Predator-induced collapse of niche structure and species coexistence.

Biological invasions are both a pressing environmental challenge and an opportunity to investigate fundamental ecological processes, such as the role of top predators in regulating biodiversity and food-web structure. In whole-ecosystem manipulations of small Caribbean islands on which brown anole lizards (Anolis sagrei) were the native top predator, we experimentally staged invasions by competitors (green anoles, Anolis smaragdinus) and/or new top predators (curly-tailed lizards, Leiocephalus carinatus). We show that curly-tailed lizards destabilized the coexistence of competing prey species, contrary to the classic idea of keystone predation. Fear-driven avoidance of predators collapsed the spatial and dietary niche structure that otherwise stabilized coexistence, which intensified interspecific competition within predator-free refuges and contributed to the extinction of green-anole populations on two islands. Moreover, whereas adding either green anoles or curly-tailed lizards lengthened food chains on the islands, adding both species reversed this effect-in part because the apex predators were trophic omnivores. Our results underscore the importance of top-down control in ecological communities, but show that its outcomes depend on prey behaviour, spatial structure, and omnivory. Diversity-enhancing effects of top predators cannot be assumed, and non-consumptive effects of predation risk may be a widespread constraint on species coexistence.

Grazing herbivores reduce herbaceous biomass and fire activity across African savannas.

Fire and herbivory interact to alter ecosystems and carbon cycling. In savannas, herbivores can reduce fire activity by removing grass biomass, but the size of these effects and what regulates them remain uncertain. To examine grazing effects on fuels and fire regimes across African savannas, we combined data from herbivore exclosure experiments with remotely sensed data on fire activity and herbivore density. We show that, broadly across African savannas, grazing herbivores substantially reduce both herbaceous biomass and fire activity. The size of these effects was strongly associated with grazing herbivore densities, and surprisingly, was mostly consistent across different environments. A one-zebra increase in herbivore biomass density (~100 kg/km2 of metabolic biomass) resulted in a ~53 kg/ha reduction in standing herbaceous biomass and a ~0.43 percentage point reduction in burned area. Our results indicate that fire models can be improved by incorporating grazing effects on grass biomass.

Large herbivores suppress liana infestation in an African savanna.

African savannas are the last stronghold of diverse large-mammal communities, and a major focus of savanna ecology is to understand how these animals affect the relative abundance of trees and grasses. However, savannas support diverse plant life-forms, and human-induced changes in large-herbivore assemblages-declining wildlife populations and their displacement by livestock-may cause unexpected shifts in plant community composition. We investigated how herbivory affects the prevalence of lianas (woody vines) and their impact on trees in an East African savanna. Although scarce (<2% of tree canopy area) and defended by toxic latex, the dominant liana, Cynanchum viminale (Apocynaceae), was eaten by 15 wild large-herbivore species and was consumed in bulk by native browsers during experimental cafeteria trials. In contrast, domesticated ungulates rarely ate lianas. When we experimentally excluded all large herbivores for periods of 8 to 17 y (simulating extirpation), liana abundance increased dramatically, with up to 75% of trees infested. Piecewise exclusion of different-sized herbivores revealed functional complementarity among size classes in suppressing lianas. Liana infestation reduced tree growth and reproduction, but herbivores quickly cleared lianas from trees after the removal of 18-y-old exclosure fences (simulating rewilding). A simple model of liana contagion showed that, without herbivores, the long-term equilibrium could be either endemic (liana-tree coexistence) or an all-liana alternative stable state. We conclude that ongoing declines of wild large-herbivore populations will disrupt the structure and functioning of many African savannas in ways that have received little attention and that may not be mitigated by replacing wildlife with livestock.

The Difficulty of Predicting Evolutionary Change in Response to Novel Ecological Interactions: A Field Experiment with Anolis Lizards.

