Nomenclatural and Taxonomic Changes in Hawaiian Alectryon (Sapindaceae)!

G EORGE L INNEy 2

Alectryon is endemic to the southwestern Pa proposal, some problems with the existmg cific, from Indonesia to Australia, the Philip nomenclature and taxonomy of the two Ha pines, Hawaii, and New Zealand. The species waiian species were found. are either shrubs or trees; all have pinnately Based on non-flowering specimens col compound leaves. Flowers, borne in panic lected around 1870 on Oahu (Makaleha Val ulate inflorescences, are less than 5 mm in ley) and Molokai (Pali of Kalaupapa), Hille diameter, apetalous, and are androdioe brand (1888) published the provisional genus cious, perhaps monoecious in A . grandifolius , "Mahoe, gen. nov. ?" Specimens have been though Radlkofer (1933) called them "falsely deposited in Berlin (B), Kew (K) , and Bishop polygamous." The fruit is 1-3 coccate, on the Museum , Honolulu,(BISH) herbaria. In No order of 10 mm in diameter in most species, vember 1889, Radlkofer saw and annotated at and lacks any obvious suture. At maturity it least one ofthe two specimens at K (Figure 1). ruptures irregularly to expose a bright red He also saw and annotated at least one of the "aril." The seed is borne distal to the "aril" and specimens at B (Figure 2). There was no date partially imbedded in it. Its exposed surface is of annotation on the sheet at B, and the sheet highly polished. Thirty-two names have been is no longer extant. An additional annotation published, including four since Alectryon on the label , probably in Radlkofer's hand was last monographed by Radlkofer in 1933. writing, reads, " Genus imperfecte cognitum The type species of the genus is A. excelsus Sapindacearum." Apparently based solely on J. Gaertner, 1788. Hillebrand's material, as no other collec Two species have been recognized from the tions are known to have existed at that time , Hawaiian Islands: A . macrococcus Radlk. Radlkofer (1890) published the new species A . (1890) and A. mahoe St. John and Frederick macrococcus in a footnote, along with A. (1949), both endemic. These are small trees strigosus and A . reticulatus, both New Guinea 3-8 m tall, rarely twice as tall, ofmesic forests. species. Alectryon strigosus and A. reticulatus Among the five major islands ofthe Hawaiian were given full Latin diagnoses, while A . ma chain, they are absent only from the island of crococcus was presented very casually, with a Hawaii. They differ from extra-Hawaiian spe reference to Hillebrand (1888), repeating in cies, as far as known, by the presence of a formation on fruit and leaflet sizes, and an hypodermis just below the upper epidermis of original contribution on the presence of the the leaf, by the very large size ofthe fruits, up hypodermis, all in German. It is possible that to 60 mm in diameter, and in having patelli he was not actually intending to publish the form as opposed to spheroid seeds. The two name at that time, only anticipating future Hawaiian species are supposedly separated by publication. Nevertheless, Radlkofer, in his the indument of the undersurface of mature later treatments, and all subsequent workers leaves: glabrate in A. mahoe, and densely to have accepted A. macrococcus as validly pub mentose in A. macrococcus. Young and unex lished in the 1890 paper. No type was des panded leaves ofboth species are more or less ignated. Indeed, no specimen at all was cited, densely sericeo-tomentose. Alectryon macro except by reference to Hillebrand, who had coccus is being proposed as rare and en cited collections. Because of this reference, dangered. During the preparation of the Hillebrand's text is to be considered part of the protologue. Upon the discovery of an additional popu

I Manuscript accepted June 1985. lation at Auwahi, Maui, by Joseph Rock in 2 Department of Botan y. 1910, Radlkofer (Radlkofer and Rock, 1911) 68 Changes in Hawaiian Alectryon-LINNEY 69

