The Subgenera of (Compositae): Hawaiian Studies 18 1

HAROLD ST. JOHN2

THE SUBGENERA OF Dubautia only 3 to 4 feet tall, has foliage similar to THE SHRUBBY or arborescent Compositae of that of D. plantaginea. Thus, if all the known the Hawaiian Islands have attracted much species are examined, it is evident that there attention and study. Amon g them are the can be no consistent groupings of the species related genera Dubautia and Railliardia, early into several genera on the basis of stature, described by Gaudichaud and maintained by foliage, or involucre. nearly every investigator to the present. They In a review of Dubautia and Railliardia, 1"ere separated by characters of the involucre, Keck ( 1936: 24':"'25 ) emphasized the-exist­ paleae, and pappus. Many of the botanists ence of species invalidating each one of the who have done field work in Hawaii got their several characters alleged to separate the 'two first concept of Duba1ttia from D. plantagic genera. The two genera had recently been nea, a vigorous small tree of the mountains monographed, but Keck asserted (p. 25) that of Oahu. Its involucre of4 to 6 firm distinct the monographer "Sherff gives an admirable bracts and its terminal plume of crowded systematic account of the species, but evades oblong-oblanceolare leaves, glabrate on the the question of 'how the two genera are to surfaces, give it a very marked habit and as­ . be distinguished by failing to raise it." pect..In Railliardia there are numerous spe­ To one acquainted with the in the cies-small bushes with sessile and coriace­ field there is no habital aspect to separatethe ous, often decussate, leaves which clothe the species into two groups and, as indicated, stem for a considerable distance. The involu­ there are no strong morphological differ­ cre is cylindric or campanulate, of several in­ ences, not even a single constant character. volucral bracts united into a rube.: Once Keck 's merging of the two genera has not formed, it is difficult to break away from this been generally adopted, inasmuch as the fol­ concept of the two old genera , but R. arborea lowing botanists have continued in the main­ Gray and R. strutbioloides Gray are trees; tenance of both Dubautia and Railliardia: and R. lonchophylla Sherff, though a Sherff (1941: 29-30); Degener (1940 and 1946 ); and Selling (1947: 330-332), who l This is the eighteenth of a series of papers de­ gave the first detailed account of the pollen signed to present descriptions, revisions, and rec­ ords of Hawaiian plants. The preceding papers "of the Hawaiian plants. Selling accepted both have been .published in Bernice P. Bishop Mus., Dubautia and Raillardia {=Railliardia}, and, Occas. Papers 10 ( 4 ) , 1933; 10 (1 2 ), 1934; 11 referring (p. 33 1) to Keck 's reduction, said (14) , 1935; 12 (8 ) ,1936; 14 ( 8 ),1938; 15(1 ), 1939; 15 (2), 1939; 15 ( 22 ), 1940 ; 15 ( 28 ), "there seems to be no proper reason for this." 1940 ; 17(12 ),1943; Calif. A cad. Sci.iProc. IV, Selling describes and illustrates pollen of the 25(16),1946; To rrey Bot. Club,Bul. 72: 22-30, two genera, describing it in almost identical 1945 ; Lloydia 7: 265-274, 1944; Pacific Sci. 1: 5-20, 1947; Brittonia 6 (4 ): 43 1-449 , 1949; phrasing. He states that one cannot separate Gray Herb., Contrib. 165 : 39-42, pl. 3, 1947; Railliardia from Dubautia on the basis of Pacific Sci. 3(4) : 296-301,1949. structure of the pollen grains. This might "Chairman, Department of Botany, University sound like a significant.fact, but also insep­ of Hawaii. Manuscr ipt received November 22, 1949. arable on the basis of pollen structure, are the

