-group.

We describe a new pair of trophobiotic partners from the ant genus Acropyga and the root mealybug genus Neochavesia. A recent field study on Acropyga ants and associated root mealybugs, conducted in the Peruvian Amazon, led to the discovery of Acropygamanuense LaPolla & Schneider, sp. nov. and its root mealybug symbiont Neochavesiapodexuta Schneider & LaPolla, sp. nov. The new root mealybug belongs to the family Xenococcidae, whose members are all obligate associates of Acropyga ants. Providing joint descriptions of new mutualist partners in the same article is a novel approach for this system, and it offers benefits to the ongoing study of mutualism and patterns of association among these symbiotic ants and scales. Here, we also begin to revise the species-group composition of Acropyga by establishing the smithii species-group, and we provide updated information to aid in identifying the new ant species and root mealybug species.

Population Genomics and Demographic Sampling of the Ant-Plant Vachellia drepanolobium and Its Symbiotic Ants From Sites Across Its Range in East Africa

The association between the African ant plant, Vachellia drepanolobium, and the ants that inhabit it has provided insight into the boundaries between mutualism and parasitism, the response of symbioses to environmental perturbations, and the ecology of species coexistence. We use a landscape genomics approach at sites sampled throughout the range of this system in Kenya to investigate the demographics and genetic structure of the different partners in the association. We find that different species of ant associates of V. drepanolobium show striking differences in their spatial distribution throughout Kenya, and these differences are only partly correlated with abiotic factors. A comparison of the population structure of the host plant and its three obligately arboreal ant symbionts, Crematogaster mimosae, Crematogaster nigriceps, and Tetraponera penzigi, shows that the ants exhibit somewhat similar patterns of structure throughout each of their respective ranges, but that this does not correlate in any clear way with the respective genetic structure of the populations of their host plants. A lack of evidence for local coadaptation in this system suggests that all partners have evolved to cope with a wide variety of biotic and abiotic conditions.

Tree resprout dynamics following fire depend on herbivory by wild ungulate herbivores

Abstract Savanna tree cover is dynamic due to disturbances such as fire and herbivory. Frequent fires can limit a key demographic transition from sapling to adult height classes in savanna trees. Saplings may be caught in a ‘fire trap’, wherein individuals repeatedly resprout following fire top‐kill events. Saplings only rarely escape the cycle by attaining a fire‐resistant height (e.g. taller than the minimum scorch height) during fire‐free intervals. Large mammalian herbivores also may trap trees in shorter size classes. Browsing herbivores directly limit sapling height, while grazing herbivores such as cattle facilitate sapling growth indirectly via grass removal. Experimental studies investigating how meso‐wildlife, megaherbivores and domestic livestock affect height of resprouts following fire are rare, but necessary for fully understanding how herbivory may reinforce (or counteract) the fire trap. In our study system, interactive fire–herbivore effects on transitions from sapling (<1 m) to adult tree (>1 m) height classes may be further influenced by plant defences, such as symbiotic ants. We used the Kenya Long‐term Exclosure Experiment (KLEE) to investigate how post‐fire resprout size of a widespread monodominant East African tree, Acacia drepanolobium was influenced by (a) herbivory by different combinations of cattle, meso‐wildlife (15–1,000 kg) and megaherbivores (>1,000 kg) and (b) the presence of acacia–ant mutualists that confer tree defences. We sampled height, stem length and ant occupancy of resprouts exposed to different herbivore combinations before and after controlled burns. Resprout height of saplings that were short prior to fire (<1 m) was reduced primarily by meso‐wildlife. Negative effects of elephants on post‐fire resprout height increased with pre‐fire tree size, suggesting that resprouts of the tallest trees (with the greatest potential to escape the fire trap cycle) were preferentially browsed and reduced in height by elephants. There were no significant cattle effects. Synthesis. We provide experimental evidence for two potential pathways through which large herbivores exert control over sapling escape from the fire trap: (a) post‐fire meso‐wildlife browsing of short (<1 m) resprouts and (b) elephant browsing of the largest size class of resprouts, which would otherwise be most likely to escape the fire trap.

