evolution: a phylogenetic test with allozymes in Sceloporus (Reptilia).
Weighted squared-change parsimony was used with the gradual simulations and unweighted squared-change parsimony was used with speciational simulations [ILLUSTRATION FOR FIGURE 2 OMITTED].
The choice of boundary options influences the distribution of the simulated changes in both speciational and gradual evolution simulations ([ILLUSTRATION FOR FIGURE 2 OMITTED], step 1).
Figure 6a compares the distributions of changes produced in the speciational Unbounded simulations (indicated by a line) with the distribution of changes reconstructed by linear or squared-change parsimony (indicated by bars).
If branch lengths are obtained by assuming a model of phenotypic evolution that makes them directly proportional to some aspect of the phylogeny (e.g., under gradual or speciational
models of change), p(BL/P) is implicitly assumed to be 1 and [Var.sub.BL] is assumed to equal 0.
For the speciational and punctuational models, the amount of character change was sampled from a Poisson distribution with its parameter adjusted to match the total expected amount of change in the phyletic model.
These authors distinguished among: (1) the phyletic context, where character state changes may occur anywhere along a given tree; (2) the speciational context, where character state changes are assumed to occur only during a speciation event and in both daughter lineages; and (3) the punctuational context, where character state changes are assumed to occur only in one daughter species at the time of a speciation event.
NJ is more accurate than the other two under the punctuated and speciational context models.