DATA FOR A STRATAPHY STRATOCLADISTIC SEARCH

Stratocladistic analysis requires additional data to supplement what is traditionally collected for cladistic analysis. Specifically, stratigraphic data and autapomorphies are necessary components of a stratocladistic data set.

As previously mentioned, one of the chief differences between stratocladistics and cladistics is the use of stratigraphic data in the process of phylogenetic inference in stratocladistics. These data are in the form of the temporal or stratigraphic intervals of first and last appearance of taxa. They are coded as discrete states in the character matrix, rather than as continuous values (e.g., absolute dates or meters of section). Taxa that span multiple intervals are coded as being polymorphic and possess all states corresponding to intervals in which they are sampled. Exactly how stratigraphic intervals should be defined for stratocladistic analysis, and how finely they should be divided remains an unresolved matter, and one for further research (see Smith 2000; Fisher et al. 2002).

Autapomorphies are often explicitly excluded from cladistic analysis, as they offer no information regarding sister-clade relationships among taxa. However, they are required for stratocladistic analysis because they potentially provide information about the optimality of hypotheses of ancestor-descendant relationships among taxa. Operationally, they affect the optimality of assignments of taxa as ancestors, in which consideration of a single taxon as an ancestor weighs the possible stratigraphic debt savings against the possible morphologic debt incurred. This morphologic debt will equal the number of autapomorphies, if character changes have a weight of one, so the exclusion of autapomorphies will bias the results toward finding more optimally assigned ancestors than would be the case if autapomorphies were included. Aside from this obvious issue of erroneously assigning ancestors, failure to include autapomorphies can be a computational challenge, as it can result in a large number of optimal trees when multiple taxa can be either ancestors or terminal taxa on a tree without a change in total debt. In such a case, for a given topology, numerous permutations of ancestral assignments of such taxa will be equally optimal. Searching vast numbers of equally optimal trees dramatically increases search times. Therefore, it is strongly suggested that users resist the temptation to append a single stratigraphic character to existing cladistic matrices for analysis in StrataPhy without first adding autapomorphies of included taxa.