Carbon dioxide and the uneasy interactions of trees and savannah grasses

Savannahs are a mixture of trees and grasses often occurring as alternate states to closed forests. Savannah fires are frequent where grass productivity is high in the wet season. Fires help maintain grassy vegetation where the climate is suitable for woodlands or forests. Saplings in savannahs are particularly vulnerable to topkill of above-ground biomass. Larger trees are more fire-resistant and suffer little damage when burnt. Recruitment to large mature tree size classes depends on sapling growth rates to fire-resistant sizes and the time between fires. Carbon dioxide (CO(2)) can influence the growth rate of juvenile plants, thereby affecting tree recruitment and the conversion of open savannahs to woodlands. Trees have increased in many savannahs throughout the world, whereas some humid savannahs are being invaded by forests. CO(2) has been implicated in this woody increase but attribution to global drivers has been controversial where changes in grazing and fire have also occurred. We report on diverse tests of the magnitude of CO(2) effects on both ancient and modern ecosystems with a particular focus on African savannahs. Large increases in trees of mesic savannahs in the region cannot easily be explained by land use change but are consistent with experimental and simulation studies of CO(2) effects. Changes in arid savannahs seem less obviously linked to CO(2) effects and may be driven more by overgrazing. Large-scale shifts in the tree-grass balance in the past and the future need to be better understood. They not only have major impacts on the ecology of grassy ecosystems but also on Earth-atmosphere linkages and the global carbon cycle in ways that are still being discovered.     

Tree size and herbivory determine below-canopy grass quality and species composition in savannahs

Large single-standing trees are rapidly declining in savannahs, ecosystems supporting a high diversity of large herbivorous mammals. Savannah trees are important as they support both a unique flora and fauna. The herbaceous layer in particular responds to the structural and functional properties of a tree. As shrubland expands stem thickening occurs and large trees are replaced by smaller trees. Here we examine whether small trees are as effective in providing advantages for grasses growing beneath their crowns as large trees are. The role of herbivory in this positive tree-grass interaction is also investigated. We assessed soil and grass nutrient content, structural properties, and herbaceous species composition beneath trees of three size classes and under two grazing regimes in a South African savannah. We found that grass leaf content (N and P) beneath the crowns of particularly large (ca. 3.5 m) and very large trees (ca. 9 m) was as much as 40% greater than the same grass species not growing under a tree canopy, whereas nutrient contents of grasses did not differ beneath small trees (<2.3 m). Moderate herbivory enhanced these effects slightly. Grass species composition differed beneath and beyond the tree canopy but not between tree size classes. As large trees significantly improve the grass nutrient quality for grazers in contrast to smaller trees, the decline of the former should be halted. The presence of trees further increases grass species diversity and patchiness by favouring shade-tolerant species. Both grazing wildlife and livestock will benefit from the presence of large trees because of their structural and functional importance for savannahs.     

Feedback of trees on nitrogen mineralization to restrict the advance of trees in C4 savannahs

Remote sensing studies suggest that savannahs are transforming into more tree-dominated states; however, progressive nitrogen limitation could potentially retard this putatively CO₂-driven invasion. We analysed controls on nitrogen mineralization rates in savannah by manipulating rainfall and the cover of grass and tree elements against the backdrop of the seasonal temperature and rainfall variation. We found that the seasonal pattern of nitrogen mineralization was strongly influenced by rainfall, and that manipulative increases in rainfall could boost mineralization rates. Additionally, mineralization rates were considerably higher on plots with grasses and lower on plots with trees. Our findings suggest that shifting a savannah from a grass to a tree-dominated state can substantially reduce nitrogen mineralization rates, thereby potentially creating a negative feedback on the CO₂-induced invasion of savannahs by trees.     

