Savannas cover a fifth of the land surface and contribute a third of terrestrial net primary production, accounting for three-quarters of global area burned and more than half of global fire-driven carbon emissions1,2,3. Fire suppression and afforestation have been proposed as tools to increase carbon sequestration in these ecosystems2,4. A robust quantification of whole-ecosystem carbon storage in savannas is lacking however, especially under altered fire regimes. Here we provide one of the first direct estimates of whole-ecosystem carbon response to more than 60 years of fire exclusion in a mesic African savanna. We found that fire suppression increased whole-ecosystem carbon storage by only 35.4 ± 12% (mean ± standard error), even though tree cover increased by 78.9 ± 29.3%, corresponding to total gains of 23.0 ± 6.1 Mg C ha−1 at an average of about 0.35 ± 0.09 Mg C ha−1 year−1, more than an order of magnitude lower than previously assumed4. Frequently burned savannas had substantial belowground carbon, especially in biomass and deep soils. These belowground reservoirs are not fully considered in afforestation or fire-suppression schemes but may mean that the decadal sequestration potential of savannas is negligible, especially weighed against concomitant losses of biodiversity and function.
The use of digital elevation models has proven to be crucial in numerous studies related to savanna ecosystem research. However, the insufficient spatial resolution of the chosen input data is often considered to be a limiting factor when conducting local to regional scale ecosystem analysis. The elevation models and orthorectified imagery created in this study represent the first wall-to-wall digital elevation data sets produced for the Kruger National Park (KNP), South Africa, at very high spatial resolution. Using colour-infrared (CIR) aerial imagery from the archives of the Chief Directorate: National Geo-spatial Information (CDNGI), Department of Agriculture, Land Reform and Rural Development (DALRRD) aerial acquisition programme, we created digital surface models (DSMs), digital terrain models (DTMs) and CIR orthomosaics covering the entire KNP with a nominal ground sampling distance of 0.25 m. Elevation information was derived using state-of-the-art stereo matching algorithms that utilised semi-global matching (SGM) as a cost aggregation function throughout the image pairing, using the Enterprise software from CATALYST. The final products were validated against reference products, and showed excellent agreement with R² values of 0.99. Further, the validation of the DTM and DSM revealed median absolute vertical height error (LE90) across all sites of 1.02 m and 2.58 m, respectively. The orthomosaics were validated with in situ ground control points (GCPs) exhibiting a horizontal Circular Probable Error (CPE) of 1.37 m. The data resulting from this work will be distributed freely with the aim of fostering more scientific studies in the African science community and beyond.
Conservation implications: Accurate information about terrain and surface height are crucial inputs to a variety of scientific analysis, which are essential in protected areas, such as flood prediction or fire hazard estimation. Elevation data sets and orthomosaics in very high resolution can therefore serve as a crucial tool to improve park management and foster positive implications on conservation efforts.
Herbivores balance forage acquisition with the need to avoid predation, often leading to trade‐offs between forgoing resources to avoid areas of high predation risk, or tolerating increased risk in exchange for improved forage. The outcome of these decisions is likely to change with varying resource levels, with herbivores altering their response to predation risk across heterogeneous landscapes. Such contrasting responses will alter the strength of non‐consumptive predation effects, but are poorly understood in multiple‐predator/multiple‐prey systems. We combined fine‐scaled spatial information on two predator and 11 herbivore species with remotely‐sensed measurements of forage quantity and vegetation structure to assess variation in herbivore response to predation risk with changing environmental context, herbivore body size, herbivore foraging strategy (browsers vs. grazers), predator type (ambush vs. coursing hunters) and group size across a South African savanna landscape. Medium‐sized herbivore species were more likely to adjust their response to risk with a changing resource landscape: warthog, nyala and wildebeest tolerated increased long‐term predator encounter risk in exchange for abundant (warthog and nyala) or preferred (wildebeest) forage, and nyala selected areas with higher visibility only in landscapes where food was abundant. Impala were more likely to be observed in areas of high visibility where wild dog risk was high. In addition, although buffalo did not avoid areas of high lion encounter risk, large buffalo groups were more frequently observed in open areas where lion encounter risk was high, whereas small groups did not alter their space use across varying levels of risk. Our findings suggest that risk effects are not uniform across landscapes for medium‐sized herbivores and large buffalo groups, instead varying with environmental context and leading to a dynamic landscape of fear. However, responses among these and other prey species were variable and not consistent, highlighting the complexities inherent to multi‐predator/multi‐prey systems.