FIGURE I. Sheet of A . macrococcus in Kew, annotated FIGURE 2. Sheet of A. macrococcus, formerl y in Berlin, by Radlk ofer, here selected as new lectotype. Annotation annotated by Radlk ofer. Annotation reads: "Alectryon reads: "Alectryon macrocarpus [sic] m. sp. nov., mox macrococus [sic] m. Radlk ." Photograph by H. St. John , edenda (Mahoe Hillebr. Fl. Hawiens. p. 86 XI. 89. used by permission. Determ. Radlkofer." Kew photograph , printed by permission. may have been responsible for all the sub sequent confusion. St. John and Frederick provided a full Latin description for A . macro (1949, here and throughout) were in error coccus. He cited the Hillebrand collections in in generally attributing dense tomentum to Band K as specimens seen, and Rock's newly plants of Mo lokai . discovered Maui material as not seen. Details After additional collections had been made of floral morphology, still unknown to Radl on Oahu and Kauai by several collectors, St. kofer , were provided by Rock. In his mono John and Frederick recognized two discrete graph of the Sapindaceae, Radlkofer (1933) entities among the collections. They described repeated his description of A . macrococcus the new species A . mahoe, comprising all of without substantial change. Neither Radlko the Oahu collections, and probably tho se fer and Rock (1911) nor Radlkofer (1933) from Kauai. At the same time, St. John and mentioned dense golden brown tomentum on Frederick undertook to lectotypify A . macro leaf undersurfaces. This is noteworthy as the coccus. The y ignored the K specimens and tomentum is the most salient feature of the perhaps were unaware oftheir existence, or at Auwahi population. On the other hand, Rock least ofthe Radlkofer annotation on the one. (1913) ascribed the dense tomentum to the They selected the sheet in B annotated by species in general, ignoring the fact that spec Radlkofer, which St. John had seen and pho imens other than those collected at Auwahi tographed in 1935. Curiously, they made their were glabrate. I believe this bias or oversight selection in the belief that the sheet had since 70 PACIFIC SCIENCE, Volume 41, 1987 been destroyed in 1943. Its destruction has and Rock (1911) and Radlkofer (1933) noted been confirmed by Dr. Paul Hiepko, Curator that the leaves were "glabra (vel subtus t. of Phanerogams, Berlin Dahlem (in litt.) Hillebr. laxe tomentosa)." If St. John and Hillebrand's original material was puta Frederick realized the contradiction, it may be tively from mixed collections (Makaleha Val that they gave weight to the annotated speci ley, Oahu, and Pali of Kalaupapa, Molokai). men over the protologue, believing the pro The leaf indument character was believed to tologue in error. This again would require that be different on the two islands (glabrate on the now lost B specimen in fact had been Oahu, tomentose on Molokai), and St. John tomentose. However, the ICBN "Guide for and Frederick recognized the two entities as the Determination of Types" (GDT) TA.(d) distinct species. Therefore, the identities ofthe indicates the protologue is given precedence two species might have been reversed, depend to annotations and other indications of au ing on the choice of lectotype. St. John and thor's intent when there is a contradiction. Frederick chose a Molokai specimen. Though Article 8.1 states that a lectotype designation they do not say specifically, for their taxono may be superseded if shown to be in serious my to be consistent, it must be presumed that conflict with the protologue. Second, as to the their lectotype possessed leaves densely to choice of the annotated B sheet over the an mentose beneath. Certainly, in their discus notated K sheet, it must be repeated that the B sion they attributed this character to A . ma sheet bore no date of annotation. The sheet crococcus S.s. However, the extant Hillebrand mayor may not have been annotated prior to material in K is glabrate, regardless of the the date of publication. Surely, Radlkofer had collection locality written on the sheet, and annotated it prior to his 1911 paper, as he might be believed to be from Oahu. Locality specifically cited it there. The K sheet bears information to the contrary might be con both a prepublication date and the note "sp. sidered to be in error. Rather, I am fully per nov., mox edenda." The purport of the note is suaded the now lost B material also was extremely powerful, as no other sheet bore glabrate, not densely tomentose. Collections anything comparable. Furthermore, this sheet unquestionably from Molokai (Degener 9534 agrees with the protologue. In view of GDT and 9536; Steve Anderson 545: all in BISH), T.4.(a) which states: "A lectotype must be are glabrate. Unfortunately, St. John's photo chosen from among the elements definitely graph of the B sheet, though excellent, pro studied by the author up to the time the name vides no assistance. In the photo the leaves of the taxon was published," it is argued appear to be quite clearly glabrate, but it is not that the K sheet was the proper choice for certain that a leaf undersurface is represented. lectotype. At any rate, A . macrococcus came to be cir The third irregularity is St. John and Frede cumscribed as having densely tomentose leaf ricks having chosen a non-existent specimen. undersurfaces, and A. mahoe glabrate. The With neither a tangible specimen, nor even a other diagnostic features noted by St. John description ofit to support their claim contra and Frederick will be considered below in the dictory to the protologue, the name is no more taxonomic treatment. fixed by their lectotypification than without There are four irregularities in St. John and it. The typification is ineffectual, and violates Frederick's lectotypification with regard to the spirit ofArticles 7.3 and 7.4 of the ICBN. the International Code of Botanical Nomen The fourth irregularity has already been pre clature (ICBN) of 1983. First, in selecting the sented. That is, there is no evidence that the B tomentose condition to represent A. macro material was densely tomentose. It is my belief coccus S.S., they have violated the protologue. that the Hillebrand material was correctly re Hillebrand described the leaves as "glabrous presented in the protologue as glabrous or or slightly tomentose underneath. . . ." This slightly tomentose, and that Molokai and accords wellwith the segregate A . mahoe but is Oahu specimens are scarcely, if at all, to be discordant with A. macrococcus s.s. Radlkofer distinguished from each other. The attribu did not mention leafindument, but Radlkofer tion of dense tomentum to plants of Molokai Changes in Ha waiian A lectryon-L INNEY 71

