In Central Amazonia, large mammals create water-filled puddles when consuming soil. These mineral licks are visited by pregnant and lactating frugivorous bats; possibly for two reasons. Frugivorous bats could supplement their mineral-depleted fruit diet by drinking salty water, or they could buffer dietary plant secondary compounds by consuming soil. We analysed bat fruits from Ecuador and showed that they are depleted in elemental concentrations (Na, K, P) compared with similar fruits collected from Costa Rica, where no mineral licks occur (n = 32). Analyses of water from Ecuador revealed that water samples from six mineral licks contained more physiologically relevant elements (Na, K, Mg, Ca) than four samples from river and stream water control sites (Mann–Whitney U-test). In support of the nutrient supplement hypothesis, we observed bats drinking mineral-enriched water at these licks (video observation). Furthermore, blood collected from 68 bats differed in composition with respect to physiologically relevant minerals (Na, K, Mg, Fe) from that of frugivorous bats captured at control sites. To test whether frugivorous bats also consumed clay for detoxification, we checked for soil tracer elements in 31 faecal samples. Soil tracers are insoluble in water and, thus, are not included in a strict fruit diet. Bats from mineral licks showed higher aluminium soil tracer concentrations in their faeces than bat species that never visit licks, suggesting that frugivorous bats take up clay material at mineral licks. Our results provide evidence that frugivorous bats ingest soluble mineral nutrients and insoluble soil by consuming soil-enriched water at mineral licks, thus supporting the hypothesis that frugivorous bats of western Amazonia may derive a dual benefit from drinking water from mineral licks.

Termites are renowned ecosystem engineers. Their mounds have been described as an important element of savanna vegetation dynamics, but little is known about their large-scale impact on vegetation composition. To investigate the influence of termite-induced heterogeneity in savannas along a climatic gradient in West Africa termite mound vegetation was compared with adjacent savanna vegetation using 256 paired plots (size of the termite mound and a corresponding savanna area) in five protected areas from northern Burkina Faso to northern Benin. On each plot vegetation and soil sampling was performed. Additionally bioclimatic variables from the WORLDCLIM database were used. The vegetation on the mounds and the surrounding savanna differed within all study sites (DCA length of gradient 3.85 SD) and showed complete turnover along the climatic gradient (DCA length of gradient 5.99 SD). Differences between mounds and savanna were significantly related to termite-induced changes in soil parameters, specifically clay enrichment and increased cation concentrations (base saturation). On a local scale, termite-induced differences in soil conditions were found to be the most important factor affecting mound vegetation, while on a regional scale, annual precipitation showed the strongest significant correlations. However, with increasing precipitation, differences between mounds and the surrounding matrix became more pronounced, and the contribution of mounds to local phytodiversity increased. Eleven plant species were identified as characteristic termite mound species. In the more humid parts of the gradient, more characteristic plant species were found that may benefit from favourable soil conditions, good water availability, and a low fire impact in the mound microhabitat.

Savannas intermingled with gallery forests are dynamic habitats typical in Africa. This study aims to determine if differences in species traits lead to non-overlapping distribution of gallery-forest and savanna species and abrupt transition between gallery forest and savanna. Tree species densities were measured in 375 plots of 1500 m2 covering a total sample area of 56.25 ha along forty 3-km transects located at right angles to a riverbed with gallery forest into surrounding savanna. Location, vegetation type, soil physical properties, erosion and fire occurrence were recorded as site factors. Data analysis included the quantification of co-occurrence patterns, threshold indicator taxa analysis and fuzzy set ordination. The gallery forest–savanna gradient predicted floristic composition of plots with a correlation of 0.595 but its accuracy was locally modified by the occurrence of fire and the physical properties of soil that covered more than 30% of the range of residuals. The distribution of gallery-forest and savanna tree species did not overlap. Along the gallery forest–savanna gradient, savanna species gradually increased in density while gallery-forest species showed a community threshold at 120 m from the river beyond the width of gallery forest. The forest species driving this trend should play an important role in the dynamics of gallery forest–savanna boundaries.

Resource depletion and associated increases in interspecific competition are likely to influence differential habitat usage amongst a guild. We tested some prominent theoretical concepts using observed differences in seasonal habitat use amongst the savanna browser guild (elephant, giraffe, impala, kudu and nyala) in Hluhluwe-iMfolozi Park, South Africa. Herbivore locations (n = 3108) were recorded over 2 y using repeated road transects and, for elephant, GPS collars (187 254 downloads). Densities were calculated using a novel GIS approach designed to be a cost-effective method for annual censuses, but also able to cope with abrupt changes in visibility. Selectivity for (Manly's α) vegetation types, and overlap (Schoener's index) in vegetation type usage were calculated. Resource depletion in the dry season resulted in all members of the guild increasing selectivity for vegetation types (sum of absolute values away from the neutral value for Manly's alpha for the guild: dry seasons 3.97, 5.16; corresponding wet seasons 3.12, 3.68), but decreasing interspecific overlap (80% of Schoener's indices lower in dry season versus wet season). These effects were more marked over the second, more severe, dry season. We found support for the niche overlap hypothesis and the niche compression hypothesis. The Jarman–Bell principle was generally supported, although unexpectedly during the severe dry season elephant showed the most selectivity for vegetation type. The greater the resource depletion, the more relevant interspecific differences in habitat usage become in relation to the differential impacts of guild members.