Termites are globally dominant and functionally important soil organisms. While their role in ecosystems is being increasingly recognised and understood, methods that adequately sample termite communities across habitats can be challenging and have not advanced at the same pace as studies of termite ecology. Moreover, the appropriateness of sampling methods varies with habitat and biogeographic region due to differences in termite communities. Focusing largely on tropical systems where the majority of termites occur, we review currently available termite sampling methods and provide recommendations for sampling across different biomes and biogeographic regions. Active searching transects are most effective in rainforest habitats, whereas reduced transects, baiting and mound surveys are more appropriate in grassy systems and seasonally dry forests. Baiting is recommended for deserts. Recent advances in termite sampling, such as the use of remote sensing and DNA metabarcoding, and outstanding challenges, such as sampling episodic grass-feeding termites, are also discussed. Improved use of standardised termite sampling methods, as we recommend, should lead to increased knowledge of the patterns and drivers of termite diversity, which will, in turn, facilitate the quantification of the influence termites have on ecosystems and lead to new insights into the functioning of tropical systems.
Global climate change is predicted to increase the frequency of droughts, with major impacts on tropical savannas. It has been suggested that during drought, increased soil moisture and nutrients on termite mounds could benefit plants but it is unclear how such benefits could cascade to affect insect communities.
Here, we describe the effects of drought on vegetation structure, the cascading implications for invertebrates and how termite mounds influence such effects. We compared how changes in grass biomass affected grasshopper and ant diversity on and off Macrotermes mounds before (2012) and during a drought (2016) at two locations that experienced large variation in drought severity (Skukuza and Pretoriuskop) in the Kruger National Park, South Africa.
The 2013–2016 drought was not ubiquitous across the study site, with rainfall decreasing at Skukuza and being above average at Pretoriuskop. However, grass biomass declined at both locations. Grasshopper abundance decreased at droughted Skukuza both on and off mounds but decreased on mounds and increased off mounds at non-droughted Pretoriuskop. Ant abundance and species richness increased at Skukuza but remained the same on mounds and decreased off mounds at Pretoriuskop.
Our results demonstrate the spatially extensive effects of drought. Despite above average rainfall in 2016 at Pretoriuskop, grass biomass decreased, likely due to an influx of large mammalian herbivores from drought-affected areas. This decrease in grass biomass cascaded to affect grasshoppers and ants, further illustrating the effects of drought on invertebrates in adjoining areas with higher rainfall. Our grasshopper results also suggest that increased drought in savannas will contribute to overall declines in insect abundance. Moreover, our recorded increase in ant abundance was primarily in the form of increases in dominant species, illustrating how drought-induced shifts in relative abundance will likely influence ecosystem structure and function.
Our study highlights the phenomenon of spill-over drought effects and suggests rather than mitigating drought, termite mounds can instead become the focus for more intense grazing, with important consequences for insect communities.
Competitively dominant carnivore species can limit the population sizes and alter the behavior of inferior competitors. Established mechanisms that enable carnivore coexistence include spatial and temporal avoidance of dominant predator species by subordinates, and dietary niche separation. However, spatial heterogeneity across landscapes could provide inferior competitors with refuges in the form of areas with lower competitor density and/or locations that provide concealment from competitors. Here, we combine temporally overlapping telemetry data from dominant lions (Panthera leo) and subordinate African wild dogs (Lycaon pictus) with high-resolution remote-sensing in an integrated step selection analysis to investigate how fine-scaled landscape heterogeneity might facilitate carnivore coexistence in South Africa’s Hluhluwe-iMfolozi Park, where both predators occur at exceptionally high densities. We ask whether the primary lion avoidance strategy of wild dogs is spatial avoidance of lions or areas frequented by lions, or if wild dogs selectively use landscape features to avoid detection by lions. Within this framework, we also test whether wild dogs rely on proactive or reactive responses to lion risk. In contrast to previous studies finding strong spatial avoidance of lions by wild dogs, we found that the primary wild dog lion-avoidance strategy was to select landscape features that aid in lion avoidance. This habitat selection was routinely used by wild dogs, and especially when in areas and during times of high lion encounter risk, suggesting a proactive response to lion risk. Our findings suggest that spatial landscape heterogeneity could represent an alternative mechanism for carnivore coexistence, especially as ever-shrinking carnivore ranges force inferior competitors into increased contact with dominant species.
