Savanna ecosystems span a diverse range of climates, edaphic conditions, and disturbance regimes, the complexity of which has stimulated long‐standing interest in the mechanisms that maintain tree–grass coexistence. One hypothesis suggests that tree establishment is strongly limited by one or several demographic bottlenecks at early stages of the tree life cycle. A major impediment to testing this hypothesis is the lack of data on the relative strengths of different bottlenecks across key environmental gradients. To identify demographic bottlenecks that limit early tree establishment (0–18 months), we conducted a series of transplant experiments with two savanna trees species (Acacia robusta and Acacia tortilis) across a natural rainfall and soil fertility gradient in the Serengeti ecosystem, Tanzania. We tested the interactive effects of precipitation, herbivory, seed scarification, grass competition, water limitation, and tree species identity on two key life stages: germination and early seedling survival (0–2 months), and juvenile seedling survival (2–18 months). Germination and early seedling survival increased as a function of rainfall, in the absence of herbivores and when seeds were scarified. Juvenile seedling survival, in contrast, decreased with rainfall but increased in the absence of herbivores. Grass removal had the single strongest (positive) effect on juvenile seedling survival of any treatment. Soil moisture monitoring and grass‐addition treatments revealed that grasses negatively affected seedlings in ways that were not necessarily linked to soil moisture. A demographic model combining all effects across early life stages showed that the strength of grass competition on juvenile seedling survival was the key factor limiting early tree establishment. While rainfall had an unexpected opposing effect on the two life stages, the net effect of mean annual precipitation on early tree establishment was positive. Synthesis. Successful tree establishment in Serengeti is maximized by a seemingly unlikely sequence of events: (a) scarification of seeds by browsers, (b) heavy rainfall to promote germination, (c) intensive grazing (but absence of browsers), and (d) dry conditions during juvenile seedling growth (>2 months) to reduce competition with grasses. By considering a wide suite of conditions and their interactions, our experimental results are relevant to ongoing debates about savanna vegetation dynamics and structural shifts in tree:grass ratios.

Questions What is the pattern of compositional similarity for woody plants across the demographic bottleneck (i.e. canopy trees >2 m vs understorey trees <2 m) commonly observed in savannas? Does compositional similarity between woody plants in the canopy and understorey change across environmental gradients, and if so, which resource or disturbance factors explain the pattern? Finally, does disturbance history, i.e. whether individuals are newly established seedlings or top‐killed resprouts, help to explain the compositional similarity between canopy and understorey woody plants across sites? Location The Serengeti ecosystem, Tanzania, East Africa. Methods We compared the observed similarity to that expected from a simulated null model, and investigated the association between compositional similarity and soil bulk density, mean annual rainfall, fire return interval and elephant damage using a model selection approach. Moreover, we asked if disturbance history, i.e. whether individuals were newly established seedlings or top‐killed resprouts, helped explain compositional similarity between the canopy and understorey. Results The composition of canopy (>2 m) and understorey (<2 m) trees was surprisingly distinct: Acacia robusta (37.9%) and Acacia tortilis (23.7%) dominated the canopy, while Ormocarpum trichocarpum (43.0%) and Dichrostachys cinerea (19.6%) dominated the understorey. Compositional similarity between canopy and understorey woody plants generally decreased with soil bulk density, while for top‐killed individuals there was support for an increase in compositional similarity with fire return interval. Conclusions Our results suggest that the similarity between the canopy and recruitment communities is primarily associated with soil factors. Our results also suggest that long fire return intervals (i.e. infrequent fires) lead to increased similarity between the composition of the canopy and the recruitment pool, suggesting that disturbance promotes heterogeneity in the community across demographic bottlenecks. Despite the prominent role that megaherbivores and disturbance play in driving structural changes in savannas, our results suggest that previously unreported but pervasive species decoupling across demographic bottlenecks is primarily driven by resource availability or other bottom‐up constraints.

Savannas are spatially diverse, variable and are susceptible to high rates of disturbance from fire and herbivory. There is significant interest in woody cover dynamics in relation to disturbance regimes. Less effort has been devoted to understand processes that drive tree community composition. In this study, tree species composition data collected at the landscape scale in the Serengeti were used to identify key environmental factors driving variation in species composition. A system of 38 plots clustered within 10 sites spanning the mean annual precipitation (MAP) gradient was used to assess the relative role of bottom‐up (precipitation, soil nutrients and soil texture) vs. top‐down factors (fire and elephant herbivory) on tree community composition. We developed candidate models relating tree species composition (based on multivariate community analysis) to different combinations of plot‐level environmental covariates. Results suggest that tree community composition is largely driven by MAP and is associated with elephant population density. Strikingly, we found no evidence that fire influences species compositional turnover. In a second analysis, we used structural equation model (SEM) to explore the possible direction of association between elephant density and tree species composition. We compared a model that included elephant effects on composition to one that included community composition effects on elephant density. Results suggest that variation in elephant population density across space is more likely to drive tree community composition and not vice versa. We propose that precipitation and herbivory, rather than fire, determine tree species composition in Serengeti Acacia tree community.

Seed production, infestation and viability in Acacia tortilis (synonym: Vachellia tortilis) and Acacia robusta (synonym: Vachellia robusta) across the Serengeti rainfall gradient.

No abstract