Reduced ant defenses in Macaranga myrmecophytes (Euphorbiaceae) infested with a winged phasmid Orthomeria cuprinus

AbstractMacaranga is a tree genus that includes many species of myrmecophytes, which are plants that harbor ant colonies within hollow structures known as domatia. The symbiotic ants (plant–ants) protect their host plants against herbivores; this defense mechanism is called ‘ant defense’. A Bornean phasmid species Orthomeriacuprinus feeds on two myrmecophytic Macaranga species, Macaranga beccariana and Macaranga hypoleuca, which are obligately associated with Crematogaster ant species. The phasmids elude the ant defense using specialized behavior. However, the mechanisms used by the phasmid to overcome ant defenses have been insufficiently elucidated. We hypothesized that O.cuprinus only feeds on individual plants with weakened ant defenses. To test the hypothesis, we compared the ant defense intensity in phasmid‐infested and non‐infested M. beccariana trees. The number of plant–ants on the plant surface, the ratio of plant–ant biomass to tree biomass, and the aggressiveness of plant–ants towards experimentally introduced herbivores were significantly lower on the phasmid‐infested trees than on the non‐infested trees. The phasmid nymphs experimentally introduced into non‐infested trees, compared with those experimentally introduced into phasmid‐infested trees, were more active on the plant surface, avoiding the plant–ants. These results support the hypothesis and suggest that ant defenses on non‐infested trees effectively prevent the phasmids from remaining on the plants. Thus, we suggest that O.cuprinus feeds only on the individual M. beccariana trees having decreased ant defenses, although the factors that reduce the intensity of the ant defenses remain unclear.

Ant-Repelling Pollinators of the Myrmecophytic Macaranga winkleri (Euphorbiaceae)

Many plants have mutualistic relationships with ants, whereby plants provide food and/or nesting sites for the symbiotic ants, and in turn the ants protect the host plants by excluding herbivores. While the ants are useful as guards, they may negatively affect host reproduction by excluding pollinators. Here we studied this potential conflict in the myrmecophytic Macaranga winkleri pollinated by the thrips Dolichothrips fialae. Behavioural responses of ant guards to pollinator thrips and their chemicals, and related chemical analyses, provide evidence that thrips deter ant-guards by secreting droplets containing ant-repelling n-decanoic acid from their anuses. This is the first report of insect pollinators repelling their host’s symbiotic guard ants to perform pollination. This is a novel strategy by which a plant host avoids interference with pollination by ant-guards in an ant–plant mutualism. The acquisition of a pollination system that is resistant to ant attacks may have facilitated the evolution of myrmecophytes in the genus Macaranga.

Various chemical strategies to deceive ants in three Arhopala species (lepidoptera: Lycaenidae) exploiting Macaranga myrmecophytes.

Macaranga myrmecophytes (ant-plants) are generally well protected from herbivore attacks by their symbiotic ants (plant-ants). However, larvae of Arhopala (Lepidoptera: Lycaenidae) species survive and develop on specific Macaranga ant-plant species without being attacked by the plant-ants of their host species. We hypothesized that Arhopala larvae chemically mimic or camouflage themselves with the ants on their host plant so that the larvae are accepted by the plant-ant species of their host. Chemical analyses of cuticular hydrocarbons showed that chemical congruency varied among Arhopala species; A. dajagaka matched well the host plant-ants, A. amphimuta did not match, and unexpectedly, A. zylda lacked hydrocarbons. Behaviorally, the larvae and dummies coated with cuticular chemicals of A. dajagaka were well attended by the plant-ants, especially by those of the host. A. amphimuta was often attacked by all plant-ants except for the host plant-ants toward the larvae, and those of A. zylda were ignored by all plant-ants. Our results suggested that conspicuous variations exist in the chemical strategies used by the myrmecophilous butterflies that allow them to avoid ant attack and be accepted by the plant-ant colonies.

