Detection of Mitochondrial COII DNA Sequences in Ant Guts as a Method for Assessing Termite Predation by Ants

Termites and ants contribute more to animal biomass in tropical rain forests than any other single group and perform vital ecosystem functions. Although ants prey on termites, at the community level the linkage between these groups is poorly understood. Thus, assessing the distribution and specificity of ant termitophagy is of considerable interest. We describe an approach for quantifying ant-termite food webs by sequencing termite DNA (cytochrome c oxidase subunit II, COII) from ant guts and apply this to a soil-dwelling ant community from tropical rain forest in Gabon. We extracted DNA from 215 ants from 15 species. Of these, 17.2 % of individuals had termite DNA in their guts, with BLAST analysis confirming the identity of 34.1 % of these termites to family level or better. Although ant species varied in detection of termite DNA, ranging from 63 % (5/7; Camponotus sp. 1) to 0 % (0/7; Ponera sp. 1), there was no evidence (with small sample sizes) for heterogeneity in termite consumption across ant taxa, and no evidence for species-specific ant-termite predation. In all three ant species with identifiable termite DNA in multiple individuals, multiple termite species were represented. Furthermore, the two termite species that were detected on multiple occasions in ant guts were in both cases found in multiple ant species, suggesting that ant-termite food webs are not strongly compartmentalised. However, two ant species were found to consume only Anoplotermes-group termites, indicating possible predatory specialisation at a higher taxonomic level. Using a laboratory feeding test, we were able to detect termite COII sequences in ant guts up to 2 h after feeding, indicating that our method only detects recent feeding events. Our data provide tentative support for the hypothesis that unspecialised termite predation by ants is widespread and highlight the use of molecular approaches for future studies of ant-termite food webs.     

Defense of termitaria by termitophilous ants

Summary Mounds of Amitermes laurensis are frequently faided by meat ants Iridomyrmex sanguineus. Of eight ant species which often cohabit with the termites, Camponotus sp. B and C were considerably dependent on the termintaria for their nest sites and effectively protected it from the attacks by meat ants. Many termite colonies cohabiting with those two ant species were vigorous, suggesting that this ant-termite relationship is mutualistic; thus, the ants were provided nest sites and probably even food and the termites were protected from destructive natural enemies.     

Community interactions between ants and arboreal-nesting termites in New Guinea coconut plantations

Summary: We investigated the incidence of inquiline ants and of arboreal-nesting ants on a community of three arboreal-nesting termites living in New Guinea coconut plantations. Inquiline ants were present in 10 % of Microcerotermes biroi nests and in 4 % of Nasutitermes princeps nests. Live termite nests inhabited by the most common inquiline ant, Camponotus sp. A, were generally left by the ant after several months. In some nests, Camponotus sp. A was observed coexisting with its host during the whole observation period (3 years). Therefore, Camponotus sp. A was apparently an opportunistic inquiline which did not affect significantly the mortality of termite colonies. The arboreal-nesting ant, Crematogaster irritabilis, was locally found occupying up to 99 % of the trees present in 1 ha plots. In such hot spots, the overall abundance of termites was approximately half that of plots devoid of Crematogaster irritabilis. The high density of Crematogaster irritabilis may be an important limiting factor for the termite assemblage, by hastening the death or hindering the establishment of arboreal termite colonies.     

Diversity of fungus‐growing termites (Macrotermes) and their fungal symbionts (Termitomyces) in the semiarid Tsavo Ecosystem,Kenya

Fungus‐growing termites of the subfamily Macrotermitinae together with their highly specialized fungal symbionts (Termitomyces) are primary decomposers of dead plant matter in many African savanna ecosystems. The termites provide crucial ecosystem services also by modifying soil properties, translocating nutrients, and as important drivers of plant succession. Despite their obvious ecological importance, many basic features in the biology of fungus‐growing termites and especially their fungal symbionts remain poorly known, and no studies have so far focused on possible habitat‐level differences in symbiont diversity across heterogeneous landscapes. We studied the species identities of Macrotermes termites and their Termitomyces symbionts by excavating 143 termite mounds at eight study sites in the semiarid Tsavo Ecosystem of southern Kenya. Reference specimens were identified by sequencing the COI region from termites and the ITS region from symbiotic fungi. The results demonstrate that the regional Macrotermes community in Tsavo includes two sympatric species (M. subhyalinus and M. michaelseni) which cultivate and largely share three species of Termitomyces symbionts. A single species of fungus is always found in each termite mound, but even closely adjacent colonies of the same termite species often house evolutionarily divergent fungi. The species identities of both partners vary markedly between sites, suggesting hitherto unknown differences in their ecological requirements. It is apparent that both habitat heterogeneity and disturbance history can influence the regional distribution patterns of both partners in symbiosis.     