Endangered Bornean elephants are severely threatened by ongoing habitat transformation and increasing levels of human-elephant conflict. Understanding how elephants move across intact and transformed landscapes, as well as within them, is therefore of vital importance for the successful implementation of conservation management initiatives. We combined remote sensing and GPS telemetry data to identify broad habitat utilization and key movement areas to aid elephant management and conflict mitigation in three spatially-isolated populations in central and eastern Sabah, Malaysian Borneo. Home ranges were estimated using Brownian Bridge Movement Models and specific behavioral movement traits were identified by pathway analyses. These behavioral traits enabled a fine-scale evaluation of movements between and adjacent to forest patches and the role of large-scale agriculture in shaping elephant movements. Both natural (topological) and anthropogenic (agricultural) landscape features were found to have a broad influence on elephant movements. All elephant populations exhibited human-mediated behavioral responses, regardless of disturbance level. Throughout their range, elephants appeared to actively select relatively degraded forests, as measured by aboveground carbon density. However, elephants actively avoided urbanized areas, including roads and villages. Throughout the elephant range, high-speed, low-trajectory movements were found at low aboveground carbon locations, with 27% of all such movements located in large-scale agriculture. Our results suggest that agriculture impacts movement strategies of elephants, with evidence of repeat agricultural use pointing towards an active rationale for this behavior. Elephants were also found to use ridgelines as movement pathways, providing further context for the protection of such forested areas. The Lower Kinabatangan population, located in small remnant forests, travelled further to meet their ecological needs, suggesting the population is under added strain. Our work represents the broadest landscape assessment of Bornean elephant movements to-date and has important implications for both future work and habitat-level protected area management strategies.
Optimal foraging theory predicts less diverse predator diets with a greater availability of preferred prey. This narrow diet niche should then be dominated by preferred prey, with implications for predator–prey dynamics and prey population ecology. We investigated lion (Panthera leo) diets in Hluhluwe–iMfolozi Park (HiP), South Africa, to assess whether lions in a site with a high density of preferred prey (prey species weighing 92–632 kg as estimated from a published meta-analysis) have a narrow diet, consisting primarily of preferred prey. HiP is a useful study site to investigate this prediction because it is a productive landscape (with a high density of prey) where lion-preferred prey constitutes up to 33% of the prey available to lions. Furthermore, to investigate whether lions in HiP exhibit sex-specific diets as documented in other southern African populations, we estimated male and female lion diets separately. We were specifically interested in testing whether traditional approaches of estimating lion diets at the population level mask sex-specific predation patterns, with possible implications for management of lions in small to medium-sized fenced reserves. Lions in HiP preferred larger prey species (63–684 kg) and had diets with a larger proportion of preferred prey than reported in an African-wide meta-analysis. However, despite the high density of preferred prey species, 36% of lion diets still consisted of typically non-preferred species such as nyala (Tragelaphus angasii). This finding suggests that lions in HiP maintain a degree of opportunism even when preferred prey are abundant. Therefore, abundant, non-preferred prey are likely to be an important resource for lion populations. Sex-specific differences in lion diets were evident in HiP, suggesting that estimation of lion resource use and carrying capacity should consider opportunistic hunting and sex-specific differences in lion diets.
Humans pose a major threat to many species through land-use change in virtually every habitat. However, the extent of this threat is largely unknown for invertebrates due to challenges with investigating their distributions at large scales. This knowledge gap is particularly troublesome for soil macrofauna because of the critical roles many of these organisms perform as ecosystem engineers. We used a combination of high-resolution airborne Light Detection and Ranging and deep learning models to map the distribution of the ecologically important termite genus Macrotermes across a South African savanna land-use gradient, quantifying the effects of land-use change on patterns of mound densities, heights and spatial patterning. Despite significant anthropogenic alteration to landscapes, termite mounds persisted and shared a number of similarities to mounds in untransformed areas. Mean mound height was not substantially reduced in transformed landscapes, and over-dispersion of mounds at localized scales was conserved. However, mound densities were partially reduced, and height distributions in transformed areas differed to those in protected areas. Our findings suggest that mound-building termites persist even in areas of relatively high human disturbance, but also highlight important differences in termite distributions that could lead to reductions in ecosystem services provided by termites in human-modified landscapes. The persistence of at least half of mounds in human-modified landscapes could serve as starting points for savanna restoration.
