High rates of seed removal can impede forest recovery, but tropical seed removal studies are few and mainly from the neotropics. Little is known about the comparative influences of active restoration (i.e. planting) and passive restoration (i.e. protection of natural regrowth) on seed removal. We conducted an evaluation of seed removal in grasslands, natural forests (tropical moist semideciduous forest), and actively (21‐, 17‐, 16‐, 11‐, 8‐, and 6‐year‐old) and passively (21‐year‐old) restored forests in Kibale National Park, Uganda. We wanted to compare the effect of vegetation type, time since restoration and restoration actions (i.e. active vs. passive) on removal of seeds of five animal‐dispersed tree species during wet and dry seasons. Seeds were either fully exposed or placed in closed mesh cages or under a mesh roof. We used differential removal rates between these treatments to attribute seed removal to different animal taxa. Seed removal rate (percentage of seed removed over a 4‐day period) was highest in passively restored forests, compared with actively restored forests, grasslands, and natural forests. We detected no significant relationship between time since restoration and seed removal rates within actively restored sites. Seed removal rate from roofed treatments was not significantly different from removal from open treatments but was significantly higher than removal from closed treatments, which we interpret as reflecting the greater effect of small mammals versus insects. Smaller seeds tended to be removed at a greater rate than larger seeds. We discuss the implications of these findings for forest regeneration.

The status of woody seedling colonization gives clues about the self‐sustainability of restored forests, a tenet of restoration success. Little is known about woody seedling colonization in restored afrotropical forests. We evaluated effects of restoration methods (active vs. passive), sampling year, restoration age, and distance from old‐growth forests on seedling colonization in restored afrotropical moist forests. Seedlings were measured in 2011 and 2014 in 71 clusters of 284 permanent sampling plots (12.6 m2 each) in actively (initially 3–16 years old) and 21 clusters of 63 plots in passively restored forests (initially 16 years old) in Kibale National Park, western Uganda. Seedlings were also measured in nearby old‐growth forests in three clusters of five plots in 2014. We determined species diversity, richness, abundance per plot, and species composition as measures of seedling colonization in restored and old‐growth forests. We found that diversity, richness, and abundance of seedlings were significantly higher in passively than actively restored forests. Diversity and richness but not abundance significantly increased between sampling years and with restoration age. Distance from old‐growth forests did not significantly affect diversity, richness, and abundance. Species composition of actively and passively restored forests was different from that of old‐growth forests after 19 years since restoration started. Our results show that passive restoration should be the preferred method for recovering afrotropical forests, and highlight the effect of continued management on biodiversity of restored forests.

Aim: There has been a debate about the effectiveness of passive restoration for recovering tropical forests, but few studies quantify the success of passive restoration. The aim of this study was to better understand tropical forest succession under a passive restoration scenario. We compared floristic and functional attributes of seedlings in a passively restored and an old-growth forest, and assessed the effect of restoration age and distance from the old-growth forest on seedling attributes. Location: Kibale National Park, western Uganda. Methods: We measured seedlings in a passively restored and an old-growth forest in 2011, 2014 and 2017. We determined species diversity, structure and composition and searched the literature for functional traits. We used generalized linear mixedeffects models to compare seedling attributes between the restored and old-growth forest and determine the influence of restoration age and distance from the oldgrowth forest. Results: Seedling species abundance, evenness, basal area and height were similar between the restored and old-growth forest. Wood density and abundance of seedlings of different dispersal modes, habitat types, fruit size categories, and regeneration strategies were also similar between the restored and old-growth forest. However, richness, diversity and composition of seedlings were different. We found a positive effect of restoration age on species abundance and abundance of non-zoochorous, medium-fruited, forest-dependent, non-pioneer light demander and shade-tolerant species, and a negative effect on evenness, wood density, abundance of pioneers and compositional dissimilarity. Basal area of seedlings and the abundance of zoochorous and forest-dependent species declined while compositional dissimilarity increased with distance from the old-growth forest. Conclusions: Our results provide empirical evidence on the potential of passive restoration to recover the structure and functionality of tropical forests in a relatively short period of time. We demonstrate that the effect of restoration age and distance from the old-growth forest is not straightforward and depends on the attributes measured.

Tropical forest management often focuses on a few high‐value timber species because they are thought to be the most vulnerable in logged forests. However, other tree species may be vulnerable to secondary effects of logging, like loss of vertebrate dispersers. We examined vulnerability of tree species to loss of vertebrate dispersers in Mabira, a heavily disturbed tropical rainforest in Uganda. Fruit characteristics and shade tolerance regimes of 269 tree species were compiled. Stem densities of tree species producing fruits of various sizes and having different shade tolerance regimes were computed for Mabira and compared with densities of conspecifics in Budongo, a less disturbed forest with similar floral composition. Seventy per cent of tree species in Mabira are animal‐dispersed, of which 10% are large‐fruited light demanders. These species are the most vulnerable because they rarely recruit beneath adult conspecifics and are exclusively dispersed by large vertebrates, also vulnerable in heavily disturbed forests. Comparison of densities between Mabira and Budongo showed that large‐fruited light demanders had a lower density in Mabira. Other categories of tree species had similar densities in both forests. It is plausible that the low density of large‐fruited light demanders is due to limited recruitment caused by dispersal limitations.

The growing trend of agricultural abandonment necessitates understanding the development of regrowth forests on old fields in the context of forest restoration. However, the successional patterns of ecosystem functioning and functional diversity of afrotropical regrowth forests are rarely examined. We assessed whether aboveground biomass (AGB) and functional diversity (FD) vary with restoration age and proximity to old-growth forests, compared AGB and FD between regrowth and old-growth forests to measure restoration success and investigated the FD – AGB relationship. We sampled trees in 63 plots (2000 m2 each) in a regrowth forest and 5 plots in an old‐growth forest in 2011, 2014 and 2017. We calculated AGB using diameter, height and wood density. We collated species functional traits (dispersal modes, habitat types, fruit sizes and regeneration guilds) and computed FD measures (richness, evenness, dispersion, divergence and RaoQ’s entropy). AGB and FD measures (richness, dispersion and RaoQ) increased with restoration age. Functional divergence declined with increasing distance to the old-growth forest. Within 22 years, regrowth forests regained 22% of the AGB and recovered all FD measures of the old-growth forest. We found positive, negative and quadratic relationships between AGB and FD depending on the FD measure and forest type. We demonstrate that regrowth forests increase ecosystem production and functional diversity in abandoned areas, however they cannot substitute old-growth forests. Considering multiple measures of functional diversity in different habitats provides a better understanding of the influence of functional diversity on ecosystem functioning.