Pacific Science (1977), vol. 31, no. 3 © 1978 by The University Press of . All rights reserved

Hybrid () in the Hawaiian Islands!

JAMES K. BAKER 2 AND SUZY ALLEN 2

ABSTRACT: First- and second-generation hybrids of Hibiscadelphus giffardianus Rock and H. hualalaiensis Rock have been found in Hawaii Volcanoes National Park, and elsewhere in the Hawaiian Islands. They are under cultivation from interspecifically cross-fertilized which occurred on parent within the park. A history of parent and hybrid species is given, and floral characteristics are analyzed. Hybrid occurrence and the implications to natural resource management in trying to preserve the integrity of native Hawaiian species and ecosystems are discussed.

HYBRID TREES OF TWO RARE, endemic Rock and W. M. Giffard. Rock (1911) des­ Hawaiian species of Hibiscadelphus (Hau cribed the species in honor of his friend kuahiwi) are growing in Hawaii Volcanoes Giffard. National Park, in arboreta and private gar­ Shortly before the died in 1930 others dens. These hybrids of H. giffardianus Rock were grown from cuttings in the community and H. hualalaiensis Rock have been des­ of Volcano a few kilometers east of the type cribed as H. x puakuahiwi· by Baker and locality (Degener 1932, Fagerlund 1944). Allen (1976a). Only one cutting survived, but before it died Hybridization occurred in the late 1950s in 1940, L. W. Bryan, Rock, and others or early 1960s, but went unnoticed for more succeeded in growing another cutting than 10 years. Investigations oftrees damaged (Fagerlund 1944) which ultimately saved the by roof rats (Rattus rattus) revealed their species from extinction. presence in 1973. It had been noticed that This cutting grew to maturity on the rats were stripping bark from Hibiscadelphus Keauhou Ranch of Herbert Shipman just trees, eating seed pods, and chewing holes in .east of Kipuka Puaulu. The 11 mature to reach the quantities ofnectar stored Hibiscadelphus giffardianus trees presently inside (Baker and Allen 1976b). Variations surviving in Hawaii are its descendents. Seven in morphology were noticed in flowers studied trees are growing at the type locality in from different trees, which led to the discovery Kipuka Puaulu, three in the Puu Mahoe of first- and second-generation hybrids. Arboretum on , and one in Wahiawa Hybridization probably originated in one Arboretum on Oahu. On three occasions the oftwo localities where the parent species were species was reduced to a single tree, and brought together in cultivation. The most today there are only 11 mature trees, which probable location is in Kipuka Puaulu (Bird makes the species one of the world's rarest. Park) in Hawaii Volcanoes National Park, In the 1950s efforts were made to save the the type locality of Hibiscadelphus giffar­ other very rare species, Hibiscadelphus dianus. The type species was known from a hualalaiensis, from extinction. Its localized, single tree discovered in 1911 by Joseph F. allopatric distribution was 72 km from Kipuka Puaulu, on Mt. Hualalai. At the time of discovery in 1909, Rock (1911,1913) I Research supported by the National Park Service estimated only a dozen trees were extant and and Hawaii Natural History Association, Hawaii Vol­ since that time the number of wild trees has canoes National Park, Hawaii. Manuscript received II August 1976. declined to two because of habitat deteriora­ 2 Hawaii Field Research Center, Hawaii Volcanoes tion by cattle ranching. National Park, Hawaii 96718. In 1953 and 1954, 12 seedlings of the 285 286 PACIFIC SCIENCE, Volume 31, July 1977

oE E F

FIGURE 1. A, Hibiscade/phus gifj'ardianus; B, hybrid H. xpuakuahiwi; C, H. hua/a/aiensis; D, E, and F, buds of A, B, and C, respectively; G, H, and I, tomentum in angles of veins of A, B, and C, respectively.

