Abstract
One of the recurring paleobiological questions over the last three decades has been whether there were any filter feeding tetrapods before whales evolved. Recently, a study proposed that a small marine reptile from the Early Triassic, Hupehsuchus nanchangensis, filter-fed in a mode similar to living right and bowhead whales (Balaenidae). The case for filter feeding was largely based on perceived similarities in dorsal-view cranial morphology between Hupehsuchus and Balaenidae, analyzed through geometric morphometrics of 2D landmarks. Here, we show that this similarity was an artifact of multiple errors, including the use of a dataset of extant cetaceans that does not match the morphology of respective species. Notably, 15 of the cetacean species examined were represented by narrow skulls reminiscent of Hupehsuchus-; without these unrealistic data points, Hupehsuchus has no morphospace overlap with any cetaceans, invalidating the proposed inference for Triassic filter-feeding. We collected a new set of landmarks using the published definitions to see how the result changes when using more accurate data along the original authors' intention. We determined that odontocetes and mysticetes do not overlap in morphospace with Hupehsuchus, which plots outside any living cetacean species. We conclude that there is insufficient evidence to suggest that Hupehsuchus was a filter feeder, in concordance with: energetic studies suggesting balaenid-style feeding would be unsustainable at the small body sizes of Hupehsuchus; the lack of an intraoral space for the baleen; and the long neck and comparatively small head that are unsuitable for continuous ram feeding to filter prey-laden volumes of water. This re-examination of Hupehsuchus highlights the challenges for inferring filter-feeding in other extinct tetrapods.