Found this on IFLS today; http://www.iflscience.com/plants-and-animals/permian-predator-returned-oceans-eat-shark Amazing!
The article didn't probe too deep into why they lived on land if they ate mostly seafood, but they hinted at it. Since the Dimetrodon both prey on and were preyed on by sea creatures, the ability to avoid aquatic predators by going on land would be invaluable. I suppose since Lustria has no shortage of terrestrial predators, Salamanders would be more likely to use water as an escape mechanism from land predators.
From what I gather, at that time bodies of water were rather well populated with a variety of animal life of greater diversity than today. I imagine it was like how fishing ducks catch fish today: developing a means to efficiently cross into another medium, while not needing to compete with the natives down there all the time. They probably needed the fins for rapid warm-up after exposure to the colder water for extended sessions.
Yeh, IFLS is more sort of general interest science rather than research journals - sometimes they have some fairly detailed stuff...