Post-fire regeneration strategies and tree bark resistance to heating in frequently burning tropical savanna woodlands and grasslands in Ethiopia

Regeneration mechanisms of vegetation and the role of tree bark resistance to frequent fire were studied in savanna woodlands and grasslands in Gambella, Western Ethiopia. Data were collected from four sites, each with three replicate plots. The variation between sites in species composition and biomass correlated with the differences in fire intensity. Foliar cover was recorded for individual plant species regenerating by sprouting from older parts of plants that had survived fire or by seedlings; records were made during the dry season and at the beginning of the wet season. Data on bark thickness and tree diameters of 12 dominant tree species were also recorded. Both facultative and obligate sprouters significantly contributed to post‐fire recovery, comprising 98.5 % of total vegetation cover. The contribution of seedlings to cover and abundance immediately following fire was negligible, but seedling density increased in the beginning of the rainy season, 4 to 5 months after fire. The importance of the sprouting and seeding strategies varied between the different plant growth forms. The highest contribution to cover and frequency was made by the most abundant grass species, which reproduced in both ways. Facultative sprouters made up 67.3 % of the vegetation cover, out of which 54 % consisted of grasses. Broad‐leaved herbs and trees/shrubs regenerating mainly by sprouting made up 31.3 % of the vegetation cover. Adaptations to fire in tree species seemed to include the development of a thick bark, once the tree has passed seedling stages. Tree bark thickness and tree diameter at breast height were strongly correlated with the time taken for cambium to reach an assumed lethal temperature of 60°C when exposed to fire, which indicated that mature trees with thick barks might resist stronger fire better than, e.g., small or young trees and trees with thin bark. However, for a given bark thickness the cambium resistance to heat varied three‐fold among species. Hence, site differences in fire intensity seemed to influence the distribution of trees depending on their bark characteristics and resistance to fire.     

Plant responses and herbivory following simulated browsing and stem cutting of Combretum apiculatum

Abstract. Time and mode of herbivory on savanna trees and their subsequent responses are dependent on, among other things, earlier herbivory and fire. We used clipping (simulated browsing) and stem cutting (simulated heavy browsing and to some extent simulated fire) to evaluate such interactions. Study organisms were a deciduous, broad‐leaved tree species, Combretum apiculatum (Combretaceae), browsing large herbivores and leaf‐eating insects. The treatments were done in the late dry season before bud break. Late in the following wet season, we recorded plant responses to treatment and browsing. The treated trees, especially the cut ones, responded by producing larger and fewer annual shoots. Compared to control trees, there was a slight increase in shoot biomass of clipped trees and a strong reduction of cut ones. Leaf area increased in clipped trees, but decreased in cut ones. A marked increase in the number of browsed trees was recorded amongst treated trees. Number of bites, consumption and utilization also increased with severity of treatment. In contrast, insect herbivory was reduced on both clipped and cut trees. The observed patterns are discussed in relation to current ideas on plant ‐ herbivore interactions.     

Effects of fire history and N and P fertilization on seedling biomass,Specific Leaf Area,and root:shoot ratios in a South African savannah

Fire exerts clear direct effects on savannah vegetation dynamics, but the indirect effects of fire, for example via changes in soil fertility, are less clear. In fire-dominated nutrient-poor savannahs, there is little evidence that fire affects plant-available nutrients, but plant performance on soils with contrasting fire histories has not been investigated. A bioassay experiment was conducted in Kruger National Park, South Africa, to test for effects of long-term fire manipulation on the ability of soils to influence tree growth. Combretum hereroense, a common savannah species, was used as a phytometer species in an experiment that combined fire history with N and P fertilization treatments. The results suggested no effect of fire history on a number of seedling traits after seven months of growth. Root mass and shoot:root ratios responded to P addition, with seedlings exhibiting lower allocation to root tissue under elevated P, particularly in the presence of N, suggesting some N and P co-limitation. Overall, it is concluded that there is little evidence that fire degrades or alters soil chemistry in a way that is likely to influence vegetation structure and dynamics in nutrient-poor savannah ecosystems.     

