Sit DL. XVII JANUARY 1963 NO. 1 3ACIFIC SCIENCE A QUARTERLY DEVOTED TO THE BIOLOGICAL AND PHYSICAL SCIENCES OF THE PACIFIC REGION HAROLD ST. JOHN Revision of the Genus Pandanus Part 14. Malaya and Singapore GEORGE W. BARLOW Species Structure of Gillichthys mirabilis TAKASHI OKUTANI Molluscan Assemblages of Izu Submarine Banks WILLIAM A. GOSLINE Osteology and Systematics of Elongate Perciform Fishes LINDSAY R. WINKLER and E. YALE DAWSON Food Habits of California Sea Hares D. L. INMAN , W. R. GAYMAN, and D. C. COX Littoral Sedimentary Processes on Kauai UNIVERSITY OF HAWAII PRESS BOARD OF EDITORS O. A. BUSHNELL, Editor-in-chief Department of Microbiology, University of Hawaii Robert Sparks, Assistant to the Editors Office of Publications and Information, University of Hawaii Thomas S. Austin Bureau of Commercial Fisheries, Hawaii Area (U. S. Fish and Wildlife Service) Honolulu, Hawaii L. H. Briggs Department of Chemistry University of Auckland Auckland, New Zealand Ai Kim Kiang Department of Chemistry University of Malaya, Singapore Gordon A. Macdonald Department of Geology University of Hawaii Donald C. Matthews Department of Zoology University of Hawaii Colin S. Ramage Department of Geology and Geophysics University of Hawaii Martin Sherman Department of Entomology University of Hawaii Donald W. Strasburg Bureau of Commercial Fisheries, Hawaii Area (U. S. Fish and Wildlife Service) Honolulu, Hawaii Albert L. Tester Department of Zoology and Entomology University of Hawaii Miklos F. Udvardy Department of Zoology University of British Columbia Vancouver, Canada i Thomas Nickerson, Managing Editor Assistant to the University Provost INFORMATION Contributions to Pacific biological and physical science will be welcomed from authors in all parts of the world. (The fields of anthropology, agriculture, engineering, and medicine are not included.) Manu- scripts may be addressed to the Editor-in-Chief, PACIFIC SCIENCE, University of Hawaii, Honolulu 14, Hawaii, or to individual members of the Board of Editors. Use of air mail is recommended for all communications. Manuscripts will be acknowledged when received and will be read promptly by members of the Board of Editors or other competent critics. Authors will be notified as soon as possible of the decision reached. FOR AUTHORS Manuscripts of any length may be submitted, but it is suggested that authors inquire concerning possi- bilities of publication of papers of over 30 printed pages before sending their manuscripts. Authors should not overlook the need for good brief papers, presenting results of studies, notes and queries, com- munications to the editor, or other commentary. PREPARATION OF MANUSCRIPT It is requested that authors follow the style of Pacific Science described herein and exemplified in the journal. Authors should attempt to conform with the Style Manual for Biological Journals , Am. Inst. Biol. Sci. Washington. ( Continued on inside back cover) PACIFIC SCIENCE A QUARTERLY DEVOTED TO THE BIOLOGICAL AND PHYSICAL SCIENCES OF THE PACIFIC REGION VOL. XVII JANUARY 1963 NO. 1 Previous issue published November 21 , 1962 CONTENTS PAGE Revision of the Genus Pandanus Stickman, Part 14 . New Species from Malaya and Singapore. Harold St. John . 3 Species Structure of the Gohiid Fish Gillichthys mirabilis from Coastal Sloughs of the Eastern Pacific. George W. Barlow 47 Preliminary Notes on Molluscan Assemblages of the Submarine Banks Around the Izu Islands. Takas hi Okutani 73 Notes on the Osteology and Systematic Position of Hypoptychus dybowskii Steindachner and Other Elongate Perciform Fishes. William A. Gosline.... 90 Observations and Experiments on the Food Habits of California Sea Hares of the Genus Aplysia. Lindsay R. Winkler and E. Yale Daivson 102 Littoral Sedimentary Processes on Kauai, a Subtropical High Island. D. L. Inman, W . R. Gayman, and D. C. Cox 106 note: Adoption of the Metric System and Celsius Scale 131 Pacific Science is published quarterly by the University of Hawaii Press, in January, April, July, and October. Subscription price is $4.00 a year; single copy, $1.25. Check or money order payable to University of Hawaii should be sent to University of Hawaii Press, Honolulu 14, Hawaii, U. S. A. Printed by Star-Bulletin Printing Company, 420 Ward Avenue, Honolulu 14, Hawaii. SMITHSONIAN' JHSTITJJT18N MAR 2 5 1963 > Revision of the Genus Pandanus Stickman, Part 14 New Species from Malaya and Singapore Harold St. John1 As A RESULT of field work in Malaya and Sing- apore and of study in the herbarium at Sing- apore, numerous new species of Pandanus have been detected. The following novelties are of this origin. Pandanus attenuatus sp. nov. (sect. Acro- stigma) Fig. 150 DIAGNOSIS HOLOTYPI: Frutex probabaliter humili, caule in apice 8 mm diametro lutescenti laevi sed cum dorsis longitudinalibus, foliis 35- 40 cm longis 18-20.5 mm latis chartaceis supra viridibus infra minime viridibus pallidioribus et proxima basem purpureo-tinctis 1-costatis et minime 2-plicatis subplanis in sectione mediali cum 14-15 nervis parallelis secundariis promi- nentibus in quoque dimidio nervis tertialis sub- visibilis infra proxima apicem, lamina ligulata subito in apice caudato 5.5-6 cm longo 0.7 mm lato contracto basi amplexicauli inermi valde nervosa sed ex 2 cm marginibus cum aculeis 0.5-1. 3 mm longis 2-6 mm distantibus subulatis adscendentibus pallidis, midnervo infra ultra mediam inermi, in sectione mediali marginibus cum aculeis 0.7-1 mm longis 1-4 mm distanti- bus subulatis valde adscendentibus, apice subu- lato cum marginibus et midnervo infra cum aculeis 0.5-1 mm longis 0.5-1 mm separatis subulatis adscendentibus, inflorescentia foeminea terminali adscendenti cum 1 syncarpio, pedun- culo 4.5 cm longo 4 mm diametro trigono brac- teato, syncarpio 3.5 cm longo 2.7 cm diametro ellipsoideo cum circa 144 drupis eis 13-13.5 mm longis 4-4.5 mm latis crassisque fusiformi- bus 5-6-angulosis corpore 6-7 mm longo, pileo 7.5-9 mm longo basi 3.5-4 mm alto ovoideo subanguloso laevi, stylo 4-6 mm longo valde proxime arcuato osseoso subulato olim in basi subanguloso, stigmate 4-4.5 mm longo late line- ari brunneo papilloso in apice extento, endo- 1 B. P. Bishop Museum, Honolulu 17, Hawaii, U. S. A. Manuscript received June 29, 1961. carpio in parte lA infera lateribus cartilagineis pallidis, seminibus 3. 5-4.5 mm longis subdo- liformatis, mesocarpio apicali hemisphaerico, mesocarpio basali sparso cum fibris fortibus in lateribus. DESCRIPTION OF ALL SPECIMENS EXAMINED: Shrub, probably low; stem at apex 7—8 mm in diameter, yellowish, smooth but with longitudi- nal ridges; leaves 35-40 cm long, 18-20.5 mm wide, chartaceous, green above, below slightly paler green and near the base suffused with purple, 1 -ribbed, slightly 2 -pleated, but nearly flat, at midsection with 14-15 parallel second- ary veins in each half, these prominent through- out, cross tertiary veins barely visible only below and near the tip, the blade ligulate, abruptly con- tracted to a 5. 5-7. 5 cm caudate tip, this 0.7 mm wide, the base amplexicaul, unarmed, very veiny, beginning 2 cm up the margins with prickles 0.5-1. 3 mm long, 2-6 mm apart, subulate, as- cending, pale; the midrib below unarmed to beyonid the middle; at midsection the margins with prickles 0.7-1 mm long, 1-4 mm apart, subulate, sharply ascending; the subulate apex having the margins and midrib below with prickles 0.5-1 mm long, 0.5-1 mm apart, sub- ulate, ascending; pistillate inflorescence terminal, ascending, bearing 1 syncarp; peduncle 3-5 — 4.5 cm long, 4 mm in diameter, 3-sided, bracted; syncarp 3-3.5 cm long, 2.7 cm in diameter, ellip- soid, bearing about 120-144 drupes, these 13- 13.5 mm long, 4-4.5 mm thick and wide, fusi- form, 5-6-angled, the body 6-7 mm long; pileus 7.5-9 mm long, the base 3.5-4 mm high, ovoid, slightly angled, smooth; style 4-6 mm long, sharply curved proximally, bony, subulate, slightly angled only on the base; stigma 4-4.5 mm long, broad linear, brown, papillose, run- ning to the tip; endocarp centering in lower but extending to the base, the walls cartilaginous, pale; seed 3. 5-4. 5 mm long, more or less barrel- shaped; apical mesocarp hemispheric, with the center hollow; basal mesocarp sparse, with strong fibers up the sides. 3 5 mm FIG. 150. Pandanus attenuatus St. John, from holotype. a, Habit, X 1; b, drupe, lateral view, X 1; c, drupe, ■ longitudinal median section, X 1; d, drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; /, drupe, apical view, X 4; g, style and stigma, oblique view, X 10; h, leaf base, lower side, X 1; i, leaf middle, lower side, X 1 \ j, leaf apex, lower side, X 1- 2 cm. Page 230: Revision of Pandanus, 14. Malaya and Singapore — St. John 5 holotypus: Malaya, Selangor, Gunong lelu Semangkok, 28 April 1922, I. H. Bur kill 8,876 (SING). SPECIMENS EXAMINED: Malaya, Selangor, Gap Sempadang, 1908, Has bin (sing). DISCUSSION : P. attenuatus is a member of the section Aero stigma, as is its closest relative, the Malayan P. unguiculatus RidL, which has the leaves 21-27 mm wide, at midsection with 17- 19 secondary veins in each half; syncarp 5.5 cm long; drupes 18-20 mm long; the seed 7-8 mm long, suprabasal; and the style 2.5-3 mm long. P. attenuatus has the leaves 18-20.5 mm wide, at midsection with 14-15 secondary veins in each half; syncarp 3-3.5 cm long; drupes 1 3— 13.5 mm long; seed 3. 5-4. 5 mm long, basal; and the style 4- 6 mm long. The new epithet is the Latin word attenuatus, drawn out, which is given in allusion to the attenuate subulate leaf apex. Pandanus caudatifolius , nom. nov. (sect. Acro- stigma) Fig. 151 P. globuliferus RidL, R. Asiat. Soc., Straits Br., Jour. 61: 42, 1912; non Thouars (1808). DIAGNOSIS OF HOLOTYPE: "Stems up to 1 m tall,” 6-7 mm in diameter, yellowish, smooth; leaves 33-37 cm long, at the middle 15-17 mm wide, near the base 10-11 mm wide, firm charta- ceous, dark green above, pale below, almost flat except at base, swordlike, tapering in lower lA to the narrower base, in upper V3 rather sharply narrowed to the 7 cm subulate apex, this 5 cm down 1 mm wide, the secondary parallel veins conspicuous and at midsection 15 in each half, no visible tertiary cross veins, the base widened and amplexicaul, unarmed, but beginning at 2-2.5 cm the margins with prickles 0.8-1 mm long, 2-4 mm apart, subulate, straight or arcuate, ascending, pale; the midrib below unarmed for lower 54; at midsection the margins with sub- ulate tipped serrae 0.2-0.3 mm long, 1.5-3 mm apart; on the caudate triangular apex the mar- gins and midrib below with prickles 0. 5-0.9 mm long, 1-3.5 mm apart, stout subulate, arcu- ate, ascending; peduncle 3 cm long; syncarp 18 mm in diameter, globose, bearing about 40 drupes, these 11-13 mm long, 3.5-4 mm wide, 2-3 mm thick, the body oblanceoloid, com- pressed, narrowed to the subulate style, 5-6- angled, the body 8—9 mm long; pileus 6-7 mm long, its base the widest part of the drupe, the lower part pyramidal-semiorbicular; style 3-4 mm long, subulate, ascending arcuate; stigma distal, 2.5-3 mm long, linear, extending to apex, brown, papillose; endocarp in lower V4, obconic, the base obtuse, stramineous, dull, cartilaginous, the walls 0.2 mm wide; apical mesocarp 3.5-4 mm long, a cavern filled with white medullary membranes and hairs; basal mesocarp sparse, the margins fibrous, the center fleshy. HOLOTYPE: Malaya, Selangor, Gunong Se- mangok, 4,000 ft. alt., April 1894, H. N. Ridley 15,612 (sing). DISCUSSION: P. caudatifolius is a member of the section Aero stigma, as is its closest relative, P. globosus St. John, a Malayan species with the leaves 11-12 mm wide, little if at all narrowed towards the base, at midsection the margins with prickles 1-1.5 mm long, 2-5 mm apart, subulate; drupes 14-17 mm long; and the style 3-6 mm long. P. caudatifolius has the leaves 15-17 mm wide near the middle, ligulate, narrowed below and near the base 10-1 1 mm wide, at midsection the margins with subulate-tipped serrae 0.2-0. 3 mm long, 1.5-3 mm apart; drupes 11-13 mm long; and the style 3-4 mm long. The epithet is derived from caudatus, having a tail, and folium, leaf, given in reference to the appearance of the leaf. Pandanus globulosus sp. nov. (sect. Acro- stigma ) Fig. 152 DIAGNOSIS HOLOTYPI: Frutex, caulibus in apice 6-7 mm diametro laevibus lucidis lute- scentibus, foliis 38-42 cm longis 11-12 mm lads firme chartaceis supra obscure viridibus infra pallide viridibus et ad basem purpureis 1-costatis licet paene planis, in sectione mediali cum 12- 14 nervis parallelis secundariis in quoque dimi- dio eis ubique prominentis, nervis tertialis nul- lis, laminis ligulatis subabrupte contractis in apice caudato subulato 5. 5-6. 5 cm longo 0.5 mm lato basi amplexicauli inermi conspicue nervosa ex 2-3 cm marginibus cum aculeis 0. 5-0.9 mm longis 2.5-7 mm distantibus subulatis proxime adscendentibus pallidis, midnervo per 30 cm in- Fig. 151. Pandanus caudatifolius St. John, from holotype. a, Habit, X 1; b, c, drupes, lateral view, X 1; d, drupe, longitudinal median section, X 1; e> drupe, lateral view, X 4; f, drupe, longitudinal median section, X 4; g, drupe, apical view, X 4; h, leaf base, lower side, X 1; i, leaf middle, lower side, X 1; j, leaf apex, lower side, X 1- io cm. Fig. 152. Pandanus globulosus St. John, from holotype. a, Habit, X 1; b, c, drupes, lateral view, X 1 \ d, drupe, longitudinal median section, X 1; e, drupe, lateral view, X 4; f, drupe, longitudinal median section, X 4; g, drupe, style, and stigma, apical view, X 4; h, leaf base, lower side, X 1; h leaf middle, lower side, X 1; j, leaf apex, lower side, X 1- 5 cwl. o 2 cm. 8 PACIFIC SCIENCE, Vol. XVII, January 1963 ermi, in sectione mediali marginibus cum acuieis 1-1.5 mm longis 2-5 mm distantibus proxime adscendentibus subulatis, in apice caudato mar- ginibus et midnervo infra cum acuieis 0.7-1 mm longis 0.7-3. 5 mm distantibus subulatis ad- scendentibus, inflorescentia foeminea terminal! erecta cum uno syncarpio, pedunculo 4—6 cm longo 3-4 mm diametro trigono bracteato, syn- carpio 2. 5-3. 5 cm longO' 2.3-3 cm diametro globoso vel elliptici-globoso cum circa 96-120 drop is eis 14-17 mm longis 3-4.5 mm latis cras- sisque fusiformibus sed apice angustiori 5-6- anguloso, corpore 9-10 mm longo oblanceoloi- deo, pileo 7-11 mm longo basi 3-5 mm alta ovoidea vel oblato-ovoidea cum 5 angulis prorni- nentibus laevibus, stylo 3-6 mm longO' proxime arcuato subulate rigido, stigmate 2-3 mm longo linear! distali brunneo papillose in apice ex - tento, endocarpio in parte V3 infera obconico lateribus 0.2 mm crassis cartilagineis stramineis, mesocarpio apicali suborbiculari cum mem- branis albis radiatis numerosis, mesocarpio bas- al! fibroso et carnoso. diagnosis OF holotype: Shrub; stems at summit 6-7 mm in diameter, smooth, shining, yellowish; leaves 38-42 cm long, 11-12 mm wide, firm chartaceous, dark green above, pale green below and with a small area near the base purple, 1 -ribbed, apparently almost flat, at mid- section with 12-14 parallel secondary veins in each half, these prominent throughout, no visible tertiary cross veins, the blade ligulate, rather abruptly contracted to a 5. 5-6. 5 cm subulate, caudate apex, this 0.5 mm wide, the base am- plexicaul and unarmed, prominently veiny, be- ginning at 2-3 cm up the margins with prickles 0. 5-0.9 mm long, 2.5-7 mm apart, subulate, closely ascending, pale; midrib unarmed for about 30 cm; at midsection the margins with prickles 1-1.5 mm long, 2-5 mm apart, closely ascending, subulate; on the caudate apex the margins and midrib below with prickles 0.7-1 mm long, 0.7-3. 5 mm apart, subulate, ascend- ing; pistillate inflorescence terminal, erect, bear- ing 1 syncarp; peduncle 4-6 cm long, 3-4 mm in diameter, 3-sided, bracteate; syncarp 2.5— 3.5 cm long, 2.3-3 cm in diameter, globose or elliptic-subglobose, bearing about 96-120 drupes, these 14-17 mm long, 3-4.5 mm wide and thick, fusiform but the tip more slender, 5-6-angled, the body 9-10 mm. long, oblanceo- loid; pileus 7-11 mm long, the base 3-5 mm high, ovoid or oblate-ovoid, with 5 prominent angles, smooth; style 3-6 mm long, curved prox- imally, subulate, rigid; stigma 2-3 mm long, linear, distal, brown, papillose; running to the tip; endocarp in lower obconic, the walls 0.2 mm thick, cartilaginous, stramineous; apical mesocarp suborbicular, with numerous radial, white membranes; basal mesocarp sparse, fibrous and fleshy. HOLOTYPUS: Malaya, Pahang, Telom ridge, Nov. 1908, H. N. R [idley\ 13,798 (sing). discussion: P. globulosus is a member of the section Aero stigma, as is its closest relative, the Malayan species P. caudatif olius St. John. Under that species is given a summary of the contrast- ing differences. The epithet is the Latin adjective globulosus , like a little sphere, and is chosen with reference to the shape of the syncarp. Pandcmus mollifoliaceus sp. nov. (sect. Acro- stigma) Fig. 153 DIAGNOSIS HOLOTYPI: Frutex, catile in apice 8 mm diametro luteo laevi, foliis 45-47 cm longis 23-29 mm latis moll iter chartaceis supra viridibus infra pallide viridibus et videtur glan- ds 1-sulcatis 2-plicatis in sectione mediali cum 17-18 nervis secundariis parallelis in quoque dimidio eis ubique prominentibus, circa apicem infra nervis tertialis reticulum cum sectionibus oblongis formantibus, lamina ligulata circa me- dian! latissima proxima basem 14-15 mm latis, in 6.5 cm apice caudato subiter diminuentibus eo 5 cm ex apice 0.7 mm lato, basi amplexicauli inermi et nervis exevidentibus, marginibus ex 3 cm cum acuieis 1—1.5 mm longis 2-5 mm dis- tantibus subulatis pallidis in 45° adscendentibus, midnervo per 35 cm inermi, in sectione mediali marginibus cum acuieis 0.5-1 mm longis 4-10 mm distantibus subulatis adpresse adscenden- tibus, in apice contractenti marginibus cum acuieis 0.8-1.7 mm longis 2. 5-3. 5 mm distan- tibus divergentibus, midnervo infra cum acuieis simulantibus 4-7 mm distantibus, in apice cau- dato marginibus et midnervo infra cum subulato- serrulis 0.5-0.91 mm longis 2-4 mm distantibus, inflorescentia foeminea terminal! erecta cum uno to cm. l cm. . Fig. 153. Pandanus mollifoliaceus St. John, from holotype. a, Habit, X 1; b, drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; ^ drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe, style, and stigma, apical view, X 4; g, leaf base, lower side, X 1; ^ leaf middle, lower side* X 1; i, leaf apex, lower side, X 1- 5 cm. 10 PACIFIC SCIENCE, VoL XVII, January 1963 capite, pedunculo 3 cm longo 3 mm diametro trigono bracteoso, syncarpio 2.7 cm longo 2.3 cm diametro subgloboso cum circa 120 drupis eis 8-10 mm longis (vel in curvatura styli sequi 12-15 mm) 4-5 mm latis 3-4 mm crassis cor- pore oblongo-ovoideo 5-anguloso lateribus stri- atis, pileo cum basi 2 mm alta pulviniformi 5-angulato minute striato, stylo 4-5 mm longo proxime in angulo recto curvato recto vel sinuoso subulato gracile firmo in dimidia infera angu- loso, stigmate 3-4 mm longo lineari distali brun- neo paene in apice extento, endocarpio in tertia infera 5-6 mm longo late obovoideo lateribus 0.2 mm crassis cum fibris longitudinalibus forti- bus et membrano gracili pallidi, mesocarpio apical i cavernoso cum membranis medullosis albis paucis, mesocarpio basali carnoso et cum fibris tenuibus. DIAGNOSIS OF HOLOTYPE: Shrub; stem 8 mm in diameter at apex, yellowish, smooth; leaves 45-47 cm long, 23-29 mm wide, soft charta- ceous, green above, pale green and apparently glaucous below, 1 -ribbed, 2-pleated, at midsec- tion with 17-18 parallel secondary veins in each side, these prominent throughout, and on the lower side near the tip with barely visible cross tertiary veins making irregular oblong meshes, the blade ligulate, though broadest at or beyond the middle, near the base 14-15 mm wide, abruptly contracted to a 6.5 cm caudate tip, this 5 cm down 0.7 mm wide, the base amplexicaul, unarmed, the veins obscure, beginning 3 cm up the margins with prickles 1—1.5 mm long, 2-5 mm apart, subulate, pale ascending at about 45°; the midrib unarmed for about 35 cm; at mid- section the margins with prickles 0.5-1 mm long, 4-10 mm apart, subulate, closely ascend- ing; on the subapical contracting region the margins with prickles 0.8-1. 7 mm long, 1.5- 2.5 mm apart, divergent, and those of the midrib below similar but 4-7 mm apart; on the caudate tip the margins and midrib below with subulate serrulations 0. 