Dichondra

Abstract: Willdenow (1806) and Kunth (in Humboldt, Bonpland, and Kunth, 1819) assumed that in Dichondra argentea H. & B. ex Willd., of which they had the original material collected by Bonpland, the fruit consisted of geminate utricles as in the classical species D. repens Forst. & Forst., D. sericea Sw., and D. carolinensis Michx. In fact, the fruit of D. argentea is entire and capsular. Authors of systems for the Convolvulaceae (Endlicher 1839; Choisy 1845; Bentham and Hooker 1876; Hallier 1893; Peter 1897; Roberty 1952) did not correct the error. On the basis of fruit-structure D. argentea cannot belong to the genus as they characterize it. Three other species with fruits essentially similar to those of D. argentea have been proposed, viz. Dichondra occidentalis (House 1906), Dichondropsis n~vea (Brandegee 1909), and Dichondra brachypoda (Wooton and Standley 1913). House did not point out the peculiarities of his species as compared to the original Dichondras, nor did Wooton and Standley for their species. Neither Van Ooststroom (1938) nor O'Donell (1957, 1959) amplified the generic description to permit inclusion of D. a rgentea and allies, nor did either exclude them from the genus. Brandegee contrasted the characters of his monotypic genus with what he supposed from descriptions were the characters of D. argentea; later (according to manuscript notes in the Gray Herbarium) he realized his error, but did not attempt to resolve the taxonomic problems. In all species under consideration, the ovary, before fertilization, is less than I mm high, subglobose, with two biovulate locules. The two locules are separated by a minute, hyaline septum at the top of which are attached the two entire styles, which in some species (or specimens) cohere toward their bases. Continued development after fertilization is not the same in all species. In D. argentea and allies the entire structure (except the styles, which eventually fall) increases in size more or less uniformly, often accompanied by maturation of both ovules in one or both locules; the wall takes on a slightly horny or cartilaginous texture; upon ripening the capsule dehisces, usually first loculicidally and then septicidally, and the four parts of the ovary wall and the several seeds fall separately, leaving the persistent septum. In D. repens and allies, the central part of the ovary including the septum remains minute, and the tyles persist; the lateral parts of the ovary increase greatly, usually accompanied by the maturation of one ovule and the abortion of the other in each locule ; the walls remain thin and membranous; upon ripening the two parts of the fruit (th \"utr ic les\") separate from the minute septum and from the minute receptaculal attachment, and each falls with its enclosed seed, leaving the septum and often the persistent styles. (The exceptions are the parts of the fruits of D. sericea, D. evolvulacea (L.f.) Britt., D. macrocalyx Meissn., and D. ~)arvifola Meissn., each of which develops independent loculicidal dehiscence.) The styles can be, and often have been, described as gynobasic, but it appears that this is a very simple ontogenetie modification of the terminal position. At the very least the above data militate against acceptance of a monotypic family Dichondraceae, as proloosed by Dumortier, Anal. Fam. P1. pp. 20, 24. 1829 (citation from Bullock 1958, 1959). They may even indicate that the genus should not constitute a subfamily or tribe of the Convolvulaceae (of. Hallier