Traditional characters and Procrustes-aligned landmark data: a sensitivity analysis in morphological data type...

Total-evidence phylogenetic analyses combine multiple data types, but the relative influence of different partitions can be difficult to assess. This is especially true in morphological studies that analyse traditional discrete characters together with geometric morphometric (GM) landmarks, because these partitions differ in structure, scale and completeness. Here, we test how varying the weight of a cranial landmark configuration relative to craniodental discrete characters affects tree length contribution and topology in parsimony analyses of extant primates and fossil hominins. We performed two iterative weighting series: one including all available GM and discrete data, and another excluding discrete characters for two fossil hominins so they were represented only by GM data. In both series, increasing GM weight produced stepwise, not continuous, topological change, with multiple iterations yielding identical most parsimonious trees before new topologies emerged. GM contribution to total tree length increased nonlinearly, suggesting diminishing returns. Topological shifts were concentrated among fossil hominins and were most pronounced when some taxa were represented only by heavily weighted landmark data. These results show that GM data can meaningfully affect parsimony-based phylogenetic inference, but their impact depends strongly on weighting and missing data distribution.