5-0.9 mm long, 2-4 mm apart; pistillate inflorescence terminal, erect, bearing one head; peduncle 3 cm long, 3 mm in diam- eter, 3 -sided, bracteate; syncarp 2.7 cm long, 2.3 cm in diameter, subglobose, bearing about 120 drupes, these 8-10 mm long (or following the curve of the style 12-15 mm), 4-5 mm wide, 3-4 mm thick, the body oblong-obovoid, 5 -angled, the sides striate; pileus with the base 2 mm high, cushion-shaped, 5 -angled, finely striate; style 4-5 mm long, proximally ascend- ing at right angles, straight or twisting, slender subulate, firm, angled in lower half; stigma 3-4 mm long, linear, distal, brown, running almost to the tip; endocarp centering in lower V3, and 5-6 mm long, broadly obovoid, the walls 0.2 mm thick, of heavy longitudinal fibers and a thin connecting membrane, pale; apical mesocarp a broad cavern with a few white medullary mem- branes; basal mesocarp fleshy and with weak fibers. HOLOTYPUS: Malaya, Negri Sembilan, Gun- ong Angsi, eastward on ridge, 2,600 ft. alt., 24 Nov. 1923, Mohamed Nur 11,691 (sing). DISCUSSION: P. mollifoliaceus is a member of the section Aero stigma, as is its closest rela- tive, P. caudatifolius St. John, of Malaya, a species with its leaves 15-17 mm wide, firm chartaceous, tapering from the middle down- wards and near the base only 10-11 mm wide, at midsection the leaf margins with subulate tipped serrae 0.2-0. 3 mm long, 1.5-3 mm apart; syncarp 18 mm in diameter, globose, bearing about 40 drupes; drupes with the body oblance- oloid, the sides smooth; pileus base pyramidal- semiorbicular; stigma 2.5-3 mm long; endocarp in lower 14, obconic; and the apical mesocarp longer than wide. P. mollifoliaceus has its leaves 23-29 mm wide, soft chartaceous, tapering downwards and near the base 14-15 mm wide, at midsection the leaf margins with prickles 0.5-1 mm long, 4-10 mm apart, subulate; syn- carp 2.7 cm long, 2.3 cm in diameter, subglobose, bearing about 120 drupes; drupes with the body oblong-obovoid, the sides striate; pileus base cushion-shaped; stigma 3-4 mm long; endocarp in lower V3, broadly obovoid; and the apical mesocarp wider than long. The new epithet is formed from the Latin mollis, soft; foliaceus, leafy, in reference to the soft texture of the leaves. Pandanus albibracteatus sp. nov. (sect. Pan- danus ) Figs. 154, 155, 169, 170^ DIAGNOSIS HOLOTYPI: Arboriformis, caulibus 8 m altis 15 cm diametro viridarium formans, cortice brunneo muricato, radicibus fulturosis 1 Fig. 154. Pandanus albibracteatus St. John, from holotype. a, Fresh drupe, lateral view, X 1; ^ dried drupe, lateral view, X 1; ^ drupe, longitudinal median section, X 1; d, drupe, apical view, X 1; e, carpel apices, stigmas, and proximal sinuses, oblique view, X 4; /, leaf base, lower side, X 1; ft leaf middle, lower side, X 1; b, leaf apex, lower side, X 1. 12 PACIFIC SCIENCE, VoL XVII, January 1963 m longis 3—4 cm diametro in lineis muricatis, foliis 1.72-1.97 m longis 5-3.5 cm latis coriaceis olivaceo-viridibus supra glaucis infra glauci- oribus supra midnervum sulcatis 2-plicatis in sectione mediali cum 41-44 nervis parallelis secundariis in quoque medio nervis tertialis nul- lis infero folii ligulato medio exteriori gladi- formati sensim in apice 40 cm longo caudato subulato trigono diminuenti eo 10 cm ex apice 0.9 mm lato, basi amplexicauli et inermi sed ex 2-6 cm marginibus cum aculeis 4.5-7 mm longis 10-45 mm separatis subulatis rectis vel arcuatis apicibus rubro-brunneis plerumque adscendenti- bus, midnervo infra cum aculeis 3.5 mm longis 20-45 mm separatis conico-subulatis reflexis, in sectione mediali marginibus cum aculeis 3-5 mm longis 7-20 mm separatis subulatis adpresse adscendentibus, midnervo infra cum aculeis 3.5- 4 mm longis 22-28 mm separatis crassiter sub- ulatis adscendentibus, in apice subulato mar- ginibus et midnervo infra cum subulato-serrulis 0.3-0. 5 mm longis 4-10 mm separatis; inflores- centia foeminea terminali arcuata cum syncarpio solitario, pedunculo 27 cm longo 13-20 mm diametro trigono folioso-bracteato, syncarpio 23 cm longo 17.5 cm diametro ellipsoido-subglo- boso cum 67 phalangibus, nucleo maturo molli et carnoso subito contracto in reliquum minp- tum, phalangibus 5. 5-6.2 cm longis dimidio supero 3.2-4 cm lato 2. 5-3. 5 cm crasso oblongo- ellipsoideo vel obovoideo infra aurantiaco-rubro supra subviridibus, deinde rosaceo 5-6-anguloso lateribus laevibus sublucidis subconvexis, suturis lateralibus plerumque nullis rariter 1 (-2), apice eonvexo, sinibus apicalibus centralibus 0.7-2 mm profundis V-formatis in fondam plerumque rectis, dimidio infero phalangiis 33-4.2 cm latis 2. 8-3. 9 cm crassis carnoso-incrassatis rosaceo- aurantiacis cuneato-oblongis humeris 3-6 mm latis 5-8-angulosis lateribus gradatim curvatis carne saccharosa innoxia sed non consumpta, carpellis 4-7 eis centralibus 2A tarn grandibus quam marginalibus, apicibus depresso-conico eis marginalibus asymmetricis et aliquis cum piano vel concavo distalo stigmatis, stigmatibus 2.5-4 mm longis cordatis vel deltoideis vel reniform- ibus pallide brunneis papillosis sulcatis obliquis centripetalibus, sinibus proximalibus Vi-Vi ad fondam extends, endocarpio minime supra- mediali osseoso mahogani-colorato intra lucido lateribus 5 mm crassis, seminibus 12-15 mm longis 5 mm diametro ellipsoideis, mesocarpio supero in quoque apice cavernam cum fibris paucis et membranis a lb is medullosis multis formantibus, mesocarpio basali grande fibroso et carnoso. DIAGNOSIS OF holotype: Treelike; stems 8 m tall, 15 cm in diameter, forming a large clump; bark brown, muricate; prop roots 1 m long, 3-4 cm in diameter, muricate in lines; leaves 1.72-1.97 m long, 5-5.5 cm wide, cori- aceous, olive green, above glaucous and more so below, broad furrowed above the midrib, and | with 2 rounded pleats, at midsection with 41-44 j secondary parallel veins in each half, no ter- tiary cross veins, lower half ligulate, outer half sword-shaped, gradually tapering to a 40 cm j caudate subulate trigonous apex, this 10 cm down only 0.9 mm wide, the base amplexicaul ! and unarmed, but beginning at 2-6 cm the mar- 1 gins with prickles 4.5-7 mm long, 10-45 mm apart, subulate, mostly ascending, straight or arcuate, red-brown-tipped; the midrib below with prickles 3-5 mm long, 20-45 mm apart, conic-subulate, reflexed; at midsection the mar- gins with prickles 3-5 mm long, 7-20 mm apart, subulate, appressed ascending; the midrib below with prickles 3.5-4 mm long, 22-28 mm apart, stout subulate, ascending; on the subulate apex the margins and midrib below with subulate- serrulations 03-0.5 mm long, 4-10 mm apart; pistillate inflorescence terminal, arching, bear- ing one syncarp; peduncle 27 cm long, 13-20 mm in diameter, 3 -sided, leafy bracted; syncarp 23 cm long, 17.5 cm in diameter, ellipsoid- subglobose, bearing 67 phalanges, the ripe core soft and fleshy and quickly shrivelling to a very small remnant; phalanges 5. 5-6.2 cm long, the upper half 3.2-4 cm wide, 2. 5-3. 5 cm thick, oblong-ellipsoid or obovoid, below orange-red, above greenish changing to pink, 5-6-angled, the sides smooth, a little shiny, gently curved, lateral sutures usually none, rarely 1 (—2), the apex convex; central apical sinuses 0.7-2 mm deep, V-shaped at the very bottom, mostly straight; lower half of phalange 33-4.2 cm wide, 2. 8-3.9 cm thick, fleshy enlarged, pinkish orange, cuneate-oblong in outline, the shoulders 3-6 mm wide, often eaten by small, black ants, and reduced to white scars, 5-8-angled, the sides Page 238: Revision of Pmdanus, 14. Malaya and Singapore — St. John 13 smooth, gently curved, the flesh sweet, inoffen- sive, but not eaten; carpels 4-7, the central ones about 2A as large as the marginal, the apices very low conic, the marginal ones asymmetric and some of them having just distal of the stigma a slight flat platform or even a concavity; stigmas 2.5-4 mm long, from cordate to deltoid or reniform, light brown, papillose, creased, oblique, centripetal; proximal sinus running Vi- 2A way to valley bottom; endocarp slightly supramedian, bony, mahogany-colored, the inner surfaces shining, the lateral walls 5 mm thick; seeds 12-15 mm long, 5 mm in diameter, el- lipsoid; upper mesocarp forming in the apex of each carpel a cavern with a few fibers but filled with aerenchyma of white medullary membranes; basal mesocarp fibrous and fleshy, extensive. DESCRIPTION OF STAMINATE TREE: Tree 7 m tall, 12 cm in diameter; bark brown, muricate; prop roots none, but tree abused; leaves 1.23 m long, 3.9-4 cm wide, coriaceous, furrowed above the midrib, the sides not plicate, at midsection with 35 secondary parallel veins in each half, no tertiary cross veins, sword-shaped, tapering into a 15-20 cm subulate trigonous apex, this 10 cm down, 3 mm wide; at 3-4.5 cm from the base the margins with prickles 4-5 mm long, 7-21 mm apart, arcuate subulate, with thickened base, ascending; the midrib below beginning at 6-7 cm with prickles 3-4.5 mm long, 4-22 mm apart, similar but reflexed; at midsection the margins with prickles 1.5-2. 5 mm long, 6-12 mm apart, subulate, closely appressed ascend- ing; the midrib below with prickles 1.5-2 mm long, 13-25 mm apart, subulate, arcuate ascend- ing; the subulate apex with margins and midrib below with prickles 0. 5-0.7 mm long, 3-7 mm apart, subulate-serrations; staminate inflor- escence single, terminal, arcuate and pendent, 50 cm long; peduncle fleshy and soon disintegrat- ing; lower bracts with lower part and the other bracts all white; lowest floral bract 48 cm long, 2.2 cm wide, the body ligulate except at base with marginal prickles 2-2.5 mm long, acerose, ascending, closely appressed, white; the apex 20-25 cm long, subulate, foliaceous; median bract 19 cm long, 2.6 cm wide, linear-lanceolate, unarmed except near the tip; spikes about 12, dense, 2.5-4 cm long, 1.5-2 cm in diameter, cylindric; staminate flowers very numerous, 1.5- 1.9 cm long; column 4-8 mm long, bearing a common section of equal length and with about 24 stamens, the free filament tips 1-2 mm long; anther body 2. 4-3. 6 mm long, linear-lanceolate, bearing an apical subulate prolongation of the connective 0.6-0. 7 mm long. HOLOTYPUS: Malaya, Penang Island, 5 miles n. of George Town, top of sea beach, June 30, I960, H. St. John 26,371 (bish). SPECIMENS EXAMINED: Malaya, Penang Is- land, Tanjong Tokong, 6 miles n. of George Town, top of sea beach, staminate, June 30, I960, H. St. John 26,370 (bish); Penang Island, Mt. Pleasure, 8 miles n. of George Town, top of quartz sand sea beach, clump with stems cut back to 1 m. by 10 cm, old phalanges from the ground, June 30, I960, H. St. John 26,369 (bish). DISCUSSION : P. albibracteatus is a member of the section Pandanus, as is its closest relative, the Singapore species P. Boryi Gaud., which has the phalanges with the central apical sinuses 1.5-2. 5 mm deep; marginal carpels with a visor projecting over the stigma; leaves 3. 5-4.4 cm wide, sword-shaped, at midsection with 39 sec- ondary parallel veins in each half, and near the base the margins with prickles 3-4.5 mm long, 7-25 mm apart. P. albibracteatus has the phal- anges with the central apical sinuses 0.7-2 mm deep; stigmas lacking a visor; leaves 5-5.5 cm wide, the lower half ligulate, the outer half sword-shaped, secondary parallel veins 41-44 at midsection in each half, near the base the mar- gins with prickles 4.5-7 mm long and 10-45 mm apart. The new epithet is formed from the Latin words albus, white; bracteatus, with bracts, and given in allusion to the white floral bracts. Pandanus ambiglaucus sp. nov. (sect. Pan- danus ) Figs. 156, 170 b DIAGNOSIS HOLOTYPI: Arbor juvenalis, trunco erecto 1 m alto 10 cm diametro, radici- bus fulturosis ad 1 m longis et 4 cm diametro brunneis in lineis sparse muriculatis, foliis 2- 2.47 m longis 5.8— 6.3 cm latis coriaceis viridi- bus et glaucis in lateribus ambis, spinis omnibus albis, lamina late sulcata et dimidiis subsigmoi- 14 PACIFIC SCIENCE, Vol. XVII, January 1963 deis in sectione mediali cum 44-46 nervis parallelis secundariis in quoque dimidio, nervis tertialis nullis, gladiformatis sensim in apice trigono subulato caudato 20-30 cm longo di- minuentibus eo 10 cm ex apice 1.8 mm lato, basi amplexicauli et inermi sed ex 5-6 cm mar- ginibus cum aculeis 4-6 mm longis 5-26 mm separatis valde subulatis arcuatis adscendenti- bus, midnervo infra ex 7. 5-8. 5 cm cum aculeis 3-4.5 mm longis 20-32 mm separatis arcuatis subulatis reflexis basibus crassis, in sectione mediali marginibus cum aculeis majoribus 2.5- 3 mm longis 18-30 mm separatis arcuatis sub- ulatis adscendenti-adpressis basibus crassis ple- rumque alternantibus cum aculeis minoribus 1- 2 mm longis, midnervo infra cum aculeis 1.5-2 mm longis 13-24 mm separatis subulatis ad- scendentibus, in apice subulato marginibus et midnervo infra cum subulato-serrulis 0.3-1 mm longis 3-8 mm separatis, inflorescentia foemi- nea terminali arcuati, pedunculo 40 cm longo 2 cm diametro trigono folioso-bracteato cum syncarpio solitario eo 21 cm longo 16 cm diam- etro ellipsoideo (submaturo et viridi) cum 85 phalangibus eis 6-6.6 cm longis 2.8-4 cm latis 2. 8-3. 6 cm crassis obovoideis ad apicem infra cuneatis, apice convexo 5-6-anguloso lateribus supra convexis infra concavis (evidente imma- turis), laevibus lucidis, suturis lateralibus nullis, sinibus apicalibus centralibus 0.5-2. 5 mm pro- fundis latis et vadosis rectis vel subcurvatis, carpelis 7-8 eis centralibus Vi-2A tarn grandibus quam illis marginalibus eis basalibus mediali- busque cum apicibus subplanis sed illis centrali- bus cum apicibus oblato-conicis, stigmatibus 2- 3 mm longis suborbicularibus vel ellipticis ob- scure brunneis obliquis sulcatis centripetalibus, ceteris marginalibus truncatis, sinibus proximal- ibus Vi ad fondam extentis, endocarpio in parte lA inf era osseoso brunneo, seminibus 12-16 mm longis ellipsoideis, mesocarpio supero in quaque carpella cavernam cum fibris et mem- branis pallidis medullosis formanti, mesocarpio basali fibroso et carnoso. DIAGNOSIS OF HOLOTYPE: Young tree, trunk erect 1 m tall, 10 cm in diameter; prop roots up to 1 m long and 4 cm in diameter, sparsely muriculate in vertical lines, brown; leaves 2- 2.47 m long, 5.8—63 cm wide, coriaceous, green and glaucous on both sides, the spines all white, the blade broad furrowed over the midrib, the sides slightly sigmoid, at midsection with 44- 46 secondary parallel veins in each half, no visible tertiary cross veins, sword-shaped, grad- ually tapering upwards into a 20-30 cm caudate, subulate, trigonous apex, this 10 cm down 1.8 mm wide, the base amplexicaul and unarmed, but beginning at 5-6 cm up the margins with prickles 4-6 mm long, 5-26 mm apart,' stout subulate, arcuate, ascending; the midrib below beginning at 7. 5-8. 5 cm with prickles 3-4.5 mm long, 20-32 mm apart, arcuate subulate, ■ heavy based, reflexed; at midsection the mar- gins with larger prickles 2.5-3 mm long, 18- 30 mm apart, arcuate subulate, heavy based, ascending appressed, mostly alternating with smaller ones 1-2 mm long; the midrib below with prickles 1.5-2 ■'mm long, 13-24 mm apart, subulate, ascending; on the subulate apex the margins and midrib below with subulate serru- lations 0.3-1 mm long, 3-8 mm apart; pistil- late inflorescence terminal, arching; peduncle 40 cm long, 2 cm in diameter, 3 -sided, leafy bracted, bearing a single syncarp, this 21 cm long, 16 cm in diameter, ellipsoid, slightly im- mature and still green, bearing 85 phalanges, these 6-6.6 cm long, 2.8-4 cm wide, 2. 8-3. 6 cm thick, obovoid above, cuneate below, the apex convex, 5— 6-angled, the sides convex- above, concave below (but this probably due to immaturity), smooth, shining; lateral sutures none; central apical sinuses 0.5-2. 5 mm deep, shallow, the line straight or gently curved; car- pels 7-8, the central ones Vl -2A as large as the marginal ones, the basal and median phalanges with the carpel apices almost flat, but those near the center with the apices oblate conic; stigmas 2-3 mm long, suborbicular to elliptic, dark brown, oblique, creased, centripetal, some of the marginal ones truncate; proximal sinus running Vz way to valley bottom; endocarp in lower J/3, bony, brown; seeds 12-16 mm long, ellipsoid; upper mesocarp forming in each car- pel apex a cavern, with fibers and pale medul- lary membranes; basal mesocarp fibrous and fleshy. HOLOTYPUS: Singapore, Woodlands Road, . 14 miles mark, opposite Kranji Nature Reserve, fresh water swamp, with Dicranopteris linearis, July 10, I960, H. St. John 26,376 (bish). Fig. 155. Pandanus albibracteaius St. John, from St. John 26,370. a, Staminal column with anthers, lateral view, X 10; h, leaf base, lower side, X 1; ^ leaf middle, lower side, X 1; d, leaf apex, lower side, X 1- FIG. 156. Pandanus ambiglaucus St. John, from holotype. a, Phalange, lateral view, X 1; b, phalange, longitudinal median section, X 1; c, phalange, apical view, X 1; ^ carpel apex, stigma, and proximal sinus of a median phalange, oblique view, X 1; e, carpel apex, stigma, and proximal sinus of a phalange near the base, oblique view, X 1; /, leaf base, lower side, X 1 \ g, leaf middle, lower side, X 1; leaf apex, lower side, XI. o 10 cm/. O 1 OCYYl. I 1 1 — . — 1 i. L — l • i I 4 e o J FIG. 157. Pandanus Boryi Gaud., from Singapore, Posher g 36,919. a, Phalange, lateral view, X 1; h, phalange, longitudinal median section, X 1; o, phalange, apical view, X 1; ^ carpel apex, stigma, and proximal sinus, oblique view, X 1; ^ leaf base, lower side, X 1 ; leaf middle, lower side, X 1; g, leaf apex, lower side, X 1- 18 PACIFIC SCIENCE, Vol. XVII, January 1963 DISCUSSION: P. ambiglaucus is a member of the section Pandanus, as is P. inclinatus St. John, also of Singapore, its closest relative, and a species with the syncarp with 44-61 phal- anges, these with the apex truncate or sub- truncate, and the lower half fleshy enlarged; endocarp median; leaves 1.5— 1.7 m long, 4.4— 4.7 cm wide, above dark olive green, below green; and the secondary veins at midsection 38-40 in each half. P. ambiglaucus has the syncarp with 85 phalanges, these with the apex convex, and the lower half cuneate; endocarp in the lower lA\ leaves 2-2 .47 m long, 5.8-63 cm wide, green and glaucous on both sides; and the secondary veins 44-46 in each half at the midsection of the leaf. The new epithet is from the Greek words ambos, both; glaukos, color of the sea, applied to the glaucous coating on both sides of the leaves. Pandanus Boryi Gaud., Bot. Voy. La Bonite, pi. 22, f. 15, 1841 Fig. 157 Description from Fosberg 36,919, Singapore: "Small tree,” leaves more than 1.64 m long, 3.5 cm wide, subcoriaceous, green above and below, channeled above the midrib, 2 -pleated, in section M-shaped, at midsection with 39 parallel secondary veins in each half, no tertiary cross veins, sword-shaped, tapering gradually to the trigonous subulate tip, more than 15 cm long, and about 10 cm down, only 2.5 mm wide, the base amplexicaul and unarmed, but begin- ning at 4-7 cm the margins with prickles 3- 4.5 mm long, 7-25 mm apart, arcuate, subu- late, ascending, pale; the midrib below be- ginning at 19 cm with prickles 3 mm long, 25-40 mm apart, similar but reflexed; at mid- section the margins with prickles 2-2.5 mm long, 8-15 mm apart, arcuate subulate, ascend- ing appressed; the midrib below bearing simi- lar prickles 1.7-2 mm long; on the subulate apex the margins and midrib below with ser- rulations 0.3-0. 5 mm long, 3-8 mm apart, red- tipped. . . fruiting head small cylindric, red when ripe”; phalange 6.3 cm long, 3.4 cm wide, 2.8 cm thick, pyriform, compressed, when dried the apex brown, but yellowish below, the sides smooth, shining, gently curved, free in upper V3, the apex low convex, lateral sutures none (from the direction of the basal fibers and from the remnant of a shoulder, it is apparent that when fresh the basal third had fleshy en- largements); central apical sinuses 1.5-2. 5 mm deep, broad; carpels 6, the apices very low, de- pressed pyramidal, the inner ones slightly the smaller; stigmas 2-2.5 mm wide, reniform, dark brown, papillose, centripetal and mostly oblique, some of the marginal ones with the apex trun- cate by an overhanging visor which terminates a distal concavity; proximal sinus running Vz way to valley bottom; endocarp median, 3 cm long, bony, massive, dark mahogany-colored, the lateral walls 3 mm thick, the inner surfaces shining; seeds 12 mm long, 3-4 mm in diam- eter, ellipsoid; upper mesocarp forming in the apex of each carpel a cavern with few fibers but filled with aerenchyma of white medullary membranes; basal mesocarp fibrous and fleshy. EXPANDED DESCRIPTION FROM OTHER SPECI- MENS: Branch tips 4.5-5 cm in diameter, brown, with crowded leaf scars; leaves 1-1.64 m long, 3. 