Termite mounds create nutrient hotspots that serve as key resource areas for savanna vegetation and mammalian herbivores. However, despite the key ecological roles performed by termite mounds, few studies have investigated their influence on invertebrate communities, and none have examined such effects across environmental gradients. We hypothesised that termite mounds would support greater numbers of invertebrates than the surrounding savanna matrix and that assemblages would differ in composition due to the enhanced nutritional quality of vegetation on mounds. We also predicted that the differences between on-mound and off-mound invertebrate diversity would be more pronounced in areas where the difference in nutritional value between mounds and the savanna matrix vegetation was most prominent. We tested these hypotheses in Kruger National Park, South Africa, by sampling ground- and grass-dwelling invertebrate herbivores, omnivores and detritivores on and at various distances away from termite mounds at three savanna sites of varying vegetation quality across a rainfall gradient. All invertebrate groups sampled responded to termite mounds, but mound influence varied across trophic groups (Orthoptera showed the clearest patterns), diversity measures (changes in abundance rather than species richness) and with mean annual rainfall (strongest effects at the highest rainfall site). Orthoptera were more abundant on mounds, particularly at the wettest site, and there was a positive relationship between mound size and Orthoptera species richness. Ant assemblage composition on mounds differed from that off mounds and beetle abundance was greater on mounds, possibly as a result of concentrated mammalian herbivore activity and faecal deposition on mounds. Our results suggest that termite mounds are not only important nutrient and foraging hotspots for vertebrates, but that they also introduce heterogeneity in invertebrate communities, especially in nutrient-poor savannas.
Using remotely sensed imagery to identify biophysical components across landscapes is an important avenue of investigation for ecologists studying ecosystem dynamics. With high-resolution remotely sensed imagery, algorithmic utilization of image context is crucial for accurate identification of biophysical components at large scales. In recent years, convolutional neural networks (CNNs) have become ubiquitous in image processing, and are rapidly becoming more common in ecology. Because the quantity of high-resolution remotely sensed imagery continues to rise, CNNs are increasingly essential tools for large-scale ecosystem analysis. We discuss here the conceptual advantages of CNNs, demonstrate how they can be used by ecologists through distinct examples of their application, and provide a walkthrough of how to use them for ecological applications.
Understanding the drivers of vegetation carbon dynamics is essential for climate change mitigation and effective policy formulation. However, most efforts focus on abiotic drivers of plant biomass change, with little consideration for functional roles performed by animals, particularly at landscape scales. We combined repeat airborne Light Detection and Ranging with measurements of elephant densities, abiotic factors, and exclusion experiments to determine the relative importance of drivers of change in aboveground woody vegetation carbon stocks in Kruger National Park, South Africa. Despite a growing elephant population, aboveground carbon density (ACD) increased across most of the landscape over the 6‐year study period, but at fine scales, bull elephant density was the most important factor determining carbon stock change, with ACD losses recorded only where bull densities exceeded 0.5 bulls/km2. Effects of bull elephants were, however, spatially restricted and landscape dependent, being especially pronounced along rivers, at mid‐elevations, and on steeper slopes. In contrast, elephant herds and abiotic drivers had a comparatively small influence on the direction or magnitude of carbon stock change. Our findings demonstrate that animals can have a substantive influence on regional‐scale carbon dynamics and warrant consideration in carbon cycling models and policy formulation aimed at carbon management and climate change mitigation.
Foraging site selection by large herbivores is influenced by multiple factors varying across landscapes and spatial scales. Termite mounds harbour highly nutritious plants compared with the savanna matrix, making them preferred foraging patches in many savannas. However, it is unknown whether termite mounds equally influence herbivore grazing intensity across geological substrates and mound sizes. These knowledge gaps hamper our ability to draw general trans-ecosystem conclusions about the effect of termite mounds for savanna herbivores. We measured grazing intensity on mounds of three different size classes (small, medium and large) across two geologies with differing soil nutrition (granite and basalt) in Gonarezhou National Park, Zimbabwe. We recorded measurements across three seasons (hot wet, cool dry and hot dry), and at multiple distances from mounds. Grazing intensity on mounds was higher on nutrient-poor granite than nutrient-rich basalt, and termite mounds of all sizes had a significant effect on grazing on granite during the cool dry season. Grazing was highest on large mounds on both geologies throughout the year. Large mounds also had the largest spatial influence on grazing in the cool dry season, up to 8 m beyond the mound edge on granite and 2 m on basalt. When scaled up to the landscape level, mounds influenced about 15% of the granite landscape, but only about 0.5% of the basalt landscape. Our results show that the positive effects of mounds on grazing intensity were pronounced on nutrient-poor soils but negligible on nutrient-rich soils, and that the magnitude of these effects varied across seasons and with mound size.
