Present-day patterns of species diversity and their ecological characteristics in the tropics result from more than 100 million years of evolution. The environmental conditions in which species evolved have left a fingerprint on their functional traits, so investigating this legacy may improve our understanding of current patterns of ecosystem function and potentially guide us in managing our resources more wisely as the climate changes. Amazonian forests are ideal for such a study as they play a major role in the global carbon cycle and harbour a remarkable diversity of angiosperm lineages and species with a broad range of ecologies. Here, I linked data from long-term forest inventory plots, environmental conditions, and a newly constructed phylogeny in order to investigate the legacy of evolution on modern-day patterns of ecosystem function and diversity. I show first that evolutionarily related taxa are more similar in their demography and carbon storage and processing ability (i.e. wood density, potential size, growth and mortality-rates) than expected by chance. Thus, the degree of evolutionary history shared between lineages is a good proxy for their carbon traits. Next, using the evolutionary relationships among lineages I find a legacy of evolutionary history on current patterns of whole ecosystem productivity across the Amazon, such that communities with more evolutionarily distinct lineages have greater wood productivity. Finally, I compare the role of heritability versus selection on shaping lineages preferences for certain environments (i.e. soils and climate). I show that there is a tendency for evolutionarily related taxa to have more similar environmental preferences than expected by chance, but that certain kinds of habitat specialisation have also occurred repeatedly and independently in many lineages. These findings are important for understanding the future of Amazonian forests under global change and support an evolutionary perspective as an important component of conservation strategies.