I suppose that the authors estimate the causal relationships between two variables and then create a table presenting all relationships in a way similar to the correlation matrix (for the economy of space reasons).
For example, there is a Granger causality relationship running from ED to EU.
Between ED and CO2 there is a bidirectional causal relationship and so on.
thank you very much for the clarification, now I understand because before I was confused, I was thinking is the causality generated through VECM models.