Our analysis yielded 106 most parsimonious trees of 408 steps, but these are similar in overall structure. A strict consensus is shown in Fig. 1. The variations involve combinations of different outgroup arrangements and different arrangements within three major annonaceous groups : the ambavioid group, the annonoid group, and the clade including the uvarioid and xylopioid groups. Rather different arrange ments of major groups and certain in dividual taxa are seen among one step less parsimonious trees, although many features are similar. The consistency index is 0.27, which is low for this num ber of taxa (Sanderson and Donoghue,
1989). This means that the level of homoplasy is high, so the details should not be taken too seriously. Many characters appear to be systematically worthless at this level of analysis, in the
CONCLUSIONS
Our conclusions are summarized in Fig. 10, with sketches of major pollen types superimposed on the character of infratectal structure.
This confirms the view that pollen is one of the most valuable sources of evidence on relationships in the Annonaceae, although it is subject to homoplasy like any other character set. Although this homoplasy is usually recognized on the basis of nonpalynological characters, sometimes, as in the case of Fusaea, it can be suspected from conflicts among the pollen characters themselves. Our results confirm that the basic pollen type in the family was boatshaped and monosulcate, with granular exine structure, although shape may have undergone reversals within the family. Whether the conclusion that granular structure is primitive can be extended to Angiosperms as a whole depends on whether the An giosperms are rooted near Magrwliales or among paleohcrbs. Monosulcates gave rise to inapcrturate pollen in two lines, possibly one. One of these developed columellae and the tetrad habit, with an interesting reversal to single grains with a reduced proximal exine in lsolona. The other, which remained granular, diversified into various ina perturates, including two independent echinate lines, and a second major tetrad group. Sulculates originated four times, three times from different monosulcates, once from inaperturates.
Because of the high level of homoplasy in the family, many of these re sults are unstable and need testing with additional data. Improvements could come from vegetative anatomy and morphology or cytology of the many genera where data on these characters are missing, from analyses of molecular sequences, or from better evidence on relationships and basic states within genera.