[339 ] 340 PACIFIC SCIENCE, Vol. IV, October, 1950 genera Lagenophora, T etram olopium, Argy­ Twenty species have the involucral bracts roxiphium, and other genera with the "Du­ united into a campanulate or funnel-form in­ bautia type pollen." So, at least, the pollen volucre, and the pappus awns long plumose, structure provides no significant differences these representing Railliardia of Gaudichaud. between Dt/b autia and RaiUiardia. Selling There remain three species, all from Kauai, twice accompanied Skottsberg on field trips which have the involucre campanulate, united to the Hawaiian Islands and Selling's rejec­ Jar from ~ to % of its length, and the pap· tion of Keck's work seems in harmony with 'pus awns short ciliate. These three perfectly that of his elder companion, Skottsberg. recombine the characters of the two older Skottsberg 0944: 510) discounted Keck's groups. As they cannot be logically fitted into revision. He discussed Keck's evaluation of either of these groups, a new subgenus, Mixta, the characters and consequent fusing of the is made to receive them; the other two groups two genera under D ubautia, and said, "I can­ are described as subgenera. not find, however, that he has brought to Keck removes the Hawaiian genera Du­ light any new circumstances, not known to batttia and Railliardia from the and, Hillebrand, Sherff, Degener and other writers quoting Bentham's opinion, refers to their be­ on this subject, and they have considered it ing likened to Robinsonia and Rhetin oden­ better to keep them separate. Nothing is dron of Juan Fernandez. The similarity is in gained by uniting them." Then Skottsberg de­ their habit of growth as small rosette trees, scribed a new species of RaiUardia {= R ail­ but in characters of ' and fruit there liardia}. To this it may be replied that the are fundamental differences from these ge­ gain attained by Keck is the delimiting of a nera of the . Ked: inclines towards genus with diagnostic characters that can be the views of Skottsberg that close relatives stated in words by one botanist and applied are Bedfordia of Australia and Bracbiono­ with the same understanding by others. The stylum of New Guinea. These are both of reduction of Railliardia to Dubautia is here the Senecioneae, and, on analysis of their accepted. The only recent authors accepting characters, are markedly different. Bedfordia the view of Keck seem to be Hartt and Neal is a genus ~f two species of Tasmania and (1940: 264), Fosberg (1943: 395-397 and Victoria in .Australia- or trees with . 1948: 115), and Neal 0948: 743) .Fos­ stellate tornentose leaves, heads axillary or berg, in his "Summary of the Hawaiian in axillary panicles, and the denticulate capil­ Plants ," places Railliardia as a synonym of lary pappi very numerous and caducous. Dttbat/tia and indicates that it is a descendant Bracbionostylurn is a monotypic genus from of the original immigrant . the high mountains of southwest New Guinea Of the several characters previously used -a shrub with the heads unisexual, hetero­ for generic separation, the best and most sig­ gamous, anthers free, and the pappus bristles nificant are those of the freedom or union of slender, capillary, roughened, and early cadu­ the involucra1 bracts and the degree of cilia­ cous. On comparison, neither of these genera tion of the pappus. Both of these characters has the habit, or similarity of and