Levels of specificity of Xylaria species associated with fungus-growing termites: a phylogenetic approach

Fungus-growing termites live in obligate mutualistic symbiosis with species of the basidiomycete genus Termitomyces , which are cultivated on a substrate of dead plant material. When the termite colony dies, or when nest material is incubated without termites in the laboratory, fruiting bodies of the ascomycete genus Xylaria appear and rapidly cover the fungus garden. This raises the question whether certain Xylaria species are specialised in occupying termite nests or whether they are just occasional visitors. We tested Xylaria specificity at four levels: (1) fungus-growing termites, (2) termite genera, (3) termite species, and (4) colonies. In South Africa, 108 colonies of eight termite species from three termite genera were sampled for Xylaria . Xylaria was isolated from 69% of the sampled nests and from 57% of the incubated fungus comb samples, confirming high prevalence. Phylogenetic analysis of the ITS region revealed 16 operational taxonomic units of Xylaria , indicating high levels of Xylaria species richness. Not much of this variation was explained by termite genus, species, or colony; thus, at level 2–4 the specificity is low. Analysis of the large subunit rDNA region, showed that all termite-associated Xylaria belong to a single clade, together with only three of the 26 non-termite-associated strains. Termite-associated Xylaria thus show specificity for fungus-growing termites (level 1). We did not find evidence for geographic or temporal structuring in these Xylaria phylogenies. Based on our results, we conclude that termite-associated Xylaria are specific for fungus-growing termites, without having specificity for lower taxonomic levels.     

Termite mound density and distribution patterns across three land-use types in the Vhembe District of the Limpopo Province,South Africa

This study investigated the effect of land-use on density and distribution patterns of termite mounds. A total area of 12 ha was investigated using four 1 ha plots from each of three land-use types (mango orchards, maize fields and communal rangelands). A total of 297 mounds from four termite species were recorded. Plotted GIS coordinates for each mound in ArcMap showed a random distribution pattern in all land-use types. The mean number of mounds per hectare was significantly higher (p < 0.001) in communal rangelands (52.5 ± 1.21), than in maize fields (14.75 ± 3.15) and mango orchards (7.5 ± 0.87), and dominated by small-sized mounds of Trinervitermes sp. Few mounds of Odontotermes sp. were found. Mounds of the edible termites, Macrotermes natalensis and M. falciger, were found in all land-use types, with the highest density for both species being in maize fields. Although the mound height for both species was similar, mound circumference for M. falciger was significantly larger (p < 0.001) which may limit land available for agricultural use. Density of mounds was influenced by land-use which may lead to changes in termite ecosystem functioning and availability of termites as a free source of protein.     

Intra- and interspecific comparisons of bacterial diversity and community structure support coevolution of gut microbiota and termite host

We investigated the bacterial gut microbiota from 32 colonies of wood-feeding termites, comprising four Microcerotermes species (Termitidae) and four Reticulitermes species (Rhinotermitidae), using terminal restriction fragment length polymorphism analysis and clonal analysis of 16S rRNA. The obtained molecular community profiles were compared statistically between individuals, colonies, locations, and species of termites. Both analyses revealed that the bacterial community structure was remarkably similar within each termite genus, with small but significant differences between sampling sites and/or termite species. In contrast, considerable differences were found between the two termite genera. Only one bacterial phylotype (defined with 97% sequence identity) was shared between the two termite genera, while 18% and 50% of the phylotypes were shared between two congeneric species in the genera Microcerotermes and Reticulitermes, respectively. Nevertheless, a phylogenetic analysis of 228 phylotypes from Microcerotermes spp. and 367 phylotypes from Reticulitermes spp. with other termite gut clones available in public databases demonstrated the monophyly of many phylotypes from distantly related termites. The monophyletic "termite clusters" comprised of phylotypes from more than one termite species were distributed among 15 bacterial phyla, including the novel candidate phyla TG2 and TG3. These termite clusters accounted for 95% of the 960 clones analyzed in this study. Moreover, the clusters in 12 phyla comprised phylotypes from more than one termite (sub)family, accounting for 75% of the analyzed clones. Our results suggest that the majority of gut bacteria are not allochthonous but are specific symbionts that have coevolved with termites and that their community structure is basically consistent within a genus of termites.     