Arthropod responses to experimental fire regimes in an Australian tropical savannah: ordinal‐level analysis

Fire is widely used for conservation management in the savannah landscapes of northern Australia, yet there is considerable uncertainty over the ecological effects of different fire regimes. The responses of insects and other arthropods to fire are especially poorly known, despite their dominant roles in the functioning of savannah ecosystems. Fire often appears to have little long‐term effect on ordinal‐level abundance of arthropods in temperate woodlands and open forests of southern Australia, and this paper addresses the extent to which such ordinal‐level resilience also occurs in Australia’s tropical savannahs. The data are from a multidisciplinary, landscape‐scale fire experiment at Kapalga in Kakadu National Park. Arthropods were sampled in the two major savannah habitats (woodland and open forest) using pitfall traps and sweep nets, in 15–20 km² compartments subjected to one of three fire regimes, each with three replicates: ‘early’ (annual fires lit early in the dry season), ‘late’ (annual fires lit late in the dry season), and ‘unburnt’ (fires absent during the five‐year experimental period 1990–94). Floristic cover, richness and composition were also measured in each sampling plot, using point quadrats. There were substantial habitat differences in floristic composition, but fire had no measured effect on plant richness, overall composition, or cover of three of the four dominant species. Of the 11 ordinal arthropod taxa considered from pitfall traps, only four were significantly affected by fire according to repeated‐measures ANOVA . There was a marked reduction in ant abundance in the absence of fire, and declines in spiders, homopterans and silverfish under late fires. Similarly, the abundances of only four of the 10 ordinal taxa from sweep catches were affected by fire, with crickets and beetles declining in the absence of fire, and caterpillars declining under late fires. Therefore, most of the ordinal taxa from the ground and grass‐layer were unaffected by the fire treatments, despite the treatments representing the most extreme fire regimes possible in the region. This indicates that the considerable ordinal‐level resilience to fire of arthropod assemblages that has previously been demonstrated in temperate woodlands and open forests of southern Australia, also occurs in tropical savannah woodlands and open forests of northern Australia.     

Effects of Prescribed Burning and Mechanical Bush Clearing on Ungulate Space Use in an African Savannah

Savannah ecosystems exhibit constant transitions between states dominated by trees and states dominated by a combination of trees and grasses. Transitions between these states are driven by interactions between fire and herbivory. Bush encroachment (i.e. an increase in the density of woody vegetation) is often caused by anthropogenic disturbance such as climate change, invasive plants, fire control, and livestock practices. As bush encroachment alters the dynamics between fire and herbivory, it may have significant impact on savannah ecosystems. Two of the most common measures to mitigate bush encroachment are prescribed burning and bush clearing by mechanical treatments. We studied the effects of these two mitigation measures on ungulate space use in Lapalala Wilderness, a private conservation area within the Waterberg Biosphere Reserve, northern South Africa. Burning and bush clearing affected both the overall abundance and the species composition of ungulates at particular patches, but these effects were influenced by habitat and the type of bush clearing treatment. Contrary to our expectations, most species occurred less frequently in burnt patches, and also less frequently in patches that had been bush cleared. Our results suggest that combined effects of fire and bush clearing may have positive effects on grazers and negative effects on browsers. Although our sampling design did not allow us to fully resolve interactive effects of burning and bush clearing treatments across habitats, our study highlights the fact that there are complex ecological consequences of habitat alterations in savannah ecosystems.     