fruit are significant and have always been used in structure, to suggest that it is a close relative the classification of these species into larger of D ubazai«. groups. However, the two pairs of characters Keck agrees with the Engler and Pranrl are not always correlated. Six species have system in placing the Hawaiian Dubeati« and the involucre of separate bracts, and the pap­ its relatives in the -Galinsoginae; pus awns short ciliate lacerate or short ciliate, that seems the best placement at present, these representing Dttb atttia of Gaudichaud. though by -its structure and habit it is .not a Subgenera of Dubautia-ST. JOHN 341 close relative of the other genera currently D. magnifolia Sherff, Amer. Jour. Bot. 20: assigned to that subtribe, and those species 6 16, 1933. with a smooth receptacle wholly lacking chaff D. microcephala Skottsb., Hort, Goroburg., are least satisfactorily placed there. Hence, Acta 2: 277-280, fig. 8, 1926. the fact that all the other genera placed there D. p1antaginea Gaud., Voy. Freycinet Uranie, are American does not necessarily imply that Bot. 469, (1826) = 1830; Atlas pI. 84, Dubauiia is of American parentage. The 1826. Galinsoginae are a group transitional to the Helenieae. The Senecioneae also has to be Subgenus Railliardiaster subgen. nov. considered, and it is worthy of note that in Railliardia Gaud., Voy. Freycinet Uranie, the Engler and Prantl key to the Cornposirae Bot. 469, (1826) = 1830; Atlas pI. 83, the old genera Dubautia and Railliardia are 1826. included in the keys of all three of these Type species: R. linearis Gaud., 1. c. tribes. Investigations in search of a closely Bracteae involucri cohaerentae, involucrum related genus have been made in these three tubulosum vel angusre carnpanularum infun­ tribes, but the writer must report that he dibuliformeve est. Aristae pappi longe plu­ failed to find one. mosae. Bracts of the involucre united into an in­ Key to Subgenera of Dubautia volucre tubular or narrowly carnpanulate or Involucral bracts distinct; pappus bristles funn elform. Pappus bristles usually long plu­ short ciliate or lacerate-ciliate . Eudubautia mose. Involucral bracts united for part of their length, Enumeration of Species Pappus bristles long plumose. . ... (Gray) Keck, Bernice P. ~ ...... Railliardiaster Bishop Mus., Occas. Papers 11 (19) : 28, Pappus bristles short ciliate . . . Mixta 1936. Raillardia arborea Gray, Amer. Acad. Arts Genus DUBA UTIA Gaud. and Sci., Proc. 5: 134, 1861. Subgenus Eudubautia subgen. nov. D. ciliolata (DC.) Keck, Bernice P. Bishop Mus., Occas. Papers 11(19) : 26, 1936. Dubautia Gaud., Voy. Freycinet Uranie, Bot. Railliarda ciliolata DC., Prodr. 6: 441, 469, (1 826) = 1830; Atlas pI. 84, 1826. 1837. Type species: D. plantaginea Gaud., 1. c. D. coriacea (Sherif) Keck, Bernice P. Bishop Bracteae involucri liberae. Arisrae pappi Mus., Occas. Papers 11 (1 9): 27, 1936. ciliati-lacerataevel breve ciliati, Raillardia coriacea Sherif, Bot. Gaz. 95: Bracts of the involucre distinct. Pappus 80, 1933; later as Railliardia coriacea bristles squamose, ciliate-lacerate, or usually Sherif, Bernice P. Bishop Mus., BuI. 135: short plumose. 122, 125, fig. 40, 1935. D.demissifolia (Sherff) Keck, Bernice P. Enumeration of Species Bishop Mus., Occas. Papers 11 (1 9 ): 27, Dubautia Knudsenii Hbd. , Fl. Haw. Is. 223, 1936. 1888. Raillardia . demissifolia Sherff, Bot. Gaz. D.laxa H. & A., Bot. Beechey Voy. 87, 1832. 95: 78, 1933; later as Railliardla demissi­ D.laevigata Gray, Amer. Acad. Arts and Sci., folia Sherff, Bernice P. Bishop Mus., Bul. Proc. 5: ·135, 1861. 135 : 118-120, fig. 38, 1935. 342 PACIFIC SCIENCE, Vol. IV, October, 1950

