Can extensive reticulation and concerted evolution result in a cladistically structured molecular data set?

Hierarchy is the main criterion for informativeness in a data set, even if no explicit reference to evolution as a causal process is provided. Sequence data (nuclear ribosomal DNA ITS) from Armeria (Plumbaginaceae) contains a certain amount of hierarchical structure as suggested by data decisiveness and distribution of tree lengths. However, ancillary evidence suggests that extensive gene flow and biased concerted evolution in these multicopy regions have significantly shaped the ITS data set. This argument is discussed using parsimony analysis of four data sets, constructed by combining wild sequences with those from different generations of artificial hybrids (wild + F-1, F-2, and backcrosses; wild + backcrosses; wild + F-1; wild + F-2). Compared to the F-1 hybrids, F-2 show a certain degree of homogenization in polymorphic sites. This effect reduces topological disruption caused by F-1 and is considered to be illustrative of how extensive gene flow and biased concerted evolution may have modeled the wild ITS data. The possibility that hierarchy has arisen as a result of-or despite a significant contribution from-those two such potentially perturbing forces raises the question of what kind of signal are we recovering from this molecular data set. (C) 2001 The Willi Hennig Society.