HYBRIDS AND PHYLOGENETIC SYSTEMATICS II. THE IMPACT OF HYBRIDS ON CLADISTIC ANALYSIS.

I examined three aspects of the cladistic treatment of a set of 17 F hybrids of known parental origin: (1) impact of hybrids on consistency index (CI) and number of most parsimonious trees (Trees), (2) placement of hybrids in cladograms, and (3) impact of hybrids on hypotheses of relationship among species. The hybrids were added singly and in randomly selected sets of two to five to a data set composed of Central American species of Aphelandra (including the parents of all hybrids). Compared to analyses with the same number of OTUs all of which were species, the analyses with hybrids yielded results with significantly higher CI. There was no difference in Trees between analyses with hybrids versus species. There was thus no evidence that hybrids would appear to be more problematic for cladistic methods than species. Accordingly, hybrids will not be readily identifiable as taxa that cause marked change in these indices. About % of the hybrids were placed as the cladistically basal members of the lineage that included the most apomorphic parent. Relatively apomorphic hybrids were placed proximate to the most derived parent (ca. 13% of hybrids). Other placements occurred more rarely. The most frequent placements of hybrids thus did not distinguish them from normal intermediate or apomorphic taxa. When analyses with hybrids yielded multiple most parsimonious trees, these were no more different from each other than were the equally parsimonious trees that resulted from analyses with species. Most analyses with one or two hybrids resulted in minor or no change in topology. When hybrids caused topological change, they frequently caused rearrangements of weakly supported portions of the cladogram that did not include their parents. When they disrupted the cladistic placement of their parents, they often caused their parents to change positions, with at least one topology bringing the parental lineages into closer proximity with the hybrid placed between them. Hybrids between parents from the two main lineages of the group caused total cladistic restructuring. In fact, the degree of relationship between a hybrid's parents (measured by both cladistic and patristic distance) was strongly correlated with CI (negatively) and with the degree of disturbance to cladistic relationships (positively). Thus, hybrids between distantly related parents resulted in cladograms with low CI and major topological changes. This study suggests that hybrids are unlikely to cause breakdown of cladistic structure unless they are between distantly related parents. However, these results also indicate that cladistics may not be specially useful in distinguishing hybrids from normal taxa. The applicability of these results to other kinds of hybrids is examined and the likely cladistic treatment of hybrids using other sources of data is discussed.