D. Hillebrandi (Mann) Keck, Bernice P. Bi­ shop Mus., Occas. Papers 11(19): 28, shop Mus., Occas. Papers 11( 19): 28, 1936. 1936 (as D. Hillebrandii). · . Raillardia platyphylla Gray, Amer. Acad. Raillardia Hillebrandi Mann, Amer. Acad. Arts and Sci., Proc. 5: 134, 1861. Arts and Sci., proc. 7: 175, 1867 (as Rail­ D. reticulata (Sherff) Keck, Bernice P. Bi­ liardia Hillebrandii Mann in Sherff, Ber­ shop Mus., Occas. Papers J 1(19) : 28, nice P. Bishop Mus., Bu!. 135: 126,1935). 1936. D. kohalae (Skottsb.) comb. nov. Raillardia reticulata Sherff, Bot. Gaz. 95: Raillardia Kohalae Skottsb., Hort. Goto- 78, 1933; later as Railliardia reticulata . burg., Acta 15: 510-511, 1944. The spe­ Sherff, Bernice P. Bishop Mus., BuI. 135: cific name is a geographic one, taken fro~ 128, 130, fig. 41, 1935. the type locality in the Kohala Mts., and D. Rockii (Sherff) Keck, Bernice P. Bishop hence, following Rec. XLIII of the Inter­ Mus.,Occas. Papers 11(19 ) : 28, 1936. national Rules, should be 'written with a Railfardia rocki Sherff, Bot. Gaz. 95: 79, small initial letter. 1933; later as RailliardiaRockii Sherff, D. latifolia (Gray) Keck, Bernice P. Bishop Bernice P. Bishop Mus., Bul. 135: 130­ 'Mus., Occas. Papers 11(19): 26, 1936. 131, 1935. The change from R; rocki to , Raillardia lati/olia Gray, Amer. Acad. Arts R. R ockii was made by Sherff himself in and Sci., Proc. 5: 132, 1861. his second discussion of the species. Though D. linearis (Gaud.) Keck, Bernice P. Bishop he quoted the original specific name in­ Mus., Occas. Papers 11 (1 9 ) : 27, 1936. correctly as Rockii, still this second one is Railliardia linearis Gaud., Voy. Freycinet an acceptable form of a personal specific Uranie, Bot. 469,(1826) = 1830; Atlas name, and it can be adopted as a correction pl. 83, 1826. by the author himself of a typographical D. lonchophylla (Sherff) Keck, Bernice P. error. D. scabra (DC.) Keck, Bernice P. Bishop Bishop Mus., Occas. Papers 11( 19 ) : 27, Mus., Occas. Papers 11(19): 26, 1936. 1936. ' ., ~~n Railliarda scabra DC., Prodr. 6: 44 1, 1837. Raillardia lonchophylla Sherff, Amer. j our. D. Sherffiana Fosb., Torrey Bot. Club, Bul. Bot. 20: 619, 1933; later as Railliardia Ion­ 70: 395-397, 1943. (See fig. 1.) chophylla Sherff, Bernice P. Bishop Mus., D. struthioloides (Gray) Keck, Bernice P. BuI. 135: 122, fig. 39, 1935. Bishop Mus., Occas. Papers 11 (19): 28, D. Menziesii (Gray) Keck, Bernice P. Bishop 1936. Mus., Occas. Papers II ( 19): 28; 1936. Raillardia struthioloides Gray; Amer. Acad. Raillardia Menziesii Gray, Amer. Acad. -Arts and Sci., Proc. 5: 134, 1861. Arts and Sci., Proc. 5: 133-134,1861. D. ternifolia (Sherff) Keck, Bernice P. Bi­ D. molokaiensis (H bd.) Keck, Bernice P. shop Mus., Occas. Papers 11( 19 ) : 27, Bishop Mus., Occas. Papers 11( 19) : 27, 1936. 1936; Raillardia terni/olia Sherff, Amer. Jour. Raillardia Molokaiensis Hbd., Fl. Haw. Is., Bot. 20: 618, 1933; later as Railliardia 226, 1888. terni/ olia Sherff, Bernice P. Bishop Mus., D. montana (Mann) Keck, Bernice P. Bishop Bul. 135: 121, 1935. Mus., Occas. Papers 11(19) :-28, 1936. D. thyrsiflora (Sherff) Keck, Bernice P. Bi­ Raillardia montana Mann, Amer. Acad. shop Mus., Occas. Papers 11 ( 19): 27, Arts and Sci., Proc. 7: 176, 1867. 1936. D. platyphylla (Gray) Keck, Bernice P. Bi- , Raillardia thyrsifiora Sherff, Amer. Jour. Subgenera of Dubautia~ST. JOHN 343

