Taxonomic decisions are critically dependent on the reliability of the analysed characters. However, when character conflicts arise, such as discrepancies between shell morphology and molecular data, relying solely on prior assumptions about the systematic significance of characters and taxa can lead to erroneous conclusions. This is because inductive reasoning, unlike deduction, produces probabilistic rather than definitive conclusions. The uncertainty inherent in induction stems from the incompleteness of our observations; thus, using a partial or selective dataset increases the likelihood of error. To address this issue, we conducted a case study using the genus Cathaica. First, we reconstructed molecular phylogenies based on all available published datasets. Next, we performed discrete and continuous character optimisations to identify apomorphic characters. Finally, we explored the correlation between molecular distances and shell morphological distances. Our character optimisation revealed that shell characters are highly variable and often discordant with molecular data, whereas discrete genital characters exhibit stronger congruence with phylogenetic relationships. Notably, molecular distances were not significantly correlated with shell landmark Euclidean distances, underscoring the mismatch between genetic and shell-based evidence. Based on these findings, we propose that C. zhangcunxiangi, C. wangjiaxunae and C. sculptilis are synonyms of C. fasciola, and C. mengi is a synonym of C. pyrrhozona. This study emphasises that relying solely on partial datasets or untested characters, without considering the probabilistic nature of non-deductive inference, can lead to misleading taxonomic interpretations.
Zhang et al. (Tue,) studied this question.