Mutualistic acacia ants exhibit reduced aggression and more frequent off‐tree movements near termite mounds

In many ant–plant mutualisms, ants establish colonies in hollow thorns, leaf pouches, or other specialized structures on their host plants, which they then defend from herbivores. Resource heterogeneity could affect the maintenance of these mutualisms if it leads to one or both partners altering their investment in the interaction. Such a phenomenon may be especially pertinent to the Acacia–ant mutualism found in East African savannas, where termite mounds have a profound effect on the spatial structuring of resources used by both plants and ants. Here, we examined whether the proximity to termite mounds of Acacia drepanolobium trees is associated with variation in the behavior of one of their ant associates, Crematogaster nigriceps. We found that ant colonies near termite mounds had decreased aggressive responses to simulated herbivory as well as increased off‐tree movement. We hypothesize that these changes are the result of resident ant colonies near termite mounds shifting investment from defense of their host plant to foraging for nearby resources.     

Intra- and Interspecific Comparisons of Bacterial Diversity and Community Structure Support Coevolution of Gut Microbiota and Termite Host

We investigated the bacterial gut microbiota from 32 colonies of wood-feeding termites, comprising four Microcerotermes species (Termitidae) and four Reticulitermes species (Rhinotermitidae), using terminal restriction fragment length polymorphism analysis and clonal analysis of 16S rRNA. The obtained molecular community profiles were compared statistically between individuals, colonies, locations, and species of termites. Both analyses revealed that the bacterial community structure was remarkably similar within each termite genus, with small but significant differences between sampling sites and/or termite species. In contrast, considerable differences were found between the two termite genera. Only one bacterial phylotype (defined with 97% sequence identity) was shared between the two termite genera, while 18% and 50% of the phylotypes were shared between two congeneric species in the genera Microcerotermes and Reticulitermes, respectively. Nevertheless, a phylogenetic analysis of 228 phylotypes from Microcerotermes spp. and 367 phylotypes from Reticulitermes spp. with other termite gut clones available in public databases demonstrated the monophyly of many phylotypes from distantly related termites. The monophyletic “termite clusters” comprised of phylotypes from more than one termite species were distributed among 15 bacterial phyla, including the novel candidate phyla TG2 and TG3. These termite clusters accounted for 95% of the 960 clones analyzed in this study. Moreover, the clusters in 12 phyla comprised phylotypes from more than one termite (sub)family, accounting for 75% of the analyzed clones. Our results suggest that the majority of gut bacteria are not allochthonous but are specific symbionts that have coevolved with termites and that their community structure is basically consistent within a genus of termites.     

Ecological context of savanna chimpanzee (Pan troglodytes verus) termite fishing at Fongoli, Senegal

Chimpanzees (Pan troglodytes verus) at Fongoli, Senegal, consume termites year-round. Understanding the ecological context behind this behavior is especially important in light of the environmental conditions at Fongoli. This mosaic savanna habitat is one of the hottest and driest sites where chimpanzees have been studied. Two genera and four species of termites were found in association with tools used by chimpanzees in a sample of 124 termite mounds that were monitored. The chimpanzees of Fongoli termite fish predominantly in woodland and forest habitat types, and, although woodland accounts for the majority of the chimpanzees' home range, forest habitat types comprise only about 4% of their range. Thus, habitat type has an influence on the Fongoli chimpanzees' termite fishing. Termite consumption to the degree seen at Fongoli may have particular significance for hominid evolution, given the expansion of Pliocene hominids into increasingly open, hot, and dry habitats.     

Geophagic termite mounds as one of the resources for African elephants in Ugalla Game Reserve,Western Tanzania

Knowledge of the distribution and nutrient values of key resources supporting the survival of wildlife species is integral for an effective conservation planning and management of the species. In the Miombo ecosystem of the Ugalla Game Reserve, African elephants (Loxodonta africana Blumenbach 1797), eat soil, that is geophagy, from certain termite mounds. We mapped that all the geophagic termite mounds are exclusively situated in the flood plain. To understand why soils from some termite mounds are eaten, we collected and analysed soil samples from 10 geophagic termite mounds, seven nongeophagic termite mounds and 13 samples from the surrounding flood plain. Percentage of clay content did not differ significantly among the soil samples. Soils from geophagic termite mounds were richer in mineral elements compared with other soil samples. The results demonstrate that the driver for geophagic behaviour is related to rich mineral element contents found in geophagic termite mounds made of the mineral‐enriching termites (Macrotermes). Thus, geophagic termite mounds play a role in elephant's dietary needs and possibly influence their movement patterns in Ugalla, as the elephants cannot obtain enough minerals from their feeds. Geophagic termite mounds should be protected from potential destructive land uses, such as airstrip construction.     