TABLE I

SUMM ARY OF DIAGNOSTIC F EATURES U SED BY ST . J OHN AND FREDERICK (1949) IN D ELIMITING THE H AWAIIAN SPECIES OF Alectryon

"A. macrococcus" "A . mahoe" leaflet length II -18.5 cm 9-26 em leaflet shape lance-ellipt ic elliptic, slightly lance- or obl ance-elliptic, or rarely oval vestiture of lower leaf surface densely soft yellowish pilosulose glabrous except for main veins calyx cam panula te knee-pan shaped base tru nca te style scarcely evident slender, evident stigma lobes short, erect, not diverging [well developed]" anther length O.5mm l.l- I.5 mm

*Not specifically stated but necessarily implied by the context. in general is in error and must be the result of entity of Maui is recognized as a distinct confusion after the discovery ofthe tomentose taxon, it is without name or type. population at Auwahi, Maui. Whether or not The taxonomic problem is just this: Is the the tomentose population is taxonomically tomentose entity a valid taxon? And if so, at distinct will be treated below, but the rationale what level should it be recognized? St. John behind St. John and Frederick's lectotypifica and Frederick recognized features other than tion is in doubt. I believe that St. John and leaf indument whereby they distinguished Frederick's type designation was arbitrary, their glabrate and tomentose species. Rather without any apparent merit, and in error. The than anyone feature, it was the sum ofa list of GOT T.4 .(e) deals with situations in which differences which justified to them the recog heterogeneous (here, in the opinion of St. nition oftwo species. Reexamination ofthese John and Frederick) elements have been seg other characters has shown that they are regated as distinct taxa . In such cases, " If it invalid or trivial. They are summarized in can be shown that the element thus selected [as Table. 1. lectotype of the original taxon] is in serious The criteria ofleaflet length and shape over conflict with the protologue, then one of the lap or are too similar to be diagnostic. As to previously segregated elements is to be se floral characteristics, there is still a paucity of lected as the lectotype." The specimen at Kew material. However, features which might be (Figure I) annotated by Radlkofer, is here interpreted intuitively are borne out by the designated as the lectotype: specimens at hand. The drawing (reproduced here as Figure 3) in St. John and Frederick's " Alectryon macro "h is Sapindaceae paper reputedly is a typical flower of the to carpus [sic] m. sp. (det.rnova Flora mentose entity from Auwahi, Maui. Intui nov., mox edenda Hawiensis [sic] Oahu[?] tively, it represents a syndrome indicative of (Mahoe Hillebr . Dr. Hillebrand 12/72" Fl. Hawiens. p. 86 immaturity or abortiveness. The calyx is unex panded, stamens small and unextended for XI. 89.Oeterm. pollen presentation, the ovary small, style and Radlkofer" stigmas unextended for pollen reception. It is Consequent to this lectotypification, the in probable that in 1949 good floral material of dum ent character of the original protologue is this entity was unavailable to St. John and reinstated for A. macrococcus, i.e., mature Frederick. Nevertheless, Rock (Radlkofer leaves "glabrous or slightly tomentose under and Rock, 1911) had previously described neath." Also in consequence, A . mahoe St. flowers from the Auwahi plants, and in form John and Frederick becomes a later synonym they agreed more with St. John and Frede of A. macrococcus. If the densely tomentose rick' s A . mahoe than with St. John and Fre- 72 PACIFIC SCIENCE, Volume 41,1987

