Secondary succession is a widespread phenomenon in the Anthropocene, primarily driven by human land-use changes and disturbance. As a result, secondary forests now comprise more than 50% of the world’s tropical forests, making them a main component of human-modified tropical landscapes and central to global restoration efforts. Yet, our ability to predict successional trajectories and to scale restoration strategies across human-modified landscapes is limited because of key knowledge gaps related to (1) secondary succession at the very start, (2) how multiple attributes change simultaneously over time, and (3) how successional trajectories vary across socio-ecological systems. Here, we analyzed secondary succession across six tropical landscapes (three countries × dry and wet forests). We established 122 permanent plots in recently abandoned agricultural fields and monitored them annually up to five years, measuring 12 attributes related to structure, diversity, functional composition, and biotic interactions. We found that (1) large variation in successional trajectories across and within sites emerged within the very first years, driven by coarse- and fine-scale variations in climate and human land-use; (2) different forest attributes followed distinct trajectories; while structure and diversity increased, functional composition and biotic interactions showed little change, indicating that forest recovery is inherently multidimensional; (3) early successional trajectories were more strongly shaped by landscape forest cover and previous land-use history than by climate, highlighting the need to integrate human activities into successional studies and restoration planning. Together, these findings demonstrate that early succession, multidimensionality, and socio-ecological context must be integrated to improve predictions of forest succession and to design effective, context-specific restoration strategies in human-modified tropical landscapes.