POLEMONIACEAE: Gilia.
CORDIEAE: Cordia.
BORAGINEAE: Pulmonaria.
VERBENACEAE: Aegiphila.
POLYGONEAE: Polygonum.
THYMELEAE: Thymelea.
MONOCOTYLEDONS.
PONTEDERIACEAE: Pontederia.
In some of these families the heterostyled condition must have been acquired at a very remote period. Thus the three closely allied genera, Menyanthes, Limnanthemum, and Villarsia, inhabit respectively Europe, India, and South America. Heterostyled species of Hedyotis are found in the temperate regions of North and the tropical regions of South America. Trimorphic species of Oxalis live on both sides of the Cordillera in South America and at the Cape of Good Hope. In these and some other cases it is not probable that each species acquired its heterostyled structure independently of its close allies. If they did not do so, the three closely connected genera of the Menyantheae and the several trimorphic species of Oxalis must have inherited their structure from a common progenitor. But an immense lapse of time will have been necessary in all such cases for the modified descendants of a common progenitor to have spread from a single centre to such widely remote and separated areas. The family of the Rubiaceae contains not far short of as many heterostyled genera as all the other thirteen families together; and hereafter no doubt other Rubiaceous genera will be found to be heterostyled, although a large majority are homostyled. Several closely allied genera in this family probably owe their heterostyled structure to descent in common; but as the genera thus characterised are distributed in no less than eight of the tribes into which this family has been divided by Bentham and Hooker, it is almost certain that several of them must have become heterostyled independently of one another. What there is in the constitution or structure of the members of this family which favours their becoming heterostyled, I cannot conjecture. Some families of considerable size, such as the Boragineae and Verbenaceae, include, as far as is at present known, only a single heterostyled genus. Polygonum also is the sole heterostyled genus in its family; and though it is a very large genus, no other species except P. fagopyrum is thus characterised. We may suspect that it has become heterostyled within a comparatively recent period, as it seems to be less strongly so in function than the species in any other genus, for both forms are capable of yielding a considerable number of spontaneously self-fertilised seeds. Polygonum in possessing only a single heterostyled species is an extreme case; but every other genus of considerable size which includes some such species likewise contains homostyled species. Lythrum includes trimorphic, dimorphic, and homostyled species.
Trees, bushes, and herbaceous plants, both large and small, bearing single flowers or flowers in dense spikes or heads, have been rendered heterostyled. So have plants which inhabit alpine and lowland sites, dry land, marshes and water. (6/3. Out of the 38 genera known to include heterostyled species, about eight, or 21 per cent, are more or less aquatic in their habits. I was at first struck with this fact, for I was not then aware how large a proportion of ordinary plants inhabit such stations. Heterostyled plants may be said in one sense to have their sexes separated, as the forms must mutually fertilise one another. Therefore it seemed worth while to ascertain what proportion of the genera in the Linnean classes, Monoecia, Dioecia and Polygamia, contained species which live “in water, marshes, bogs or watery places.” In Sir W.J. Hooker’s ‘British Flora’ 4th edition 1838, these three Linnean classes include 40 genera, 17 of which (i.e. 43 per cent) contain species inhabiting the just-specified stations. So that 43 per cent of those British plants which have their sexes separated are more or less aquatic in their habits, whereas only 21 per cent of heterostyled plants have such habits. I may add that the hermaphrodite classes, from Monandria to Gynandria inclusive, contain 447 genera, of which 113 are aquatic in the above sense, or only 25 per cent. It thus appears, as far as can be judged from such imperfect data, that there is some connection between the separation of the sexes in plants and the watery nature of the sites which they inhabit; but that this does not hold good with heterostyled species.)