5- 4.5 cm wide, at midsection with 39—45 parallel secondary veins in each half, near the base the marginal prickles 3-5 mm long, those of the midrib below 3-5 mm long; peduncle terminal, 22 cm long, 15 mm in diameter, 3 -sided, leafy bracted, bearing a single syncarp, this 16 cm long, 14 cm in diameter, broadly el- lipsoid, bearing numerous phalanges; phalanges 5.5- 6 cm long, 2-3.4 cm wide, 2-2.8 cm thick, pyriform to wedge-shaped, compressed; carpels 6-10. STANDARD specimen: Singapore: edge of "dry” mangrove swamp, April 17, 1956, F. R. Fosberg 36,919 (us). SPECIMENS EXAMINED: Singapore, Sungei Gunong, Feb. 1893, H. N. Rlidley'} 3,003 (sing); Singapore, Dec. 15, 1904, W. Fox 12,373 (sing). DISCUSSION: P. Boryi Gaud, is a member of the section Pandanus. The type specimen is a single phalange in the Paris museum, collected by Gaudichaud, and left without data. It was a misfortune that he abandoned his taxonomic work half done and devoted his last years to morphogenesis, producing lengthy papers that are now forgotten. P. Boryi was published with the binomial and with a figure showing a o T 10 ovn. FIG. 158. Pandanus carnosus St. John, from holotype. a, Fresh lateral phalange, lateral view, X 1; b, dried subbasal phalange, lateral view, X lj G phalange, longitudinal median section, X lj d, lateral phalange, apical view, X 1; e, subbasal phalange, apical view, X 1 \ f, apex of carpel in lateral phalange, oblique view, X 4; g, apex of carpel in subbasal phalange, oblique view, X 4; h, leaf base, lower side, X lj h leaf middle, lower side, X 1; j, leaf apex, lower side, X 1- 20 PACIFIC SCIENCE, VoL XVII, January 1963 lateral view of a phalange, nothing more. Since the genus existed, this was valid publication. Brongniart reviewed the Pandanus specimens collected by Gaudichaud and documented them as far as possible (Ann. Sci. Nat. Bot. VI, 1: 290, 1875). For this species he postulated that it came from He de la Reunion. However, in the recent detailed revision of the Mascarene species (Linn. Soc. Lond. Bot. Jour. 55: 1-32, 1953) Vaughan and Wiehe found no local species to match it. It is of interest, then, to note that the recent collection from Singapore, Posher g 36,919, is a perfect match for Gaudichaud’s illustration of P. Boryi . This collection consists of one phalange and one leaf minus the tip, so it is scarcely more adequate than the Gaudichaud collection. Since it is a perfect match for the Gaudichaud figure, we offer here a description of this phalange and leaf^as a standard description to supplement the minimal publication by Gaudichaud. In the Singapore herbarium there are also two collec- tions from Singapore that fall into this same ■species. One has six phalanges and two' leaves minus the tips; the other had a branch, many leaf bases, peduncle, core of the syncarp, and eight phalanges. Together they contribute to our knowledge of the population. No one of them furnishes all the details, but of all, the Fosberg 36,919 seems best to take as a standard. The vessel "La Bonite,” on which Charles Gaudichaud was the botanist, circumnavigated the globe in 1836 and 1837, and stopped at Singapore. Pandanus is still fairly common on Singapore Island and occurs there in several spe- cies. One of the tidal estuaries not far from the harbor is Sungei Pandan, and the general ver- nacular name of the genus in Malayan is "pan- dan.” This was the source of our generic name, Pandanus, latinized and first published by Rum- phius, later adopted and made valid by Stickman. Pandanus carnosus sp. nov. (sect. Pandanus) Fig. 158 DIAGNOSIS HOLOTYPI: Arbor 10 m alta 15 cm diametro, cortice pallide brunneo sparse muri- cato, radicibus fulturosis 1-2 m longis 3-3.5 cm diametro pallide brunneis in lineis muriculatis, foliis 1.8-2 .4 m longis 5-5.2 cm latis supra olivaceo-viridibus et subglaucis infra pallide vir- idibus coriaceis cum midnervo forti et in centra late sulcatis in sectione mediali cum 45 nervis parallelis secundariis in quoque dimidio' nervis tertialis nullis laminis gladiformatis cum apici- bus 15-30 cm longis subulatis trigonis eo 10 cm ex apice 0.9- 1.3 mm lato basi amplexicaule et inermi sed ex 10 cm marginibus cum spinis 5- 6.5 mm longis 11-30 mm separatis valide subulatis arcuatis adscendentibus pallidis, mid- nervo infra ex 22 cm cum spinis 6 mm longis 15-35 mm separatis simulantibus sed reflexis, in sectione mediali marginibus cum aculeis 3.5- 4 mm longis 11-18 mm separatis subulatis valde adpresse adscendentibus apicibus rubris, mid- nervo infra cum aculeis 3 mm longis 15-25 mm separatis simulantibus adscendentibus, in apice subulate marginibus et midnervo infra cum subulato-serris 0.5-0.8 mm longis 2-11 mm separatis, syncarpio solitario terminali 18 cm longo 1 5 cm diametro subgloboso cum phalangi- bus numerosis eis 6.7-73 cm longis 2.7-3. 5 cm latis 2.4-3. 1 cm crassis apice in parte viridi in parte rubro-aurantiaco 5-6-anguloso, parte supera libera oblongo-elliptica vel oblongo- obovoideo apice depresso-convexo lateribus sub- curvatis rugulosis sed sublucidis, suturis laterali- bus nullis, sinibus centralibus apicalibus 4.5- 5.5 mm profundis et V-formatis in apicibus anguste conicis (sed in phalangibus paucis proba- baliter eis subbasalibus apicibus demissiter pyra- midalibus sinibus 2-3.5 mm profundis), carpelis 6- 10 apicibus plerumque anguste conicis eis marginalibus lA latioribus asymmetricalibus et plus minusve distaliter compressis erectis vel subdivergentibus, parte lA inf era phalangis carnoso-incrassata 3-4 mm et deinde 25-35 mm diametro, stigmatibus 3-3.5 mm longis ellipticis sulcatis brunneis papillosis, sinibus proximalibus V5-V2 ad fondam extensis, endocarpio medi- ali 2.5-3 cm longo mahogani-brunneo osseoso solido lateribus 3-4 mm crassis intra lucidis, se.mi.oi bus 13-18 mm longis 4.5 mm diametro fusiformibus, mesocarpra supero in apice car- pel! quique cavernam formanti eis lateralibus 2.5 cm longis cum membranis albis medullosis, mesocarpio basal! fibroso et carnoso. DESCRIPTION OF ALL SPECIMENS EXAMINED: Tree 10 m tall, 15 cm in diameter; bark light brown, sparsely muricate; prop roots 1-2 m long, 3-3.5 cm in diameter, pale brown, muricu- Page 246: Revision of Pandanus, 14. Malaya and Singapore — St. John 21 late in longitudinal lines; leaves 1. 8-2.4 m long, 4.4- 5. 2 cm wide, above olive green and some- what glaucous, below paler green, coriaceous, with a strong midrib and above it broadly chan- neled, at midsection with 45 secondary parallel veins in each half, no visible tertiary cross veins, sword-shaped, gradually tapering to the 15-30 cm subulate trigonous apex, this 10 cm down 0.9-1. 3 mm wide, the base amplexicaul and unarmed, but beginning at 10 cm the margins with spines 5-6.5 mm long, 11-30 mm apart, stout subulate, arcuate ascending, pale; the mid- rib below beginning at 22 cm with spines 6 mm long, 15-35 mm apart, similar but reflexed; at midsection the margins with prickles 3.5-4 mm long, 11-18 mm apart, subulate, closely ap- pressed ascending, reddish-tipped; the midrib below with similar ascending prickles 3 mm long, 15-25 mm apart; on the subulate apex the margins and midrib below with subulate- serrae 0. 5-0.8 mm long, 2-11 mm apart; syn- carp single, terminal, 18 cm long, 15 cm in diameter, subglobose, bearing numerous pha- langes, these 6.7-73 cm long, 2.7-3. 5 cm wide, 2.4- 3. 1 cm thick, the apex partly green, the remainder red-orange, 5-6-angled, the upper 2A free, oblong-elliptic or oblong-obovoid, the apex low convex, the sides gently curved, rugulose but even so somewhat shining, lateral sutures none; central apical sinuses 4.5-53 mm deep and V-shaped when with narrowly conic apices (but on the occasional phalange, probably a subbasal one, with low pyramidal apices, the sinuses only 2-33 mm deep); carpels 6-10, the apices usually narrowly conic, the outer ones of the same thickness but about lA wider, asym- metric and more or less flattened distally, erect or slightly outwardly curved; the lower of phalange fleshy distended with shoulders dis- tended to 3-4 mm all around, thus 25-35 mm in diameter; stigmas 3-33 mm long, elliptic, creased, brown, papillose; proximal sinus run- ning Vz-Vz way to valley bottom; endocarp median, 23-3 cm long, dull mahogany-brown, bony, massive, the lateral walls 3-4 mm thick, the inner surfaces shining; seeds 13-18 mm long, 43 mm in diameter, fusiform; upper meso- carp forming in each carpel apex a cavern, the lateral ones 23 cm long, filled with the white membranes of an aerenchyma; basal mesocarp fibrous and fleshy. HOLOTYPUS: Singapore, Kranji village, cut- over thicket near house, 2 m. alt., July 10, I960, H. St. John 26,375 (bish). specimens EXAMINED: Malaya, Pahang, Te- lok Sisik, Kuantan, near the sea, 4 Dec. 1924, I. H. Bur kill & Md . Haniff 17,346 (sing). DISCUSSION: P. carnosus is a member of the section Pandanus . It is one of the curious Asiatic species with fleshy, enlarged bases of the pha- langes. None of the described species are close relatives, but it is related to an undescribed one from Okinawa, represented by the collection St. John 25,950. The new epithet is the Latin adjective car- nosus, fleshy, in allusion to the fleshy base of the phalange. Pandanus glohosus sp. nov. (sect. Pandanus) Fig. 159 DIAGNOSIS HOLOTYPI: Arbor ? ramis 3 cm diametro brunneis rugosis, foliis 81-90 cm longis 33 cm latis supra viridibus infra pallidi- oribus et apparente glands late sulcatis coriaceis in sectione mediali cum 38 nervis parallelis secundariis in quoque medio nervis tertialis nul- lis laminis gladiformatis sensim in apice 20 cm longo subulato trigono diminuentibus eo 10 cm ex apice 2 mm lato basi amplexicauli inermi sed ex 5 cm marginibus cum aculeis 33-5 mm longis 5-12 mm separatis arcuatis subulatis ad- scendentibus apicibus brunneis, midnervo infra ex 5 cm cum aculeis 2-3 mm longis 6-18 mm separatis simulantibus sed reflexis, in sectione mediali marginibus cum aculeis 2.8-3 mm longis 5-12 mm separatis subulatis valde adpressis adscendentibus, midnervo infra cum aculeis 0.8-1 mm longis 5-12 mm separatis subulatis adscendentibus, in apice subulato marginibus et midnervo infra cum aculeis 03-0.8 mm longis 4-8 mm separatis subulatis adpresse adscenden- tibus, inflorescentia foeminea terminal! cum syncarpio solitario, pedunculo 9 cm longo 8 mm diametro trigono folioso-bracteato, syncarpio 8.8 cm diametro globoso cum 26 phalangibus eis 3. 6-3 .9 cm longis 23-2.8 cm latis 2.2-23 cm crassis breviter crasse pyriformatis 6-angulosis lateribus planis vel subcurvatis laevibus lucidis in sicco pallide brunneis parte 2A supera libera apice depresse convexo, suturis lateralibus nul- lis, sinibus apicalibus centralibus 2-33 mm pro- "I 10 cm. scm. Fig. 159. Pandanus globosus St. John, from holotype. a, Syncarp, X Vl\ b, phalange, lateral view, X 1; c, phalange, longitudinal median section, X 1; d, phalange, apical view, X 1; e, carpel apex, stigma, and prox- imal sinus, oblique view, X 4; f, leaf base, lower side, X 1; S> leaf middle, lower side, X 1; ^ leaf apex, lower side, X 1- I — 1 — — — » Fig. 160. Pandanus inclinatus St. John, from holotype. a, Habit, X 1/200; b, phalange, lateral view, X 1; c, phalange, longitudinal median section, i< 1; d, phalange, apical view, X 1; e> f> carpel apices, stigmas, and proximal sinuses, oblique view, X 4; g, bark with adventitious rootlets, X 1- o.2 mm.. o 5cm. 24 PACIFIC SCIENCE, Vol. XVII, January 1963 fundis late V-formatis rectis vel subcurvatis, carpelis 5-6 apicibus conicis angulatis basi ro- tundato apice rugoso eis centralibus quando sunt minoribus, stigmatibus 2.5-3 mm longis cordato- orbicularibus vel ellipticis brunneis papillosis prominentibus obliquis centripetalibus, sinibus proximalibus Vi —2A ad fondam extentis, endo- carpio minime submediali brunneo osseoso la- teribus 1.5-2 mm crassis intra obscure brunneis lucidis, seminibus 10-14 mm longis 3.5-4 mm diametro ellipsoideis vel obliquiter ellipsoideis, mesocarpio supero in apice quaque carpellae cavernam cum fibris paucis et membranis pal- lidis medullosis formanti, mesocarpio basali fib- roso et carnoso. diagnosis OF HOLOTYPE: Apparently tree- like; branch 3 cm in diameter, brown, roughened by crowded leaf scars; leaves 81-90 cm long, 3.5 cm wide, green above, below paler green and apparently glaucous, with a broad V-shaped furrow above the midrib, coriaceous, at mid- section with 38 secondary veins in each half, no visible tertiary cross veins, sword-shaped, taper- ing gradually to the 20 cm subulate, trigonous apex, this 10 cm down 2 mm wide, the base am- plexicaul, unarmed, but beginning 5 cm up the margins with prickles 3.5-5 mm long, 5-12 mm apart, arcuate subulate, fully ascending, brown- tipped; the midrib below beginning at 5 cm with prickles 2-3 mm long, 6-18 mm apart, similar but reflexed; at midsection the margins with prickles 2.8-3 mm long, 5-12 mm apart, subulate, flat appressed, ascending; the midrib below with prickles 0.8-1 mm long, 5-12 mm apart, subulate, ascending; on the subulate tip the margins and midrib below with subulate, appressed ascending prickles 0. 5-0.8 mm long, 4-8 mm apart; pistillate inflorescence terminal, bearing one syncarp; peduncle 9 cm long, 8 mm in diameter, 3 -sided, leafy bracted; syncarp 8.8 cm in diameter, globose, bearing about 26 pha- langes, these 3-6— 3-9 cm long, 2. 5-2.8 cm wide, 2.2-2. 5 cm thick, short, thick pyriform, 6-angled, the sides plane or gently curved, smooth, shin- ing, when dried light brown, upper 2A free, the apex low convex; lateral sutures none; central apical sinuses 2-3.5 mm deep, wide V-shaped, straight or gently curved; carpels 5-6, the apices with a rounded base and conic, angled, wrinkled tip, the central ones, when present, slightly the smaller; stigmas 2.5-3 mm long, cordate- orbicular to -elliptic, brown, papillose, prom- inent, oblique, centripetal; proximal sinus running l/2-2A way to valley bottom; endocarp' slightly inframedian, brown, bony, the lateral walls 1.5-2 mm thick, the inner surfaces pol- ished, dark brown; seeds 10-14 mm long; 3.5-4 mm in diameter, ellipsoid or obliquely so; upper mesocarp forming in the apex of each carpel a cavern with a few fibers and with pale, medul- lary membranes; basal mesocarp fibrous and fleshy. HOLOTYPUS: Malaya, Kedah, w. coast, P. Dayang Banting, Pfulau] Langkawi, sea level, on limestone, 27 Nov. 1934, M. R. Henderson 29J33 (SING). SPECIMENS EXAMINED: Borneo: Karimata groep, Poelau, Karimata, 23/3/1931, Mondi 133 (BO). Celebes: Kota Menado, strand, 31 /XII/ 1894, Ko orders 1 8,462 (bo). DISCUSSION: P. globosus is a member of the section Pandanus, as is its closest relative, the Vietnamese species P. subulatus St. John, a spe- cies which has the syncarp suborbicular-ellipsoid, 10-10.5 cm in diameter; carpel apices pyramidal or oblate-pyramidal; stigmas 1.5-2 mm long, black; endocarp supramedian; the leaf apex at the point 10 cm down 3.5 mm wide; the mar- gins near the base with spines 5-6 mm long, ascending at 45°; P. globosus has the syncarp globose, 8.8 cm in diameter; carpel apices low convex; stigmas 2.5-3 mm long, brown; endo- carp inframedian; leaf apex at the point 10 cm down 2 mm wide; the margins near the base with prickles 3.5-5 mm long, fully ascending. The new epithet is the Latin adjective globo- sus, spherical, and is given in reference to the spherical head of fruits. Pandanus inclinatus sp. nov. (sect. Pandanus). Figs. 160, 170c diagnosis HOLOTYPI: Arbor 8 m alta 13 cm diametro, cortice pall id e brunneo muricato et cum radicillis adventivis adpresse adscendentibus ad 1 cm longis, radicibus fulturosis nullis, foliis- 1. 5-1.7 m longis 4.4-4.7 cm lads coriaceis supra obscure olivaceo-viridibus infra viridibus spinis omnibus albis cum apicibus brunneis lamina 1-sulcatis et lateribus semirevolutis in sectione Page 250: Revision of Pandanus , 14. Malaya and Singapore — St. John 25 mediali cum 38-40 nervis parallelis secundariis in quoque dimidio, nervis tertialis nullis, lamina gladiformi in apice caudato trigono subulato 15-20 cm longo diminuenti eo 10 cm ex apice 1 mm lato, basi amplexicauli et inermi sed ex 3- 7 cm marginibus cum aculeis 3.5-6 mm iongis 10-30' mm separatis crassiter subulatis subarcu- atis adscendentibus, midnervo infra ex 7-8 cm cum aculeis 3-5 mm Iongis 20-50 mm separatis crassiter subulatis valde reflexis, in sectione mediali marginibus cum aculeis 3-3.5 mm iongis 7-17 mm separatis crassiter subulatis ad- scendentibus valde adpressis, midnervo infra cum aculeis 2.5-3 mm Iongis 8-23 mm separatis subulatis adscendentibus, in apice subulato mar- ginibus et midnervo infra cum subulato serrulis 0.5-1 mm Iongis 5-10 mm separatis vel in loculis nullis, inflorescentia foeminea terminali cum syncarpio uno, pedunculo 30 cm longo 13- 15 mm diametro trigono folioso-bracteato, syn- carpio 16-23 cm longo 13-20 cm diametro cum 44-61 phalangiis eis 6-6.8 cm Iongis, climb dio supero 3.2— -4.3 cm lato 23-3.7 cm crasso oblongo-obovoideo truncate vel subtmncato aurantiaco-rubro apice subviridi 5-7-anguloso lateribus subcurvatis vel subplanis laevibus et lucidis, suturis lateralibus nullis, sinibus centrali- bus apicalibus 0.7-2 mm profundis rectis vel subcurvatis, dimidia infera phalangis turbinata carnosa humeris carnosis 3-6 mm latis carne subsucrato comes to aurantiaco-rubro, carpelis 6-12 plerumque 7-9 (sed in gemine 18 et pha- langio 5.5 cm lato 4.4 cm crasso) interioribus paene minoribus, apicibus planatis sed pauce oblato-pyramidalibus, stigmatibus 2-3 mm Ion- gis eilipticis vel late ellipticis obliquis brunneis sulcatis pluribus rnarginalium truncatis ex piano oblique et cum concavite parvo distali, sinibus proximalibus profundis Vi-Vd ad fondam ex- tent is, endocarpio 3 cm longo mediali osseoso obscure mahogani-colorato intra lucido lateribus 2.5 mm crassis, seminibus 14-17 mm Iongis 4- 5.5 mm diametro ellipsoideo, mesocarpio supero in quoque carpelo cavernam cum fibris panels et aerenchyma cum membranis albis formanti, mesocarpio basali fibroso et carnoso. DIAGNOSIS OF holotype: Tree, 8 m tall, 13 cm in diameter; bark light brown, muricate, and with a few appressed ascending rootlets up to 1 cm in length; prop roots none; leaves 1.5- 1.7 m long, 4.4-47 cm wide, coriaceous, above dark olive green, below green, the spines all white with minute brown tips, blade broad fur- rowed above the midrib, the sides down curved, at midsection with 38-40 secondary parallel veins in each half, no tertiary cross veins, sword- shaped, tapering upwards into a 15-20 cm sub- ulate, trigonous, caudate apex, this 10 cm down 1 mm wide, the base amplexicaul and unarmed, but starting at 3-7 cm up the margins with prickles 3.5-6 mm long, 10-30 mm apart, stout- subulate, slightly arcuate, ascending; the midrib below beginning at 7-8 cm with prickles 3-5 mm long, 20-50 mm apart, stout subulate, sharply reflexed; at midsection the margins with prickles 3-3.5 mm long, 7-17 mm apart, stout subulate, ascending, closely appressed; the mid- rib below with prickles 2.5-3 mm long, 8-23 mm apart, subulate, ascending; on the subulate apex the margins and midrib below with sub- ulate serrulations 0.5-1 mm long, 5-10 mm apart or even lacking for distances on the mar- gins; pistillate inflorescence terminal, bearing one syncarp; peduncle 30 cm long, 13-15 mm in diameter, 3 -sided, leafy bracted; syncarp lb- 23 cm long, 13-20 cm in diameter, bearing 44-61 phalanges, these 6-6.8 cm long, the upper half 3.2-4 .3 cm wide, 2.3-37 cm thick, oblong obovoid, truncate or subtruncate, orange-red, the color becoming paler upwards to the partly greenish apex, 5-7-angled, the sides gently curved or almost plane, smooth and shining; lateral sutures none; central apical sinuses 07-2 mm deep, the bottom straight or gently curved, V-shaped, then immediately wide flaring; lower half of phalange turbinate above the truncate base, fleshy enlarged, the shoulders 3-6 mm wide, the flesh slightly sweet, nonirritating, orange-red; carpels 6-12, mostly 7-9 (but in a double 18 and this one 5.5 cm wide, 4.4 cm thick), the central ones only slightly the smaller, the apices flattened but perceptibly oblate pyra- midal; stigmas 2-3 mm long, elliptic or broadly so, oblique, brown, creased, many of the mar- ginal ones truncate by the visor-like edge of a distal oblique plane terminating in a small con- cavity; proximal sinus deep, running Vi-Vi way to valley bottom; endocarp median, 3 cm long, bony, dark mahogany-colored, the inner sufaces shining, the lateral wall 2.5 mm thick; seeds 26 PACIFIC SCIENCE, Vol. XVII, January 1963 14—17 mm long, 4-5.5 mm in diameter, ellip- soid; upper mesocarp forming in the apex of each carpel a cavern with a few fibers and an aerenchyma of white medullary membranes; basal mesocarp fibrous and fleshy. HOLOTYPUS: Singapore, Kranji Nature Re- serve, brackish swamp with Sonneratia caseola- ris, Hibiscus tiliaceus, Acrostichum aureum, July 10, I960, H. St. John 26,378 (bish). DISCUSSION : P. inclinatus is a member of the section Pandanus, as is its closest relative, P. vietnamensis St. John, of Vietnam, a species with the syncarp 27 cm long, with 106 phal- anges, these 5.8-6 cm long, the apex low con- vex; stigma 3.5-5 mm long or wide, black; leaves 0.99-1.49 m long, 5-6 cm wide, the mar- gin near the base with spines 4-11 mm long, the nearby midrib below with spines 6-8 mm long. P. inclinatus has the syncarp 16-23 cm long, with 44-6 1 phalanges, these 6-6.8 cm long, the apex truncate or subtruncate; stigmas 2-3 mm long, brown; leaves 1.5-1. 