Total ecosystem carbon storage has frequently been found to increase with woody encroachment in savannas. However the loss of grass roots associated with woody encroachment can lead to a decrease in below-ground carbon storage which is not compensated for by an increase in above-ground carbon. To investigate how the extent of total woody cover affected ecosystem carbon, soil and above-ground carbon storage along eight thicket–savanna and five forest–grassland boundaries were measured. To investigate whether changes in soil carbon concentrations were related to the percentage of C4 (grass) roots to total roots and root quantity and quality, we measured fine-root biomass, root C : N ratios, root N, and % C4 roots at three different depths across thicket patches of different ages (n = 189). Forests contained significantly more carbon than adjacent grasslands in both above-ground carbon (mean difference 12.1 kg m−2) and in the top 100 cm of the soil (mean difference 4.54 kg m−2). Thickets contained significantly more above-ground carbon than adjacent savannas (3.33 kg m−2) but no significant differences in soil carbon were evident. Total fine-root biomass appeared to be more important than root quality (root C : N) in determining soil carbon concentrations during the encroachment process (i.e. in thicket of different ages). Similarly for thickets, the % C4 roots had no significant effect on soil carbon concentrations. In conclusion, thicket invading into open savanna vegetation did not lead to significant gains in ecosystem carbon at this study site. Significant gains were only evident in mature forest, suggesting that the process may take place very slowly.

Community-wide seasonal fruiting and seed dispersal by the common brown lemur (Eulemur fulvus fulvus) was investigated in a tropical dry forest, north-western Madagascar. The brown lemur is the sole disperser of large seeds > 10 mm in diameter. Considering the limited assemblage of dispersers, large-seeded plants should display distinctive fruiting patterns to acquire dispersers. To compare fruiting patterns and seed dispersal by the brown lemur between small-seeded and large-seeded plants, fruiting conditions for 432 trees on a transect, feeding activity over 1212 h of observation, and the composition of 1126 dung samples were recorded for 1 y. Seeds of a total of 52 species were identified through both observations and faecal analysis. As rainfall increased, larger numbers of species and individuals of small-seeded plants fruited. Among the 52 species, the brown lemur dispersed 29 and 13 species with small and large seeds, respectively, during the rainy season. High moisture levels probably favoured seed germination and seedling establishment in various species. During the dry season, although small-seeded species rarely fruited, a few large-seeded species, particularly Vitex beraviensis, formed a long-term fruiting peak and provided essential food resources for the brown lemur. Because seeds of these large-seeded plants were frequently dispersed by the brown lemur, dry-season fruiting seemed to be favourable to avoid competition for dispersers with other plant species.

Interactions between ants and non-myrmecochorous diaspores occur frequently on the forest floor and the results of these interactions (e.g. diaspore cleaning or removal) will often depend on the morpho-chemical characteristics of the diaspores. We conducted two different experiments with seven plant species in the north-eastern Atlantic forest, Brazil. To evaluate whether ant interactions decrease diaspore persistence time on the forest floor, we established sample stations by placing five diaspores of each species spaced every 10 m along a transect and monitored ant interactions over a 24-h period. We also compared diaspore removal by ants and vertebrates through a paired experiment. We monitored removal of the seven plant species in treatments where ants were excluded and treatments where vertebrates were excluded. We recorded 332 ant–diaspore interactions, most resulting in ants cleaning diaspores in situ. Persistence of diaspores on the forest floor varied greatly among studied species and was less than 50% for three species. Ants also removed more diaspores than did vertebrates. Number of diaspores removed by ants was greater for small diaspores and with high lipid concentration. Ant–diaspore interactions may not always exert a positive influence on plant recruitment for diaspores with poor nutrient concentration. Consequently, ants may play a disproportionately important role as secondary dispersers in tropical forests.

Freshwater migratory shrimps, an important component of tropical aquatic ecosystems, are vulnerable to land-use change during their upstream and downstream migrations. At La Selva Biological Station in the Sarapiquí region of Costa Rica, shrimp population data were collected between 1988 and 1989, before massive land-use change occurred downstream that could potentially affect shrimp recruitment upstream. Using generalized linear models and a Bayesian inference framework, the relative abundance of Macrobrachium olfersi between recent (2008–2011) and historical time periods (1988–1989) was compared in three stream reaches. Shrimp relative abundance in two stream reaches within the protected area of La Selva was relatively constant yearly and between recent post-disturbance (2008–2011) and historical pre-disturbance (1988–89) time periods. In contrast, a stream reach bordered by pasture accessible to fishermen, showed an 87% decrease in relative abundance between recent and historical time periods suggesting site-level disturbance, possibly from fishing. The lack of change between historical and contemporary sampling periods within interior-forest stream reaches suggests that shrimp populations in protected forested reaches are resistant or resilient to certain land-use changes occurring downstream.