A selection of lepospondyl amphibians from the family Diplocaulidae, to scale.
Amphibians are a diverse group, including the earliest ones which were little more than walking fish, giant crocodile-like carnivores, armoured landlubbers and forerunners of the amniotes (reptiles, birds, mammals). Lepospondyls were one group which were probably part of the reptiliomorph line (although they are so derived, that their connections may lie elsewhere) which consist of predominantly smaller species that have a passing resemblence, although not closely related, to salamanders and even eels! Nectrideans are one diverse order that range through the Carboniferous and Permian. Most species retain the unspecialised salamandroid form, apart from one remarkable family. With some species looking much like a lizardy pancake that had mistakenly swallowed a boomerang, the aquatic diplocaulids are one of the more recognisable of the fossil amphibians.
So what is the purpose of such a wide head? Much confusion/argument has reigned since the original discovery of Diplocaulus (the best known member of the family) in 1877, and various theories have been proposed. They include a hydrofoil head to aid swimming in swift water, while the sheer width of the skull makes it harder to swallow by predators. The head may have ploughed the bottom substrate stirring up prey. As the prongs were thick extensions of the skull and consist of dense bone, it is entirely plausible they were used for territorial aquatic combat. One caveat is that these ideas can only be applicable to the adults, as young individuals had typical rounded heads and, interestingly, relatively larger orbits. During adolescence the rear corners of the skull grew at a speedier rate than the rest of the body. Some individuals of D. magnicornis have skulls sized around 40 cms wide, with partial remains suggesting potentially even bigger specimens. A trace fossil known as Hermundurichnus fornicatus is attributable to a resting Diplocaulus and seems to show that the the ends of the prongs were connected to the body by skin flaps.
Fossils of Diplocaulus are found in wet environments subject to seasonal aridity. Many specimens have been found curled up in mud burrows, a behaviour seen in modern-day amphibians to survive dry periods. Ultimately some individuals didn't survive the dry season, while others suffered a more grusome fate. A few specimens were discovered dismembered with pieces bitten off and chew marks. Sharp teeth shed from the culprit were found alongside the victims. It seems that the unlucky amphibians were predated by the Early Permian's apex predator Dimetrodon.
Even more extreme within the 'horned' species was
Diploceraspis, which closely resembled
Diplocaulus, but was smaller with relatively longer prongs. A minor difference is the absence of a flange on the underside of the prongs which would have assisted lift in the water, indicating that it probably lived in slower streams compared to it's larger relative. In contrast, early species from the Carboniferous are fairly unremarkable and resemble long-tailed salamanders with two stubby prongs on the back of the skull. One species,
Keraterpeton has the distinction of the most number of vertebrae (up to 26) known in an amphibian. The last known lepospondyl and only Gondwanan member was a species of
Diplocaulus known as
D. minimus. There is some argument that it should be either placed in it's own genus or lumped in with it's closer relative
Diploceraspis. Apart from being the only lepospondyl from the Late Permian,
D.minimus had an unusually asymmetrical skull with the left prong long and tapered, while the right one was short and blunt. Why this should be is difficult to intrepret, but it has been seen in multiple skulls so it was a normal oddity and not due to distortion during fossilisation.

https://www.deviantart.com/artbyjrc/art/Boomerang-heads-Diplocaulids-856073958