7 m long, 4.4-4.7 cm wide, the margin near the base with prickles 3.5-6 mm long, and the nearby midrib below with prickles 3-5 mm long. The new epithet is the Latin participle in- clinatus, inclined, given in reference to the ob- lique stigmas. Pandanus incrassatus sp. nov. (sect. Pandanus) Figs. 161, 171 DIAGNOSIS HOLOTYPI: Arbor 7-8 m alta 8- 10 cm diametro, cortice griseo remote muricato in ramis vetustioribus et trunco cum radicillis adventivis paucis adpresse adscendentibus, radi- cibus fulturosis 5-10 dm longis 10-13 mm diametro pallide brunneis cum radicillis brevi- bus spiniformatis in lineis, foliis 1.45-1.7 m longis 4.2-4.3 cm latis coriaceis supra olivaceo- viridibus infra viridibus et glaucis spinis albis deinde apicibus brunneis, laminis 1-sulcatis et cum plicis duobus rotundatis in sectione mediali cum 34-35 nervis secundariis in quoque medio nervis tertialis nullis gladiformatis sensim in apice 30 cm longo subulato trigono diminuen- tibus eo 10 cm ex apice 1.1 mm lato, basi arn- plexicauli et inermi albo sed ex 7—9 cm mar- ginibus cum spinis 3.5-5 mm longis 15-25 mm separatis arcuatis crassiter subulatis adscenden- tibus, midnervo infra ex 9-12 cm cum spinis 3-4 mm longis 20-28 mm separatis valide sub- ulatis reflexis, in sectione mediali marginibus cum aculeis 2.5-3 mm longis 10-20 mm sepa- ratis arcuatis subulatis adpresse adscendentibus, midnervo infra cum aculeis 1.5-2 mm longis 12-20 mm separatis arcuatis subulatis adscen- dentibus, in apice subulato marginibus et mid- nervo infra cum subulato-serrulis 0. 5-0.7 mm longis 3-5 mm separatis, inflorescentia foeminea terminali cum syncarpio solitario pendenti, pedunculo 23 cm longo 1.5 cm diametro trigono folioso-bracteato, syncarpio 15 cm longo 14.5 cm diametro subgloboso cum 54 phalangibus eis 4.6-5 .2 cm longis media sup era 2.5-3. 1 cm longa, 3-3.6 cm lata 2.3-3 cm crassa obovoidea plano-truncata viridi deinde rubro-aurantiaca 5- 6-angulosa lateribus subcurvatis infra laevibus et lucidis supra rugulosis et sublucidis, suturis lateralibus nullis, sinibus apicalibus centralibus plerumque (et in phalangibus lateralibus api- calibusque omnibus) 1-2 mm profundis late V- formatis rectis vel subcurvatis (vel in phalangi- bus subbasalibus paucis 2-3 mm profundis), media infera phalangis carnoso-incrassata 2.8- j aurantiaco-rubra infra pallidiori, humeris 2-5 mm latis undulatis carne saccharosa sed non consumpta, carpelis 5-8 centralibus l/2~2A tarn j, grandibus quam marginalibus, phalangibus di- | midii superi cum apicibus carpeli subtruncatis j vel oblatiori-pyramidalibus sed eis dimidii inferi cum apicibus oblato-pyramidalibus, stigmatibus j 2-3 mm longis ellipsoideis obscure brunneis papillosis sulcatis obliquis prominentibus centri- petalibus eis marginalibus plerumque cum supra stigmatem galea distali cartilaginea, sinibus proximalibus Vi~2A ad fondam extentis, endo- carpio 2.5 cm longo mediali osseoso obscure brunneo intra lucido lateribus 2.5 mm crassis, seminibus 12 mm longis 4 mm diametro ellip- soideis, mesocarpio supero in apice quique car- peli cavernam cum membranis albis lucidis medullosis formanti, mesocarpio basali fibroso et carnoso. DIAGNOSIS OF HOLOTYPE: Tree 7-8 m tall, 8-10 cm in diameter; bark gray, remotely muri- cate, and the older trunk with a few appressed ascending adventitious rootlets; prop roots 5-10 dm long, 10-13 mm in diameter, light brown, with remote, short, spinelike rootlets in vertical Fig. 161. Pandanus incrassatus St. John, from holotype. a, Habit, X 1/100; b, dried phalange, lateral view, X 1; c, phalange, longitudinal median section, X 1| ^ phalange, apical view, X 1; ^ carpel apex, stigma, and proximal sinus, oblique view, X 4; /, prop root with spinelike rootlets, X lj S> leaf base, lower side, X 1> h, leaf middle, lower side, X 1; h leaf apex, lower side, X 1- 5 vnw,. 28 PACIFIC SCIENCE, Vol. XVII, January 1963 rows; leaves 1.43-1.7 m long, 4.2-4.3 cm wide, coriaceous, above olive green, below green and glaucous, the spines white, becoming brown- tipped, 1 -ribbed and with two rounded pleats, at midsection with 34-35 secondary parallel veins in each half, no visible tertiary cross veins, sword-shaped, long tapering to a 30 cm subu- late trigonous apex, this 10 cm down 1.1 mm wide, the base amplexicaul and unarmed, white, beginning at 7-9 cm the margins with spines 3.5- 5 mm long, 15-25 mm apart, arcuate, heavy subulate, ascending; the midrib below begin- ning at 9-12 cm with spines 3-4 mm long, 20— 28 mm apart, stout subulate, reflexed; at mid- section the margins with prickles 2.5-3 mm long, 10-20 mm apart, arcuate subulate, ap- pressed ascending; the midrib below with prickles 1.5-2 mm long, 12-20 mm apart, arcu- ate subulate, ascending; near the apex the mar- gins and midrib below with subulate serrula- tions 0.5-0. 7 mm long, 3-5 mm apart; pistillate inflorescence terminal, bearing 1 pendent syn- carp; peduncle 23 cm long, 1.5 cm in diameter, 3 -sided, leafy-bracted; syncarp 15 cm long, 14.5 cm in diameter, subglobose, bearing 54 phal- anges, these 4.6-5. 2 cm long, the upper half 2.5- 3. 1 cm long, 3-3.6 cm wide, 2.3-3 cm thick, obovoid, flat truncate, green, turning red- orange, the shade deepening downwards, 5-6- angled, the sides gently curving, below smooth and shining, above rugulose and less shiny; lateral sutures none; central apical sinuses mostly (that is, for the lateral and apical phal- anges) 1-2 mm deep, wide V-shaped, straight or gently curving (on a few subbasal ones 2-3 mm deep); lower half of phalange fleshy en- larged, 2. 8-3. 6 cm wide, 2.4-3. 1 cm thick, wide cuneate, vermilion above, gradually paling be- low, the shoulders 2—5 mm wide, scalloped; pulp sweetish, innocuous, but not eaten; carpels 5-8, the central 1-2 being Vl-2A the size of the marginal ones; on the phalanges of the upper half the carpel apices subtruncate or very oblate pyramidal, on those of the lower half the carpel apices oblate-pyramidal; stigmas 2-3 mm long, ellipsoid, dark brown, papillose, creased, oblique, prominent, centripetal, the marginal ones mostly with a distal, cartilaginous visor partly project- ing over the stigma; proximal sinus running Vl—Vi way to valley bottom; endocarp median 2.5 cm long, bony, dark brown, the inner sur- faces shining, the lateral walls 2.5 mm thick; seeds 12 mm long, 4 mm in diameter, ellipsoid; upper mesocarp forming a cavern in each carpel apex with white, shining, medullary mem- branes; basal mesocarp fibrous and fleshy. holotypus: Singapore, Kranji Nature Re- serve, mangrove swamp, with Sonneratia caseo- laris, Hibiscus tiliaceus, Acanthus ilicifolius, near sea level, 7 July I960, H. St. John 26,374 (BISH). DISCUSSION: P. incrassatus is a member of the section Pandanus, as is its closest relative, the Vietnamese species P. reversispiralis St. John, a species with the syncarp oval-subglo- bose; phalanges 4.3-4.6 cm long, orange, the apex low convex; stigmas all terminal, exposed; leaves 70-130 cm long, 4.8-6 cm wide, slightly glaucous above, at midsection the margins with prickles 3—4 mm long. P. incrassatus has the syncarp subglobose; phalanges 4.6-5. 2 cm long, red-orange, the apex truncate; marginal stigmas covered by a visor; leaves 145-170 cm long, 4.2-4. 3 cm wide, olive green, not glaucous above, and at midsection the margins with prickles 2.5-3 mm long. The new epithet is the Latin adjective incras- satus, thickened, and is given with allusion to the thickened base of the phalange. Pandanus ohtusus sp. nov. (sect. Pandanus ) Fig. 162 DIAGNOSIS HOLOTYPI: Arbor ? foliis 85-88 cm longis 3.4 cm latis coriaceis 1-sulcatis lateri- bus arcuatis in sectione mediali cum 37-39 nervis parallelis secundariis in quoque medio eis infra conspicuis, nervis tertialis nullis, lami- nis gladiformatis sensim in apice 15 cm longo subulato trigono diminuentibus eo 10 cm ex apice 2.5 mm lato, basi amplexicauli et inermi sed ex 5 cm marginibus cum aculeis 2.5-4 mm longis 8-14 mm separatis arcuatis subulatis pal- lidis in 45° adscendentibus, midnervo infra ex 8.5 cm cum aculeis 3.5-4 mm longis 12-20 mm separatis subulatis valde reflexis basibus crassiter conicis, in sectione mediali marginibus cum aculeis 3-3.5 mm longis 6-11 mm separatis arcuatis subulatis adscendentibus apicibus ru- bris, midnervo infra angusto salienti cum acu- leis 1.5-2 mm longis 7-10 mm separatis arcu- o 5 cm. Fig. 162. Pandanus obtusus St. John, from holotype. a, Syncarp, X Vl\ b, phalange, lateral view, X 1; i phalange, longitudinal median section, X 1; d; phalange, apical view, X 1; leaf apex, lower side, X 1. smm. 30 PACIFIC SCIENCE, Vol. XVII, January 1963 atis subulatis adscendentibus, in apice subulato marginibus et midnervo infra cum subulato- serrulis 0.3-07 mm longis 3-7 mm separatis, inflorescentia foeminea cum syncarpio unico terminali, pedunculo 23 cm et plus longo cla- vato in apice 11 mm diametro subtrigono, syn- carpio globoso ? 12 cm diametro cum circa 38 phalangibus eis 5-5.5 cm longis 3-3.8 cm latis 27-3.3 cm crassis cuneatis 5-6-angulosis in sicco pallide brunneis lateribus laevibus lucidis subcurvatis vel subplanis parte l/$ supera libera apice convexo vel alte convexo suturis lateral i- bus nullis, sinibus centralibus apicalibus olim lineas brunneas, carpelis 8-10 apicibus rotun- datis interioribus minoribus et l/z tarn grandibus quam rnarginalibus, stigmatibus 2-2.5 mm longis cordatis vel reniformibus obliquis centri- petalibus, sinibus proximalibus brevibus lA~lA ad fondam extentis, endocarpio mediali 2.5-3 cm longo osseoso extra brunneo intra pallidi lateribus 2 mm crassis, seminibus 12 mm longis 3.5 mm diametro obliquiter ovoideis, mesocar- pio supero in apice carpello quoque cavernam formanti eis lateralibus majoribus cum fibris longitudinalibus et membranis albis medullosis, mesocarpio basali fibroso et carnoso. DESCRIPTION OF ALL SPECIMENS EXAMINED: Probably treelike; leaves 85-130 cm long, 3 A cm wide, coriaceous, furrowed above the mid- rib, the two halves gently arching, at midsec- tion with 21-39 secondary parallel longitudinal veins in each half, these conspicuous below, no visible cross veins, blade swordlike, tapering upwards and gradually narrowed to a 15 cm subulate, trigonous apex which 10 cm down is 2.5 mm wide, the base amplexicaul and unarmed, but beginning at 5 cm the margins with prickles 2.5-4 mm long, 8-14 mm apart, arcuate subulate, pale, ascending at 45°; the midrib below beginning at 8.5 cm with prickles 3. 5-4. 5 mm long, 12-20 mm apart, the base heavy conic, the rest subulate, fully reflexed; at midsection the margins with prickles 3-3.5 mm long, 6-11 mm apart, arcuate subulate, ascend- ing, reddish-tipped; the midrib below narrow and salient, with prickles 1.5-2 mm long, 7-10 mm apart, arcuate subulate, ascending; on the subulate apex the margins and midrib below with subulate-tipped serrations 0.3-07 mm long, 3-7 mm apart; pistillate inflorescence with a single, terminal syncarp; peduncle more than 23 cm in length, clavate and at apex 11 mm in diameter, somewhat 3 -sided; syncarp apparently globose and 12 cm in diameter, and bearing about 38 phalanges, these 5-5.5 cm long, 3-3.8 cm wide, 27-3.3 cm thick, cunei- form, 5-6 angled, when dried pale brown, the sides smooth and shiny, gently curving or flat- tish, free in upper Vd, the apex convex or high convex, lateral sutures none, central apical siT nuses mere brown lines on the obtuse apex; carpels 8-10, the apices rounded, not raised, the inner ones smaller, about Vi the size of the marginal ones; stigmas 2-2.5 mm long, cordate to reniform, oblique, centripetal; proximal sinus a very short crack or hole by the stigma, run- j ning V4-V3 way to the valley bottom; endocarp j median, 2.5-3 cm long, bony, pale within, dark j brown near the margins, the seed cavities shin- j ing, the lateral walls 2 mm thick; seeds 12 mm j long, 3.5 mm in diameter, obliquely ovoid; upper mesocarp forming in the apex of each carpel a cavern, the lateral ones thrice the larger, with longitudinal fibers and white medullary I membranes; basal mesocarp fibrous and fleshy. HOLOTYPUS: Malaya, Pahang, Beserah, Kuan- tan, sea level, 18 Aug. 1929, M. R, Henderson 22,746 (sing): SPECIMENS EXAMINED: Singapore, 15 Feb. 1870, J. S. Gfoodenough}, (SING). This has leaves only 2 cm wide, and with 21 secondary veins on a side, but the phalanges seem iden- tical. DISCUSSION: P. obtusus is a member of the section Pandanus. In that section the most simi- lar species seems to be a Marquesan one, P, tectorius Soland. var. uapensis F. Br., which has the phalanges 5.4-6 cm long, 4.5 cm wide, the apical and lateral sutures very shallow or super- ficial; apical central sinuses 1 mm deep; stigmas 3 mm long; endocarp supramedian, but the leaves and syncarp are unknown. P. obtusus has the phalanges 5-5.5 cm long, 3-3.8 cm wide, the lateral sutures none, the apical central si- nuses mere tessellate rulings; stigmas 2-2.5 mm j long; and the endocarp median. The new epithet is the Latin word obtusus , j blunt or obtuse, and is given with reference to the obtuse phalange and carpel apices. Fig. 163. Pandanus rubricoloratus St. John, from holotype. a, Syncarp, X V2; dried phalange, lateral view, XI \ c, phalange, longitudinal median section, X 1; d, phalange, apical view, X 1; carpel apex, stigma, and proximal sinus, oblique view, X 4; f, leaf base, lower side, X 1; ^ leaf middle, lower side, X 1; ^ leaf apex, lower side, X 1- 32 PACIFIC SCIENCE, Vol. XVII, January 1963 Pandanus rubric olor atm sp. nov. (sect. Pan- danus) Fig. 163 DIAGNOSIS HOLOTYPI: Arboriformis 3-5 m alta, foliis 66-73 cm longis 2.8-3 cm latis cori- aceis supra viridibus infra pallidioribus supra midnervum sulcatis in sectione mediali cum 27 nervis parallelis secundariis in quoque medio nervis tertialis nullis laminis gladiformatis in apice 12-15 cm iongo subulato sensim diminu- entibus eo 10 cm ex apice 4 mm lato basi am- plexicauli et inermi sed ex 3—4 cm marginibus cum aculeis 2.5-5 mm longis 4-18 mm sepa- rates subulatis pallidis adscendentibus, mid- nervo infra ex 8-10 cm cum aculeis 2.5-3 mm longis 16-30 mm separatis simulantibus sed adpresse reflexis, in sectione mediali marginibus cum aculeis 2-2.5 mm longis 4-12 mm sepa- ratis subulatis adscendentibus apicibus brun- neis, midnervo infra cum aculeis simulantibus 1.5-2 mm longis 6-13 mm separatis; in apice subulato marginibus et midnervo infra cum subulato-serrulis 0.6-1 mm longis 2-6 mm sepa- ratis; inflorescentia foeminea terminal! cum syncarpio solitario, pedunculo 15 cm et plus longo 8 mm diametro exclavatis trigono folioso- bracteato, syncarpio 9 cm longo 8.5 cm diametro globoso cum circa 34 phalangibus, els 3.6 cm longis 2 .4-2 .7 cm latis 2-2.4 cm crassis rubris parte 5/9 supera libera, hac late ellipsoidea 5- 6-angulosa apice depresse convexo lateribus sub- curvatis laevibus suturis lateralibus nullis, sini- bus apicalibus centralibus 0.5-2 mm profundis vadosis, carpelis 6-9 (in gemine 12 et phalangio 3.3 cm lato) apicibus depresse pyramidalibus angulatis rugosis (in phalangibus basalibus apicibus depresse convexis) parte 4/9 inf era carnoso-incrassata, stigmatibus 2-3.5 mm latis cordatis griseis horizontalibus vel obliquis cen- tripetalibus, sinibus proximalibus latis Vl ad fondam extentis, endocarpio mediali 15 mm longo osseoso obscure brunneo lateribus 1.5 mm crassis intra papillosis et sublucidis, semi- nibus 8-11 mm longis obliquiter ellipsoideis, mesocarpio supero in apice quoque carpelo cavernam cum fibris paucis et membranis brun- neis medullosis, mesocarpio basali fibroso et carnoso. diagnosis of holotype: Treelike, 3-5 m tall; leaves 66-73 cm long, 2.8-3 cm wide, coriaceous, green above, paler beneath, fur- rowed above the strong midrib, the sides arch- ing, at midsection with 27 parallel secondary veins in each half, no tertiary cross veins, blade sword-shaped, tapering upwards and gradually tapering to the 12-15 cm subulate apex which 10 cm down is 4 mm wide, the base amplexi- caul and unarmed, but beginning 3-4 cm up the margins with prickles 2.5-5 mm long, 4-18 mm apart, subulate, ascending, pale; midrib below beginning at 8-10 cm with prickles 2.5- 3 mm long, 16-30 mm apart, similar but re- flexed, appressed; at midsection the margins with prickles 2-2.5 mm long, 4-12 mm apart, subulate, ascending, brown-tipped; the midrib below with similar prickles 1.5-2 mm long, 6-13 mm apart; on the subulate tip the mar- gins and midrib below with subulate serrations 0.6-1 mm long, 2-6 mm apart; pistillate in- florescence terminal, bearing one syncarp; pe- duncle more than 15 cm long, 8 mm in diam- eter, ' not ' clavate, 3-sided, leafy bracted; syn- carp 9 cm long, 8.5 cm in diameter, globose, bearing about 34 phalanges, these 3.6 cm long, 2.4-27 cm wide, 2-2.4 cm thick, red, upper 5/9 free, this part broadly ellipsoid, 5-6-angled, the apex low convex, the sides gently curved, smooth, lateral sutures none; central apical si- nuses 0.5-2 mm deep, shallow; carpels 6-9 (in a double 12, and the phalange 3.3 cm wide), the apices low pyramidal, angled and wrinkled (on the basal phalanges the carpel apices low convex); lower 4/9 of phalange fleshy enlarged as shown by persistent shoulders; stigmas 2-3.5 mm wide, cordate, gray, horizontal or oblique, centripetal; proximal sinus wide, running V2 way to valley bottom; endocarp median, 15 mm long, bony, blackish brown, the lateral walls 1.5 mm thick, the inner surfaces papillose, a little shiny; seeds 8-11 mm long, oblique ellipsoid; upper mesocarp forming in the apex of each carpel a cavern with a few fibers and an aeren- chyma tissue of brownish medullary mem- branes; basal mesocarp fibrous and fleshy. holotypus: Malaya, Pahang, Pul.au Tioman, Telok Paya, sea level, 19 May 1927, M. R. Hen- derson 18,444 (sing). DISCUSSION: P. rubricoloratus is a member of the section Pandanus , as is its closest relative, the Vietnamese species P. reversispiralis St. lent. FIG. 164. Pandanus brevicornutus St. John, from holotype. a, Syncarp, X 1; ^ drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; ^ drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe apex and style, apical view, X 4; g, style, distal view, X 10; h, style and stigma, proximal view, X 10; i, style and stigma, lateral view, X 10; j, style and stigma, proximal view, X 10; k, l, style, distal view, X 10; m, leaf base, lower side, XI \ n> leaf middle, lower side, X 1; o, leaf apex, lower side, X !• 34 PACIFIC SCIENCE, Vol. XVII, January 1963 John, which has the syncarp 11-14 cm in diam- eter, oval-subglobose; phalanges 4.3 -4.6 cm long, orange; central apical sinuses 3.5-5 mm deep; endocarp lateral walls 2-3 mm thick; and the leaves 4.8-6 cm wide, at midsection the margins with prickles 3-4 mm long, and 5-17 mm apart. P. rubricoloratus has the syncarp 8.5 cm in diameter, globose; phalanges 3.6 cm long, red; central apical sinuses 0.5-2 mm deep; endocarp lateral walls 1.5 mm thick; and the leaves 2.8-3 cm wide, at midsection the mar- gins with prickles 2—2.5 mm long, and 4—12 mm apart. The epithet is formed from the Latin words ruber, red; coloratus, colored, in reference to the red color of the fruit. Pandanus brevicornutus sp. nov. (sect. Rykia) Fig. 164 DIAGNOSIS HOLOTYPI: Frutex, ramis assur- gentibus in apice 17 mm diametro luteis lucidis, foliis 41-47 cm longis 27-29 mm latis in medio sed in basi 18-22 mm latis ligulatis chartaceis proxima basem supra midnervum sulcatis alibi planis in sectione mediali cum 19-21 nervis parallelis secundariis in quoque medio nervis tertialis transversis semievidentis proxima api- cem in apice 3.5 cm longo 1.5 mm lato trigono subulato semiabrupte diminuentibus, basi am- plexicauli inermi sed ex 3-4 cm marginibus cum aculeis 1.5—3 mm longis 3—7 mm separatis subulatis adscendentibus obscure mahogani- coloratis lucidis, midnervo infra ex 9-10 cm cum aculeo unico 2.5 mm longo arcuato subu- lato reflexo, in sectione mediali marginibus cum aculeis 0.5-1. 