The available evidence is used to produce a revised taxonomy for Onagraceae tribe Epilobieae, with two genera, Epilobium and Boisduvalia. Zauschneria is a specialized, bird-pollinated taxon closely related to one group of Epilobium, and it is consequently reduced to the status of a section of that genus. It is treated as comprising a single species with three tetraploid (n = 15) and three octoploid (n = 30) subspecies, between which extensive intergradation occurs. Chamerion (Chamaenerion), a Eurasian group of seven species, two reaching North America, is closely related to and doubtless shares a common ancestor with Epilobium sect. Epilobium; the other four sections are more distantly related. They include a total of seven species, all xerophytes confined to western North America, the probable place of origin of the tribe. Boisduvalia is treated as comprising two sections in place of the three recognized earlier, based upon further morphological studies and evidence obtained from artificial hybridization. The taxonomy proposed for the tribe Epilobieae appears to provide a balanced classification for the group that brings it into line with the other five tribes and 15 genera of the family. The tribe Epilobieae, with some 210 species, differs from the other five tribes in the family Onagraceae (Raven, 1964b) in its dotlike, almost entirely heteropycnotic chromosomes that persist as chromatic dots through interphase (Kurabayashi et al., 1962); basic chromosome numbers x = 9, 10, 12, 13, 15, 16, and 18; occurrence of many species in moist places; and, in all but a few species, the habit of shedding the mature pollen as tetrads. It may be related to the less specialized group Jussiaeeae, consisting of the genus Ludwigia (including Jussiaea and Isnardia), which resembles it in having dotlike, heteropycnotic chromosomes; in growing in moist places; and, in many species, in shedding its mature pollen in tetrads. The basic chromosome number of Ludwigia, however, is x 8; its tetrads are made up of pollen grains that differ greatly from those of Epilobieae and have probably evolved independently (Skvarla et al., 1975); and it lacks interxylary phloem, present in Epilobieae (Carlquist, 1975). Ludwigia seems to be a relatively generalized genus within the family, resembling Fuchsia in its retention of primitive characteristics, as inferred by Eyde & Morgan (1973), and it probably is not directly related to Epilobieae, which then appear as a rather isolated group within the family. Although the fossil record is badly in need of reevaluation, Epilobieae probably extend back to Paleogene time (see discussion in Eyde & Morgan, 1973: 785). Within this tribe, the genera are closely related to one another. This paper is directed to the following question: how many genera and sections, in the sense of Lewis & Lewis (1955) and Raven (e.g., 1963, 1969), is it useful to recognize? Boisduvalia (Raven & Moore, 1965) includes six species of annual plants of western North and South America. Collectively, they are distinguished from 1 I am grateful to the U. S. National Science Foundation for a series of grants in support of my study of Onagraceae, to Steven R. Seavey for useful comments on this paper, and to G. Perraudin for advice on the progress of his hybridization experiments. 2 Missouri Botanical Garden, 2345 Tower Grove Avenue, St. Louis, Missouri 63110. ANN. MissouRI BOT. GARD. 63: 326-340. 1976. This content downloaded from 157.55.39.104 on Mon, 20 Jun 2016 05:38:00 UTC All use subject to http://about.jstor.org/terms 1976] RAVEN--EPILOBIEAE 327 Epilobium by their annual habit and lack of a coma, the tuft of long trichomes at the chalazal end of the seed that aids in dispersal. No species of Epilobium is both annual and lacks a coma, and no species is transitional between these two genera. In aspect, Boisduvalia is sharply distinct from Epilobium. Plants of Boisduvalia germinate in moist conditions, often actually submerged, and initially produce large, glabrous, opposite leaves similar to those of Epilobium. Later, they begin to produce alternate, sometimes densely pubescent, hard leaves, and when they produce flowers and fruits, they are often growing under decidedly xeric conditions. Boisduvalia includes two species with a gametic chromosome number of n = 10, one with n = 9, one with n = 19, and two with n = 15. The last is the only one of these chromosome numbers that also occurs in Epilobium; but, as we shall see, the species of these two genera with n= 15 seem to be unrelated. As to the remainder of the tribe, the bird-pollinated, orange-red-flowered Zauschneria, comprising a single polytypic species of western North America, has been recognized as distinct since it was described by Presl in 1831. As early as 1806, Salisbury (1806: pl. 58), in describing Chamaenerium halimifolium, used the existence of Zauschneria, known to him from plants gathered by Archibald Menzies along the coast of northern California, as an argument for recognizing Chamaenerion as a genus distinct from Epilobium. Zauschneria was obviously generically distinct, yet had the coma of Epilobium: why not Chamaenerion also? In fact, Zauschneria is more closely related to Epilobium sect. Cordylophorum, which I shall discuss below, than to Epilobium s. str. All species of both groups have a gametic chromosome number of x = 15, unique in Epilobium (Lewis & Raven, 1961); most have a relatively long floral tube; and all grow in xeric sites and are somewhat woody at the base. Their wood anatomy (Carlquist, 1975) is virtually identical. The single species of Zauschneria (see p. 335) shares with Epilobium nevadense and E. nivium a characteristic unusual for the tribe: a prominent apiculus of brown oil cells at the tip of each leaf (Brandegee, 1892; Stein, 1915; Munz, 1929; Raven, 1962a). Their seeds, like those of Epilobium sect. Xerolobium (comprising one highly polymorphic xerophytic annual species), are large, obovoid to clavate, and prominently constricted at the micropylar end (Seavey, Magill & Raven, 1977). Furthermore, Zauschneria, like Boisduvalia and the two sections of Epilobium just mentioned, as well as a third, Epilobium sect. Crossostigma, has large endexine channels in the distal pollen walls, which are therefore different from all other pollen walls in the family Onagraceae and a clear indication of relationship (Skvarla et al., 1976). The viscin threads in their pollen are thick and fluted (incised compound), unlike the less sharply ridged, tightly compound ones of sects. Chamaenerion and Epilobium (Skvarla et al., 1977). Zauschneria differs from the three species of Epilobium sect. Cordylophorum only in its longer floral tubes, orange-red flowers, and possession of a scale at the base of each stamen within the narrow part of the floral tube. These scales are highly variable and sometimes reduced to an irregular line (Curran, 1888). They appear to be homologous with the ring inside the floral tube in Boisduvalia (Curran, 1888: 255; Raven & Moore, 1965: 239, figs. This content downloaded from 157.55.39.104 on Mon, 20 Jun 2016 05:38:00 UTC All use subject to http://about.jstor.org/terms 328 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VOL. 63 2-4) and in many species of Epilobium, such as E. obcordlatum A. Gray. In the hummingbird-pollinated flowers of Zauschneria, this ring has apparently been somewhat elaborated during the course of evolution in relation to the protection of the abundant nectar from potential nectar thieves of low energetic requirements (Heinrich & Raven, 1972). Zauschneria is pollinated by hummingbirds and separated by the syndrome of characteristics typical of bird pollination from its bee-pollinated relatives, but overwhelmingly similar to them in all other re-