Tiktaalik roseae is a lovely example of a transitional fossil - it has a number of morphological features that clearly place it on the 'fish-to-tetrapod' transition. The type specimen is a remarkably well-preserved fossil that's been very carefully analysed. A recent paper by Jason Downs & his co-workers described the results of their examination of the fossil's cranium (Downs et al. 2008).
While Tiktaalik - which lived during the late Devonian period - is definitely a lobe-finned fish, it also posseses a number of features also seen in tetrapods. In fact, when Tiktaalik was first described, it was called the 'fish-a-pod' because of the presence of a wrist-joint in its pectoral fins.
This feature would certainly have been advantageous in a lineage that ultimately colonised the land. But life on land requires a number of other functional adaptations - not only for gas exchange (lungs must have ante-dated terrestrial life) & support and movement, but also for feeding. After all, in the water an animal needs only to open its mouth, and food may well be brought in by the inrush of water. But on land, an animal needs greater mobility in the neck department, enabling greater movement of the head to capture passing prey.
Various features of Tiktaalik's morphology place it between the lobe-finned fish Panderichthys and the early amphibian Acanthostega in the tetrapod phylogenetic tree. Its excellent preservation, and the increasing number of fossils of this species, allowed Downs & his colleagues to examine the cranial features associated with a more mobile head - previously little was known about changes in this feature during tetrapod evolution.
Among other things, they found that the general shape of Tiktaalik's cranium was closer to that of early tetrapods than to Panderichthys and other lobe-finned fish. Changes in its head shape seem to have been related to changes in a particular bone, the hyomandibula. In fish this is an important part of the articulation between cranium and jaw, while in amphibians and reptiles the hyomandibula has become the columella, the bone that transmits vibrations between the ear drum & the middle ear. In Tiktaalik the hyomandibula no longer braced the upper jaw, & this change - along with the loss of other bones - meant that the animal would have had greater head mobility than earlier organisms in this lineage. Like lungs & limbs, the beginnings of a neck seem to have pre-dated movement onto land. But like lungs & limbs, the presence of a neck would also be advantageous, particularly in animals living in shallow waters at the land's edge. As Down et al. (2008) note, [fish] move and feed in three-dimensional space and are readily able to orient the body in order to position the mouth towards prey. A neck is advantageous in seetings where teh body is relatively fixed, as is the case in shallow pools of water or in marginal settings where the body is supported by appendages planted against a substrate. Interestingly, the team suggest that the way Tiktaalik's hyomandibula is oriented also suggests that the animal relied less than other fish on pumping water over its gills during respiration.
Fish and tetrapods also differ in the degree to which their cranial bones can move. There's quite a bit of 'intracranial kinesis' in fish, where the bones move against each other as the animal feeds and respires. This is not the case for most tetrapods - and Tiktaalik shows an intermediate level of fusion in the cranium & between cranium & upper jaw.
The team conclude:
[with] appendages able to support the weight of the body, a more consolidated skull, ribs capable of sustaining the trunk under a gravitational load, a respiratory system less reliant on water breathing, a head capable of independent motion, and cranial proportions approaching those of tetrapods, Tiktaalik reveals an important stage in the origin of terrestrial vertebrates.
J.P. Downs, E.B. Daeschler, F.A. Jenkins Jr & N.H. Shubin (2008) The cranial endoskeleton of Tiktaalik roseae. Nature 455: 925-928