5 mm longis 6-11 mm separatis graciliter subulatis proxime adpressi-adscenden- tibus, midnervo infra cum aculeis 2-2.5 mm longis paucis remotis arcuatis subulatis reflexis vel adscendentibus, in apice subulato marginibus cum serrulis 0.2— 0.5 mm longis 0.5— 1.5 mm separatis, midnervo infra cum serrulis simulan- tibus 2-3 mm separatis, infructescentia cum syncarpio unico terminal! erecto, pedunculo 10 cm longo 6 mm diametro cum folioso-bracteis multis, syncarpio 5.2 cm longo 4 cm diametro late ellipsoideo cum circa 960 drupis eis 16— 18.5 mm longis 2-2.5 mm latis 1.5-2 mm crassis anguste cylindricis compressis 5-angu- losis corpore 14-15 mm longo, pileo 5-6 mm longo basi 2 mm alta semiorbiculari, stylo 3-4 mm longo lanceolato osseoso brunneo lucido proxime curvato integro vel aliquis bifidis, stig- mate 1-1.5 mm longo elliptico vel late elliptico sulcato brunneo papilloso paene ad apicem ex- tento, endocarpio in parte 14 infera cartilagineo brunneo lateribus 0.1 mm crassis intra pallidis exlucidis, semine 3-4 mm longo ellipsoideo truncato, mesocarpio apicali cavernoso angusto ellipsoideo 9 mm longo, mesocarpio basali in lateribus fibroso intra carnoso 4-5 mm longo. DESCRIPTION OF ALL SPECIMENS EXAMINED: Shrub; branches assurgent, up to 1.3 m long, sprawling and forming a thicket, 17 mm in diameter at apex, yellowish, shining; leaves 35- 50 cm long, 24-29 mm wide at the middle, 18- 22 mm wide near the base, ligulate, chartaceous, yellowish green, not glaucous beneath, near the base sulcate over the midrib, elsewhere flat, at midsection with 18-21 secondary parallel nerves in each half, the tertiary cross veins barely visi- ble only near the tip, near the tip rather ab- ruptly tapering into a 3-5-7 cm subulate trigon- ous apex 1.5 mm wide, the base amplexicaul and cream-colored, unarmed, but beginning at 3-4 cm the margins with prickles 1.5-3 mm long, 3-7 mm apart, subulate, ascending, dark mahogany colored, shining; the midrib below 9-10 cm up with a prickle 2.5 mm long, arcu- ate subulate, reflexed; at midsection the mar- gins with prickles 0.5-1. 5 mm long, 6-11 mm apart, slender subulate, closely appressed ascend- ing; the midrib below with a few remote prickles 2-2.5 mm long, arcuate subulate, re- flexed or ascending; on the subulate apex the margins with serrulations 0.2-0. 5 mm long, 0.5-1. 5 mm apart; the midrib below with simi- lar ones 2-3 mm apart; pistillate infructescence with a single, terminal, erect syncarp (rarely a smaller, lateral, secondary one, globose and 2 cm in diameter); peduncle 10-15 cm long, 4-6 mm in diameter, with numerous leafy bracts; syncarp 3. 5-5. 2 cm long, 3-4 cm in diameter, when young cream colored, broadly ellipsoid, bearing 600-960 drupes, these 16-18.5 mm long, 2-2.5 mm wide, 1.5-2 mm thick, nar- rowly cylindric, compressed, 5 -angled, upper 14 free, the body 14-15 mm long; pileus 5-6 mm long, the base 2 mm high, semiorbicular, crowned by a style 3-4 mm long, prolonged Page 260: Revision of Pandanus, 14. Malaya and Singapore — St. John 35 into a lanceolate, entire or less commonly bifid beak, bony, brownish, shining, flattened, proxi- mally curved; stigma 1-1.5 mm long, elliptic or broadly so, creased, brown, papillose, running almost to the apex; endocarp in lower V3, carti- laginous, brown, walls 0.1 mm thick, the inner surface pale, dull; seed 3-4 mm long, ellipsoid, ! truncate; apical mesocarp a slender cavity 9 mm long; basal mesocarp fibrous up the sides, fleshy within, 4-5 mm long. HOLOTYPUS: Malaya, Johore, Sungei Kayu, in swamp, 17 Oct. 1936, Kiah bin Haji Salleh 32,102 (SING). SPECIMENS EXAMINED: Malaya, Sfungei] Kayu, Mawai-Jemaluang Road, in swamp, evi- dently fruits commonly, 11 Oct. 1936, E. ]. H. Corner 32,460 (SING), in large part, but not the loose fruit), (sing). DISCUSSION : P. brevicornutus is a member of the section Rykia, as is its closest relative, the Malayan species P. kedahensis St. John, which has the syncarp 12.5 cm long, 11 cm in diam- eter; drupes 4. 2-4.4 cm long, 8-13 mm wide, narrowly oblanceoloid; style 4-5 mm long, bifurcate, the lobes mostly divergent; endocarp j with apical narrow pyramidal projection; seeds 15- 16 mm long; leaves 4.5-4.6 cm wide, and at midsection with 43 secondary parallel veins in each half. P. brevicornutus has the syncarp 3. 5-5.2 cm long, 3-4 cm in diameter; drupes 16- 18.5 mm long, 2-2.5 mm wide, narrowly 1 cylindric; style 3-4 mm long, lanceolate, less commonly unevenly bifid; endocarp apex trun- cate; seeds 3-4 mm long; leaves 24-29 mm wide, and at midsection with 18-21 secondary parallel veins in each half. The new epithet is taken from the Latin words brevis, short; cornutus, horned, and is chosen in reference to the short, hornlike style of this species. Pandanus distentus sp. nov. (sect. Rykia) Fig. 165 DIAGNOSIS HOLOTYPI: Planta cum foliis 1.8- 3 m longis 9 cm latis crassiter coriaceis 1-sul- ; catis 2-plicatis in sectione M-formatis in sec- tione mediali cum 61 nervis parallelis secunda- riis prominentibus in quoque medio nervis ter- 1 tialis fortibus in dimidia ultima reticulam for- mantibus lamina ligulata valde contracta in apice 29 cm longo subulate caudato trigono eo 10 cm ex apice 1 mm lato basi non conservati, proxima basem marginibus cum spinis 5-7 mm longis 11-24 mm separatis deltoideo-subulatis complanatis subarcuato-adscendentibus pallidis sed apicibus nigris, midnervo infra cum spinis teretihus 5-6 mm longis 12-30 mm distantibus subulatis arcuatis reflexis nigris basibus crassis, in sectione mediali marginibus cum spinis 4- 5.5 mm longis crasse subulatis arcuatis adscen- dentibus saepe tota pallidis, midnervo infra cum spinis 4 mm longis 4.5-6 cm separatis crasse subulatis arcuatis complanatis valde reflexis, in sectione diminuenti marginibus cum subulato- serris 1.2-2 mm longis 1-4 mm separatis arcu- atis, midnervo infra angusto salienti cum aculeis simulantibus, in apice caudato marginibus et midnervo infra cum serrulis 0.2-0.4 mm longis 1-4 mm separatis, inflorescentia foeminea "circa 30 cm longa,” pedunculo 10 mm diametro trigono folioso-bracteato probabaliter cum syn- carpio solitario, syncarpio circa 15 cm longo et 10-11 cm diametro ellipsoideo, drupis multis 34-36 mm longis (vel stylo sequi 37-41 mm) 8-12 mm latis 7-12 mm crassis anguste oblan- ceoloideis subcompressis 5-7-angulosis corpore 30-32 mm longo anguste ellipsoideo vel fili- form! lateribus subcurvatis vel planis, pileo 9- 12 mm longo basi pyramidali-semiorbiculari laevi, stylo 3-5 mm longo osseoso rigido lucido compresso proxime curvato basi late oblonga parte supera acute bifida cum lobis subulatis sed eis cum lateribus exterioribus extensis undulatis, stigmate 3-4 mm longo late ovoideo bifido ad apices extenso brunneo papilloso, endocarpio submediali osseoso obscure brunneo lateribus 1 mm crassis intus lucido in apice cum projectione central! anguste conico, semine 15 mm longo 5 mm diametro ellipsoideo obtuso, mesocarpio apicali cavernoso cum fibris et membranis, mesocarpio basali in lateribus fibro- so intra carnoso. diagnosis OF HOLOTYPE: Stature not re- corded; leaves 1.8-3 m long, 9 cm broad, thick coriaceous, 1 -ribbed and 1 -furrowed, 2 -pleated, in section M-shaped, at midsection with 61 parallel secondary veins in each half, these rather prominent except at base, the tertiary cross veins strong, transverse, forming oblong meshes, these at least from the middle outwards giving lo cm, FIG. 165. Pandanus distentus St. John, from holotype. a, Phalange, lateral view, X 1; ^ phalange, longi- tudinal median section, X 1; c-k, drupe and style, apical view, X 1; h stigma, proximal view, X 4; n, o, style, distal view, X 4; p, leaf base, lower side, X 1; leaf apex, lower side, X 1; r, leaf venation, lower side, near base, X 4. (e, f, h, i, k, Frequent and most typical styles; c, d , g, j, infrequent forms of style.) o 10 cm. Fig. 166. Pandanus inundatus St. John, from holotype. a, Infructescence, habit, X 1; ^ drupe, lateral view, X h e, drupe, longitudinal median section, X 1; ^ drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe and style, apical view, X 4; g, drupe apex and style, apical view, X 10; h, style and stigma, proximal view, X 10; i, leaf base, lower side, X 1; j, leaf middle, lower side, X 1; k, leaf apex, lower side, X 1. 38 PACIFIC SCIENCE, Vol. XVII, January 1963 both leaf surfaces a reticulate appearance, the blade ligulate, rather sharply contracted to a 29 cm subulate, caudate, trigonous apex, this at 10 cm down from the tip 1 mm wide, the base not preserved, near the base the margins with thorns 5-7 mm long, 11-24 mm apart, deltoid, subu- late, flat, slightly arcuate ascending, pale except for tiny blackish tips; the midrib below with heavy based, terete spines 5-6 mm long, 12-30 mm apart, subulate, arcuate, reflexed, black; at lower third the margins with spines 4-5.5 mm long, 13—29 mm apart, heavy subulate, arcuate, ascending, pale throughout or with minute blackish tips; the midrib with spines 4 mm long, 4.5-6 cm apart, flat, heavy subulate, arcu- ate, strongly reflexed; on the contracting area near the tip the margins with prickles 1.2-2 mm long, 1-4 mm apart, subulate-tipped ser- rae, arcuate; the midrib below narrow, sharp, salient, bearing similar prickles; on the caudate apex the margins and midrib below with ser- rulations 0.2— 0.4 mm long, 1—4 mm apart; pis- tillate inflorescence "about 30 cm. long,” pe- duncle about 10 mm in diameter, 3-sided, leafy bracted, apparently bearing a single syncarp, this apparently about 15 cm long and 10 or 11 cm in diameter, ellipsoid (an estimate based upon a part of the core, two slabs of drupes, and many separate ones) ; drupes numerous 34- 36 mm long (or following the curve of the style 37-41 mm), 8-12 mm wide, 7-12 mm thick, narrowly oblanceoloid, a little com- pressed, 5-7-angled, the body 30-32 mm long, narrowly ellipsoid or fusiform, the sides gently curved or plane; pileus 9-12 mm long, the base pyramidal-semiorbicular, smooth; style 3-5 mm long, bony, rigid, shining, flattened, proximally curved, the base broad oblong, the upper part sharply bifid with subulate tips but the outer sides with broad, more or less undulate, ex- panded margins; stigma 3-4 mm long, broad ovoid, bifid, running to the apices, brown, papil- lose; endocarp submedian, bony, dark brown, the lateral walls 1 mm thick, the inner surface shining, with a narrow conical projection run- ning upward into the beak; seed 15 mm long, 5 mm in diameter, ellipsoid, obtuse; apical mesocarp cavernous, but with fibers and mem- branes; basal mesocarp fibrous up the margins, fleshy within. HOLOTYPUS: Malaya, Pulau Penang, Telok ! Aling, native, common, 23 Oct. 1951, /. Sinclair 39,382 (sing). DISCUSSION: P. distentus is a member of the section Rykia, as is its closest relative, the Ma- layan species P. longicaudatus Holttum & St. John, a species with the drupes 41-44 mm long, the lateral ones with styles 4-8 mm long, , bifurcate, the forks subulate; endocarp walls 1.5-2 mm thick; leaves 4.7-6 cm wide, the caudate apex 15-22 cm long, secondary nerves . 48 in each half, and near the base the margins , with thorns 3-5 mm long, 6-14 mm apart. P. distentus has the drupes 37-41 mm long, the styles sharply bifid, but the lobes with undulate, expanded outer margins; endocarp walls 1 mm i thick; leaves 9 cm wide, the caudate apex 29 1 cm long, the secondary nerves 61 in each half, and near the base the margins with thorns 5-7 i; mm long, 1 1-24 mm apart. The new epithet is the Latin participle dis- tentus, distended, and is given in reference to ; the lateral extensions of the style forks. Pandanus inundatus sp. nov. (sect. Rykia ) Fig. 166 diagnosis holotypi: Frutex, caule in apice 8 mm diametro, foliis 37-42 cm longis 30-32 mm latis in medio sed proxima basem 20 mm latis chartaceis et sulcatis minime 2-plicatis ligu- latis ad apicem diminuentibus in apice trigono subulato 4 cm longo 1-1.5 mm lato, in sectione mediali cum 19-20 nervis parallelis secundariis conspicuis in quoque dimidio, nervis tertialis : nullis, basi amplexicauli et inermi sed ex 3-3.5 cm marginibus cum aculeis 2. 5-3. 5 mm longis i 4-8 mm separatis subulatis arcuatis brunneis, midnervo infra ex 10 cm cum aculeis simulanti- • bus 3 mm longis reflexis remotis paucis, in sec- I done mediali marginibus cum aculeis 0.8-1 mm i longis 7-20 mm separatis subulatis adpresse ad- scendentibus, midnervo infra cum aculeo unico i vel pluribus remotis 1.5 mm longis compressis i arcuatis adscendentibus, in apice subulato mar- ginibus et midnervo infra cum aculeis 0.3-0. 5 mm longis 0.5- 1.5 mm separatis subulatis ad- f scendentibus, infructescentia terminali cum cap- |l ite unico, pedunculo 12 cm longo 4 mm diame- tro trigono folioso-bracteato, syncarpio 6 cm longo 4.5 cm diametro ellipsoideo sed minime Page 264: Revision of Pmdanus, 14. Malaya and Singapore — St. John 39 3-planato cum circa 832 drupis eis 17—19 mm longis 2-2.5 mm latis 1.5-2 mm crassis viridi- bus corpore 14-15 mm longo anguste cylindrico compresso 5-6-anguloso, pileo 6-8 mm longo basi 2-2.5 mm alta ellipsoidea laevi, stylo 5-6 mm longo plerumque proximo frequente subre- flexo, stigmate 2-3.5 mm longo anguste ellip- tico brunneo papillose ex apice distanto, endo- carpio in parte lA infera cartilagineo pallide brunneo lateribus 0.1 mm crasso intra exlucido, semine 2.7 mm longo ellipsoideo, mesocarpio apicali cavernoso 8 mm longo, mesocarpio basali in lateribus fibroso intra carnoso. DESCRIPTION OF ALL SPECIMENS EXAMINED: Shrub up to 1.7 m tall, 8-15 mm in diameter, often decumbent; bark dark mahogany-colored, shining; internodes 7-11 mm long; prop roots several and as much as 40 cm long, 7-8 mm in diameter, brown, with loose spirals of adventi- tious rootlets 2-5 mm long, divergent, prickle- like; leaves 37-65 cm long, 25-34 mm wide at the middle, 20-24 mm wide near the base, chartaceous, above glossy dark green, below yel- lowish green, at base channeled above the mid- rib, slightly 2 -pleated, ligulate, the apex nar- rowed to a 4-5 cm trigonous subulate tip 1-1.5 mm wide, at midsection with 18-22 secondary parallel conspicuous veins in each half, but no clearly visible tertiary veins, the base amplexi- caul, unarmed, but beginning at 3-3.5 cm the margins with prickles 2. 5-3. 5 mm long, 4-8 mm apart, subulate, bent upwards, brownish; the midrib below beginning at 10 cm with a few remote similar prickles 3 mm long, reflexed; at midsection the margins with prickles 0.8-1 mm long, 7-20 mm apart, subulate, appressed as- cending; the midrib below with one or several remote prickles 1.5 mm long, compressed, heavy based, arcuate ascending; on the subulate tip the margins and midrib beneath with prickles 0.3- 0.5 mm long, 0.5-1. 5 mm apart, subulate, as- cending; infructescence terminal, with a single head, or rarely two heads; peduncle 10-17 cm long, 4-6 mm in diameter, trigonous, leafy bracted, sometimes pendent; syncarp 4-6 cm long, 3-4.5 cm in diameter, ellipsoid but slightly 3-sided, bearing 832-1,040 drupes, these 17-19 mm long, 2-2.5 mm wide, 1.5-2 mm thick, green, the body 14-15 mm long, narrowly cylin- dric, compressed, 5-6-angled; pileus 6-8 mm long, the base 2-2.5 mm high, ellipsoid, smooth; style 5-6 mm long, mostly proximal in attach- ment, often somewhat reflexed; stigma 2-3.5 mm long, narrowly elliptic, brown, papillose, remote from the apex; endocarp in lower cartilaginous, pale brown, the walls 0.1 mm thick, the inner surface dull; seed 2.7 mm long, ellipsoid; apical mesocarp a cavern 8 mm long; basal mesocarp fibrous up the sides, fleshy within. HOLOTYPUS: Malaya, Johore, Sungei Sedili, 30 July 1939, Kiah bin Haji Salleh 63,926 (SING). SPECIMENS EXAMINED: Malaya, Johore, 5Vl miles, Kota Tinggi-Mawai Road, in very swampy places in backwaters of streams, forming more or less pure stands, 13 April 1935, E. J. H. Corner 29,231 (BO, sing); Johore Sfungei] Berassau, Mawai-Jemaluang Road, frequent in swampy forest, 6 Feb. 1935, E. J. H. Corner 28,744 (sing); ditto, flopping, gregarious in wet hollows in swampy forest, 7 Feb. 1935, Corner s. n. (SING); ditto, gregarious in standing water, low elevation, 28 April 1935, Corner 29,361 (BO, sing); ditto, in standing swamp by the stream, 5 Jan. 1936, Corner 29,991 (sing). discussion: P. inundatus is a member of the section Rykia and in that section is its closest relative, P. Scortechinii Martelli, a species with drupes 12-13 mm long; style terminal, 4-5 mm long; stigma linear; endocarp slightly subme- dian; seed 6 mm long; leaves 25-27 cm long, near the base 14 mm wide, at midsection the secondary nerves 15-16 in each half, tertiary cross veins visible in the outer half, at midsec- tion the margins with weak prickles 0.2-0. 3 mm long, flat appressed, close or remote or none. P. inundatus has the drupes 17-19 mm long; style of lateral drupes markedly lateral, 5-6 mm long; stigma narrowly elliptic; endocarp in lower Vy seed 2.7 mm long; leaves 37-65 cm long, near the base 20 mm wide, at midsection the secondary nerves 18-22 in each half, no clearly visible cross veins; at midsection the margin with prickles 0.8-1 mm long, 7-20 mm apart, subulate, appressed ascending. This was probably the species illustrated by Martelli (Webbia 4(2): t. 32, f. l-3a, 1914) as his P. johorensis Martelli. Martelli did not have the habit of indicating the specimen from 40 PACIFIC SCIENCE, Vol. XVII, January 1963 FIG. 167. Pandanus kedahensis St. John, from holotype. a, Syncarp, Y Vl\ b, drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; d, lateral drupe and style, apical view, X 1; ^ apical drupe and style, apical view, X 1 \ f, g, stigma, proximal view, X 4; h, leaf base, lower side, X 1; i, leaf middle, lower side, X 1 ; j, leaf apex, lower side, X i. Page 266: Revision of Pandanus, 14. Malaya and Singapore — St. John 41 which he made his illustrations. Evidence is now accumulating to show that he did not consist- ently illustrate the types, but rather that he used the best specimen at hand when making the drawing. Having seen and illustrated the type number of P. joborensis, it is evident to the writer that Martelli’s figures on his plate 32, published 10 years after he published the spe- cies, were drawn from another specimen, not the holotype. This other specimen seems to be like those here described as the new species P. inundatus. The new epithet is the Latin participle inun- datus, flooded, given in reference to the habitat of the species. Pandanus kedahensis sp. nov. (sect. Rykia ) Fig. 167 DIAGNOSIS HOLOTYPI: Planta cum "foliis grandibus,” pedunculo 22 cm vel plus longo 12 mm diametro trigono folioso-bracteato, bracteis 47-70 cm longis 4. 5-4.6 cm latis coriaceis 1-costatis 2-plicatis in sectione mediali cum 43 nervis parallelis secundariis in quoque latere eis supra evidentibus et infra prominentibus, nervis tertialis prominentibus et reticulum cum sec- tionibus oblongis formantibus lamina ligulata ad basem spathacea sed ex 14 cm marginibus et midnervo infra cum aculeis 0.5-1 mm longis 1-4 mm separatis subulatis adscendentibus api- cibus brunneis, in sectione mediali marginibus cum aculeis 2-3 mm longis 2-10 mm separatis arcuatis subulatis adscendentibus adpressis semi- sombratis in margine undulato, midnervo infra angusto salienti cum aculeis 1 mm longis 1-4 cm separatis arcuatis subulatis adscendentibus, in apice caudato marginibus et midnervo infra cum subulato-serrulis 0.5-0.8 mm longis 1-2 mm separatis, syncarpio 12.5 cm longo 11 cm diametro late ovoideo cum circa 544 drupis eis 4.2-4.4 cm longis 8-13 mm latis 7-12 mm crassis anguste oblanceoloideis 5-6-angulatis parte 2A supera libera corpore 3.8-4 cm longo, pileo 13-14 mm longo pyramidali-ovoideo 5-6- angulato laevi plerumque ex apice styli planato proxime inclinato truncate, stylo 4-5 mm longo oblongo osseoso brunneo lucido bifurcato lobis plerumque divergentibus, stigmate 2-3 mm longo proximo late ovato bifido brunneo papil- loso paene in apicibus extento, endocarpio in parte 2A infera obscure brunneo osseoso cum projectione angusto pyramidali apical i lateribus 2-3 mm crassis intra rugosis lucidisque, semine 15-16 mm longo 2 mm diametro, mesocarpio apicali cavernoso et cum membranis albis, meso- carpio basali fibroso et carnoso. DIAGNOSIS OF HOLOTYPE: Stature not re- corded; "large leaved”; peduncle more than 22 cm long, 12 mm in diameter, 3 -sided, leafy bracted, these bracteal leaves 47-70 cm long, 4. 