An analysis of structure: biomass structure relationships for characteristic species of the western Kalahari,Botswana

Savannah ecosystems are important carbon stocks on the Earth, and their quantification is crucial for understanding the global impact of climate and land‐use changes in savannahs. The estimation of aboveground/belowground plant biomass requires tested allometric relationships that can be used to determine total plant biomass as a function of easy‐to‐measure morphological indicators. Despite recent advances in savannah ecology, research on allometric relations in savannahs remains confined to a few site‐specific studies where basal area is typically used as the main morphometric parameter with plant biomass. We investigate allometric relations at four sites along a 950‐km transect in the Kalahari across mean rainfall gradient 170 mm yr^?1–550 mm yr^?1. Using data from 342 harvested trees/shrubs, we relate basal area, height and crown diameter to aboveground biomass. These relationships are strongest in trees and weakest in small shrubs. Strong allometric relationships are also determined for morphologically similar groups of woody vegetation. We show that crown diameter can be used as an alternative to basal area in allometric relationships with plant biomass. This finding may enhance the ability to determine aboveground biomass over large areas using high‐resolution aerial or satellite imagery without requiring ground‐based measurements of basal area.     

The importance of low atmospheric CO2 and fire in promoting the spread of grasslands and savannas

The distribution and abundance of trees can be strongly affected by disturbance such as fire. In mixed tree/grass ecosystems, recurrent grass‐fuelled fires can strongly suppress tree saplings and therefore control tree dominance. We propose that changes in atmospheric [CO₂] could influence tree cover in such metastable ecosystems by altering their postburn recovery rates relative to flammable herbaceous growth forms such as grasses. Slow sapling recovery rates at low [CO₂] would favour the spread of grasses and a reduction of tree cover. To test the possible importance of [CO₂]/fire interactions, we first used a Dynamic Global Vegetation Model (DGVM) to simulate biomass in grassy ecosystems in South Africa with and without fire. The results indicate that fire has a major effect under higher rainfall conditions suggesting an important role for fire/[CO₂] interactions. We then used a demographic model of the effects of fire on mesic savanna trees to test the importance of grass/tree differences in postburn recovery rates. We adjusted grass and tree growth in the model according to the DGVM output of net primary production at different [CO₂] relative to current conditions. The simulations predicted elimination of trees at [CO₂] typical of the last glacial period (180 ppm) because tree growth rate is too slow (15 years) to grow to a fire‐proof size of ca. 3 m. Simulated grass growth would produce an adequate fuel load for a burn in only 2 years. Simulations of preindustrial [CO₂] (270 ppm) predict occurrence of trees but at low densities. The greatest increase in trees occurs from preindustrial to current [CO₂] (360 ppm). The simulations are consistent with palaeo‐records which indicate that trees disappeared from sites that are currently savannas in South Africa in the last glacial. Savanna trees reappeared in the Holocene. There has also been a large increase in trees over the last 50–100 years. We suggest that slow tree recovery after fire, rather than differential photosynthetic efficiencies in C₃ and C₄ plants, might have been the significant factor in the Late Tertiary spread of flammable grasslands under low [CO₂] because open, high light environments would have been a prerequisite for the spread of C₄ grasses. Our simulations suggest further that low [CO₂] could have been a significant factor in the reduction of trees during glacial times, because of their slower regrowth after disturbance, with fire favouring the spread of grasses.     

The influence of tree canopies and elephants on sub-canopy vegetation in a savannah

The apparent influence of elephants on the structure of savannahs in Africa may be enhanced by management activities, fire and other herbivores. We separated the effect elephants have on grasses, woody seedlings (<0.5 m) and saplings (0.5–2 m) from the effect of tree canopies (canopy effect), and herbivory (park effect). We defined the canopy effect as the differences between plant abundances and diversity indices under tree canopies and 20 m away from these. Our testing of the park effect relied on the differences in the sub-canopy plant indices inside and outside a protected area that supported a range of herbivores. We based our assessment of the elephant effect on sub-canopy vegetation indices associated with elephant induced reductions in tree canopies. The park and canopy effects were more pronounced than the elephant effect. The park effect suppressed the development of woody seedlings into saplings. Conditions associated with tree canopies benefited woody plants, but not the grasses, as their indices were lower under trees. Elephants reducing canopies facilitated grass species tolerant of direct solar radiation. We concluded that management should consider other agents operating in the system when deciding on reducing the impact that elephants may have on vegetation.     

