We present the first cross‐continental comparison of the flowering and fruiting phenology of tropical forests across Africa. Flowering events of 5446 trees from 196 species across 12 sites and fruiting events of 4595 trees from 191 species across 11 sites were monitored over periods of 6 to 29 years and analyzed to describe phenology at the continental level. To study phenology, we used Fourier analysis to identify the dominant cycles of flowering and fruiting for each individual tree and we identified the time of year African trees bloom and bear fruit and their relationship to local seasonality. Reproductive strategies were diverse, and no single regular cycle was found in >50% of individuals across all 12 sites. Additionally, we found annual flowering and fruiting cycles to be the most common. Sub‐annual cycles were the next most common for flowering, whereas supra‐annual patterns were the next most common for fruiting. We also identify variation in different subsets of species, with species exhibiting mainly annual cycles most common in West and West Central African tropical forests, while more species at sites in East Central and East African forests showed cycles ranging from sub‐annual to supra‐annual. Despite many trees showing strong seasonality, at most sites some flowering and fruiting occurred all year round. Environmental factors with annual cycles are likely to be important drivers of seasonal periodicity in trees across Africa, but proximate triggers are unlikely to be constant across the continent.

No abstract

No abstract

Considerable areas dominated by bracken Pteridium aquilinum (L.) Kuhn occur worldwide and are associated with arrested forest recovery. How forest recovery is impeded in these areas remains poorly understood, especially in the African highlands. The component processes that can lead to recruitment limitation—including low seed arrival, availability and persistence—are important determinants of plant communities and offer a potential explanation for bracken persistence. We investigated key processes that can contribute to recruitment limitation in bracken‐dominated clearings in the Bwindi Impenetrable National Park, Uganda. We examined if differences in seed rain (dispersal limitation), soil seed bank, or seed removal (seed viability and persistence) can, individually or in combination, explain the differences in tree regeneration found between bracken‐dominated areas and the neighboring forest. These processes were assessed along ten 50‐m transects crossing the forest–bracken boundary. When compared to the neighboring forest, bracken clearings had fewer seedlings (bracken 11,557 ± 5482 vs. forest 34,515 ± 6066 seedlings/ha), lower seed rain (949 ± 582 vs. 1605 ± 335 tree seeds m−2 year−1), comparable but sparse soil seed bank (304 ± 236 vs. 264 ± 99 viable tree seeds/m2), higher seed removal (70.1% ± 2.4% vs. 40.6% ± 2.4% over a 3‐day interval), and markedly higher rodent densities (25.7 ± 5.4 vs. 5.0 ± 1.6 rodents per 100 trapping sessions). Camera traps revealed that rodents were the dominant animals visiting the seeds in our seed removal study. Synthesis: Recruitment limitation contributes to both the slow recovery of forest in bracken‐dominated areas, and to the composition of the tree species that occur. Low seed arrival and low persistence of unburied seeds can both explain the reduced density of seedlings found in bracken versus neighboring forest. Seed removal, likely due to rodents, in particular appears sufficient to constrain forest recovery and impacts some species more severely than others.

Elephants are locally concentrated in Bwindi Impenetrable National Park. Vegetation damage attributable to elephants appears to be increasing and may result in the modification of the forest. We examined the implied selectivity of stem damage due to elephants. We followed 26.84 km of recent elephant trails and used 122 plots to document tree damage in relation to species, stem sizes and locations. Of 897 trees (DBH ≥2 cm), 542 (60.4%) were intact, 22 (2.5%) debarked, 274 (30.5%) toppled and 172 (19.2%) had broken branches. Small trees were more likely to be pushed over or have their branches broken, whereas large trees were more commonly debarked. The species most frequently selected for damage included mid‐successional species such as Newtonia buchananii, Myrianthus holstii and Chrysophyllum albidum. These species may be vulnerable to increasing elephant numbers. Our analyses using general linear models indicate that elephants are selective concerning where, how and what tree stems they damage. We found a higher incidence of elephant damage per‐tree stem in open areas than in more closed areas, suggesting feedback in which elephants maintain open habitats that may be conducive for other species such as mountain gorillas. More work is needed to better determine how changing elephant numbers may influence Bwindi's conservation values.