Disruption of a protective ant-plant mutualism by an invasive ant increases elephant damage to savanna trees.

Invasive species can indirectly affect ecosystem processes via the disruption of mutualisms. The mutualism between the whistling thorn acacia (Acacia drepanolobium) and four species of symbiotic ants is an ecologically important one; ants strongly defend trees against elephants, which can otherwise have dramatic impacts on tree cover. In Laikipia, Kenya, the invasive big-headed ant (Pheidole megacephala) has established itself at numerous locations within the last 10-15 years. In invaded areas on five properties, we found that three species of symbiotic Crematogaster ants were virtually extirpated, whereas Tetraponera penzigi co-occurred with P. megacephala. T. penzigi appears to persist because of its nonaggressive behavior; in a whole-tree translocation experiment, Crematogaster defended host trees against P. megacephala, but were extirpated from trees within hours. In contrast, T. penzigi retreated into domatia and withstood invading ants for >30 days. In the field, the loss of defensive Crematogaster ants in invaded areas led to a five- to sevenfold increase in the number of trees catastrophically damaged by elephants compared to uninvaded areas. In savannas, tree cover drives many ecosystem processes and provides essential forage for many large mammal species; thus, the invasion of big-headed ants may strongly alter the dynamics and diversity of East Africa's whistling thorn savannas by disrupting this system's keystone acaciaant mutualism.

Host-plant use by twoOrthomeria(Phasmida: Aschiphasmatini) species feeding onMacarangamyrmecophytes

Myrmecophytes depend on symbiotic ants (plant-ants) to defend against herbivores. Although these defensive mechanisms are highly effective, some herbivorous insects can use myrmecophytes as their host-plants. The feeding habits of these phytophages on myrmecophytes and the impacts of the plant-ants on their feeding behavior have been poorly studied. We examined two phasmid species, Orthomeria alexis and O. cuprinus, which are known to feed on Macaranga (Euphorbiaceae) myrmecophytes in a Bornean primary forest. Our observations revealed that: (i) each phasmid species relied on two closely-related myrmecophytic Macaranga species for its host-plants in spite of their normal plant-ant symbioses; and (ii) there was little overlap between their host-plant preferences. More O. cuprinus adults and nymphs were found on new leaves, which were attended by more plant-ants than mature leaves, while most adults and nymphs of O. alexis tended to avoid new leaves. In a feeding choice experiment under ant-excluded conditions, O. alexis adults chose a non-host Macaranga myrmecophyte that was more intensively defended by plant-ants and was more palatable than their usual host-plants almost as frequently as their usual host-plant, suggesting that the host-plant range of O. alexis was restricted by the presence of plant-ants on non-host-plants. Phasmid behavior that appeared to minimize plant-ant attacks is described.

Exploitation of Food Bodies onMacarangaMyrmecophytes by Larvae of a Lycaenid Species,Arhopala zylda(Lycaeninae)

ABSTRACT. Larvae of Arhopala zylda (Lycaenidae) feed on food bodies (FBs) produced by two Macaranga (Euphorbiaceae) myrmecophytic species, M. beccariana and M. hypoleuca. We examined their feeding behavior in detail via field observations and rearing experiments in the field and laboratory. Larvae of A. zylda fed only on FBs and not leaves during the first through third instars; during the fourth (final) instar, they ate both FBs and leaves of the host plants. The larvae actively fed on FBs on young leaves, which were always attended by many plant symbiotic ants. These results suggested that A. zylda larvae depend entirely on FBs for food, except late in the final instar, and that the FB-feeding habit is associated with special traits that enable the larvae to evade ant aggression, which usually functions as an effective anti-herbivore defense for the host plants.