Tolerance of a perennial herb, <Emphasis Type="Italic">Pimpinella saxifraga</Emphasis>, to simulated flower herbivory and grazing: immediate repair of injury or postponed reproduction?

Perennial, polycarpic herbs can respond to herbivory either by (1) regrowth in the same season in order to compensate for lost reproductive structures or by (2) postponing reproduction until the following growing season. We tested these response patterns with the perennial umbellifer Pimpinella saxifraga by simulating flower herbivory and shoot grazing both in the field and in a common garden experiment. In the field, both simulated flower herbivory and grazing effectively suppressed current reproduction, whereas no statistically significant effects of previous-year treatments on growth or reproduction were found in the following year. In the common garden, in the first year the species fully compensated for simulated flower herbivory in vegetative parameters but seed set was reduced by 26%. After 2 years of flower removal, the plants overcompensated in shoot and root biomass by 47 and 46%, respectively, and compensated fully in reproductive performance. Simulated grazing resulted in 21% lower shoot biomass in the first season, but the root biomass was not affected. In the second season the root biomass increased by 43% as compared to the control plants. However, regrowth following simulated grazing resulted in a significant delay in flowering with the consequence that the seed yield of fertile plants was reduced by 55% as compared to the control plants. These results suggest that in resource-rich garden conditions P. saxifraga may immediately repair injuries caused by flower herbivory, but repairs more extensive shoot injury less successfully. Delayed phenology decreases the benefits of immediate repair. In resource-poor conditions, the benefits of regrowth can be negligible. Accordingly, in our field population, the plants postponed their reproduction until the following year in response to simulated grazing and frequently in response to flower removal. When the plants gain very little from regrowth, the costs of reproduction would select for postponed reproduction in response to injury.     

Model highlights likely long‐term influences of mesobrowsers versus those of elephants on woodland dynamics

The potential long‐term influences of mesobrowsers versus those of savannah elephants on woodland dynamics have not been explored. This may be a critical omission especially in southern African savannahs, where efforts to preserve existing woodlands are typically directed at elephant management. We describe a simple browse–browser model, parameterized from an extensive review of the literature and our own data, including quantitative assessment of impala impact, from the study site, iMfolozi Park, South Africa. As there is a paucity of species‐specific demographic data on savannah woody species, we modelled, in a novel approach, functional groups of plant species typical of Acacia woodlands. Outputs suggest that over the long term (100 years), low‐to‐moderate densities of impala will have a similar impact on woodland structure, in terms of density of adult trees, as low‐to‐moderate densities of elephant. Further, the outputs highlight the apparently strong synergistic effect impala and elephant impacts combined have on woodland dynamics, suggesting that reduction or removal of either impala or elephant will radically reduce long‐term destruction of savannah woodlands. Recorded changes in adult tree numbers in iMfolozi broadly supported the model's outputs.     

Tree Species Composition and Harvest Intensity Affect Herbivore Density and Leaf Damage on Beech,Fagus sylvatica,in Different Landscape Contexts

Most forests are exposed to anthropogenic management activities that affect tree species composition and natural ecosystem processes. Changes in ecosystem processes such as herbivory depend on management intensity, and on regional environmental conditions and species pools. Whereas influences of specific forest management measures have already been addressed for different herbivore taxa on a local scale, studies considering effects of different aspects of forest management across different regions are rare. We assessed the influence of tree species composition and intensity of harvesting activities on arthropod herbivores and herbivore-related damage to beech trees, Fagus sylvatica, in 48 forest plots in three regions of Germany. We found that herbivore abundance and damage to beech trees differed between regions and that – despite the regional differences - density of tree-associated arthropod taxa and herbivore damage were consistently affected by tree species composition and harvest intensity. Specifically, overall herbivore damage to beech trees increased with increasing dominance of beech trees – suggesting the action of associational resistance processes – and decreased with harvest intensity. The density of leaf chewers and mines was positively related to leaf damage, and several arthropod groups responded to beech dominance and harvest intensity. The distribution of damage patterns was consistent with a vertical shift of herbivores to higher crown layers during the season and with higher beech dominance. By linking quantitative data on arthropod herbivore abundance and herbivory with tree species composition and harvesting activity in a wide variety of beech forests, our study helps to better understand the influence of forest management on interactions between a naturally dominant deciduous forest tree and arthropod herbivores.     