AbstractDetermining whether and how evolution is predictable is an important goal, particularly as anthropogenic disturbances lead to novel species interactions that could modify selective pressures. Here, we use a multigeneration field experiment with brown anole lizards (Anolis sagrei) to test hypotheses about the predictability of evolution. We manipulated the presence/absence of predators and competitors of A. sagrei across 16 islands in the Bahamas that had preexisting brown anole populations. Before the experiment and again after roughly five generations, we measured traits related to locomotor performance and habitat use by brown anoles and used double-digest restriction enzyme-associated DNA sequencing to estimate genome-wide changes in allele frequencies. Although previous work showed that predators and competitors had characteristic effects on brown anole behavior, diet, and population sizes, we found that evolutionary change at both phenotypic and genomic levels was difficult to forecast. Phenotypic changes were contingent on sex and habitat use, whereas genetic change was unpredictable and not measurably correlated with phenotypic changes, experimental treatments, or other environmental factors. Our work shows how differences in ecological context can alter evolutionary outcomes over short timescales and underscores the difficulty of forecasting evolutionary responses to multispecies interactions in natural conditions, even in a well-studied system with ample supporting ecological information.

Host phylogeny and functional traits differentiate gut microbiomes in a diverse natural community of small mammals.

Differences in the bacterial communities inhabiting mammalian gut microbiomes tend to reflect the phylogenetic relatedness of their hosts, a pattern dubbed phylosymbiosis. Although most research on this pattern has compared the gut microbiomes of host species across biomes, understanding the evolutionary and ecological processes that generate phylosymbiosis requires comparisons across phylogenetic scales and under similar ecological conditions. We analysed the gut microbiomes of 14 sympatric small mammal species in a semi-arid African savanna, hypothesizing that there would be a strong phylosymbiotic pattern associated with differences in their body sizes and diets. Consistent with phylosymbiosis, microbiome dissimilarity increased with phylogenetic distance among hosts, ranging from congeneric sets of mice and hares that did not differ significantly in microbiome composition to species from different taxonomic orders that had almost no gut bacteria in common. While phylosymbiosis was detected among just the 11 species of rodents, it was substantially weaker at this scale than in comparisons involving all 14 species together. In contrast, microbiome diversity and composition were generally more strongly correlated with body size, dietary breadth, and dietary overlap in comparisons restricted to rodents than in those including all lineages. The starkest divides in microbiome composition thus reflected the broad evolutionary divergence of hosts, regardless of body size or diet, while subtler microbiome differences reflected variation in ecologically important traits of closely related hosts. Strong phylosymbiotic patterns arose deep in the phylogeny, and ecological filters that promote functional differentiation of cooccurring host species may disrupt or obscure this pattern near the tips.

Mutualism disruption by an invasive ant reduces carbon fixation for a foundational East African ant-plant.

Invasive ants shape assemblages and interactions of native species, but their effect on fundamental ecological processes is poorly understood. In East Africa, Pheidole megacephala ants have invaded monodominant stands of the ant-tree Acacia drepanolobium, extirpating native ant defenders and rendering trees vulnerable to canopy damage by vertebrate herbivores. We used experiments and observations to quantify direct and interactive effects of invasive ants and large herbivores on A. drepanolobium photosynthesis over a 2-year period. Trees that had been invaded for ≥ 5 years exhibited 69% lower whole-tree photosynthesis during key growing seasons, resulting from interaction between invasive ants and vertebrate herbivores that caused leaf- and canopy-level photosynthesis declines. We also surveyed trees shortly before and after invasion, finding that recent invasion induced only minor changes in leaf physiology. Our results from individual trees likely scale up, highlighting the potential of invasive species to alter ecosystem-level carbon fixation and other biogeochemical cycles.

Experimental evidence that effects of megaherbivores on mesoherbivore space use are influenced by species' traits.