Protection Mutualisms and the Community: Geographic Variation in an Ant-Plant Symbiosis and the Consequences for Herbivores

Protection mutualisms mediate trophic interactions in many systems, but their effects on the surrounding community are rarely studied. Ant-plant symbioses are classic examples of protection mutualisms: myrmecophytic plants provide nesting space and food for symbiotic ants in exchange for ant defense. Ant defense should thus reduce the abundance of herbivores, but studies of ant-plant symbioses usually measure damage to the plant without quantifying the herbivores themselves. In this study, we investigated whether geographic variation in the quality of ant defense in a symbiotic mutualism between Cordia alliodora trees and Azteca ants was associated with the abundance and species richness of plant herbivore communities. In three tropical-dry-forest sites in Middle America, we found that the density of Azteca ants within trees was negatively associated with the levels of leaf herbivory. At sites where ants were effective tree defenders, tree herbivores were less abundant and herbivore assemblages on trees exhibited lower species richness than at a site where ants were poor defenders. In addition, in a site where ants reduced herbivory, herbivore communities were less abundant and diverse in the presence of ants than in their absence, where as in a site where ants did not reduce herbivory, there were no differences in herbivore abundance or richness between trees with or without ants. We conclude that geographic variation in the quality of ant defense drives variation in myrmecophytic-plant herbivore communities. Moreover, ant-plant protection mutualisms should have important but rarely considered effects on herbivore population dynamics and food-plant specialization.

Myrmecoxeny inArhopala zylda(Lepidoptera, Lycaenidae) Larvae Feeding onMacarangaMyrmecophytes

ABSTRACT Some species in the tree genus Macaranga (Euphorbiaceae) in the Southeast Asian tropics are myrmecophytic; they have highly species-specific mutualisms with symbiotic ants (plant-ants), which defend them from herbivores. However, larvae of some Arhopala (Lycaenidae, Lycaeninae) species can elude the ants. Here we demonstrated that Arhopala zylda larvae showed myrmecoxeny on their myrmecophytic Macaranga host plants; they had no stable association with the plant-ants. Despite the presence of many plant-ants, A. zylda larvae were rarely attended or attacked by ants on their host plants. The plant-ants of three other myrmecophytic Macaranga species (non-hosts to A. zylda) also paid little attention to experimentally introduced A. zylda larvae. The myrmecoxeny seen in A. zylda is notable among lycaenid larvae that feed on myrmecophytes, because almost all are obligate intimate myrmecophiles.

Effects of Dam‐Induced Landscape Fragmentation on Amazonian Ant–Plant Mutualistic Networks

Mutualistic networks are critical to biological diversity maintenance; however, their structures and functionality may be threatened by a swiftly changing world. In the Amazon, the increasing number of dams poses a large threat to biological diversity because they greatly alter and fragment the surrounding landscape. Tight coevolutionary interactions typical of tropical forests, such as the ant-myrmecophyte mutualism, where the myrmecophyte plants provide domatia nesting space to their symbiotic ants, may be jeopardized by the landscape changes caused by dams. We analyzed 31 ant-myrmecophyte mutualistic networks in undisturbed and disturbed sites surrounding Balbina, the largest Central Amazonian dam. We tested how ant-myrmecophyte networks differ among dam-induced islands, lake edges, and undisturbed forests in terms of species richness, composition, structure, and robustness (number of species remaining in the network after partner extinctions). We also tested how landscape configuration in terms of area, isolation, shape, and neighborhood alters the structure of the ant-myrmecophyte networks on islands. Ant-myrmecophytic networks were highly compartmentalized in undisturbed forests, and the compartments had few strongly connected mutualistic partners. In contrast, networks at lake edges and on islands were not compartmentalized and were negatively affected by island area and isolation in terms of species richness, density, and composition. Habitat loss and fragmentation led to coextinction cascades that contributed to the elimination of entire ant-plant compartments. Furthermore, many myrmecophytic plants in disturbed sites lost their mutualistic ant partners or were colonized by opportunistic, nonspecialized ants. Robustness of ant-myrmecophyte networks on islands was lower than robustness near lake edges and in undisturbed forest and was particularly susceptible to the extinction of plants. Beyond the immediate habitat loss caused by the building of large dams in Amazonia, persistent edge effects and habitat fragmentation associated with dams had large negative effects on animal-plant mutualistic networks.