Estimating and Analyzing Savannah Phenology with a Lagged Time Series Model

Savannah regions are predicted to undergo changes in precipitation patterns according to current climate change projections. This change will affect leaf phenology, which controls net primary productivity. It is of importance to study this since savannahs play an important role in the global carbon cycle due to their areal coverage and can have an effect on the food security in regions that depend on subsistence farming. In this study we investigate how soil moisture, mean annual precipitation, and day length control savannah phenology by developing a lagged time series model. The model uses climate data for 15 flux tower sites across four continents, and normalized difference vegetation index from satellite to optimize a statistical phenological model. We show that all three variables can be used to estimate savannah phenology on a global scale. However, it was not possible to create a simplified savannah model that works equally well for all sites on the global scale without inclusion of more site specific parameters. The simplified model showed no bias towards tree cover or between continents and resulted in a cross-validated r² of 0.6 and root mean squared error of 0.1. We therefore expect similar average results when applying the model to other savannah areas and further expect that it could be used to estimate the productivity of savannah regions.     

Do <Emphasis Type="Italic">Polylepis australis</Emphasis> trees tolerate herbivory? Seasonal patterns of shoot growth and its consumption by livestock

Browsing is one of the main factors determining survival, growth rate, woodland structure, and distribution of the high mountain tree Polylepis australis. This species has a substantial regrowth capacity, which may function as a mechanism to tolerate herbivory, but it is unknown to what extent it may compensate for the impact of herbivory. In 15 low-density tree stands subject to exclusion, moderate, and heavy livestock pressure, we selected 12 P. australis individuals <2 m tall, tagged four new shoots per tree and measured shoot length every month during a year. At the stand and at the tree level, we analyzed monthly dynamics of growth and browsing, and the annual output in terms of total browsing and total gross and net growth (not discounting and discounting consumption, respectively). In addition, we assessed the influence of stand, tree and microsite characteristics on growth and browsing patterns. Polylepis australis fully compensated for herbivory in terms of shoot gross growth at moderate, but not at heavy livestock pressure. In terms of net growth, this species did not fully compensate for herbivory at any stocking rate. We found a strong coupling between browsing and growth along the year, suggesting that regrowth attracts browsing, and browsing promotes regrowth. At the stand level, annual gross growth was not affected by habitat characteristics, while at the tree level, annual gross growth decreased on more rocky microsites for browsed but not for unbrowsed trees. We concluded that stocking densities should be limited to allow for a reasonable annual net growth, as its nitrogen rich leaves are a valuable food resource and P. australis forests provide important ecosystem services.     

Modelling ungulate dependence on higher quality forage under large trees in African savannahs

In African savannahs, large trees improve grass quality, particularly in dry and nutrient poor areas. Enhanced below-canopy grass nutrients, such as nitrogen and phosphorus contents should therefore attract and benefit grazers. To predict whether ungulates really need these forage quality islands we focused on four grazer species, i.e., zebra, buffalo, wildebeest, and warthog, differing in body size and digestive system. We confronted literature estimations of their feeding requirements with forage availability and quality, observed in three South African savannah systems, through linear modelling. The model predicted the proportion of below-canopy grass that grazers should include in their diet to meet their nutritional requirements.During the wet season, the model predicted that all animals could satisfy their daily nutrient requirements when feeding on a combination of below- and outside-canopy grasses. However, wildebeest, having relatively high nutrient demands, could meet their nutrient requirements only by feeding almost exclusively below canopies.During the dry season, all animals could gain almost twice as much digestible protein when feeding on below – compared to outside-canopy forage. Nonetheless, only warthogs could satisfy their nutrient requirements – when feeding almost exclusively on below-canopy grasses. The other ungulate species could not meet their phosphorus demands by feeding at either site without exceeding their maximum fibre intake, indicating the unfavourable conditions during the dry season.We conclude that grazing ungulates, particularly warthog, zebra, and buffalo, actually depend on the available below-canopy grass resources. Our model therefore helps to quantify the importance of higher quality forage patches beneath savannah trees. The composition of grazer communities depending on below-canopy grasses can be anticipated if grazer food requirements and the abundance of large trees in savannahs are known. The model suggests that the conservation of large single-standing trees in savannahs is crucial for maintenance of locally grazing herbivores.     