Bot. 20 : 618 , 1933; later as Railliardia NOTES ON Dttb autia Sherffiana th yrsifiora Sherif, Bernice P. Bishop Mus., The type collection of Dubautia Sherffiana Bul. 135: 121-122,1935. Fosb. is H. St. John & F. R. Fosberg 12,161, Omitted from this enumeration of species from Oahu, W aianae Mts., brushy ridge, east are numerous described varieties and forms. of 2nd gulch east of Kaupakuhale, Mokuleia, Also omitted are the several interspecific hy­ 2,500 ft. alt., Oct. 23, 1 932 . At the same brids and intergeneric hybrids described by time and locality, only '200 feet higher up Sherif or by Degener and Sherff. :These puta­ the same ridge, another collection of this nov­ tive hybrids have not been evaluated and elty was made (St. John & Fosberg 12,162); allocated by the writer. but this was not listed in Fosberg's paper. The species of the subgenus Railliardiaster Subgenus Mixta subgen. nov. are not common on Oahu, and not until Type species: Dubautia railliardioides September 18, 1949, did the writer find an­ Hbd.; the reference follows. other colony. When climbing one of the nar­ row rocky ridges leading to Puu Kanehoa in Bracteae involucri plusminusve cohaeren­ the W aianae Mountains, he saw below .hirn rae, involucrum campanulatum est. Aristae a patch of unfamiliar yellow flowers. By pappi brevi ciliati, grasping hands with a student he was lowered Bracts of the involucre more or less united; over the brink and he grabbed a few sprigs involucre campanulate;pappus bristles short of what appeared to be Railliardiaster. Re­ ciliate. turning the next week, he was roped by his companions, M. Canoso and C. E. St. John, Enumeration of Species and lowered 20 feet down the vertical basalt cliff. There, on a dirt covered ledge so narrow Dubautia •paleata Gray, Amer. Acad. Arts as to deny a foothold to man, were three and Sci., Proe. 5 : 135, 1861. vigorous bushes 1-1.5 meters tall , many­ D. railliardioides Hbd.(as D. raillardioides), branched, with the numerous cymes in full Fl. Haw. Is. 224, 1888, emend. Sherif, golden flower. A.fter taking photographs and Bernice P. Bishop Mus., Bul. 135: 107, gathering full specimens, the collector was 1935. Hillebrand published his new spe­ hauled again to a foothold on the knife-edged raill~rdioides cies as D. and discussed its ridge. The data for these specimens are: Oahu, similarity to the next genus, Raillardia of Waianae Mts., southeast ridge of South Peak Gaudichaud. We must accept this generic of Puu Kanehoa, on face of basalt cliff, 20 name as Railliardia, which was the orig­ ft. below crest of sharp ridge, 2,600 ft. alt., inal, well-formed, and valid name. Hence, in open sunny spot, at top of thicket of Metro­ though there is no covering rule, Sherif took the logical, 'and, we believe, correct sideros and Euphorbia, Sept. 25, 1949, H. St. course in changing the specific name to be John 23,924; and also, 23,922 and 23,923. in harmony with the correct spelling of the Though collected in the southern part of the generic name. He was not justified in cap­ Waianae Mountains, these specimens proved italizing the specific name D. railliardioides to be of the same species, Dubautia Sherffiana under the 1935 International Rules of Bo­ Fosberg, earlier described from the northern tany, Ree. XLIV, examples, and Art. 3. part of the range. After studying all of the D. waialeale Rock, Torrey Bot. Club,Bul. specimens, including an isorype, the following 37: 303- 304, f. 5, 1910. changes in description are proposed: blades 344 'PACIFIC. SCIENCE, VoI. IV, October, 1950