The cost of myrmecophytism: insights from allometry of stem secondary growth

Plant defence traits against herbivores incur production costs that are usually difficult to measure. However, estimating these costs is a prerequisite for characterizing the plant defence strategy as a whole. Myrmecophytes are plants that provide symbiotic ants with specialized nesting cavities, called domatia, in exchange for protection against herbivores. In the particular case of stem domatia, production of extra wood seems to be the only associated cost, making this indirect defence trait a particularly suitable model for estimating the cost of defence. Measurements were made of growth pattern and cumulative production cost of domatia over secondary growth in the myrmecophyte Leonardoxa africana subsp. africana, whose internodes display both a solid basal segment and a hollow distal part (the domatium), thus allowing paired comparison of investment in wood. Previous studies showed that 'overconstruction' of the hollow part of internodes during primary growth is needed for mechanical support. In this study, it is shown that the relationship between the woody cross-sectional area of the solid and hollow parts of internodes is negatively allometric at the beginning of secondary growth and nearly isometric later on. Thus, in hollow stems, the first phase of slow secondary growth compensates for the 'overconstruction' of the ring of wood during primary growth. Moreover, the cumulative production cost of a domatium (estimated as the additional volume of wood required for a hollow stem compared with a solid one) is very high at the beginning of secondary growth and then quickly tends to zero. Making domatia incurs high costs early in ontogeny, costs that are then amortized later in development of stems and of individual plants. Characterizing ontogenetic variation of the net cost of this peculiar defence mechanism will help us build more accurate theoretical models of resource allocation in myrmecophytes.

Change in biomass of symbiotic ants throughout the ontogeny of a myrmecophyte, Macaranga beccariana (Euphorbiaceae)

Macaranga myrmecophytes (ant-plants) provide their partner symbiotic ants (plant-ants) with food bodies as their main food, and they are protected by the plant-ants from herbivores. The amount of resource allocated to food bodies determines the plant-ant colony size and consequently determines the intensity of ant defense (anti-herbivore defense by plant-ants). As constraints in resource allocation change as plants grow, the plant-ant colony size is hypothesized to change with the ontogenesis of Macaranga myrmecophyte. To determine the ontogenetic change in the relative size of the plant-ant colony, we measured the dry weights of the whole plant-ant colony and all of the aboveground parts of trees at various ontogenetic stages for a myrmecophytic species (Macaranga beccariana) in a Bornean lowland tropical rain forest. Ant biomass increased as plant biomass increased. However, the rate of increase gradually declined, and the ant biomass appeared to reach a ceiling once trees began to branch. The ant/plant biomass ratio consistently decreased as plant biomass increased, with the rate of decrease gradually accelerating. We infer that the ontogenetic reduction in ant/plant biomass ratio is caused by an ontogenetic change in resource allocation to food rewards for ants related to the physiological changes accompanying the beginning of branching.

The Evolution of Communication in Two Ant-Plant Mutualisms

Myrmecophytes are plants that provide nesting sites and food to ants that protect them against herbivores. Plant signals function to synchronize ant patrolling with the probability of herbivory. We compared the communication signals in two symbioses involving ant and plant pairs that are closely related. The two plants emitted the same volatile compounds upon damage. These compounds are simple molecules common in the plant kingdom. Electroantennography revealed that the two symbiotic ants, as well as several other ant species, were able to perceive these compounds. However, workers of one species responded only to hexanal, while those of the other species responded mostly to methyl salicylate. The two signals involved in the focal symbioses are ‘cheap’ (low metabolic cost), which is consistent with theoretical predictions for the evolution of signalling between partners with convergent interests. They are also not specific, which is expected between plants and broad-spectrum predators such as ants. The fact that different signals are used in the two sister symbioses suggests different mechanisms underlying similar adaptations in the evolution of communication.