The extinction of 80% of megaherbivore (>1,000 kg) species towards the end of the Pleistocene altered vegetation structure, fire dynamics and nutrient cycling world-wide. Ecologists have proposed (re)introducing megaherbivores or their ecological analogues to restore lost ecosystem functions and reinforce extant but declining megaherbivore populations. However, the effects of megaherbivores on smaller herbivores are poorly understood. We used long-term exclusion experiments and multispecies hierarchical models fitted to dung counts to test (a) the effect of megaherbivores (elephant and giraffe) on the occurrence (dung presence) and use intensity (dung pile density) of mesoherbivores (2-1,000 kg), and (b) the extent to which the responses of each mesoherbivore species was predictable based on their traits (diet and shoulder height) and phylogenetic relatedness. Megaherbivores increased the predicted occurrence and use intensity of zebras but reduced the occurrence and use intensity of several other mesoherbivore species. The negative effect of megaherbivores on mesoherbivore occurrence was stronger for shorter species, regardless of diet or relatedness. Megaherbivores substantially reduced the expected total use intensity (i.e. cumulative dung density of all species) of mesoherbivores, but only minimally reduced the expected species richness (i.e. cumulative predicted occurrence probabilities of all species) of mesoherbivores (by <1 species). Simulated extirpation of megaherbivores altered use intensity by mesoherbivores, which should be considered during (re)introductions of megaherbivores or their ecological proxies. Species' traits (in this case shoulder height) may be more reliable predictors of mesoherbivores' responses to megaherbivores than phylogenetic relatedness, and may be useful for predicting responses of data-limited species.

Density dependence and the spread of invasive big-headed ants (Pheidole megacephala) in an East African savanna.

Supercolonial ants are among the largest cooperative units in nature, attaining extremely high densities. How these densities feed back into their population growth rates and how abundance and extrinsic factors interact to affect their population dynamics remain open questions. We studied how local worker abundance and extrinsic factors (rain, tree density) affect population growth rate and spread in the invasive big-headed ant, which is disrupting a keystone mutualism between acacia trees and native ants in parts of East Africa. We measured temporal changes in big-headed ant (BHA) abundance and rates of spread over 20 months along eight transects, extending from areas behind the front with high BHA abundances to areas at the invasion front with low BHA abundances. We used models that account for negative density dependence and incorporated extrinsic factors to determine what variables best explain variation in local population growth rates. Population growth rates declined with abundance, however, the strength of density dependence decreased with abundance. We suggest that weaker density dependence at higher ant abundances may be due to the beneficial effect of cooperative behavior that partially counteracts resource limitation. Rainfall and tree density had minor effects on ant population dynamics. BHA spread near 50 m/year, more than previous studies reported and comparable to rates of spread of other supercolonial ants. Although we did not detect declines in abundance in areas invaded a long time ago (> 10 years), continued monitoring of abundance at invaded sites may help to better understand the widespread collapse of many invasive ants.

Aridity weakens population-level effects of multiple species interactions on Hibiscus meyeri.

Predicting how species' abundances and ranges will shift in response to climate change requires a mechanistic understanding of how multiple factors interact to limit population growth. Both abiotic stress and species interactions can limit populations and potentially set range boundaries, but we have a poor understanding of when and where each is most critical. A commonly cited hypothesis, first proposed by Darwin, posits that abiotic factors (e.g., temperature, precipitation) are stronger determinants of range boundaries in apparently abiotically stressful areas ("stress" indicates abiotic factors that reduce population growth), including desert, polar, or high-elevation environments, whereas species interactions (e.g., herbivory, competition) play a stronger role in apparently less stressful environments. We tested a core tenet of this hypothesis-that population growth rate is more strongly affected by species interactions in less stressful areas-using experimental manipulations of species interactions affecting a common herbaceous plant, Hibiscus meyeri (Malvaceae), across an aridity gradient in a semiarid African savanna. Population growth was more strongly affected by four distinct species interactions (competition with herbaceous and shrubby neighbors, herbivory, and pollination) in less stressful mesic areas than in more stressful arid sites. However, contrary to common assumptions, this effect did not arise because of greater density or diversity of interacting species in less stressful areas, but rather because aridity reduced sensitivity of population growth to these interactions. Our work supports classic predictions about the relative strength of factors regulating population growth across stress gradients, but suggests that this pattern results from a previously unappreciated mechanism that may apply to many species worldwide.