& 'S* Q#!§l##@1 i!Ji!!$$'hffi5M Mi¥W ; liMA WI ;!Ji z;m ..44&£j Hybrid Hibiscadelphus in the Hawaiian Islands-BAKER AND ALLEN 287

Hualalai species were planted in Kipuka ANALYSlS OF Hibiscadelphus HYBRIDS Puaulu (National Park Service planting records) in the belief that the national park Comparative morphology of parent was a proper refuge for rare and endangered species and Fl hybrids is shown in Figure 1. island flora. Six of the 12 trees were planted The following floral characteristics were within 30 meters of a Hibiscadelphus giffar­ examined: (I) number of , (2) length dianus tree which for a number of years had of shortest , (3) length of longest bract, been a primary source ofseed for that species, (4) length of calyx, (5) length of peduncle, though relatively few seedlings survived. (6) length of corolla, (7) length of staminaI We believe that the history of H. x column, (8) lengths of styles and stigmas, puakuahiwi began when park management (9) length of longest , (10) length of personnel unknowingly collected inter­ shortest petal, (11) width of shortest petal, specifically cross-fertilized seed from this (12) width oflongest petal, and (13) direction tree. Several seedlings were planted in nearby of corolla whorl. Kipuka Ki, but only two of the unsuspected Criteria 3-7 were selected to plot potential hybrids survived. Both grew adjacent to a key characters to separate parents from road. hybrids. The plots of criteria 4-7 in Figure 2 Because of their easy roadside access and show that the degree of overlap of these because the two FI hybrids demonstrated measurements make them unreliable charac­ such vigor in producing flowers and seed, ters. Criterion 3 is reliable but caution must both became primary sources for herbarium be taken to determine that longest bracts specimens, and seed for propagation of "H. are not stunted or shortened due to necrosis giffardianus" trees. Consequently, many or insect damage. specimens were identified improperly, and Figures 1 and 3 show that Hibiscadelphus numerous F2 hybrids were grown unknow­ giffardianus bracts are the longest of the three ingly in the park, Hawaii, and world arboreta taxa, while H. hualalaiensis flowers nearly (Baker and Allen 1977). On Maui a second­ lack them. Figure 3 shows no overlap in generation tree was used unwittingly to prop­ twice the standard deviations in lengths of agate a large number of F3 seedlings for longest bracts, and this character alone will perpetuation of H. giffardianus on that usually separate parents and hybrids. island. When this error was noted, the F3 A peculiarity of bract morphology of seedlings were destroyed so as not to com­ Bishop and Herbst pound the hybrid dilemma. (1973), will easily distinguish this other extant The second, but less feasible source of the taxon. Bracts of H. distans are connate two FI hybrids may have been trees from through one-third their length, while bracts Puu Mahoe (FlemingjVodtrock) Arboretum of the other taxa are not. on Maui, where both parent species have In analyzing the direction ofcorolla whorl, been cultivated since the 1950s. In 1960, two we found the following (c, clockwise; cc, seedling "H. giffardianus" trees were given counterclockwise) : to Hawaii Volcanoes National Park by the Territorial Division of Forestry, but existing H. giffardianus H. hualalaiensis records indicate only that the trees came from 38 percent c 63 percent c Maui. We assume they came from Puu 62 percent cc 37 percent cc Mahoe. These seedlings may have been hy­ FI hybrids F2 hybrids brids, but the exact location of their origin 43 percent c 53 percent c and planting in the park is unknown. 57 percent cc 47 percent cc As we know of no other FI trees outside the two in Kipuka Ki, and since the plantings When viewed apically, a is considered of seedlings in Kipuka Ki are well docu­ left-handed ifthe overlap in a clockwise mented, we believe the origin of the two FI direction, and right-handed if the petals trees to be cross-fertilized seed taken from overlap in a counterclockwise direction Kipuka Puaulu. (compare parts Band C, Figure 1). 288 PACIFIC SCIENCE, Volume 31, July 1977 r------E===;::==3-- F2 1--E=~~=3----H. ·h.

