Direct and indirect human impacts may be causing widespread vegetation changes African tropical dry forests (TDFs). This study provides the first field-based large-scale assessment of vegetation changes for the miombo region, using nation-wide re-measured permanent plots for the Republic of Zambia. Using path analysis, a technique used to describe and quantify causal relationships, we investigated the drivers of change for the 2006-2014 period, under different land-use and productivity trajectories. We assessed the change in vegetation metrics representing stand structure and tree diversity, identified causal factors for species richness, basal area, and productivity and compared the biomass change of different species. We assessed carbon emissions and absorptions from forestlands and used error propagation and sensitivity analysis to quantify uncertainty. Our results suggest that Zambia's TDFs are resilient in the face of human activities, with significant biomass gains observed in the re-measured plots over the country. However, the proximity to roads, fragmentation by other land uses, and to a lesser extent fire occurrence were found to negatively affect productivity. We found that biomass gains were concentrated in several dominant species, mostly belonging to a single subfamily of non-nodulating legumes (Fabaceae, subfamily Caesalpinioideae) characteristic of the region. Our results indicate that Zambia's TDFs have been acting as an overall carbon sink, despite large carbon emissions from land-cover change. Decline in biomass for certain dominant species signal a risk of over-exploitation. We also identified important differences in plant diversity and functional traits between miombo woodlands and other types of African savanna vegetation, signaling differences in ecological processes at play. These results illustrate the ecological complexity and diversity of Africa's vegetation, and caution against overgeneralizations of ecological processes in the context of global change and carbon management. Future research should focus on understanding the observed species-specific biomass gain and identifying its potential drivers.