| Abstract: | Varied approaches to estimating confidence intervals for immunological and hybridization distances can be uniformly applied to any matrix of distances. One procedure bootstraps the pairwise dissimilarities between the distances of every pair of taxa to all others, creating a derived matrix of distances for which dispersions can be estimated. Another approach bootstraps the sample of differences between pairwise homologous branch lengths concerning each pair of taxa and between asymmetric halves of the matrix, to find a standard error of the dispersions. This allows comparison of the robustness of trees among different sources of data. DNA hybridization, transferrin immunology and protein immunodiffusion matrices all yield much the same result once standard deviations of dissimilarities are acknowledged: namely, unresolvable trichotomies among the human-chimp-gorilla clade and among this clade with orang and gibbon; conventional relationships among hominoids, cercopithecoids, ceboids and strepsirhines; and a polychotomy among anthropoids, strepsirhines, tarsiers, tupaiids and dermopterans. |