Influence of Colophospermum mopane canopy cover on litter decomposition and nutrient dynamics in a semi‐arid African savannah

The objective of this study was to investigate the influence of mopane canopy cover on litter decomposition in a semi‐arid African savannah. We used a randomized block design with five blocks of 100 × 100 m demarcated in a 10‐ha pocket of open mopane woodland. Litterbags were placed beneath large (8.3 m crown diameter) and small mopane trees (2.7 m crown diameter) and in the intercanopy area. Decomposition was fastest in the intercanopy area exposed to solar radiation (k = 0.35 year^?1), intermediate beneath small trees (k = 0.28 year^?1) and slowest beneath large trees (k = 0.23 year^?1). Soil temperatures beneath small and large trees were 3–5 and 6–9°C lower than in the intercanopy area, respectively. Bacterial and fungal counts were significantly higher (P < 0.05) beneath large than small trees and in the intercanopy area. The amount of N and P released did not vary significantly among sampling sites. Soil moisture in the dry season was similar among sampling sites but rainy‐season soil moisture was significantly greater (P < 0.05) beneath large than small trees and in the intecanopy area. Mopane canopy cover retarded litter decomposition suggesting that photodegradation could be an important factor controlling carbon turnover in semi‐arid African savannahs.     

Termites, Large Herbivores, and Herbaceous Plant Dominance Structure Small Mammal Communities in Savannahs

Large herbivores and termites are important functional groups in African savannahs. Both groups affect small mammals, which are also important determinants for savannah structure and function. Because vegetation on Macrotermes mounds are preferentially grazed by large herbivores, and mounds represent resource-rich distinct habitat patches for small mammals in relatively resource-poor savannahs, termite mounds are ideal sites for studies of how grazing by large mammals and productivity affect communities of small mammals. We conducted an experiment in Lake Mburo National Park, Uganda, with four treatments: large vegetated Macrotermes mounds (with and without large herbivores) and adjacent savannah areas (with and without large herbivores). We replicated the treatment blocks nine times and trapped small mammals regularly over a period of almost 2 years. Small mammal species assemblages differed considerably between mounds and savannah areas. Grazing had a substantial effect on small mammal species assemblages in the resource-poor savannah, but not in the relatively resource-rich termitaria. Small mammal species abundance, biomass, and richness were higher on termite mounds than adjacent savannah areas. Excluding large herbivores caused a major increase in species abundance, biomass, and richness both on savannah and termitaria. Herbaceous plant species evenness was an important determinant of the small mammal community. Small mammal biomass increased with high plant dominance, indicating that a few dominant plant species are important for biomass production of small mammals. Small mammal diversity was not related to any of the treatments, but increased with plant species evenness as well as richness. Fencing increased species dominance in the small mammal community on both savannah and termitaria, probably because competitive patterns shift from inter-guild (that is, between large and small mammals) to intra-guild (that is, between small mammals) when large mammals are excluded. The study highlights the complex interactions among large herbivores, termites, herbaceous plants, and small mammals in African savannahs. When studying the structure and function of small mammal communities it is therefore important to consider several coexisting functional groups.