The Epigenetic Signature of Colonizing New Environments in Anolis Lizards.

Founder populations often show rapid divergence from source populations after colonizing new environments. Epigenetic modifications can mediate phenotypic responses to environmental change and may be an important mechanism promoting rapid differentiation in founder populations. Whereas many long-term studies have explored the extent to which divergence between source and founder populations is genetically heritable versus plastic, the role of epigenetic processes during colonization remains unclear. To investigate epigenetic modifications in founding populations, we experimentally colonized eight small Caribbean islands with brown anole lizards (Anolis sagrei) from a common source population. We then quantitatively measured genome-wide DNA methylation in liver tissue using reduced representation bisulfite sequencing of individuals transplanted onto islands with high- versus low-habitat quality. We found that lizard sex and habitat quality explained a significant proportion of epigenetic variation. Differentially methylated cytosines mapped to genes that encode proteins with functions likely to be relevant to habitat change (e.g., signal transduction, immune response, circadian rhythm). This study provides experimental evidence of a relationship between epigenetic responses and the earliest stages of colonization of novel environments in nature and suggests that habitat quality influences the nature of these epigenetic modifications.

Economy of scale: third partner strengthens a keystone ant-plant mutualism.

While foundation species can stabilize ecosystems at landscape scales, their ability to persist is often underlain by keystone interactions occurring at smaller scales. Acacia drepanolobium is a foundation tree, comprising >95% of woody cover in East African black-cotton savanna ecosystems. Its dominance is underlain by a keystone mutualistic interaction with several symbiotic ant species in which it provides housing (swollen thorns) and carbohydrate-rich nectar from extra-floral nectaries (EFN). In return, it gains protection from catastrophic damage from mega-herbivores. Crematogaster mimosae is the ecologically dominant symbiotic ant in this system, also providing the highest protection services. In addition to tending EFN, C. mimosae tend scale insects for carbohydrate-rich honeydew. We investigated the role of scale insects in this specialized ant-plant interaction. Specifically, does this putatively redundant third partner strengthen the ant-plant mutualism by making the ant a better protector of the tree? Or does it weaken the mutualism by being costly to the tree while providing no additional benefit to the ant-plant mutualism? We coupled observational surveys with two scale-manipulation experiments and found evidence that this third partner strengthens the ant-plant mutualism. Trees with scale insects experimentally removed experienced a 2.5X increase in elephant damage compared to trees with scale insects present over 10 months. Reduced protection was driven by scale removal causing a decrease in ant colony size and per capita baseline activity and defensive behavior. We also found that ants increased scale-tending and the density of scale insects on trees when EFN were experimentally reduced. Thus, in this system, scale insects and EFN are likely complementary, rather than redundant, resources with scale insects benefitting ants when EFN production is low (such as during annual dry periods in this semi-arid ecosystem). This study reveals that a third-partner strengthens an ant-plant mutualism that serves to stabilize a whole ecosystem.

Good neighbors make good defenses: associational refuges reduce defense investment in African savanna plants.

