Tropical forests are one of the most biodiverse, carbon-dense, but also threatened ecosystems in the world. Hence, promoting their conservation and ensuring the success of the successional process are important steps to maintain ecosystem functioning and effectively mitigate biodiversity loss and climate change. Here we aimed to determine the main drivers of forests attributes (e.g., structure, diversity and functional composition) that are informative of community assembly and ecosystem functioning and understand the functional vulnerability of high-diversity tropical rainforests situated on nutrient-poor soils and within highly fragmented human-modified landscapes. For that, we developed two independent but complementary studies: in the first (C1), we used data from 1.9 ha old-growth and 1.6 ha second-growth forest plots to assess the effects of landscape structure (matrix openness, patch density) on nine forest attributes related to structure (basal area, maximum height, structural heterogeneity), diversity (species richness, Simpson diversity, functional richness), and composition (community weighted-mean animal-dispersal shade-tolerance and seed size) over four different spatial scales (400 to 3200 m buffer radius) in three different regions (i.e., landscape contexts). We found that landscape openness negatively affected most forest attributes, but the magnitude of effect varied with i) landscape attribute, ii) forest attribute, and iii) region (i.e. landscape context): deforestation had a negative stronger effect than fragmentation; landscape openness reduced most strongly forest structure and animal-dispersed species; landscape openness had different effects in different regions, and affected forest structural attributes at all spatial scales, animal dispersed species at larger spatial scales, and seed size at small spatial scales. These findings indicate that landscape effects on forest attributes in human-modified landscapes cannot be generalized as they depend on the landscape context. In the second study (C2), we used data from 2.7 ha second-growth forest plots to assess the how stand age, landscape structure (matrix openness, patch density), seasonal water availability (climatic water deficit), and soil properties (sum of base, nitrogen, organic carbon and clay contents) predict aboveground carbon stock and functional composition (community weighted-mean maximum stem diameter, maximum stem height and species-specific wood density), redundancy, and vulnerability of second-growth Atlantic forests. We found that forest age and soil properties were the main drivers of aboveground carbon stock, functional redundancy and vulnerability; that seasonal water availability predicted species growth and wood density; while soil properties predicted species growth, but not wood density; and finally, that stands in highly fragmented and deforested landscapes can harbor carbon-dominant species. These findings advance our understanding of the factors that drive carbon stock and vulnerability in second- growth Atlantic forests and highlight the importance of considering local site conditions in forest restoration and climate change mitigation efforts. Keywords: Climate Change. Functional traits. Land use Change. Landscape context. Dispersal. Soil Properties. Spatial Scale. Succession