I am under the impression that the covalent character in the bonding of transition metal compounds increases across a group; in fact I think this is why they're called transition metals (a transition from purely ionic compound shown by the pre transition metals to covalent bonding in the non-metals). I think d orbital splitting and ligand field effects become more significant across the series (please confirm or deny) which would support this assertion. Is the reason for this that the d orbitals decrease in energy across the series, as a result of increasing effective nuclear charge, which results in a better energy match with ligand orbitals?
However, could one not argue that the covalency should decrease across the series as the d orbitals become more contracted and orbital overlap with ligands is less efficient. One trend that would seem to support this view is that the metal carbon bonds in organometallic complexes with alkyl and aryl ligands tend to decrease in strength across the series.