cence, always white or silvery when fresh, turns rusty or yellowish when dry. The hairs of Marina are relatively stiff, always short, never spiral, and unchanged by drying. The surface of the Dalea leaflet is uniformly green (beneath the vesture, if present), or may be minutely pallid-dotted, but the dots never run together into linear patterns. In Marina both faces of the leaflet are marked with fine sinuous lineoles, pallid against a green field, which ascend at a more or less acute angle from near the midrib toward the margin. The calyx- ribs of almost all genuine Daleae anastomose in the teeth and are produced beyond the triangular or deltate base of the tooth into a distinct point, often prolonged into a subulate or setiform tip. In Marina the ribs take the form of alternate I and Y, but the branches of the Ys peter out into the plane herbaceous blade of the calyx-tooth and fail to anastomose with the adjacent Is to form the gallery of closed Gothic arches characteristic of Dalea. The two genera are alike in having one gland situated behind each petiolule along the leaf- rachis (these exceptionally impressed or obsolete in both); but while in Dalea there are (nearly always) a pair of glands or spicules situated between the petiolules ventrally, in Marina there is but one. Reduction of the glands in some species of each genus render this character negative in a few individual cases, but when present it is decisive. The array of differential characters found in foliage, flower, and caryotype enable one to recognize a Marina at any stage of growth.
The flower of Marina is always pedicelled, but this is not a diagnostic character, for several genuine Daleae, while unlike Marina in other characters, have racemose and not the more usual spicate inflorescence. Pedicelled flowers are shared by the genera Marina and Psorothamnus, which are further alike in their plane, herbaceous calyx-teeth and, so far as known, in the basic chromosome number of ten. The inner petals of Psorothamnus, however, are inserted on the hypanthium rim and the ovary contains at least two collateral ovules, sometimes an additional one or two pairs. The leaflets of Psorothamnus lack the lineoles found in Marina, and there are no intrapetiolular glands or spicules. In both these genera the blades of the keel overlap in vernation and adhere by their faces. In almost all groups of Dalea where the keel-petals are united at anthesis, the union of the blades is by their exterior edges (valvate not imbricate). The relationship of Marina to Psorothamnus is obviously a close one, perhaps closer than to Dalea sens, restr.
Two main divisions of Marina are defined below in the rank of section. The much greater, sect. Marina, which has two primary centers of speciation, one around the Gulf of California, the other in the Balsas Depression in southern Mexico, is characterized by a compressed pod, obliquely obovate, half-circular, or harp-shaped in profile, with terminal but often strongly eccentric style-base and a cavity substantially larger than the seed. The glands on the faces of the pod are here arranged, except in the anomalous M. gemmea, in transverse or obliquely descending crescentic patterns. The section consists of three main series and three ancillary monotypic ones: the large ser. Chrysorrhizae, concentrated around the Gulf of California, with mostly short, firm pedicels, a pubescent calyx, and gland-sprinkled banner; the smaller ser. Unifoliolatae, ranging through southern Mexico to Guatemala, and differing in capillary pedicels and glandless banner, generally associated with a reduced number of leaflets; the ser. Crenulatae, centering in Jalisco, with many leaflets, capillary pedicels, and gland-sprinkled banner, but totally glabrous, thus essentially like sect. Carroa except for the compressed pod; and ser. Gemmeae, Marina, and Vetulae, each of one highly specialized species, the two first probably derived from ser. Unifoliolatae, the third from ser. Chrysorrhizae. The sect. Carroa, dispersed over the whole range of the genus except that it is absent from Baja California, has a nutlike pod much less strongly compressed than that of sect. Marina, commonly subglobose or pyriform, with lateral style-base and valves closely investing and finally adherent to the seed. The pod’s valves are charged with blister-glands arranged in no obvious pattern and are as a rule uniformly pustulate over the whole surface.
The ser. Chrysorrhizae, ser. Crenulatae, and sect. Carroa each present a series of life- forms ranging from shrub to herb, and count among their herbaceous members both perennial and monocarpic species. The remaining series within sect. Marina consist of herbs only, among which obligate annuals preponderate. The species of short duration appear always more highly specialized in the context of their group, and it is clear that the monocarpic condition is the terminus of several parallel evolutionary paths.
Since its description in 1853 Marina has been maintained distinct from Dalea (or Parosela) only by Hutchinson (1964, p. 419) who, however, misinterpreted the flower and referred the genus to his tribe Psoraleeae as though it had petals inserted on the hypanthium rim and not elevated on the staminal column. The close affinity of the generitype, M. gracilis, independently described as Parosela delicata Rose, to several annual, uniovulate paroselas found in southern Mexico was perceived by Rydberg (1919, p. 51) who treated it as forming by itself a group Delicatae, distinguished from Chrysorrhizae by its sharply dentate calyx-teeth. While certainly a strongly marked species, M. gracilis is unquestionably congeneric with M. holwayi and some others each in its own way uniquely modified.
An earlier name for Marina as defined in these pages is Trichopodium C. K. Presl (1844), unfortunately a later homonym. The generitype of Trichopodium is one of the totally glabrous species with capillary pedicels and nutlike pods that formed part of Parosela sect. Diffusae Rydb. Rydberg (1919, p. 49) treated Trichopodium as a subgenus of Parosela which included Marina and was supposed to differ from genuine Parosela in having flowers at once pedicelled and reflexed at anthesis. As already noted, this is not a practical diagnostic character. However, Trichopodium sensu Rydberg, if shorn of its sects. Lasiostomae and Lachnostachyae, which have pedicellate flowers but all other characters of Dalea, is equivalent of Marina sensu latiori of this account. Because the name Trichopodium was applied to an unnatural concept by Rydberg and thus became ambiguous in its application I have taken up, for the small sectional group of Marina which includes the original T. glandulosum Presl, the name Carroa Presl (1858, pro gen., substituted for Trichopodium Presl non Lindl.).
Division of the genus into two sections characterized by a fruiting character, while theoretically satisfactory, raises a difficulty in practice. The majority of perennial and shrubby marinas flower in fall and winter, and their pods ripen slowly, often not until all flowers are past anthesis. At this stage of growth the plants are inconspicuous and rarely collected. The identity of the average specimen, which is likely to bear flowers but at best quite unripe pods, could not be determined by means of a key based on the fundamental sectional characters. As an alternative to an ideal systematic key, I present a synopsis of the genus which stresses fundamental characters but ignores uncommon exceptions. Following this will be found a pragmatic key to the species based primarily on more easily observed, specific rather than sectional characters.