----E=;::::3----- FI STAMINAL COLUMN H. g.-----E=:::::;==:r-- 40 45 50 55 60 65 70 75 80 85 88

---£=~;::==3---F2

--E=:;:==3----- H. h.

FI PEDUNCLE -----1=::;:=3--- H. g.

10 15 20 25 30 35 40 45 50 55

--E~~3--F2

---E;::::l------...:---- H. h.

--E:;:=I- FI CALYX --E::;:::3---- H. g.

10 15 20 25 30 35 40 45 50 55

-----f===:;::==3--- F2 -----1==:==::;:===:3--- H. h.

----f=;:=3--- FI COROLLA ----E==:::;::==:3-- H. g.

35 40 45 50 55 60 70 75 80 LENGTHS IN MM

FIGURE 2. Ranges, means, and one standard deviation of measurements for staminal column, peduncle, calyx, and corona lengths. Horizontal lines are ranges, vertical lines are means, and boxes are one standard deviation on either side of the mean. H. g., Hibiscadelphus giffardianus; H. h., H. hualalaiensis; F I, first-generation hybrid; F2, second-generation hybrid. Hybrid Hibiscadelphus in the Hawaiian Islands-BAKER AND ALLEN 289

20

F2 10 N-IOO

0 5 25 r$ 40 a UJ 0: ::> (f) 0 - 5 10 15 25 30 35 I

I

20

FI N = 100 10

o •I I I•• • 5 10 15 20 LENGTHS OF LONGEST BRACTS (mm)

FIGURE 3. Ranges, means, and two standard deviations ofmeasurements for lengths of longest bracts. Horizontal lines are ranges, vertical lines are means, and boxes are two standard deviations on either side of the mean. Bar graphs represent numbers of flowers. H. g., Hibiscade/phus giffardianus; H. h., H. hua/a/aiensis; Fl. first-generation hybrid; F2, second-generation hybrid; N. sample size. . 290 PACIFIC SCIENCE, Volume 31, July 1977

Davis (1964) and Davis and Selvaraj (1964) DISCUSSION OF HYBRID PROBLEMS showed that among 34 species of 12 genera (11 species of ), only three species In view of the apparent fertility of some could be considered predominantly left­ F2 hybrids, and occurrence of some F3s, it handed and none could be considered right­ appears that gene flow between Hibisca­ handed. The others simply showed no specific delphus species is possible. These factors must tendency. In view of these data, petal asym­ be taken into consideration in any program metry in Hibiscadelphus is of interest. designed to protect the genetic integrity of In Figure 1 it can be seen that tomentum Hibiscadelphus taxa. In other words, the taxa is located in the angles of veins on the under­ should not be grown together. sides of , which can be useful in keying Discovery of hybrid Hibiscadelphus, and species when buds or flowers are not avail­ the situation that brought about their oc­ able. Rock (1911,1913), Degener (1932), and currence, has created much interest and Bishop and Herbst (1973) also mention this controversy among those interested in main­ characteristic. Leaves of Hibiscadelphus taining native species and ecosystem integ­ giffardianus have more tomentum than FI rity. Hibiscadelphus hybrids have become a leaves, whereas the tomentum is almost case in point of problems of hybridization nonexistent on leaves of H. hualalaiensis. and genetic swamping brought about by man. However, the presence of tomentum should Hybridization of Hibiscadelphus occurred not be used to key leaves from young trees, when species were cultivated outside their sucker growth, or limbs of any aged tree not natural distributions in close proximity to producing flowers. one another, whereas their natural allopatric Coloration of petals can be used to dis­ distributions would have prevented inter­ tinguish between parent and Fl taxa but not specific cross-fertilization of seed. Through­ F2s. Flowers of Hibiscadelphus giffardianus out Hawaii numerous native species, many are magenta (rarely splotched with green) with very limited natural ranges, are being throughout the 3 to 5 day life-span of the brought together under artificial conditions blossom. Flowers of H. hualalaiensis are in gardens, arboreta, and parks. yellow-green when 1 to 3 days old but Persons concerned with propagation of may tend to tum purplish in 4- to 5-day­ native for perpetuation of species old flowers. Flower color of Fl hybrids should be cautious or suspicious of or is usually magenta with greenish splotches seedlings produced from genetically com­ in petals at the apex. Flower colors of patible taxa when grown together un­ F2 hybrids range from magenta to yellow­ naturally. Propagation should always be green. through cuttings or pollination under con­ The appearance or condition of anthers trolled conditions. and may sometimes distinguish F2 In the case of Hibiscadelphus, the un­ hybrids which closely resemble parent suspected gathering of cross-fertilized seed species. Anthers of parent and Fl taxa fully and the transplanting of seedlings gathered dehisce, exposing amber-colored pollen, from beneath hybrid trees have compounded whereas anthers of some F2 flowers the hybrid problem. may not dehisce and contain abnormally small, colorless pollen grains that lack cytoplasm. ACKNOWLEDGEMENTS This may be a further indication ofreduced fertility in some F2 flowers, a situation already We wish to thank Ishmael Trono of the documented in hybrid Hibiscadelphus by Department ofBotany, University ofHawaii, Carr and Baker (1977). Meiotic analyses Honolulu, for the drawings in Figure 1. We showed some F2 trees to be little affected by also wish to thank O. Degener, D. Herbst, hybrid breakdown, while others exhibited C. Lamoureaux, H. St. John, and C. Smith moderate to severe meiotic disturbances. for reviewing the manuscript.