Intraspecific variation in plant defense phenotype is common and has wide-ranging ecological consequences. Yet prevailing theories of plant defense allocation, which primarily account for interspecific differences in defense phenotype, often fail to predict intraspecific patterns. Furthermore, although individual variation in defense phenotype is often attributed to ecological interactions, few general mechanisms have been proposed to explain the ubiquity of variable defense phenotype within species. Here, we show experimentally that associational refuges and induced resistance interact to create predictable intraspecific variation in defense phenotype in African savanna plants. Physically defended species from four families (Acanthaceae, Asparagaceae, Cactaceae, and Solanaceae) growing in close association with spinescent Acacia trees had 39-78% fewer spines and thorns than did isolated conspecifics. For a subset of these species, we used a series of manipulative experiments to show that this variability is maintained primarily by a reduction in induced responses among individuals that seldom experience mammalian herbivory, whether due to association with Acacia trees or to experimental herbivore exclusion. Unassociated plants incurred 4- to 16-fold more browsing damage than did associated individuals and increased spine density by 16-38% within one month following simulated browsing. In contrast, experimental clipping induced no net change in spine density among plants growing beneath Acacia canopies or inside long-term herbivore exclosures. Associated and unassociated individuals produced similar numbers of flowers and seeds, but seedling recruitment and survival were vastly greater in refuge habitats, suggesting a net fitness benefit of association. We conclude that plant-plant associations consistently decrease defense investment in this system by reducing both the frequency of herbivory and the intensity of induced responses, and that inducible responses enable plants to capitalize on such associations in heterogeneous environments. Given the prevalence of associational and induced defenses in plant communities worldwide, our results suggest a potentially general mechanism by which biotic interactions might predictably shape intraspecific variation in plant defense phenotype.

Skin shedding and tissue regeneration in African spiny mice (Acomys).

Evolutionary modification has produced a spectrum of animal defence traits to escape predation, including the ability to autotomize body parts to elude capture. After autotomy, the missing part is either replaced through regeneration (for example, in urodeles, lizards, arthropods and crustaceans) or permanently lost (such as in mammals). Although most autotomy involves the loss of appendages (legs, chelipeds, antennae or tails, for example), skin autotomy can occur in certain taxa of scincid and gekkonid lizards. Here we report the first demonstration of skin autotomy in Mammalia (African spiny mice, Acomys). Mechanical testing showed a propensity for skin to tear under very low tension and the absence of a fracture plane. After skin loss, rapid wound contraction was followed by hair follicle regeneration in dorsal skin wounds. Notably, we found that regenerative capacity in Acomys was extended to ear holes, where the mice exhibited complete regeneration of hair follicles, sebaceous glands, dermis and cartilage. Salamanders capable of limb regeneration form a blastema (a mass of lineage-restricted progenitor cells) after limb loss, and our findings suggest that ear tissue regeneration in Acomys may proceed through the assembly of a similar structure. This study underscores the importance of investigating regenerative phenomena outside of conventional model organisms, and suggests that mammals may retain a higher capacity for regeneration than was previously believed. As re-emergent interest in regenerative medicine seeks to isolate molecular pathways controlling tissue regeneration in mammals, Acomys may prove useful in identifying mechanisms to promote regeneration in lieu of fibrosis and scarring.

Interacting effects of wildlife loss and climate on ticks and tick-borne disease.

Both large-wildlife loss and climatic changes can independently influence the prevalence and distribution of zoonotic disease. Given growing evidence that wildlife loss often has stronger community-level effects in low-productivity areas, we hypothesized that these perturbations would have interactive effects on disease risk. We experimentally tested this hypothesis by measuring tick abundance and the prevalence of tick-borne pathogens (Coxiella burnetii and Rickettsia spp.) within long-term, size-selective, large-herbivore exclosures replicated across a precipitation gradient in East Africa. Total wildlife exclusion increased total tick abundance by 130% (mesic sites) to 225% (dry, low-productivity sites), demonstrating a significant interaction of defaunation and aridity on tick abundance. When differing degrees of exclusion were tested for a subset of months, total tick abundance increased from 170% (only mega-herbivores excluded) to 360% (all large wildlife excluded). Wildlife exclusion differentially affected the abundance of the three dominant tick species, and this effect varied strongly over time, likely due to differences among species in their host associations, seasonality, and other ecological characteristics. Pathogen prevalence did not differ across wildlife exclusion treatments, rainfall levels, or tick species, suggesting that exposure risk will respond to defaunation and climate change in proportion to total tick abundance. These findings demonstrate interacting effects of defaunation and aridity that increase disease risk, and they highlight the need to incorporate ecological context when predicting effects of wildlife loss on zoonotic disease dynamics.

