Much of the ecological discourse surrounding the polarising theropod Spinosaurus has centred on qualitative discussions. Using a quantitative multivariate data analytical approach on size-adjusted linear measurements of the skull, we examine patterns in skull shape across a range of sauropsid clades and three ecological realms (terrestrial, semi-aquatic, and aquatic). We utilise cluster analyses to identify emergent properties of the data which associate properties of skull shape with ecological realm occupancy. Results revealed terrestrial ecologies to be significantly distinct from both semi- and fully aquatic ecologies, the latter two were not significantly different. Spinosaurids (including Spinosaurus) plotted away from theropods in morphospace and close to both marine taxa and wading birds. The position of nares and the degree of rostral elongation had the greatest effect on categorisation. Comparisons of supervised (k-means) and unsupervised clustering demonstrated categorising taxa into three groups (ecological realms) was inappropriate and suggested instead that cluster division is based on morphological adaptations to feeding on aquatic versus terrestrial food items. The relative position of the nares in longirostrine taxa is associated with which skull bones are elongated. Rostral elongation is observed by either elongating the maxilla and the premaxilla or by elongating the maxilla only. This results in the nares positioned towards the orbits or towards the anterior end of the rostrum respectively, with implications on available feeding methods. Spinosaurids, especially Spinosaurus, show elongation in the maxilla-premaxilla complex, achieving similar functional outcomes to elongation of the premaxilla seen in birds, particularly large-bodied piscivorous taxa. Such a skull construction would bolster "stand-and-wait" predation of aquatic prey to a greater extent than serving other proposed feeding methods.