4$ Hybrid Hibiscadelphus in the Hawaiian Islands-BAKER AND ALLEN 291

Volcanoes National Park. Pac. Sci. 31: LITERATURE CITED 191-194. BAKER, J. K., and S. ALLEN. 1976a. Hybrid DAVIS, T. A. 1964. Aestivation in Malvaceae. Hibiscadelphus (Malvaceae) from Hawaii. Nature 20 (4918):515-516. Phyt. 33(4):276. DAVIS, T. A., and J. C. SELVARAJ. 1964. Floral ---. 1976b. Studies on the endemic asymmetry in Malvaceae. Jour. Bombay Hawaiian genus Hibiscadelphus (Hau­ Nat. Hist. Soc. 61(2):402-409. kuahiwi). Proc. First Conf. in Nat. Sciences DEGENER, O. 1932. Flora Hawaiiensis. Sec­ Hawaii Volcanoes Natl Pk.; Coop. Parks tion on Hibiscadelphus, Family Ma1vaceae, Study Unit, Univ. ofHawaii ; C. W. Smith, No. 221, Book 3. Honolulu. Editor, pp. 19-22. FAGERLUND, G. O. 1944. Hau Kuahiwi, sole ---. 1977. Status and distribution of survivor. Paradise ofthe Pacific [Honolulu] Hibiscadelphus (Malvaceae) species and 56(1):21,32. collections. Unpub. manuscript, Natl. Park NATIONAL PARK SERVICE. Tree planting rec­ Serv., Hawaii Volcanoes Natl. Park. ords. Unpub. manuscript, Natl. Park Serv., BISHOP, L. E., and D. HERBST. 1973. A new Hawaii Volcanoes Natl. Park. Hibiscadelphus (Malvaceae) from . ROCK, J. F. 1911. Hibiscadelphus. In L. Rad1­ Brittonia 25(3): 290-293. koffer and J. F. Rock, New and noteworthy CARR, G. D., and J. K. BAKER. 1977. Cyto­ Hawaiian plants. Hawaiian Bd. Agric. For. genetics of Hibiscadelphus (Malvaceae): A Bot. Bull. 1: 8-14. meiotic analysis of hybrids in Hawaii ---. 1913. The indigenous trees of the Hawaiian Islands. Pp. 297-302. Honolulu.