Influence of neighboring plants on the dynamics of an ant-acacia protection mutualism.

Ant-plant protection symbioses, in which plants provide food and/or shelter for ants in exchange for protection from herbivory, are model systems for understanding the ecology of mutualism. While interactions between ants, host plants, and herbivores have been intensively studied, we know little about how plant-plant interactions influence the dynamics of these mutualisms, despite strong evidence that plants compete for resources, that hosting ants can be costly, and that host-plant provisioning to ants can therefore be constrained by resource availability. We used field experiments in a semiarid Kenyan savanna to examine interactions between the ant-plant Acacia drepanolobium, neighboring grasses, and two species of symbiotic acacia-ants with divergent behaviors: Crematogaster mimosae, an aggressive symbiont that imposes high costs to host trees via consumption of extrafloral nectar, and Tetraponera penzigi, a less-protective symbiont that imposes lower costs because it does not consume nectar. We hypothesized that by competing with acacias for resources, neighboring grasses (1) reduce hosts' ability to support costly C. mimosae, while having little or no effect on the ability of hosts to support low-cost T. penzigi, and (2) reduce sapling growth rates irrespective of ant occupant. We factorially manipulated the presence/absence of grasses and the identity of ant occupants on saplings and evaluated effects on colony survivorship and sapling growth rates over 40 weeks. Contrary to prediction, the high-cost/high-reward nectar-dependent mutualist C. mimosae had higher colony-survival rates on saplings with grass neighbors present. Grasses appear to have indirectly facilitated the survival of C. mimosae by reducing water stress on host plants; soils under saplings shaded by grasses had higher moisture content, and these saplings produced more active nectaries than grass-removal saplings. Consistent with prediction, survival of low-cost/low-reward T. penzigi did not differ significantly between grass-removal treatments. Saplings occupied by low-cost/low-reward T. penzigi grew 100% more on average than saplings occupied by high-cost/high-reward C. mimosae, demonstrating that mutualist-partner identity strongly and differentially influences demographic rates of young plants. In contrast, contrary to prediction, grass neighbors had no significant net impact on sapling growth rates. Our results suggest that neighboring plants can exert strong and counterintuitive effects on ant-plant protection symbioses, highlighting the need to integrate plant-plant interactions into our understanding of these mutualisms.

Climatic variation modulates the indirect effects of large herbivores on small-mammal habitat use.

Large mammalian herbivores (LMH) strongly shape the composition and architecture of plant communities. A growing literature shows that negative direct effects of LMH on vegetation frequently propagate to suppress the abundance of smaller consumers. Indirect effects of LMH on the behaviour of these consumers, however, have received comparatively little attention despite their potential ecological significance. We sought to understand (i) how LMH indirectly shape small-mammal habitat use by altering the density and distribution of understorey plants; (ii) how these effects vary with climatic context (here, seasonality in rainfall); and (iii) the extent to which behavioural responses of small mammals are contingent upon small-mammal density. We tested the effects of a diverse LMH community on small-mammal habitat use using 4 years of spatially explicit small-mammal trapping and vegetation data from the UHURU Experiment, a replicated set of LMH exclosures in semi-arid Kenyan savanna. Small-mammal habitat use was positively associated with tree density and negatively associated with bare (unvegetated) patches in all plots and seasons. In the presence of LMH, and especially during the dry season, small mammals consistently selected tree cover and avoided bare patches. In contrast, when LMH were excluded, small mammals were weakly associated with tree cover and did not avoid bare patches as strongly. These behavioural responses of small mammals were largely unaffected by changes in small-mammal density associated with LMH exclusion. Our results show that LMH indirectly affect small-mammal behaviour, and that these effects are influenced by climate and can arise via density-independent mechanisms. This raises the possibility that anthropogenic LMH declines might interact with changing patterns of rainfall to alter small-mammal distribution and behaviour, independent of numerical responses by small mammals to these perturbations. For example, increased rainfall in East Africa (as predicted in many recent climate-model simulations) may relax constraints on small-mammal distribution where LMH are rare or absent, whereas increased aridity and/or drought frequency may tighten them.

