Abstract
Even in the era of large molecular datasets, morphological evidence makes an integral contribution to reconstructing evolutionary history. Whereas traditional phylogenetic methods treat all morphological characters as if they evolve in the same way, we introduce a new analytical framework that explicitly distinguishes transformational characters, which document variation in form, from neomorphic characters, which record the presence or absence of discrete traits. Using 69 datasets that span major groups of animals and other organisms, we show that transformational characters typically change twice as fast as neomorphic characters; and that losses of neomorphic features occur more than twice as frequently as gains. Evolutionary models assuming long-term equilibrium are violated in the majority of datasets, suggesting ongoing directional trends. Accounting for these asymmetries improves model fit and can substantially alter inferred evolutionary relationships, sometimes producing posterior samples that occupy distinct regions of tree space. These results provide broad empirical evidence for directional biases in morphological evolution, supporting the view that developmental and ecological constraints favour trait loss over gain. By explicitly distinguishing different types of morphological change, this framework enhances model realism, recovers different phylogenies, and opens new avenues for interpreting macroevolutionary dynamics.