DEVELOPMENTAL MORPHOLOGY OF FEMALE FLOWERS OF GYROSTEMON AND TERSONIAAND FLORAL EVOLUTION AMONG GYROSTEMONACEAE

Floral simplifications and specializations in the evolution of Gyroste monaceae have confused the systematics of the family. Recent phylogene tic analyses have demonstrated their placement among Capparales. This investigation presents a phylogenetic analysis of Gyrostemonaceae, dem onstrating that Codonocarpus and Gyrostemon form a clade that is the s ister group of Cypselocarpus, Tersonia, and Walteranthus. These phylog enetic results and data on development of Gyrostemon and Tersonia are used to discuss the morphology and evolutionary diversification of fem ale flowers of Gyrostemonaceae. The uniseriate perianth of Gyrostemona ceae consists of four to eight tepals with an unusual lateral to media n developmental sequence. The female flowers of Gyrostemon and Tersoni a display no distinctive evidence of an androecium, although the forme r has late-forming, primordium-like structures positioned between the tepals and gynoecium that may be the vestiges of either a second peria nth series or the androecium. The gynoecium of Gyrostemonaceae is sync arpous, although the two main clades in the family differ in the expre ssion of ovarian synorganization. The Codonocarpus-Gyrostemon clade is unusual in having largely separate carpels that are only syncarpous b ecause the ventral side of each is formed by the flank of the floral a pex. All Gyrostemonaceae, however, incorporate the flank of the floral apex as the ventral side of the carpel, and this is the location of o vule development. On the basis of its placement in a clade that includ es Tersonia and Walter-anthus, the uniloculate and uniovulate gynoeciu m of Cypselocarpus may be pseudomonomerous. All Gyrostemonaceae have l arge stigmas that are typical of anemophilous taxa, and they differ fr om most other Capparales in this attribute. Among Capparales, Gyrostem onaceae may be most similar to Ochradenus (Resedaceae), which also app ears to be anemophilous. It is unclear whether the similarities of Och radenus and Gyrostemonaceae are homologies, indicative of a close rela tionship between the two groups, or evolutionary parallelisms associat ed with separate shifts to anemophily.