Why does magnesium have a lower melting point than both calcium and beryllium? It does not seem to fit into the group trend.
Following up on my comment, even though I'm not convinced that this is a 'real' answer to the question. Consulting the SGTE database of elemental free energies (AT Dinsdale, CALPHAD 15(4) 317-425 (1991) with various updates over the years), one can obtain the melting temperatures for the hcp, fcc, and bcc phases of Be, Mg, Ca, Sr, and Ba. (Rough guesses looking at plots - consider good to ~5K):
Element hcp bcc fcc melting points (K), phases at RT and melt:
Be 1544, 1560, 1018, hcp at RT, bcc at melt
Mg 922, 755, 705, hcp at RT, hcp at melt
Ca 940, 1114, 1063, fcc at RT, bcc at melt
Sr 900, 1050, 1030, fcc at RT, bcc at melt
Ba 635, 1000, 709, bcc at RT and at melt
Several things to note, amongst them the variety of which crystal structure is most stable for the different elements at different temperatures. Also, the hcp phase's melting point does almost(!) monotonically decrease going down the table. The biggest anomaly would actually seem to be the relative instability of the bcc and fcc phases of magnesium (it is the only one that remains hcp up to the melting point). Overall, it would appear that the column 2 elements display a wide variety of preferences for crystal structure, and hence bonding energy configurations. Is Mg unusual? Not really...