Strong positive effects of termites on savanna bird abundance and diversity are amplified by large herbivore exclusion

Vast areas of the African savanna landscapes are characterized by tree‐covered Macrotermes termite mounds embedded within a relatively open savanna matrix. In concert with termites, large herbivores are important determinants of savanna woody vegetation cover. The relative cover of woody species has considerable effects on savanna function. Despite the potentially important ecological relationships between termite mounds, woody plants, large herbivores, and birds, these associations have previously received surprisingly little attention. We experimentally studied the effects of termites and large herbivores on the avian community in Lake Mburo National Park, Uganda, where woody vegetation is essentially limited to termite mounds. Our experiment comprised of four treatments in nine replicates; unfenced termite mounds, fenced mounds (excluding large mammals), unfenced adjacent savanna, and fenced savanna. We recorded species identity, abundance, and behavior of all birds observed on these plots over a two‐month period, from late dry until wet season. Birds used termite mounds almost exclusively, with only 3.5% of observations occurring in the treeless intermound savanna matrix. Mean abundance and species richness of birds doubled on fenced (large herbivores excluded) compared to unfenced mounds. Feeding behavior increased when large mammals were excluded from mounds, both in absolute number of observed individuals, and relative to other behaviors. This study documents the fundamental positive impact of Macrotermes termites on bird abundance and diversity in an African savanna. Birds play crucial functional roles in savanna ecosystems, for example, by dispersing fruits or regulating herbivorous insect populations. Thus, the role of birds in savanna dynamics depends on the distribution and abundance of termite mounds.     

成都市白蚁天敌资源——蚂蚁种类调查研究

部分蚂蚁种类是白蚁的重要天敌。为了探明成都市白蚁天敌——蚂蚁的常见种类及其生境,2011—2015年通过对成都市房屋建筑、园林树木、古建筑以及安装在住宅小区绿地的地下型白蚁监测装置等生境中调查白蚁与蚂蚁种类及数量,共鉴定白蚁9种,隶属于3科4属,黑胸散白蚁Reticulitermes chinensis占比最高(83.9%);蚂蚁29种,隶属于4亚科15属,中国短猛蚁Brachyponera chinensis占比最高(38.2%)。相关性分析表明,黑毛蚁Lasius niger与尖唇散白蚁R.aculabialis、新中华散白蚁R.neochinensis之间,莱曼氏蚁Formica lemani与黑胸散白蚁、尖唇散白蚁之间,敏捷扁头猛蚁Pachycondyla astuta、铺道蚁Tetramorium caesputum与尖唇散白蚁之间均呈显著负相关。黑毛蚁、莱曼氏蚁、敏捷扁头猛蚁、铺道蚁的种群数量大,在白蚁生物防治中具有一定的应用潜力。鉴于蚂蚁对白蚁的控制作用受环境、人为等多种因素的影响,建议进一步加强蚂蚁对白蚁自然控制作用的系统研究。     

Termite Feeding by <Emphasis Type="Italic">Gorilla gorilla gorilla</Emphasis> at Bai Hokou,Central African Republic

Though insectivory by large-bodied gorillas may be unexpected, researchers have reported it in all populations of gorillas studied to date. Our study of 2 well monitored groups of western gorillas (Gorilla gorilla gorilla) at Bai Hokou in Dzanga-Ndoki National Park, Central African Republic provides information on frequency and variability of termite consumption (the most commonly eaten insect) as well as some of the first direct observations of the behavior. Pooled data from both groups indicate termite feeding on 34% and 83% of days, through fecal analysis and feeding trails, respectively. Direct observations revealed that termite feeding occurred on 91% of the days for 1 group, in which the silverback fed on termites during 13% of all feeding scans, making termites the most commonly observed food item. The group that had a higher density of termite mounds in its home range consumed termites more frequently than the other group did. A higher proportion of fecal samples from the silverbacks contained termite remains than the ones from adult females and juveniles. Termite consumption was lower during the dry season, but it does not correlate with rainfall, measures of fruit availability, or fruit consumption. Displacements at termite mounds occurred more than expected, indicating that they are a patchy, sought-after food resource. Gorillas did not use tools to extract termites, but they used 2 different techniques to remove them from the cells. Though culture or social traditions may cause the variation in termite consumption across sites, further investigation of termite availability and consumption is necessary to rule out ecological and methodological explanations for observed variations.     