The protection of biodiversity is critical to ecosystem function and is a primary management goal for conservation areas globally. Maintaining a current inventory of known diversity is a central component of achieving this goal and serves as an essential starting point for future research endeavours. Since the first published survey of termites in South Africa’s Kruger National Park (KNP) over 55 years ago, our understanding of termite diversity has expanded sufficiently to merit an update and formal checklist. Here we revise the inventory of termite diversity in KNP and summarise the taxonomic and functional diversity of termites in the park. A thorough review of recent termite research in KNP added 6 new genera and 13 species to what was found in Coaton’s original survey, with one genus, Anenteotermes, recorded for the first time in southern Africa. Based on the updated species checklist, the majority of genera in the park belong to Feeding Group II (39%) and the Termitidae family (75%). Conservation implications: In savannas, termites play crucial roles in nutrient cycling, water redistribution and plant dynamics. Systematically cataloguing termite diversity and assemblage composition in the park provides an essential baseline for scientific research, aids biodiversity conservation efforts and encourages scientists and managers to consider termites in ecosystem functioning and management. Having more detailed descriptions of genera, species and feeding groups allows for more tangible, ecologically relevant attributions of termite influence, facilitates enhanced inquiry and allows for more realistic quantification of termite roles in key ecosystem processes. Keywords: termite; Kruger National Park; feeding groups; diversity; savanna.
Primary tropical forests are becoming increasingly disturbed and fragmented, making it critically important to understand the conservation value of degraded forests. Many populations of even the largest and most iconic species are now found outside of primary habitats, and the long-term survival of these and many other species depends on appropriate management of degraded areas, whether protected or not. However, for conservation in degraded habitats to be successful, an adequate understanding of the minimal ecological requirements necessary for species persistence within them is required. We combined ground and helicopter nest surveys of critically endangered Bornean orangutans with high-resolution measurements of forest canopy structure from airborne Light Detection and Ranging (LiDAR) to understand orangutan nest site selection across multiple spatial scales in degraded forests of the Lower Kinabatangan region, Malaysian Borneo. We found orangutans to be selective when choosing nest sites, with nests more likely to be observed in canopies of tall and uniform height and closer to full canopy gaps, which was consistent across spatial scales and orangutan age and sex classes. These sites likely offer orangutans an improved vantage point and/or shelter from wind and rain. In contrast, no discernible relationships between nest site selection and canopy complexity, or nest abundance and landscape forest structure or aboveground carbon density were recorded. Our findings suggest that although orangutans do nest across a range of forest conditions, their optimum requirement for nesting strongly depends on forest patches with sufficient tall canopy of uniform height. These results serve to inform degraded forest conservation strategies across Borneo, particularly where orangutans are a focal species.
Predicting and managing the global carbon cycle requires scientific understanding of ecosystem processes that control carbon uptake and storage. It is generally assumed that carbon cycling is sufficiently characterized in terms of uptake and exchange between ecosystem plant and soil pools and the atmosphere. We show that animals also play an important role by mediating carbon exchange between ecosystems and the atmosphere, at times turning ecosystem carbon sources into sinks, or vice versa. Animals also move across landscapes, creating a dynamism that shapes landscape-scale variation in carbon exchange and storage. Predicting and measuring carbon cycling under such dynamism is an important scientific challenge. We explain how to link analyses of spatial ecosystem functioning, animal movement, and remote sensing of animal habitats with carbon dynamics across landscapes.
The disappearance of an endangered African wild dog population from Serengeti National Park (SNP) led to international debate centered around one question: were researchers to blame? The “Burrows' hypothesis” postulated that stress induced by research‐related immobilization and handling reactivated a latent rabies virus, eliminating the population. Insufficient data inhibited hypothesis testing, but since wild dogs persisted alongside SNP and have been studied since 2005, the hypothesis can be tested 25 years after its proposition. To be supported, wild dog immobilization interventions should have resulted in high mortality rates. However, 87.6% of 121 handled wild dogs (2006–2016) survived >12 months post‐handling. Some argued that viral reactivation would necessitate long‐term stress. Following immobilization, 67 animals were captured, transported, and held in a translocation enclosure. Despite the longer‐term stress, 95.5% survived >12 months. Furthermore, the stable number of wild dog packs in the ecosystem over the past decade, and lack of recolonization of SNP, strongly oppose Burrows' hypothesis. Instead, factors such as heightened levels of interspecific competition are likely to have contributed to the wild dog disappearance and subsequent avoidance of the Serengeti plains. Handling and radio telemetry are invaluable when studying elusive endangered species, yielding information pertinent to their conservation and management, and had no effect on Serengeti wild dog survival.