5-4.6 cm wide, coriaceous, 1 -ribbed, 2 -pleated, at midsection with 43 parallel secondary nerves in each half, these visible above, prominent be- low, the tertiary cross veins conspicuous, at right angles, forming meshes mostly oblong, the blade ligulate, towards the base spathaceous enlarged; at 14 cm above the base the margins and midrib below with prickles 0.5-1 mm long, 1-4 mm apart, subulate, ascending, brown-tipped; at mid- section the margins undulate and with partly countersunk prickles 2-3 mm long, 2-10 mm apart, arcuate subulate, flat appressed, ascend- ing, brown-tipped; midrib below narrow, salient, with prickles 1 mm long, 1-4 cm apart, arcuate subulate, ascending; the caudate apex having the margins and midrib below with subulate tipped serrulations 0. 5-0.8 mm long, 1-2 mm apart; syncarp 12.5 cm long, 1 1 cm in diameter, broadly ovoid, bearing about 544 drupes, these 4.2-4.4 cm long, 8-13 mm wide, 7-12 mm thick, nar- rowly oblanceoloid, 5-6-angled, upper 2A free, the body 3.8-4 cm long; pileus 13-14 mm long, pyramidal-ovoid, 5-6-angled, smooth, mostly truncate by the flattened upper side of the sharply proximally bent style, this 4-5 mm long, oblong, bony, brown, shining, bifurcate, the lobes mostly diverging; stigma 2-3 mm long, proximal, broad ovate, bifid, brown, papillose, extending almost to the apices; endocarp in lower 2A, dark brown, bony, with a narrow pyramidal apical extension, the lateral walls 2-3 mm thick, in inner surface rugose, shining; seed 15-16 mm long, 2 mm in diameter; apical meso- carp one large cavern with white membranes; basal mesocarp fibrous and fleshy. HOLOTYPUS: Malaya, Kedah, Kedah Peak, in low forest, not far from summit, 3,500 ft. alt., 1 April 1925, R. E. Holttum 14,858 (sing). DISCUSSION: P. kedahensis is a member of the section Rykia, as is its closest relative, the Fig. 168. Pandanus serratus St. John, from holotype. a, Syncarp profile, X Vl\ b, syncarp, outline of trans- verse section, X Vl\ c, drupe, distal view, X 1; drupe, longitudinal median section, X 1> e—k, lateral drupes and styles, apical view, X 1; h apical drupe, apical view, X 1; m, carpel apex and style, distal view, X 4; n, carpel apex, proximal view, X 4; o, leaf base, lower side, X 1; P> leaf middle, lower side, X l> lea^ aPex> lower side, X 1; r, venation near leaf base, lower side, X 4. Page 268: Revision of Pmdanus, 14. Malaya and Singapore— St. John 43 Fig. 169. Pandanus albibracteatus St. John, from holotype and paratype. a, Clump of pistillate holotypic trees at top of beach, with figure of Cheang Kok Choy; b, branch with syncarp; c, foliage of paratype, St. John 26,370 ; d, staminate paratype, St. John 26,370, with figure of Cheang Kok Choy. Malayan species P. longicaudatus Holttum & St. John, a species with the peduncle 22 mm in diameter; syncarp 16=17 cm long, with about 900 drupes; pileus pyramidal-hemispheric; stigma 3—3.5 mm long; endocarp walls 1.5-2 mm thick, the inner surface smooth; and the seed 6 mm in diameter. P, kedabensis has the peduncle 12 mm in diameter; syncarp 12.5 cm long, with about 544 drupes; pileus pyramidal- ovoid; stigma 2=3 mm long; endocarp walls 2=3 mm thick, the inner surface rugose; and the seed 2 mm in diameter. The new epithet is a Latinization of the name of the type locality. Pandanus serratus sp. nov. (sect. Rykia ) Fig. 168 NOM. VERN. : "mengkuang ayer.” DIAGNOSIS HOLOTYPI: Licet aquatica in aqua vadosa, fo.li.is 1.5 m longis in media 8.5 cm latis ad terminos ambos diminuentibus crassiter cori- aceis supra viridibus infra minirne pallidiori- bus 1-sulcatis 2-plicatis in sectione depresse AA-formatis in sectione mediali cum 75 nervis parallelis secundariis in quaque dimidia in ap ice longe subulato diminuentibus basi amplexicauli inermi sed ex 20 cm marginibus cum aculeis 1-2 mm longis 5=10 mm distantibus valde PACIFIC SCIENCE, VoL XVII, January 1963 midnervo angusto salienti infra cum serris 2-2.5 mm longis 10-15 mm separatis adpresse adscen- dente subulatis, proxima apicem marginibus cum serrulis 0.5-1 mm longis 4-8 mm separatis, midnervo infra cum serrulis minutis remotis paucis; inflorescentia foeminea erecta cum 1 syncarpio, pedunculo 14 cm longo 1 cm diame- tro trigono bracteato, syncarpio 20.5 cm longo et proxima apicem 11 cm diametro, in media 9.5 cm diametro et basi subcylindrica, syncarpio 3-laterato apice obtuso cum circa 4,300 drupis eis 28-37 mm longis eis medialis basalisque 5-8 mm latis 4-6 mm crassis sed illis superis 9-12 mm latis 7-9 mm crassis oblanceoloideis, cor- pore 23-32 mm longo 5-6-angulato, pileo 4-8 mm longo basi semiorbiculari-pyramidali laevi, stylo 3-5 mm longo osseoso brunneo lucido eis lateralibus et inferioribus 2.5-4 mm latis ob- longis in basi latioribus gradatim proxime cur- vatis apice bifurcato lobis 1-3 mm longis ple- r unique divergentibus eis terminalibus cum stylo corniformi integro, stigmatibus 3-4 mm longis lanceolatis in lobis sed infra connatis brunneis papillosis proximis olim ad apices continuenti- bus, endocarpio in parte infera osseoso ob- scure mahogani-colorato lateribus 1 mm crassis pagina, interiori lucida apice cum extensioni subulato centrali, semine 1 cm longo anguste obovoideo, mesocarpio apicali cavernoso 17 mm longo fibris marginalibus multis centralibus paucis et cum medulla, mesocarpio basali fibroso et carnoso. DIAGNOSIS OF HOLOTYPE: Apparently aqua- tic, rooting in shallow fresh water; leaves about 1.5 m long, at the middle 8.5 cm wide, tapering towards both ends, thick coriaceous, green above, slightly paler below, 1 -ribbed, 2 -pleated, in sec- tion low M-shaped, at midsection with 75 sec- ondary parallel veins in each half, tapering towards the tip into a long subulate apex, but the actual tip not preserved, the base amplex- icaul and unarmed, but the margins beginning at 20 cm up with prickles 1-2 mm long, 5-10 mm apart, stout arcuate subulate, pale, ascend- ing; the nearby midrib below unarmed; at mid- section the margins with spines 6.5-7 mm long, 7-16 mm apart, heavy, arcuate subulate, ascending, the base much thickened, 4-5 mm wide, the tips reddish; the midrib below with appressed ascending subulate tipped serrae 2-2.5 44 arcuato-subulatis pallidis adscendentibus, mid- nervo infra inermi, in sectione mediali margini- bus cum spinis 6.5-7 mm longis 7-16 mm separatis fortiter arcuato-subulatis adscendenti- bus basi incrassata 4-5 mm lata apicibus rubris, Fig. 170. a, Staminate inflorescence and leaf of Pandanus albibracteatus , paratypic specimen, St. John 26,370; b, Pandanus ambiglaucus St. John, holotype, branch, foliage, and syncarp; c, Pandanus inclinatus St. John, holotype, branch, foliage, and syncarp. Page 270: Revision of Pandanus, 14. Malaya and Singapore — St. John 45 mm long, 10-15 mm apart, on the narrow 1 syncarp; peduncle 14 cm long, 1 cm in diam- salient midrib; near the apex the margins with eter, 3-sided, bracteate; syncarp 20.5 cm long, serrulations 0.5-1 mm long, 4-8 mm apart; the 11 cm in diameter near the tip, 9.5 cm at the midrib below with a few, tiny, very remote ser- middle, and near the base subcylindric, 3 -sided, rulations; pistillate inflorescence erect, bearing the apex obtuse, bearing about 4,300 drupes, Fig. 171. Pandanus incrassatus St. John, holotype, St. John 26,374. a, Habit, foliage, and syncarp, with figure (right) of Paul Weissich; h, trunk; c, trunk and prop roots; d, branch, foliage, and syncarp. 46 PACIFIC SCIENCE, Vol. XVII, January 1963 these 28-37 mm long, the middle and lower ones 5-8 mm wide, 4-6 mm thick; the upper and apical ones 9—12 mm wide, 7—9 mm thick, oblanceoloid; the body 23-32 mm long, 5-6- angled; pileus 4-8 mm long, the base semi- orbicular-pyramidal, smooth; style 3-5 mm long, bony, brown, shining, of the lateral and lower ones 2.5-4 mm wide, oblong, broadening at base, gently proximally curved and at apex bifurcate, the lobes 1-3 mm long, mostly diver- gent, but the terminal drupes with simple, horn- like style; stigmas 3-4 mm long, lanceolate on the lobes but united below, brown, papillose, proximal, running almost to the point of the style; endocarp centering in lower V3, bony, dark mahogany-colored, the walls 1 mm thick, the inner surface shining, the apex with a cen- tral, subulate prolongation; seed 1 cm long, nar- rowly obovoid; apical mesocarp a cavern 17 mm long, with fibers up the margin and a few within, and with traces of a medullary pith; basal meso- carp fibrous and fleshy. HOLOTYPUS: Malaya, Pahang, Sfungei] Bera, w. Tasek Bera, low elevation, 15 Oct. 1930, M. R. Henderson 24,137 (sing). DISCUSSION: P. serr at us is a member of the section Rykia , as is its closest relative, P. hetero- r stigma (Martelli) Martelli, a Sumatran species that has a raceme with 7-10 heads; syncarps 12-15 cm long, 9-19 cm in diameter; drupes 25 mm long, 7-8 mm wide; style spiniform, subu- late, or narrowly oblong and bifurcate; and the leaves 2 m or more in length, 10 cm wide. . P. serratus has the syncarp solitary, 20.5 cm long, 11 cm in diameter; drupes 28-37 mm long, 5-12 mm wide; style normally oblong, bifurcate but the apical ones stout, hornlike; and the leaves about 1.5 m long, 8.5 cm wide. As is the case with many members of the section Microstigma, this species has the drupes near the apex of the syncarp markedly longer and wider than those of the lower parts of the syncarp. The new epithet is the Latin adjective ser- ratus, saw-toothed, given in reference to the leaf margins. Species Structure of the Gobiid Fish Gillichthys mirabilis from Coastal Sloughs of the Eastern Pacific1 George W. Barlow2 A RECENT ISSUE of Systematic Zoology ( I960, vol. 9, nos. 3, 4) was devoted entirely to a symposium entitled "The Biogeography of Baja California and Adjacent Seas.” One recurring theme was the affinity between forms occurring on the Pacific Coast in the Californian province (Hubbs, I960: 134), and those in the northern part of the Gulf of California. Between these areas, in the main part of the Gulf, the fauna was said to differ, being primarily Panamic in origin (Garth, I960; Hubbs, I960; Walker, I960). The outer-coast species found in the upper Gulf of California are Pleistocene relicts, now trapped in a cul-de-sac of warm, tropical water (Hubbs, 1948: 463; I960: 137). According to Walker (I960: 130), most of the northern species found there differ from their cognates on the Pacific Coast. As the symposium re- vealed, however, there has been no extensive study comparing populations of one species on the Pacific Coast with those of the same species in the Gulf of California. The immediate objective of this investiga- tion, although it was initiated before the sym- posium was conceived, was to establish the nature of the variation between populations of Gillichthys mirabilis Cooper. The goal was to ascertain the degree of differentiation of the form from the upper Gulf in relation to the over-all structure of the species. A brief study (Barlow, 1961^) of G. seta (Ginsburg) has served as comparative material to gauge the differences observed in G. mirabilis. G. seta is a close but distinct species restricted to an ex- 1 A revised portion of the dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy at the University of Cali- fornia, Los Angeles, California. Manuscript received September 21, 1961. 2 Department of Zoology, University of Illinois, Urbana, Illinois. treme rocky habitat found only in the upper Gulf. G. mirabilis has been cited as one example of a species from the Pacific Coast which is also found in the upper reaches of the Gulf of Cali- fornia (Fig. 1). It occurs in the central region of the Gulf as well. There is a break in its distribution around Cabo San Lucas, however, separating the species into a Gulf group and a Pacific Coast group (details in Barlow, 1961^). At the inception of the present investigation G. mirabilis from the northern part of the Gulf of California was considered by some ichthyolo- gists to be a distinct species, G. detrusus Gilbert and Scofield. Differences of opinion existed (Barlow, 19 6D), and the issue was clouded by comparisons with a third species, the then un- described Gillichthys seta. As will be seen, the evidence supports the conclusion of Starks and Morris (1907: 227) that detrusus should be regarded as a synonym of mirabilis. In 1930 G. mirabilis from the San Diego area in southern California was introduced into the Salton Sea. The fish has become well estab- lished in that odd saline lake in the Colorado Desert of California. This has proved to be a fortunate experiment. ACKNOWLEDGMENTS With pleasure I acknowledge the timely as- sistance of the following people in gathering material for the study: H. P. Arai, W. J. Bald- win, Raymond Cannon, L. H. Carpelan, G. L. Durall, W. I. Follett, R. H. Linsey, J. B. Makern- son, F. W. Munz, K. S. Norris, P. E, Pickens, J. H. Prescott, J. C. Quast, E. S. Reese, R. H. Rosenblatt, Octavio Salazar, George Sims, R. R. Whitney, and John Wintersteen. Charles Baxter kindly prepared the chart of the study region, and Gerta M. Barlow assisted in the preparation of the manuscript and many of the illustrations. 47 48 The preliminary manuscript was read by George A. Bartholomew, Henry W. Thompson, and John N. Belkin, all of the University of Cali- fornia, Los Angeles, and I thank them for their helpful suggestions. The research and writing were done under the supervision of Boyd W. Walker, and I am grateful to him for his guid- ance and aid. ECOLOGY AND NATURAL HISTORY The typical habitat of mirabilis is the inter- tidal coastal slough with its extensive mud flats. Less often it is a swampy lagoon, or slough, at the mouth of a dry river which most of the time is closed off from the ocean by a sand bar. Salinity The water in these habitats is of moderate to high salinity, hypersaline conditions being com- mon. The maximum salinity where mirabilis has been recorded was 82.5 %o at "El Marino,” north of San Felipe in the upper Gulf of Cali- fornia ( Lars Carpelan, personal communica- tion). The fish usually are absent where the water is fresh or only slightly brackish. Those kept in fresh-water aquaria were listless and died within 1-2 weeks, but this may have been due to the quality of the tap water. The activa- tion time of the spermatozoa is quickest in water having a salinity of 35 %o (Weisel, 1948: 47). Temperature Aside from illumination, temperature is the one parameter of the environment to which it is related that changes more or less regularly with the latitude of the habitat. Local condi- tions, of course, may distort this temperature gradient. Furthermore, the Gulf of California has a climate different from that of the Pacific Coast. For these reasons, and in order to point out the extreme situation found at the Salton Sea, the meager temperature data available have been assembled and are presented in Table 1. Air temperatures have been used instead of sea surface temperatures. In the backwaters and ponds where mirabilis occurs, the surface water temperatures tend to be correlated with air tem- peratures (Carpelan, 1957: 376; Barlow, 1958: PACIFIC SCIENCE, Vol. XVII, January 1963 582 ) , even when moderate tidal exchange takes place (personal observations). Throughout the year on the Pacific Coast, air temperatures increase from north to south. The increase is more pronounced in summer than in winter. For example, San Diego is warmer than San Francisco by about 4-5 C dur- ing the summer, but is only 2-3 C warmer in the winter. The protected Alviso ponds of southern San Francisco Bay, with their increased insolation, exhibit a special climatological situation (Car- pelan, 1957: 37 6) that removes them from the normal gradient of temperature change asso- ciated with latitude. Air temperatures at the ponds during 1 year averaged almost 7 degrees higher than the long-range monthly means for San Francisco shown in Table 1. The climate at Alviso ponds evidently is warmer than at many truly coastal sloughs much further south. In the Gulf of California region air tempera- tures increase from north to south during much of the year. But during the late, hot, summer season the situation is reversed and the mean air temperatures decrease, though only slightly, from north to south. However, the mean sea surface temperatures for August (Roden and Groves, 1959: 14) show no latitudinal trend whatsoever. Chiefly because of the reversal in the gradient of the air temperatures, the annual range of means is low in the south but is pro- gressively greater to the north in the Gulf. In the tidal sloughs of the Gulf of California where mirabilis normally is found, the summer increase from south to north in the range of water temperatures probably is augmented by the nature of the tidal change. The tidal range increases dramatically from south to north up the Gulf, reaching a maximum of 10 m at the head of the Gulf (Roden and Groves, 1959: 28). The sills of the sloughs usually are high. Thus as one proceeds north in the Gulf, the sloughs daily have a progressively longer pond phase. This means that the buffering effect of the tidal exchange will be less, and as a result the water temperatures in the sloughs will reg- ister still greater fluctuations. The winter mean air temperatures at San Diego on the Pacific Coast, the central and northern Gulf of California areas, and the Salton GilUchthys mirabilis — Barlow 49 TABLE 1 Monthly Mean Air Temperatures PACIFIC SALTON GULF OF CALIFORNIA3 COAST1 SEA2 West Coast East Coast MONTH San Francisco 37° 48'N San Diego 32° 34'N 33° 18'N Mulege 26° 53'N La Paz 24° 10'N Guaymas 27° 55'N Topolo- bampo 25° 36'N Mazatlan 23° ll'N January 93 12.5 12.0 14.0 18.2 17.7 18.6 19.3 February 11.6 13.5 13.0 16.0 19.2 18.9 19.7 19-4 March 12.9 14.4 17.5 17.8 21.3 20.7 20.1 19.7 April « 13.4 15.2 20.5 20.1 23.2 22.8 22.2 21.3 May 14.1 16.9 24.5 22.9 25.5 25.5 25.0 23.8 June 14.9 17.8 28.0 27.2 27.2 29.1 29.1 26.4 July 14.8 19.9 30.5 30.5 30.0 29.8 29.8 27.5 August 15.1 20.3 31.5 30.4 30.3 30.5 29.7 27.6 September 16.7 19.4 29.5 29.0 29.0 30.2 29.7 27.5 October 15.7 17.8 24.5 24.6 27.0 27.3 27.8 26.7 November 14.0 15.9 18.5 19.5 23.2 22.6 24.0 23.5 December 10.4 13.6 13.0 14.9 20.0 18.6 19.7 20.6 Range 7.4 7.8 19.0 16.5 12.1 12.8 11.2 8.3 1 Hubbs, 1948. 2 Carpelan, 1958. 3 Roden, 1958. Sea are surprisingly similar. The summer tem- peratures, in marked contrast, diverge greatly. San Diego and San Francisco remain moderate, but the mean air temperatures exceed 30 C in the Gulf and Salton Sea areas. The climate of the Salton Sea resembles that of the northern part of the Gulf of California. Both lie in the same climatological region, but the Salton Sea is a relatively small and shallow body of water surrounded by desert (Carpelan, 1958). Consequently, its climate is even more extreme than that of the upper Gulf. Dispersal Pelagic larvae constitute the most likely ave- nues of genetic exchange between disjunct pop- ulations of mirabilis. The adults are closely restricted to their specialized and discontinuous habitats. At least some of the larvae must be swept out to sea where they could be distrib- uted to other bays by the oceanic currents. There is some evidence that the larvae of this genus are not adapted to a pelagic life as prolonged as those of the related genera, Quie- tula, Clevelandia, and llypnus (but not Eucyclo- gobius ). This conclusion is based on differences in postlarval pigmentation and behavior. Pelagic larvae and nektonic postlarvae characteristically are translucent with scattered melanophores, while the completely pigmented condition is found first in the benthonic stage. Postlarvae of GilUchthys mirabilis as small as 8-12 mm in length often are relatively heavily pigmented. At a comparable length, postlarvae of the other genera mentioned above are much more trans- lucent. Postlarvae of mirabilis in a slough in the northern Gulf of California, and at the Salton Sea, were seen to drift or swim into the current in slowly moving water. The postlarvae de- scended to the bottom and held station when the current strength increased. Postlarvae of Quietula guaymasiae Jenkins and Evermann were observed to form small groups that swam vigorously in a strong tidal current in the Gulf, darting from side to side at the surface in a manner reminiscent of young atherines; they seemed well adapted for life in open water. The larvae of GilUchthys mnabilis evidently are not dispersed as readily as are those of the re- lated genera just mentioned. 