' ,, 0 2 , d, In- " mm dX 5' . Ole" ' bm habit X Y2 ; c, f X l ' '---'--'c achene with P'PP'" " l-....L.-JD /fiana Fosb. a, 3 924. b lea " FIG. 1. ttbaersttt faXSb5er. From Sf. John 2 J vo Iuc re and flow . , Subgenera of Dttbautia-ST. JOHN 345

5(-7)-nerved; pappus bristles 16-25 in REFERENCES number, 2.5-3.5 mm. long. DEGENER, OTTO. 1940. Flora Hawaiiensis, Though the subgenus Railliardiaster is rare fam. 344: Railliardia demissifolia, May 20, on Oahu, there are to be considered two pub­ 1940; ed. 2: in 9/15/'46. lished records of its occurrence. Sherff (1935: 113) summarizes the range of Railliardia FOSBERG, F. RAYMOND. 1943. Notes on icabra as "Hawaii, ; and Oahu." This plants of the Pacific islands-III. Torrey species is common and Sherff lists nearly a Bot. Club , Btt!. 70: 386-397. whole page of specimens, but no collection --- 1948. Derivation of the flora of the from Oahu is in his list. None is in the Hawaiian Islands. In: Zimmerman, El­ Bishop Museum; nor has any other published wood C. 1948. Insects of Hawaii 1: 107­ record been seen, so Sherff's record seems to 119. University of Hawaii Press, Honolulu. be an error. Sherif also lists 0935: 125-126) Rail­ HARTT, CONSTANCE E., and MARIE C. NEAL. 1940. The plant ecology of Mauna liardia linearis Gaud. as on "eastern Maui , Kea, Hawaii. Ecology 21: 237-266, figs. western Oahu, Lanai, and Hawaii." For Oahu 1-22. he cites, U. S. Exploring Expedition, Kaala (Waianae) mountains, Oahu, 1840 (Gray, KECK, DAVID D. 1936. The Hawaiian silver­ New York ). Since 1840 no other collector swords : systematics, affinities, and phyto­ has found this species on Oahu, but many geographic problems of the genus Argy­ have done. so on the three more southerly roxiphium. Bernice P. Bishop Mus., Occss. islands. The flora of Oahu is the best known Papers 11(19 ): 1-38, pIs. 1-9. of any island in the group and its montane NEAL, MARIE C. 1948. In gardens of Ha­ flora is weU preserved. The U. S. Exploring waii. Bernice P. Bishop Mus., Spec. Pub. Expedition- also obtained specimens of this 40 : 1-805, figs. 1-312. species on Maui and Hawaii. This expedition PIPER, CHARLES V. 1906. Flora of the state was large and well organized but inharmoni­ of If/nsbington. U. S. N at!' Mus., U. S. ous. The commander, Capt. Charles Wilkes, Nat!' Herbarium, Contrib. 11: 1-637, pls, had trouble with his officers on the trip and 1-22, map. after it, and with the scientists for decades afterwards while acting as editor of the "pub­ SELLI N\., OLOF B. 1947. Studies in Ha­ lications of the expedition. In the botanical w'1iian pollen statistics, Part 2. The pollens collections made by this expedition, it is now of the Hawaiian Phanerogams. Bernice P. known that there is much confusion, as often Bishop Mus., Spec. Pub. 38: 1-430, pls. 1-58. rhe loose labels with the data were inserted with the wrong specimens. Piper (1906: 15) SHERFF, EARL EDWARD. 1941. New or discussed this in detail , and the evidence is otherwise noteworthy plants from the Ha­ well known. It is probable that the U. S. waiian Islands. A m er. Jour. Bot. 28: 18­ Exploring Expedition specimen came not 31. from Oahu, but from Maui or Hawaii with SKOTTSBERG, CARL. 1944. Vascular plants their other collections of these species. In from the Hawaiian Islands IV. Phanero­ any case, their record from Oahu is question­ gams collected during the Hawaiian Bog able until confirmed by a well-substantiated Survey 1938. Hart. Gotoburg., Acta 15: collection. 275-531, figs. r-556.