Termite mounds perform important roles in savanna ecosystems, generating heterogeneity and influencing ecosystem processes across multiple trophic levels. However, the influence the environment and neighboring termite colonies have on mound spatial patterning and structure is poorly understood, despite the profound implications such dynamics can have on ecosystems. To better understand these drivers, we mapped the spatial distribution and size of active and inactive Macrotermes mounds in eight 1-km2 plots on contrasting geologies, nutrient-rich granite and nutrient-poor basalt, in a semi-arid Zimbabwean savanna. Although mound density was not significantly different between basalt (5.5 mounds/ha) and granite (6.1 mounds/ha), termite mound structural attributes and spatial distribution patterns varied greatly between geologies. Mound size distributions differed between the geologies and mounds were 2.6 times taller and 3.9 times wider and had 15 times greater lateral surface area on granite. Subsequently, 6% of the total landscape was covered by mounds on granite compared with only 0.4% on basalt. On granite, large mounds exhibited significant over-dispersion at scales below 30 m, signifying density-dependent thinning. Furthermore, small mounds were clustered around large mounds, likely a result of the budding of new colonies comprising fully fledged castes less vulnerable to competition. In contrast, random patterning was evident on comparably homogenous basalt. Our results demonstrate the powerful influence geological substrate has on mound spatial patterning and structure, suggesting that the importance of termite mounds for ecosystem functioning is more pronounced on nutrient-poor granitic substrates than basalts because of the pronounced over-dispersion, which maximizes mound production per unit area, and much larger mound sizes here.
Woody encroachment can lead to a complete switch from open habitats to dense thickets, and has the potential to greatly alter the biodiversity and ecological functioning of grassy ecosystems across the globe. Plant litter decomposition is a critical ecosystem process fundamental to nutrient cycling and global carbon dynamics, yet little is known about how woody encroachment might alter this process. We compared grass decay rates of heavily encroached areas with adjacent nonencroached open areas in a semi-arid South African savanna using litterbags that allowed or excluded invertebrates. We also assessed the effect of woody encroachment on the activity of termites- dominant decomposer organisms in savanna systems. We found a significant reduction in decomposition rates within encroached areas, with litter taking twice as long to decay compared with open savanna areas. Moreover, invertebrates were more influential on grass decomposition in open areas and termite activity was substantially lower in encroached areas, particularly during the dry season when activity levels were reduced to almost zero. Our results suggest that woody encroachment created an unfavourable environment for invertebrates, and termites in particular, leading to decreased decomposition rates in these areas. We provide the first quantification of woody encroachment altering the functioning of African savanna ecosystems through the slowing of aboveground plant decomposition. Woody encroachment is intensifying across the globe, and our results suggest that substantial changes to the carbon balance and biodiversity of grassy biomes could occur.
Megafauna strongly affect vegetation structure and composition, often leading to management concern. However, the extent of their influence across large scales and varying ecosystems remains largely unknown. Using high resolution airborne Light Detection and Ranging (LiDAR), we investigated landscape-scale changes in vegetation height and three-dimensional (3D) structure across landscapes of varying elephant densities and presence over time, and in response to surface water distribution and terrain variability in the heavily managed thicket biome of the Addo Elephant National Park, South Africa. Elephants caused up to a fourfold reduction in vegetation height and altered the vertical profile, but increased vegetation height variability. Vegetation height also increased with elevation and distance from water, particularly in areas that elephants had long occupied at high densities. Slope had opposing effects on vegetation height, with height increasing with slope in areas long exposed to elephants, but decreasing where elephants had only recently been granted access. Our results suggest that elephants are the primary agents of vegetation change in this ecosystem, but that the strength of their effects varies across the landscape, enabling management to use water and terrain as mitigation tools. We further highlight the necessity of landscape-level experimental studies on megafaunal effects to untangle mechanisms and establish causality.