Large herbivores promote habitat specialization and beta diversity of African savanna trees.

Edaphic variation in plant community composition is widespread, yet its underlying mechanisms are rarely understood and often assumed to be physiological. In East African savannas, Acacia tree species segregate sharply across soils of differing parent material: the ant-defended whistling thorn, A. drepanolobium (ACDR), is monodominant on cracking clay vertisols that are nutrient rich but physically stressful, whereas poorly defended species such as A. brevispica (ACBR) dominate on nutrient-poor but otherwise less-stressful sandy loams. Using a series of field experiments, we show that large-mammal herbivory interacts with soil properties to maintain this pattern. In the absence of large herbivores, transplanted saplings of both species established on both soil types. Browsers strongly suppressed survival and growth of ACDR saplings on sandy soil, where resource limitation constrained defensive investment. On clay soil, ACBR saplings established regardless of herbivory regime, but elephants prevented recruitment to maturity, apparently because trees could not tolerate the combination of biotic and abiotic stressors. Hence, each tree species was filtered out of one habitat by browsing in conjunction with different edaphic factors and at different ontogenetic stages. Browser abundance was greater on sandy soil, where trees were less defended, consistent with predicted feedbacks between plant community assembly and herbivore distributions. By exploring two inversely related axes of soil "quality" (abiotic stress and nutrient content), our study extends the range of mechanisms by which herbivores are known to promote edaphic specialization, illustrates how the high cost of a protection mutualism can constrain the realized niche of host trees, and shows that large-scale properties of savanna ecosystems are shaped by species interactions in cryptic ways that mimic simple abiotic determinism. These results suggest that ongoing declines in large-herbivore populations may relax spatial heterogeneity in plant assemblages and reduce the beta diversity of communities.

Elephants in the understory: opposing direct and indirect effects of consumption and ecosystem engineering by megaherbivores.

Positive indirect effects of consumers on their resources can stabilize food webs by preventing overexploitation, but the coupling of trophic and non-trophic interactions remains poorly integrated into our understanding of community dynamics. Elephants engineer African savanna ecosystems by toppling trees and breaking branches, and although their negative effects on trees are well documented, their effects on small-statured plants remain poorly understood. Using data on 117 understory plant taxa collected over 7 yr within 36 1-ha experimental plots in a semi-arid Kenyan savanna, we measured the strength and direction of elephant impacts on understory vegetation. We found that elephants had neutral effects on most (83-89%) species, with a similar frequency of positive and negative responses among the remainder. Overall, estimated understory biomass was 5-14% greater in the presence of elephants across a range of rainfall levels. Whereas direct consumption likely accounts for the negative effects, positive effects are presumably indirect. We hypothesized that elephants create associational refuges for understory plants by damaging tree canopies in ways that physically inhibit feeding by other large herbivores. As predicted, understory biomass and species richness beneath elephant-damaged trees were 55% and 21% greater, respectively, than under undamaged trees. Experimentally simulated elephant damage increased understory biomass by 37% and species richness by 49% after 1 yr. Conversely, experimentally removing elephant damaged branches decreased understory biomass by 39% and richness by 30% relative to sham-manipulated trees. Camera-trap surveys revealed that elephant damage reduced the frequency of herbivory by 71%, whereas we detected no significant effect of damage on temperature, light, or soil moisture. We conclude that elephants locally facilitate understory plants by creating refuges from herbivory, which countervails the direct negative effects of consumption and enhances larger-scale biomass and diversity by promoting the persistence of rare and palatable species. Our results offer a counterpoint to concerns about the deleterious impacts of elephant "overpopulation" that should be considered in debates over wildlife management in African protected areas: understory species comprise the bulk of savanna plant biodiversity, and their responses to elephants are buffered by the interplay of opposing consumptive and non-consumptive interactions.

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.