Prevalence of a koinobiont endoparasitoid Misotermes mindeni (Diptera: Phoridae) in colonies of the fungus-growing termite Macrotermes gilvus (Blattodea: Termitidae) in Malaysia

A survey of the infestation rate of colonies of Macrotermes gilvus (Hagen) (Termitidae: Macrotermitinae) with the koinobiont endoparasitoid Misotermes mindeni Disney & Neoh (Diptera: Phoridae) was conducted in Malaysia from September 2009 to January 2011 in the states of Kedah, Penang, Perak, Selangor, Kuala Lumpur, Johor, Terengganu, and Sarawak. Of the 1,125 M. gilvus mounds surveyed, 12.4% contained termites parasitized by M. mindeni and these mounds occurred only in the states of Penang and Perak. High frequencies of mounds containing parasitized termites were found at sites in Penang: Bayan Lepas (21.1%), Minden Campus of Universiti Sains Malaysia ([USM]; 24.5%), Teluk Bahang (28.0%), and Bukit Mertajam (35.0%); the lowest frequency (4.0%) was recorded from Gelugor. The parasitized colonies at all sites were classified as healthy, with exception of several from the Minden Campus of USM (96.4% healthy) and Ayer Itam (87.5% healthy). Most parasitized colonies (71.2%) had a low level of M. mindeni infestation. Only 16.7 and 12.1% of the infested colonies had moderate or high parasite infestation levels, respectively. The height of infected mounds was significantly higher than that of the healthy mounds, but there was no difference between the mound diameters of infested and uninfested mounds. Parasite infestation level was not significantly correlated with mound height or mound diameter. The ambient light intensity at sites with infested mounds was significantly lower than that of uninfested mounds. There was also a significant negative relationship between light intensity and degree of parasitism.     

Termites as bioindicators of habitat quality in the caatinga, Brazil: is there agreement between structural habitat variables and the sampled assemblages?

The composition of termite assemblages was analyzed in three caatinga sites of the Esta??o Ecológica do Seridó, located in the municipality of Serra Negra do Norte, in the state of Rio Grande do Norte, Brazil. These sites have been subjected to selective logging, and cleared for pasture and farming. A standardized sampling protocol for termite assemblages (30h/person/site) was conducted between September 2007 and February 2009. At each site we measured environmental variables, such as soil pH and organic matter, necromass stock, vegetation height, stem diameter at ankle height (DAH) and the largest and the smallest crown width. Ten species of termites, belonging to eight genera and three families, were found at the three experimental sites. Four feeding groups were sampled: wood-feeders, soil-feeders, wood-soil interface feeders and leaf-feeders. The wood-feeders were dominant in number of species and number of encounters at all sites. In general, the sites were not significantly different in relation to the environmental variables measured. The same pattern was observed for termite assemblages, where no significant differences in species richness, relative abundance and taxonomic and functional composition were observed between the three sites. The agreement between composition of assemblages and environmental variables reinforces the potential of termites as biological indicators of habitat quality.     

The distribution,abundance, and the effects of fire on mound building termites (Trinervitermes and Cubitermes spp., Isoptera: Termitidae) in northern guinea savanna West Africa

Summary Termite mound densities in typical guinea savanna, Detarium, and grassland (boval) habitats in northern guinea savanna were determined by random quadratting of 2–3 sites in each habitat (100, 10x10 m quadrats per habitat). Dominant species in guinea savanna were T. geminatus (46 mounds ha^-1) and T. oeconomus (21 mounds ha^-1), in Detarium T. geminatus (59 mounds ha^-1) and C. curtatus (45 mounds ha^-1) and in boval C. curtatus (72 mounds ha^-1) and T. geminatus (22 mounds ha^-1). Only C. curtatus densities and total densities differed significantly between sites within habitats, but all species differed significantly in abundance between habitats. The composition of each community was related to general environment but no particular environmental variable was shown to be a major determinant of termite distribution. Evidence for the limitation of termite populations was obtained from indirect evidence of competition between colonies in Detarium, and by experimental manipulation of fire regimes in the typical guinea savanna habitat. Harvester termites increased four-five fold over two years in fire-protected plots as a result of increased food supplies. Total termite densities in the fire-protected community equilibrated to the new population density (100 mounds ha^-1) after only two-three years.     

