Megaherbivores modify forest structure and increase carbon stocks through multiple pathways.

Megaherbivores have pervasive ecological effects. In African rainforests, elephants can increase aboveground carbon, though the mechanisms are unclear. Here, we combine a large unpublished dataset of forest elephant feeding with published browsing preferences totaling nearly 200,000 records covering >800 plant species and with nutritional data for 145 species. Elephants increase carbon stocks by: 1) promoting high wood density trees via preferential browsing on leaves from low wood density species, which are more palatable and digestible; and 2) dispersing seeds of trees that are relatively large and have the highest average wood density among tree guilds based on dispersal mode. Loss of forest elephants could cause an increase in abundance of fast-growing low wood density trees and a 6% to 9% decline in aboveground carbon stocks due to regeneration failure of elephant-dispersed trees. These results demonstrate the importance of megaherbivores for maintaining diverse, high-carbon tropical forests. Successful elephant conservation will contribute to climate mitigation at a globally-relevant scale.

Trait matching and sampling effort shape the structure of the frugivory network in Afrotropical forests.

Frugivory in tropical forests is a major ecological process as most tree species rely on frugivores to disperse their seeds. However, the underlying mechanisms driving frugivore-plant networks remain understudied. Here, we evaluate the data available on the Afrotropical frugivory network to identify structural properties, as well as assess knowledge gaps. We assembled a database of frugivory interactions from the literature with > 10 000 links, between 807 tree and 285 frugivore species. We analysed the network structure using a block model that groups species with similar interaction patterns and estimates interaction probabilities among them. We investigated the species traits related to this grouping structure. This frugivory network was simplified into 14 tree and 14 frugivore blocks. The block structure depended on the sampling effort among species: Large mammals were better-studied, while smaller frugivores were the least studied. Species traits related to frugivory were strong predictors of the species composition of blocks and interactions among them. Fruits from larger trees were consumed by most frugivores, and large frugivores had higher probabilities to consume larger fruits. To conclude, this large-scale frugivory network was mainly structured by species traits involved in frugivory, and as expected by the distribution areas of species, while still being limited by sampling incompleteness.

Local people enhance our understanding of Afrotropical frugivory networks.

Afrotropical forests are undergoing massive change caused by defaunation, i.e., the human-induced decline of animal species,1 most of which are frugivorous species.1,2,3 Frugivores' depletion and their functional disappearance are expected to cascade on tree dispersal and forest structure via interaction networks,4,5,6,7 as the majority of tree species depend on frugivores for their dispersal.8 However, frugivory networks remain largely unknown, especially in Afrotropical areas,9,10,11 which considerably limits our ability to predict changes in forest dynamics and structures using network analysis.12,13,14,15 While the academic workforce may be inadequate to fill this knowledge gap before it is too late, local ecological knowledge appears as a valuable source of ecological information and could significantly contribute to our understanding of such crucial interactions for tropical forests.16,17,18,19,20,21 To investigate potential synergies between local ecological knowledge and academic knowledge,20,21 we compiled frugivory interactions linking 286 trees to 100 frugivore species from the academic literature and local ecological knowledge coming from interviews of Gabonese forest-dependent people. Here, we showed that local ecological knowledge on frugivory interactions was substantial and original, with 39% of these interactions unknown by science. We demonstrated that combining academic and local ecological knowledge affects the functional relationship linking frugivore body mass to seed size, as well as the network structure. Our results highlight the benefits of bridging knowledge systems between academics and local communities for a better understanding of the functioning and response to perturbations of Afrotropical forests.