LAUSR

Abstract Cacao agroforestry have been considered as biodiversity‐friendly farming practices by maintaining habitats for a high diversity of species in tropical landscapes. However, little information is available to evaluate whether this agrosystem can maintain functional diversity, given that agricultural changes can affect the functional components, but not the taxonomic one (e.g., species richness). Thus, considering functional traits improve the understanding of the agricultural impacts on biodiversity. Here, we measured functional diversity (functional richness‐FD, functional evenness‐FEve, and functional divergence‐Rao) and taxonomic diversity (species richness and Simpson index) to evaluate changes of bird diversity in cacao agroforestry in comparison with nearby mature forests (old‐growth forests) in the Brazilian Atlantic Forest. We used data from two landscapes with constraining areas of mature forest (49% Una and 4.8% Ilhéus) and cacao agroforestry cover (6% and 82%, respectively). To remove any bias of species richness and to evaluate assembly processes (functional overdispersion or clustering), all functional indices were adjusted using null models. Our analyses considered the entire community, as well as separately for forest specialists, habitat generalists, and birds that contribute to seed dispersal (frugivores/granivores) or invertebrate removal (insectivores). Our findings showed that small cacao agroforestry in the forested landscape sustains functional diversity (FD and FEve) as diverse as nearby forests when considering the entire community, forest specialist, and habitat generalists. However, we observed declines for frugivores/granivores and insectivores (FD and Rao). These responses of bird communities differed from those observed by taxonomic diversity, suggesting that even species‐rich communities in agroforestry may capture lower functional diversity. Furthermore, communities in both landscapes showed either functional clustering or neutral processes as the main driver of functional assembly. Functional clustering may indicate that local conditions and resources were changed or lost, while neutral assemblies may reveal high functional redundancy at the landscape scale. In Ilhéus, the neutral assembly predominance suggests an effect of functional homogenization between habitats. Thus, the conservation value of cacao agroforestry to harbor species‐rich communities and ecosystem functions relies on smallholder production with reduced farm management in a forested landscape. Finally, we emphasize that seed dispersers and insectivores should be the priority conservation targets in cacao systems.