This study assesses the impact of four fire treatments applied yearly over 3 y, i.e. early fire, mid-season fire, late fire and no fire treatments, on the grass communities of Lamto savanna, Ivory Coast. We describe communities of perennial tussock grasses on three replicated 5 × 5-m or 10 × 5-m plots of each fire treatment. Tussock density did not vary with fire treatment. The relative abundance of grass species, the circumference of grass tussocks and the probability of having a tussock with a central die-back, varied with fire treatment. Mid-season fire had the highest proportion of tussocks with a central die-back while the late fire had the smallest tussocks. Tussock density, circumference, relative abundance and probability of having a central die-back varied with species. Andropogon canaliculatus and Hyparrhenia diplandra were the most abundant of the nine grass species. They had the largest tussocks and the highest proportion of tussock with a central die-back. Loudetia simplex was the third most abundant species but was very rare in no fire plots. The distribution of tussock circumferences was right skewed and dominated by small tussocks. The proportion of the tussocks with a central die-back strongly increased with circumference, which could lead to tussock fragmentation. Taken together, this study suggests that fire regimes impact grass demography and that this impact depends on grass species and tussock size.

Measurements were conducted within a fence-exclosure between February 2008 and July 2009 to investigate the influence of soil moisture on ecosystem CO2 fluxes in a Themeda triandra-dominated grassland of a humid Kenyan savanna. Rainout shelters were constructed to reduce ambient rainfall by 0%, 10% and 20% respectively to attain variable soil water content (SWC) during plant growth. SWC within the top 30 cm layer, above-ground biomass, soil and plant nitrogen (N) concentrations were assessed monthly alongside CO2 fluxes. Net ecosystem CO2 exchange (NEE) and ecosystem respiration (Reco) were measured with closed chambers while carbon (C) partitioning during the wet and dry seasons were assessed through pulse 13C labelling. There were significant seasonal and between plot differences in SWC, above-ground biomass, canopy light utilization efficiency (α), CO2 fluxes and C allocation pattern resulting from differences in SWC. The ecosystem was a net C sink during the wet and C neutral during the dry seasons. The study showed strong seasonal fluctuations in ecosystem CO2 fluxes and underscores the significant role of the savanna grasslands in regional C balance due to its expansive nature. The savanna grassland is however vulnerable to low soil moisture, with significant reduction in CO2 uptake during drought.

Borassus aethiopum is a dioecious palm tree of African savannas. A stage-classified matrix population model has been parametrized with field data (Lamto reserve, Ivory Coast). It enabled the study of the persistence of the population, and analysis of its sensitivity to different vital rates. Age of palms in each stage were estimated to complete the description of the palm life-history. The main results are: (1) The studied populations are very close to the equilibrium but the stable stage distribution (predicted by the model) and the observed distribution are significantly different indicating a former change in the vital rates. (2) Reproduction seems to be highly delayed (first reproduction on average at 114 y), while the estimated duration of the reproductive part of the life-cycle is relatively short (22 y).