Driver Ants Invading a Termite Nest: Why Do the Most Catholic Predators of All Seldom Take This Abundant Prey?

Driver ants ( i.e. , epigaeic species in the army ant genus Dorylus , subgenus Anomma ) are among the most extreme polyphagous predators, but termites appear to be conspicuously absent from their prey spectrum and attacks by driver ants on termite nests have not yet been described. Here, we report a Dorylus ( Anomma ) rubellus attack on a colony of the fungus-growing termite Macrotermes subhyalinus that was observed during the dry season in a savannah habitat in Nigeria's Gashaka National Park. It was estimated that several hundred thousand termites (probably more than 2.4 kg dry mass) were retrieved. The apparent rarity of driver ant predation on Macrotermes nests may be explained by different habitat requirements, by the fact that these ants mostly forage aboveground, by efficient termite defense behavior and nest architecture that make entry into the nest difficult, and finally by driver ant worker morphology, which differs remarkably from that of subterranean Dorylus species that regularly invade and destroy termite colonies.     

Farmers’ perception of termites in agriculture production and their indigenous utilization in Northwest Benin

Background

Although termites are considered as agricultural pests, they play an important role in maintaining the ecosystem. Therefore, it matters to investigate the farmers’ perception of the impacts of the termites on the agriculture and their indigenous utilization.

Methods

A semi-structured questionnaire was used to interview 94 farmers through 10 villages of Atacora department, in the northwestern region of Benin, to obtain information for the development of successful strategies of termite management and conservation. Their perceptions on the importance and management of termites along with the indigenous nomenclature and utilization of termite mounds were assessed. Termite species identified by farmers were collected and preserved in 80% alcohol for identification.

Results

Eight crops were identified by farmers as susceptible to termites with maize, sorghum, and yam as being the most susceptible. According to farmers, the susceptibility to termites of these crops is due to their high-water content and sweet taste. A total of 27 vernacular names of termites were recorded corresponding to 10 species, Amitermes evuncifer, Macrotermes subhyalinus, and Trinervitermes oeconomus being the most damaging termite species. All the names given to termite species had a meaning. The drought was identified by farmers as the main factor favouring termite attacks. Demolition of termite mounds in the fields was the most commonly reported control method. Salt and other pesticides were commonly used by farmers to protect stored farm products. The lack of effective control methods is the main constraint for termite management. In northwestern Benin, farmers reported different purpose utilizations of termite mounds and termites.

Conclusions

The study has shown that farmers perceived termites as pests of several agricultural crops and apply various indigenous control practices whose efficiency need to be verified. Utilization of termites and termite mound soil as food and medicinal resources underlines the need for a more focused approach to termite control for the conservation of non-pest termite species. The sensitization of farmers on the importance of termites as well as the development of an integrated control method to combat termite pests proved necessary.

     

Termites facilitate and ungulates limit savanna tree regeneration

Both large herbivores and termites are key functional groups in savanna ecosystems, and in many savanna areas, large termite mounds (termitaria) are associated with distinct woody clusters. Studies on the effect of large mammals on tree regeneration are few, and the results are conflicting. Large herbivores have been found to be important seedling predators in some areas, but facilitate tree regeneration by outcompeting small mammals and reducing grass cover in other areas. Through the use of the experimental fencing of termite mounds and adjacent savanna areas in this study, we investigated how termites and large herbivores influence tree regeneration. Termite mounds had a higher number of seedlings, more species richness, more alpha diversity (OD) and lower evenness (E) than savanna plots. Large herbivores did not significantly affect overall seedling density, species richness, OD or E. Beta diversity was higher in savanna areas than on termitaria, and beta diversity decreased in savanna areas when herbivores were excluded. Herbivore exclusion increased the density of the 12 (40 %) most common seedling species, representing 79 % of all seedlings, and fenced plots had relatively taller seedlings than open plots. Thus, termites were the main determinants of tree regeneration in our study area, but large mammals regulated the most common species. Although our study confirms previous work suggesting that large herbivores affect tree regeneration, we found that termites were an even more important determinant. Termite impacts on tree regeneration deserve increased attention by savanna ecologists.