50 PACIFIC SCIENCE, Vol. XVII, January 1963 Reproduction Spawning apparently commences in January throughout the range of mirabilis, but the length of the season is uncertain. In southern California spawning takes place throughout the period from January to July ( Weisel, 1947: 82). My observations indicate that spawning also starts in January in the Alviso ponds; fe- males hatched there in January and February, however, are mature and full of ripe eggs in August and September of the same year. It is Fig. 1. Chart of the study region; latitude N, longitude W. Gillichthys mirabilis— B arlow 51 not known whether or not they spawn this late in the season. In the Salton Sea, the first spawn- ing occurs in January and the last in May or June. Postlarvae of mirabilis have been collected in early February in the northern and central Gulf of California. There are no data available on the length of the spawning season in Gulf populations. Depending on the size of the female, 4,000- 9.000 eggs are laid at each spawning (Weisel, 1947: 80). Weisel believed that each female spawns once, rarely twice, in a season. Exami- nation of ovaries of Salton Sea fish leads me to believe that each female mirabilis oviposits 2 or 3 times each season, with an interval of 40-50 days between spawnings. MacGinitie (1939: 500) reported that the related blind goby, Typhlogobius calif orniensis Steindachner, spawns twice during one season, with an inter- val of about 2 months. Females of a more dis- tantly related goby, the mapo, Bathygobius sopo- rator (C and V.), can repeat a spawning in 7-16 days (Tavolga, 1954: 432). If the above esti- mates are correct, each female Gillichthys mira- bilis has a potential production of about 8, GOO- 27. 000 eggs in one season. The young of mirabilis reach maturity by the end of their first year. Rate of growth was in- vestigated only at the Salton Sea. It is clear from these data and from the changes in size composition of collections from other locations taken at different times of the year, that most fish attain adulthood and a standard length of 100-140 mm by the end of their first year. Nothing is known of their life expectancy. MATERIALS AND METHODS In the preliminary analysis of mirabilis and seta, certain counts and measurements were se- lected as being especially promising (Barlow, 1961a). While investigating these, other useful characters also came to light. All counts and measurements, except one, were carried out in accordance with the methods of Hubbs and Lagler ( 1949) . The exception, the height of the anal fin, is defined here as the distance from the base of the last element of the anal fin to the tip of the second to last ray, when the fin is erect. The study was based almost entirely on speci- mens deposited in the Fish Collections of the University of California, Los Angeles. About 4,000 specimens of mirabilis were examined. The fish from San Francisco Bay ( Alviso ponds ) were obtained from Sim’s Bait Shop, Long Beach, California, and care was taken to verify the original source of this material. After fixa- tion in 10% formalin, the specimens were washed and placed in 46% isopropyl alcohol. Statistical Methods The purpose of each statistical test in this paper is solely to accept or reject a hypothesis. The decision to accept or reject a hypothesis was based on the 5% critical level, to minimize type II errors (Hoel, 1954: 33). Acceptance of a hypothesis such as equal means, however, merely indicates no detectable differences and does not imply that the populations in question are proven to be the same. Critical values usually are given; the subscript notation 0.95 refers to a one-tail test, and 0.975 to a two-tail test. Four kinds of tests were employed. The F-test was used when the hypothesis involved the testing of two variances. Covariance was applied to problems when two or more regressions ( fit- ted by least squares) were tested at the same time. Decisions to accept or reject differences between two means were determined by /-tests. Correlation coefficients were evaluated as out- lined by Hoel (1954: 124). Obvious differences were not tested. When several variances or means were compared, the two most divergent values were examined first. If no significant differences were revealed, the others were not tested. The statistical parameters of the fin ray counts are presented graphically. The format is that of Hubbs and Hubbs (1953). Unless the sep- aration between means was patent, no statistical inferences have been drawn from these figures. RESULTS Head The head of fish from the Gulf of California, as seen from the side, is more pointed than in Pacific Coast specimens. In this respect the Gulf populations approach seta. The difference in 52 PACIFIC SCIENCE, Vol. XVII, January 1963 TABLE 2 Fin Counts for Gillichthys mirabilis, MATERIALS AND LOCALITIES °N LOCALITY COLLECTED FIN COUNTS n (SIZE RANGE) Pacific Coast 37°27' San Francisco Bay June 1957 30(39-86) 35°20' Morro Bay August 1957 30(86-132) 34°24' Goleta September 1957 30(64-125) 34°00' Venice June 1957 28(83-138) 33°37' Newport Bay October 1955 25(56-154) 33°01' San Elijo Lagoon June 1956 30(58-142) 32°47' Mission Bay May 1946 30(38-54) 31°43' Estero Punta Banda July 1955, 1956 20(80-124) 30°30' Bahia San Quintin July 1956 14(91-142) 26°45' Punta Abreojos May 1948 16(10-20) 24°47' Bahia Sta. Maria April 1955 30(22-38) Salton Sea 33° 18' NW Shore July 19531 30(55-109) NE Shore July 19541 30(40-78) NW Shore July 1955 30(35-74) Gulf of California 31°2 r Bahia Cholla January 1949 30(20-30) Bahia Cholla April 1954 30(48-88) 31 ° 18' San Felipe April 1957 31(25-50) 27°57' Estero Soldado January 1952 31(15-20) 26° 54' Mulege March 1957 30(35-50) 26°22' Bahia Agiabampo January 1951 10(66-100) 1 Not included in Figures 6-12. shape is due largely to the more depressed head of the Gulf fish, and was one of the main reasons for naming this form as a new species. In their description, Gilbert and Scofield (1898: 498) commented on the depressed head of Gillichthys detrusus ( = G. mirabilis Cooper ) . This condi- tion prevails in all the populations investigated from the Gulf. The mean head depth for fish from "El Marino” slough, just north of San Fe- lipe, is 15.6% (13-18%; n = 10) of standard lenth, which agrees nicely with the value for Bahia Agiabampo fish (Barlow, 196L*). It should be noted at this point that the position of the opercles and the branchiostegals of pre- served gobies can cause considerable variation in measurements of head depth and especially of head width. The head shape of Salton Sea fish is more like that of specimens from the Pacific Coast than of those from the Gulf. The measured head depth, though, is closer to that of Gulf populations (Barlow, 196 1^). The regression of head length on standard length is slightly, but significantly, greater for the Pacific Coast populations than for the fishes from the Gulf when the data are pooled (Fig. 2, Table 3). Along the Pacific Coast the five populations also differ significantly from one another ( Table 3 ) . The differences between Gulf populations are of approximately the same degree as those between the Pacific Coast groups, but this was not tested statistically. Fish from the Salton Sea have the shortest heads, which is immediately noticeable when inspecting large adults. Relative growth differences are seen best by plotting the logarithms of head and standard length (Martin, 1949: 8). The regressions for all populations so presented are nearly parallel (no figure) . These curves for relative growth of the head, however, are slightly but clearly curvi- linear; the slopes steepen with increasing stand- ard length. The relative degree of difference between any of the lines is the same for the Gillichthys mirabilis — BarloW 53 FIG. 2. Head length versus standard length. Gulf of California populations: open triangle, San Felipe (El Marino); open square, Mulege; open circle, Bahia Cholla, 1954; closed triangle, Guaymas (Estero Boco- chibampo); closed circle, Agiabampo. Pacific Coast populations: closed circle, San Francisco Bay (Alviso ponds); open triangle, Goleta; closed square, Newport Bay; open circle, San Elijo Lagoon; closed triangle, Estero Punta Banda; open square, Bahia San Quintin. Salton Sea population: open triangle, 1953; open circle, 1954; open square, 1955. smallest and largest fish measured, i. e., the lines are parallel. Therefore, differences between slopes of arithmetic plots presented here are of no biological significance. Upper Jaw The analysis of upper jaw measurements is complicated by the curvilinear relationship be- tween jaw length and standard length. (Here- after, upper jaw is referred to as jaw.) This dif- ficulty is alleviated somewhat by fitting two re- gressions through the data; one line for fish greater than 90 mm, and another for the smaller specimens (Fig. 3). The change in slope is an artifact of the arithmetic plot; relative growth is only slightly curvilinear. The situation is also complicated by the sexual dimorphism of jaw length in adult fish. The average length of the jaw is slightly greater in males, even though the range is about the same in both sexes. Conse- quently, the results of the covariance tests (Table 3) should be regarded with skepticism, since some of the assumptions underlying the test are not entirely satisfied. The jaw of fish from the Gulf is about equal to, or slightly larger than, the jaw of Pacific Coast fish when specimens of the same length are compared. On the same basis, the jaw is much shorter in fish from the Salton Sea. If jaw lengths are compared as a function of head length instead of standard length, the jaw is still relatively longest in Gulf fish and shortest Fig. 3. Upper jaw length versus standard length. Gulf of California populations: open triangle, San Felipe (El Marino); open square, Mulege; open circle, Bahia Cholla, 1954; closed triangle, Guaymas (Estero Bocochibampo) ; closed circle, Agiabampo. Pa- cific Coast populations: closed circle, San Francisco Bay (Alviso ponds); open triangle, Goleta; closed square, Newport Bay; open circle, San Elijo Lagoon; closed triangle, Estero Punta Banda; open square, Ba- hia San Quintin. Salton Sea population: open triangle, 1953; open circle, 1954; open square, 1955. 54 PACIFIC SCIENCE, VoL XVII, January 1963 TABLE 3 Results of Co-Variance Tests on Differences in Morphology of Gillichthys mirabilis (Hypothesis tested is equal means after adjustment for standard length. Key: F, F ratio; d.f., degrees of freedom; F.05, 5% critical value.) F d.f. F .95 DECISION Head length Gulf/Pacific/Salton Sea 80.3 2,299 3.03 reject Gulf/Pacific 8.87 1,201 3.89 reject Pacific (5 populations) 14.8 4,128 2.44 reject Salton Sea ( 1953-55) 10.1 2,97 3.09 reject Salton Sea (1954-55) 336 1,66 3.99 accept Upper jaw length > 90 mm Gulf/Pacific/Salton Sea 10.6 2,87 3.10 reject < 90 mm Gulf/Pacific/Salton Sea 41.6 2,208 3.04 reject Salton Sea (1953-55) 7.99 2,84 3.10 reject Salton Sea (1954-55) <0.58 1,66 3.99 accept Anal fin height Gulf/Pacific/Salton Sea 37.8 2,291 3.03 reject Pacific ( 6 populations ) 19.1 5,129 2.29 reject Salton Sea (1953-55) 3.13 2,95 3.10 reject in those from the Salton Sea. The jaw is 81, 74, and 7 1 % of head length for Gulf, Pacific Coast and Salton Sea fish, respectively. Thus, the jaw of the Salton Sea fish is shorter in regard to head length as well as standard length. Median Fin Height The soft dorsal and the anal fins are higher in fish from the Gulf than in those from Cali- fornia and the Salton Sea (Barlow, 1961*, table 1 ) . Only the variation in anal fin height is re- ported in detail. The anal fin height varies significantly from population to population along the Pacific Coast (Fig. 4, Table 3). San Francisco fish, for in- stance, often fall below the fitted regression; Newport Bay and Estero Punta Bunda speci- mens usually are above it. The data are too few to make an unequivocal statement about trends, but the fins of northern fish tend to be lower than those from the south. Gilbert and Scofield (1898: 498) mentioned the higher anal fin of detrusus ( = mirabilis Cooper) when they described it as a new species. The type locality is the tidal region of the Colo- rado River delta, but high anal fins are also typical of the populations to the south in the Gulf (Fig. 4). The regression of fin length on standard length for Gulf forms is higher and steeper than those for the populations from the Pacific Coast and the Salton Sea. The variation between Gulf populations is assumed to be significant. No trend with latitude is apparent. One sample from the northwestern region of the Gulf (an isolated salt pond 25 km north of San Felipe) has remarkably low median fins, and the dorsal fin profile is less rounded than in the typical Gulf form (Fig. 5) - The height of the anal fin averages 8.8% (8. 3-9.5%) of the standard length (based on 10 specimens, 61- 102 mm in length). Thus the anal fin height and the dorsal fin shape resemble those of fish from the Pacific Coast. These data were not in- cluded in the regression of fin height versus standard length because the specimens had not been examined at the time the calculations were carried out and the figure was prepared. The fish of this population are thought to be not repre- sentative of the usual Gulf condition. Gillichthys mirabilis — Barlow 55 Fig. 4. Anal fin height versus standard length. Gulf of California populations: open triangle, San Felipe (El Marino); open square, Mulege; open circle, Bahia Cholla, 1954; closed triangle, Guaymas (Estero Boco- chibampo); closed circle, Agiabampo. Pacific Coast populations: closed circle, San Francisco Bay (Alviso ponds ) ; open triangle, Goleta; closed square, Newport Bay; open circle, San Eli jo Lagoon; closed triangle, Estero Punta Banda; open square, Bahia San Quintin. Sal ton Sea population: open triangle, 1953; open square, 1954; open circle, 1955. The anal fin height of Salton Sea fish (Fig. 4) is considerably lower than in Gulf groups. Compared to Pacific Coast populations, the fin heights are similar but nevertheless lower. The variation in height from year to year in the Salton Sea fish is just significant at the 5% level. The profile of the second dorsal fin is more rounded in specimens from the Gulf of Califor- nia than in those from the outer coast or the Salton Sea. This results from the relatively greater increase in length of the centermost rays in the higher fins. Besides the change in shape, the anterior height of the second dorsal fin equals or exceeds that of the first dorsal in Gulf forms (Gilbert and Scofield, 1898: pi. 38). In fish from the Pacific Coast and the Salton Sea, the anterior rays are lower and the central rays are equal to or only slightly higher than the spiny dorsal. Distance between Dorsal Fins The distance between dorsal fins is greatest in fish from the northern part of the Gulf of California, ranging from 1.0 to 1.8 times the eye diameter. The distance between fins of speci- mens from the central and southern Gulf is about the same as that of fish from central Cali- fornia, being 0.7-1. 1 times the eye diameter. The distance between dorsal fins was one of the diagnostic characters used by Gilbert and Scofield (1898: 498) to separate detrusus from mirabilis. In Salton Sea fish, the distance between fins is less than one eye diameter. The posterior margin of the first dorsal sometimes touches the anterior margin of the second dorsal. A greater number of elements in both dorsal fins as well as an encroachment on the space by two to three supernumerary spines on the leading edge of the second dorsal fin cause the decrease in distance. Squamation The scales are slightly larger in specimens of mirabilis from the Gulf of California than they are in those from the outer coast and the Salton Sea. The counts range from 60 to 80 irregular vertical rows of scales along the body in Gulf forms, while there are about 80-100 rows in fish from the Pacific Coast and the Salton Sea. The ventral squamation is more extensive in fish from the Gulf than in those from the Pacific Coast or the Salton Sea. The scales extend an- teriorly around the base of the pelvic fin in about 30% of the specimens from Bahia Agiabampo (southern Gulf), in about 10% from Guaymas (central Gulf), and in about 80% from Bahia Cholla (northern Gulf). In fish from the Pa- cific Coast, the scales only occasionally reach anteriorly beyond the lateral insertion of the pelvic fin base. In many respects the scalation is less exten- sive in Salton Sea fish, although the scales are about the same size as those on fish from the Pacific Coast. The ventral, anterior margin of the scaled area usually just reaches the insertion of the pelvic fins, but not beyond. The scaleless area directly under the pelvic fins is greater than in naturally occurring populations. The naked 56 area extends posteriorly to just beyond the tip of the pelvic fins but is not as wide as the pelvic fins; the size of the area is variable but is always larger than in other populations. A vertical strip in the axil, about one-half an eye diameter wide, is also without scales. Further, the strip of scales on the nape is shorter than in the other popula- tions (Barlow, 1961a) . The base of the caudal fin rays has two to six vertical rows of scales on it. In this respect only, the Salton Sea fish occasionally are more fully scaled than those of the other populations. PACIFIC SCIENCE, Vol. XVII, January 1963 populations tends to converge in the lower lati- tudes. The mean number of dorsal fin rays in the Salton Sea population is greater by about one ray than in the other populations. The large range and variation of the New- port Bay population are caused chiefly by one specimen with only eight rays. The fin does not look aberrant but it must be considered atypical. Anal Fin First Dorsal Fin The average spine number is about six for all populations, although some means deviate appreciably from this value (Fig. 6). The count never was found to be less than six in Gulf populations. Five spines, however, are not un- common in other populations. The only geo- graphic trend discernible concerns the variances of the populations; this is reported below. Second Dorsal Fin The mean number of rays in the second dor- sal fin is clearly higher in Pacific Coast than in Gulf populations (Fig. 6). On the Pacific Coast the counts are lower in the south, while in the Gulf the counts are more nearly uniform, or slightly higher in the south. Thus the average number of fin rays for Gulf and Pacific Coast On the Pacific Coast the mean number of anal fin rays increases from the north, south to San Diego, then decreases further south. In the Gulf, where samples are less complete, the means are about the same in all populations, or are a little higher in the south. The means of the Gulf and Pacific Coast populations con- verge in the lower latitudes (Fig. 7). The mean number of anal fin rays in the fish from the Salton Sea is conspicuously higher than in any of the naturally occurring popula- tions ( Fig. 7 ) . Pectoral Fin The mean number of pectoral fin rays in- creases progressively from north to south on the Pacific Coast, but in the Gulf of California the average number decreases to the south. Con- sequently, the means of Pacific Coast and Gulf FIG. 5. Gillichthys mkabilis, female, standard length 100 mm, from "El Marino” slough north of San Felipe, Baja California. The small black spot on the side of the specimen is a blemish. Gillichthys mirabilis ■ — -BARLOW 57 Fig. 6. Fin rays in the first (spiny) and second (segmented only) dorsal fins, from left to right, pre- sented as a function of °N latitude. Materials and place names in Table 2. Large rectangle, one standard deviation on either side of mean (triangular promi- nence) ; blackened space, two standard errors on either side of mean; lower horizontal line, range of counts. Open rectangle, Pacific Coast populations; vertical lines in rectangle, Salton Sea population; oblique lines in rectangle, Gulf of California populations. populations overlap in the southern latitudes (Fig. 8). The average pectoral fin count of the Salton Sea fish again is much higher than in Pacific Coast animals. The number of fin rays, how- ever, is similar to that found in northern Gulf populations (Fig. 8). The Venice population, near Los Angeles on the Pacific Coast, is extremely variable. Some pectoral fins are deformed and obviously are aberrant. This habitat is polluted by sewage and industrial waste products. Combined Fin Counts Some of the trends and differences observed in the study of individual fins disappear when the fin counts are combined and plotted as a function of latitude (Fig. 9). At different lati- tudes the averages of the combined fin counts are about the same, except in California where a slight increase from north to south occurs. Otherwise there is no trend in the mean num- ber with latitude. When comparisons are made at the same latitudes, the Pacific Coast popula- tions and all of the Gulf populations, except for San Felipe, broadly overlap. Contrariwise, some divergences are enhanced. The Salton Sea and San Felipe populations are obvious deviants. Further discussion of combined fin elements is reserved for the section on annual variation. Fin Ray Variance Over the geographic range of mirabilis there are dissimilarities in the variances of each of the fins. The difference between the extreme variances of the first dorsal fin on the Pacific Fig. 7. Fin rays in the anal fin (segmented only) presented as a function of °M latitude. Materials and place names in Table 2. Large rectangle, one standard deviation on either side of mean (triangular promi- nence); blackened space, two standard errors on either side of mean; lower horizontal line, range of counts. Open rectangle, Pacific Coast populations; vertical lines in rectangle, Salton Sea population; oblique lines in rectangle, Gulf of California populations. 