FIGURES 3-7. Flowers of Hawaiian Alectryon, all approximately X 10. Figure 3. "Representative" flower of A . macrococcus sensu St. John and Frederick; J. F. Rock, Nov. 1910, Haleakala, Maui , 2600 ft. Figure 4. Bud and mature flower from Auwahi , Maui, plant, St. John 26870. Figure 5. Mature flower and detached stamen from Auwahi, Maui, plant, Little 31122. Figure 6. Young flower from A . mah oe holotype, Frederick and Sakimura 185. Figure 7. Older flower from A. mahoe ho!otype. Figures 3, 6, and 7 reprinted from Pacific Science by permission of University of Hawaii Press. Changes in Hawaii an Alectryon-LINNEY 73 derick 's A . macrococcus. Two post-1949 col Aug 1976 E. L. LITTLE, Jr. 31122. (BISH lections in BISH (St. Jo hn 26870, and E. L. 423454). Little, Jr. 31122), both from Auwahi, have Synonymy: Alectryon macrococcus sensu St. well-developed flowers. Figures 4 and 5 show John and Frederick, Pac. Sci. 3(4) :300, line-reductions ofphotographs offlowers from 301. 1949, non Radlk., 1890. these two collections. Figures 6 and 7 are The name A . mahoe St. John and Frederick, flowers of the A . mahoe holotype. From Fig reduced to synonymy with A. macrococcus ures 3-7 it may be seen that mature flowers of on nomenclatural grounds above, is thus to the two entities are virtually indistinguishable. be identified with the autonymic variety on The immature or abortive nature of St. John taxonomic grounds. and Frederick's A . macrococcus flower is borne out. Floral material, though still mea ger, indicate no feature which might be tax ACKNOWLEDGMENTS onomically useful. If such exists, it must be This work was begun under USFWS Coop very subtle indeed. In the end, St. John and erative Agreement # 14-16-0001-79096. I Frederick's list reduces to the single character wish to express my gratitude to C. W. Smith ofleafundersurface indument. This character for the use of his word processing facilities serves only to distinguish Auwahi, Maui, used in the preparation of the manuscript; plants from those of all other sites. As men also to Karen Asherman for assistance in tioned earlier, young and unexpanded leaves the photo-mechanical transfer preparation of of both entities are densely tomentose. Even the figures; and to G . D . Carr and Charles among mature leaves of the glabrate entity, Lamoureux for their comments and advice on scarcely any specimen has proved to be per the text. fectly glabrous. All bear at least a few hairs on the lower order veins and venules, and man y specimens, including the A. mahoe holotype, LITERATURE CITED bear hairs to varying degrees in the areoles as HILLEBRAND, W. 1888. Flora of the Ha waiian well. Islands. Lubrecht and Cramer, New York. I am satisfied that the Auwahi population is RADLKOFER, L. 1890. Uber die gliederung der truly distinct from all other kno wn popula familie der Sapindaceen. Sitzungsber. tions. However, because of the factors men Konigl. Bayer. Akad. Wiss., Math.-Phys. tioned abo ve (dense tomentum on young CI. 20(1) :255. leaves of all, variability in tomentosity in gla ---. 1933. Sapindaceae. 62. Alectryon brate plants, and lack of other useful distin Gaertn. In: A. Engler (ed.), Das Pjlanzen guishing features), I can consider them no reich 98(1) :983-1002. more than varieties of the same species. The ---, and J. ROCK. 1911. New and note glabrate taxon then is A . macrococcus Radlk. worthy Hawaiian plants. Bot. Bull. Terr. var. macrococcus, while the tomentose taxon Haw . Bd.Ag. For., Div. For. 1:3-4. is here described as the new variety: ROCK, J. 1913. The Indigenous Trees of the Alectryon macrococcus Radlk. var. auwa Hawaiian Islands. Charles E. Tuttle, Rut hiensis Linne y, var. nov. A varietate macro land, Vermont. cocco foliolis maturis dense aureis indutis ST. JOHN, H ., and L. FREDERICK. 1949. A sec infra differt. Eight miles east of Ulupala ond Hawaiian species of Alectryon (Sapin kua, dr y forest on lava. 3000-3500 ft. Sma 11 daceae). Hawaiian Plant Studies 17. Pac. tree 12 ft, 4 inches in diameter. Five trees. 26 Sci. 3(4):296- 301.