Hypotheses about constraints, selection, and other evolutionary processes often predict different rates of change among different characters. For example, it is thought that gross soft anatomical characters change less frequently than do general shell characters. However, some shell characters on primitive gastropods, such as the sinus and selenizone, are thought to be linked to soft anatomy. If both premises are true, then selenizone/sinus characters should change less frequently than other characters. Workers also have documented active trends among gastropod shell characters. One explanation is a driven trend, where the rate of change in one direction (i.e., gain or loss, increase or decrease) is greater than the rate of change in the opposite direction.
Hypotheses about relative rates and biased change are tested here for lophospiroid gastropods. Likelihood analyses test whether hypotheses positing many different rates predict data significantly better than do hypotheses positing few rates. One tree-based approach assumes that a phylogeny is known and thus treats a tree as a model. A second tree-based approach treats phylogeny as an additional hypothesis. This multiparameter approach allows competing rate hypotheses to “assume” a phylogeny that maximizes their likelihoods.
Likelihood tests reject hypotheses of low rate heterogeneity among lophospiroid characters. The most likely hypothesis (which treats phylogeny as an unknown) posits seven rates, with biased changes among three characters (preferential reduction of both sinus depth and sinus width and preferential addition of ornament). The distribution of rates among different character classes is consistent with the prediction that characters associated with internal anatomy should show generally lower rates of change than those associated with gross shell morphology. Evaluating significance when contrasting hypotheses posit different phylogenies is problematic, but the difference in support (log-likelihood) is overwhelming. Even if a model phylogeny is used, we still reject all hypotheses using fewer than six rates and not invoking trends in sinus characters. Thus, it is difficult to avoid rejecting the hypothesis that shell characters are uniformly plastic.