Important caveats: data and phylogenetically controlled models alone don't tell you the causal mechanism. That's why formal theory is useful - you can model assumptions and logic for possible mechanisms. A lot of the paper explains the role of theory in helping us interpret non-causal analyses. 10/