Indirect benefits of symbiotic coccoids for an ant-defended myrmecophytic tree

The net benefits of mutualism depend directly on the costs and effectiveness of mutualistic services and indirectly on the interactions that affect those services. We examined interactions among Cordia alliodora myrmecophytic trees, their symbiotic ants Azteca pittieri, coccoid hemipterans, and foliar herbivores in two Neotropical dry forests. The tree makes two investments in symbiotic ants: it supplies nesting space, as domatia, and it provides phloem to coccoids, which then produce honeydew that is consumed by ants. Although higher densities of coccoids should have higher direct costs for trees, we asked whether higher densities of coccoids can also have higher indirect benefits for trees by increasing the effectiveness of ant defense against foliar herbivores. We found that trees benefited from ant defense against herbivores. Ants defended trees effectively only when colonies reached high densities within trees, and ant and coccoid densities within trees were strongly positively correlated. The benefits of reduced foliar herbivory by larger ant colonies were therefore indirectly controlled by the number of coccoids. Coccoid honeydew supply also affected per capita ant aggression against tree herbivores. Ants experimentally fed a carbohydrate-rich diet, analogous to sugar obtained from coccoids, were more aggressive against caterpillars per capita than ants fed a carbohydrate-poor diet. Ant defense was more effective on more valuable and vulnerable young leaves than on older leaves. Young domatia, associated with young leaves, contained higher coccoid densities than older domatia, which suggests that coccoids may also drive spatially favorable ant defense of the tree. If higher investments by one mutualistic partner are tied to higher benefits received from the other, there may be positive feedback between partners that will stabilize the mutualism. These results suggest that higher investment by trees in coccoids leads to more effective defense by ants against the tree's foliar herbivores.

Plant-ants feed their host plant, but above all a fungal symbiont to recycle nitrogen

In ant-plant symbioses, plants provide symbiotic ants with food and specialized nesting cavities (called domatia). In many ant-plant symbioses, a fungal patch grows within each domatium. The symbiotic nature of the fungal association has been shown in the ant-plant Leonardoxa africana and its protective mutualist ant Petalomyrmex phylax. To decipher trophic fluxes among the three partners, food enriched in (13)C and (15)N was given to the ants and tracked in the different parts of the symbiosis up to 660 days later. The plant received a small, but significant, amount of nitrogen from the ants. However, the ants fed more intensively the fungus. The pattern of isotope enrichment in the system indicated an ant behaviour that functions specifically to feed the fungus. After 660 days, the introduced nitrogen was still present in the system and homogeneously distributed among ant, plant and fungal compartments, indicating efficient recycling within the symbiosis. Another experiment showed that the plant surface absorbed nutrients (in the form of simple molecules) whether or not it is coated by fungus. Our study provides arguments for a mutualistic status of the fungal associate and a framework for investigating the previously unsuspected complexity of food webs in ant-plant mutualisms.

Defensive Plant-Ants Stabilize Megaherbivore-Driven Landscape Change in an African Savanna

Tree cover in savanna ecosystems is usually regarded as unstable, varying with rainfall, fire, and herbivory. In sub-Saharan Africa, elephants (Loxodonta africana) suppress tree cover, thereby maintaining landscape heterogeneity by promoting tree-grass coexistence. In the absence of elephants, tree encroachment may convert savannas into closed-canopy woodlands; when elephants increase in abundance, intensified browsing pressure can transform savannas into open grasslands. We show that symbiotic ants stabilize tree cover across landscapes in Kenya by protecting a dominant tree from elephants. In feeding trials, elephants avoided plants with ants and did not distinguish between a myrmecophyte (the whistling-thorn tree [Acacia drepanolobium]) from which ants had been removed and a highly palatable, nonmyrmecophytic congener. In field experiments, elephants inflicted severe damage on whistling-thorn trees from which ants had been removed. Across two properties on which elephants increased between 2003 and 2008, cover of whistling-thorn did not change significantly inside versus outside large-scale elephant exclusion fences; over the same period of time, cover of nonmyrmecophytes differed profoundly inside versus outside exclusion fences. These results highlight the powerful role that symbioses and plant defense play in driving tree growth and survival in savannas, ecosystems of global economic and ecological importance.