As Yggdrasil has mentioned, d and f orbitals are far less shieling than s and p orbitals. You would think that they shield better, for example for scanadium, you fill the 4s then the 3d orbitals and 3d orbitals are closer to the nucleus since they have a lower principle quantum number. But d- orbitals are not perfect shielders due to where their electron densities are, take a look at pictures of them and you will see what I mean. So, once they are filled up and you begin putting an electron in p-orbitals, the effective nuclear charge (this charge takes sheilding from other electron orbitals into account) actually increases due to the d orbitals inperfection in shielding. Since the 3d orbitals are introduced in period four, you see this trend breaker when you hit Galium. For the same reason you see this again in period six for thalium when you introduce f orbitals.
This is a pretty complicated explanation, I actually consulted with my research advisor before answering your question and this is what he told me pretty much. There are alot of other factors that contribute to this, like relativity and I don't wish to go into it since I don't really get it myself. So, simply put, when d and f orbitals get filled, their shielding effects do not do a great job in sheilding the increase in effective nuclear charge, so you get an break in the downward IE trend in, not only group 3A but also the group 2B metals too.
Hope this helped, even in the tiniest bit