58 PACIFIC SCIENCE, Vol. XVII, January 1963 Fig. 8. Fin rays in the left pectoral fin presented as a function of °N latitude. Materials and place names in Table 2. Large rectangle, one standard deviation on either side of mean (triangular prominence); black- ened space, two standard errors on either side of mean; lower horizontal line, range of counts. Open rectangle, Pacific Coast populations; vertical lines in rectangle, Salton Sea population; oblique lines in rectangle, Gulf of California populations. Coast (San Francisco Bay and Bahia Santa Maria), for instance, is highly significant: F = 16.9, F0.975 (29, 29) = 2.1. The intrinsic variability in the number of elements in a given fin is associated with the latitude where the population is situated. In- creasing variability is associated with higher latitudes (Fig. 10). All the fins manifest this increase in variability from south to north, though not equally well. If one were to rate the correlations subjectively from good to poor, they would stand in the order ( 1 ) first and second dorsal fins, (2) combined counts, (3) anal fin, and (4) pectoral fin. Only the data from the first dorsal fin are presented here (Fig. 10); the other data are available on re- quest. There also may be a positive correlation between latitude and variance of the vertebrae (see below). The correlation between degrees of latitude and variance of the first dorsal fin has a coeffi- cient of 0.848, which differs significantly from a coefficient of zero. This high degree of cor- relation doubtless could be increased further by reducing that portion of the intra-population variation resulting from sampling techniques, thereby improving the estimate of the real vari- ance. As will be seen further on, comparisons of relatively homogeneous year classes from suc- ceeding years show that even though the mean number of elements in the various fins may change, the variance remains the same. Some Fig. 9- Combined elements, segmented and unseg- mented, in the first and second dorsal, anal, and left and right pectoral fins, presented as a function of °N latitude. Materials and place names in Table 2. Large rectangle, one standard deviation on either side of mean (triangular prominence); blackened space, two standard errors on either side of mean; lower hori- zontal line, range of counts. Open rectangle, Pacific Coast populations; vertical lines in rectangle, Salton Sea population; oblique lines in rectangle, Gulf of California populations. Gillichthys mirahilis — Barlow 59 Fig. 10. Correlation between population variances (±) of the first dorsal fin, and °N latitude. Open circle, Pacific Coast populations; closed triangle, Salton Sea population; closed circle, Gulf of California populations. of the samples represented by a single value in Figure 10 are heterogeneous, consisting of more than one year class. Similar dispersion in each year class, but about different means, would increase the apparent variation. Fin Ray Correlations The segmented rays of the dorsal and anal fins are positively correlated. When all the counts are combined, exclusive of the Salton Sea, the correlation coefficient is 0.603, signifi- cantly greater than zero. A positive correlation also is evident within and between year classes. The number of elements in the pectoral fin compared to that of the anal, and of the soft dorsal fin, reveals negative correlations, but only when means of different populations are com- pared ( Fig. 11). The correlation coefficient for the pectoral fin with the second dorsal fin is -0.737, and with the anal fin it is -0.659; both values differ significantly from zero. The data for the sample from San Felipe are shown in Figure 11 (the highest pectoral fin mean) but are not included in the computations of the correlation coefficients; as mentioned before, the counts are atypical. The inclusion of these counts does not affect the conclusion that the pectoral and median fin counts are significantly correlated. Data from samples from the Salton Sea also are not included. Fin Ray Index Probably the most useful statistic obtained from the fin counts is a derived index. The soft rays of the dorsal and anal fins are summed, then divided by the number of rays in the left pectoral fin and multiplied by 100. This index reflects the change in relationship between the fin ray numbers of pectoral and median fins at different latitudes (Fig. 12). On the Pacific Coast, not including San Fran- cisco and Venice, the index decreases progres- sively from north to south. In the Gulf of Cali- fornia the opposite is the case; the index in- creases to the south. The indices of Gulf and outer-coast populations consequently converge at low latitudes, being slightly higher in outer- coast populations. Noteworthy is the excellent agreement of index values for samples from Bahia Cholla and San Felipe, situated at similar latitudes but on opposite coasts of the Gulf. This is especially striking, considering that the San Felipe fish have unusually high atypical counts. The most revealing index is that of the Salton Sea population. It agrees with values obtained from samples taken in the vicinity of San Diego, the source of the Salton Sea stock. Further, the index is considerably higher than in any of the Gulf populations. n < io z < 9 19 20 21 22 23 PECTORAL Fig. 11. Correlation between population means of number of pectoral fin rays, and number of segmented rays in the anal (closed circles) and dorsal (open circles) fins. 60 PACIFIC SCIENCE, Vol. XVII, January 1963 Fig. 12. Fin ray index: combined number of seg- mented rays in the second dorsal fin and anal fin, divided by the number of rays in the left pectoral fin, multiplied by 100. Open circles, means of Pacific Coast populations; closed circles, means of Gulf of California populations; closed triangle, mean of Salton Sea population; horizontal lines, ranges of values. Pectoral Fin Asymmetry Species of fishes are known to differ in de- gree of asymmetry between the number of ele- ments in the right and left pectoral fins ( Hubbs and Hubbs, 1945). The numerical asymmetry between right and left pectoral fin counts in mirabilis was investigated to determine if in- dividual populations, or geographic regions, could be characterized by this feature (Table 4). Only the total amount of asymmetry, right plus left, was found to be useful in this respect. Values for the Pacific Coast and Salton Sea pop- ulations are usually higher (mean about 30%) than those from the Gulf (mean about 20%). The amount of asymmetry in the two most southerly populations on the Pacific Coast, how- ever, is the same as that of the Gulf forms. The degree of asymmetry is equally divided between right and left. To facilitate compari- sons, each population can be characterized ac- cording to whether there are more fish with higher counts in the right fin than in the left fin, or vice versa. If a sample has more fish with right fins of a higher count than the left, it is tallied as a "dextral”; and with more lefts than rights, as a "sinistral.” Among collections from the Pacific Coast 5 out of 11 are dextral. In the Gulf, 3 out of 6 are dextral, 1 is sinistral, and 2 are equal. The Salton Sea population vacillates from year to year between dextral and sinistral. Therefore, the distribution of rights and lefts within and between populations is typical of a 50-50 chance situation, such as the toss of a coin. Vertebrae The number of vertebrae was compared in five populations (Table 5). In every sample the mode is 32, and the average is approximately the same value. Only the mean of the Salton Sea specimens differs appreciably from 32; it is significantly higher (t = 2.02; to. 95 (29) = 1.70). In contrast to the relative uniformity of the means, there are patent differences between the variances; they are greater in the Salton Sea and Pacific Coast groups than in those from the Gulf. Each of the variances of the San Francisco and Salton Sea samples was tested against that of the northern Gulf collection; the F-ratios are F = 13.6, F0.975 (18, 39) = 2.4, and F = 8.6, F 0.97 5 (29, 39) = 2.1, respectively. The variances of the San Francisco and Salton Sea groups are not significantly different: F = 1.6, F0.975 (18, 29) = 2.5. The variances of the vertebrae decrease from high to low latitudes. If the variances are listed from north to south, instead of as shown in Table 5, the following series of values is ob- tained: 0.577, 0.461, 0.182, 0.157, 0.341, 0.0. Considering the few populations represented, the correlation is surprisingly good. The pres- ence of a similar positive correlation between latitude and fin ray variance (see above) in- creases the confidence in the trend indicated by the variances of the vertebrae. Gillichthys mirabilis — Barlow Annual Variation Material for the study of annual variation came from the Salton Sea and from Bahia Cholla (northern Gulf of California). Each sample was from a year class hatched in the year consid- ered. This affords a rough estimate of the sta- bility of the statistical parameters used in the comparison of separate populations. The Salton Sea was sampled during the same month on three successive years. The fish of one sample are larger than the others ( Table 6 ) and probably were hatched earlier in the year than those of the other two samples. In addi- tion, the 1954 collection was from a different part of thd Sea, but I doubt that this is of im- portance. Both body measurements and fin counts were taken. The two samples from Bahia Cholla came from year classes that had hatched under strik- 61 ingly different temperature conditions. The first were embryos during an unusually cold period late in the winter of 1949. A moderate fish kill occurred in the northern Gulf at that time. The second collection, 1954, consists of fish that had developed in a relatively normal temperature situation. Only counts were taken because the specimens are small. Slight, but significant, differences occur be- tween the means of fin counts of different year classes from the Salton Sea (Table 6). In the 1955 class, all counts are higher than in the previous two years (except for the first dorsal fin: 1955 vs. 1953, t = 0.67, *0.95 (58) = 1.67). The mean of the second dorsal fin ex- ceeds that of the anal fin by about one ray in each year class. The ratios of median fin rays to pectoral rays, as indicated by the fin indices, remain ap- proximately the same. Thus while the means TABLE 4 Bilateral Asymmetry of Pectoral Fins Arranged by Regions, from North to South PERCENTAGES AREA n L > R R > L R + L Pacific Coast San Francisco 30 17 20 37 Morro Bay 30 23 10 33 Goleta 30 10 27 37 Venice1 28 29 25 54 Newport Bay 25 4 12 16 San Elijo Lagoon 30 13 23 36 Mission Bay 30 3 13 16 Estero Pta. Banda 20 25 10 35 Bahia San Quintin 14 14 21 35 Pta. Abreojos 16 19 6 25 Bahia Sta. Maria 30 13 3 16 Salton Sea 1953 30 17 13 30 1954 30 7 10 17 1955 30 20 13 33 Gulf of California Bahia Cholla — 1949 30 10 17 27 Bahia Cholla — 1954 30 10 10 20 San Felipe 31 10 3 13 Estero Soldado 31 13 16 29 Mulege 30 10 10 20 Bahia Agiabampo 10 0 10 10 1 Aberrant; see pectoral fin section. 62 PACIFIC SCIENCE, Vol. XVII, January 1963 TABLE 5 Vertebral Counts for Gillichthys mirabilis (Key: x, mean; s, one standard deviation.) NO. OF VERTEBRAE STATISTICAL PARAMETERS AREA YEAR 31 32 33 n X ±s Pacific Coast San Francisco 1955 3 13 3 19 32.00 0.577 Pta. Abreojos 1948 2 14 1 16 31.88 0.341 Gulf of California Bahia Cholla 1949 - 29 1 30 32.03 0.182 San Felipe 1955 - 39 1 40 32.03 0.157 Est. Soldado 1952 - 30 - 30 32.00 0 Salton Sea 1955 1 23 6 30 32.17 0.461 of fin ray counts may change, the relationships between them are relatively unaffected. The divergence of means between the year class samples from Bahia Cholla is greater than that recorded in the Salton Sea material. The mean number of rays for each fin, except the first dorsal, is higher in the "cold” year than in the "normal” year fish. The differences are sig- nificant at the 5% critical level. The outstand- ing deviation is in the mean of the second dorsal fin which is higher by one full ray. The relationship between the counts of the fin rays also is different in the 1949, or "cold” year class from Bahia Cholla. This results from the disproportionate increase of the soft dorsal fin count, which exceeds the anal fin by 1.7 rays. In most populations, as in the 1954 collection from Bahia Cholla, the average number of rays in the second dorsal fin surpasses that in the anal by only about one ray (0.8 to 1.3). The fin index also reflects the change in the relationship between the fin counts. It is higher by 4% in the 1949 sample, probably due to the unusual rise in the number of dorsal fin rays. Although the average number of fin rays and their interactions may change from year to year, the variance of each fin remains the same. F- tests were performed using the maximum and minimum variance of each fin within the year classes of each population; all F-ratios were below the 5% level of significance. The direction of the annual variation in fin counts is of particular importance to conclusions drawn later. In each instance when a year class deviates to a noticeable degree, the direction of the changes are the same for all fins with seg- mented rays. Increases in the dorsal and anal fin counts, as just shown, are paralleled by an in- crease in the number of pectoral fin rays; i. e., they are positively correlated. Within any one year class, however, there is no correlation what- soever between the number of segmented rays in median fins and in the pectoral fins (Table 7 ) . The pectoral fin counts are the same, on the average, whether a fish has many or few median fin rays, when from the same year class. The bilateral asymmetry of the number of pectoral fin rays fluctuates between samples of year classes from the Salton Sea and from Bahia Cholla. Both the amount and the direction of the asymmetry vary (Table 4). The total asym- metry is maximal in the Salton Sea sample for 1955 and in the Bahia Cholla collection taken in 1949. The head and jaw length and the anal fin heights are about the same in the three year classes collected from the Salton Sea. When the three groups are tested together, the differences are significant (Table 2), although the com- parison of anal fin heights for all three samples just exceeds the 5% level of significance. If only the 1954 and 1955 samples are compared, the differences no longer are significant (Table 3). The size ranges of the 1954 and 1955 sam- ples nearly coincide, while that from the 1953 year class contains larger fish. The inclusion in Gillichthys mirabilis— Barlow 63 the test of the sample with the larger fish ap- parently causes the significant differences be- tween the groups. Hence the differences between the three year classes seem to be caused by unequal size groups. Peculiarities of G. mirabilis from the Salton Sea The mirabilis introduced into the Salton Sea differ from naturally occurring populations in several ways. Most of these differences seem to indicate an interference with the normal de- velopmental physiology. Some of the unusual features of the Salton Sea fish have already been discussed in earlier sections, and they are enumerated below. All the counts of fin rays and vertebrae are higher in these fish than in any others. The relation- ship between the fin ray numbers as reflected by the fin index, however, has remained the same as that of the parental stock. The squamation is less extensive, and the head and upper jaw are disproportionately small. The formation of the interorbital pore is delayed. The Salton Sea fish differ from naturally oc- curring forms in still other ways not yet de- scribed. The otoliths are much coarser and more opaque than those of specimens from the outer coast. The outline of the otoliths, as seen from either flat side, is less angular and is often al- most round. The edges of the otoliths usually are only weakly crenulate or smooth, and the char- acteristic notch is indistinct. In gobiid fishes, the segmented rays of the soft dorsal and anal fin typically are preceded by one slender spine. This holds for mirabilis in naturally occurring populations, with rare exceptions. Supernumerary spines, however, have been found on two specimens of mirabilis from San Eli jo Lagoon and one from Newport Bay, and also on one specimen of seta from San TABLE 6 Annual Variation of Fin Counts of Gillichthys mirabilis (Key: x, mean; s, one standard deviation; r, range.) COLLECTED SALTON SEA GULF OF CALIFORNIA NW Shore 18 July 1953 NE Shore 17 July 1954 NW Shore 13 July 1955 Bahia Cholla 27 January 1949 Bahia Cholla 27 April 1954 Size range in mm 55-109 40-78 35-74 20-30 48-82 n 30 30 30 30 30 First dorsal X 6.23 6.27 6.27 6.07 6.07 ±s 0.504 0.451 0.521 0.255 0.253 r (5-7) (6-7) (5-7) (6-7) (6-7) Second dorsal X 12.43 12.33 12.57 11.33 10.37 ±s 0.679 0.547 0.679 0.480 0.555 r (11-14) (12-14) (11-14) (11-12) (9-11) Anal X 11.27 11.33 11.60 9.60 930 ±s 0.583 0.480 0.814 0.564 0.702 r (10-12) (11-12) (11-14) (9-11) (8-11) Pectoral (left) X 21.33 21.33 21.47 21.30 20.97 c±S 0.679 0.480 0.628 0.625 0.616 r (20-23) (21-22) (20-22) (20-23) (20-22) Combined X 74.93 74.67 75.70 71.73 69.67 ±s 1.68 1.18 1.95 1.53 1.65 r (72-79) (73-77) (71-80) (69-76) (66-72) Index X 112 111 113 98 94 64 PACIFIC SCIENCE, Vol. XVII, January 1963 TABLE 7 Relationship between the Pectoral Fin Rays and the Sum of the Segmented Rays of the Second Dorsal Fin and Anal Fin in Gillichthys mirabilis DORSAL PLUS ANAL SEGMENTED RAYS 17 18 19 20 21 22 23 24 25 26 27 San Eli jo Lagoon mean pectoral rays (n) 20.0 (9) 20.0 (10) 20.0 (9) 20.0 (2) Salton Sea 1953 mean pectoral rays (n) 21.0 (2) 21.1 (ID 21.2 (9) 21.7 (6) 21.0 (2) 1954 mean pectoral rays (n) 21.5 (17) 21.1 (8) 21.2 (6) 1955 mean pectoral rays (n) 22.0 (1) 21.5 (ID 21.3 (6) 21.5 (8) 21.0 (2) 22.0 (2) Bahia Cholla mean pectoral rays (n) 21.0 (1) 20.0 (1) 21.0 GO) 21.0 (15) 21.0 (3) Felipe. But in fish from the Salton Sea, the segmented rays of each of these fins are com- monly preceded by two, or occasionally three, spines. When two spines are present in the soft dorsal or anal, the anterior-most spine is short. Usually it is less than one-half the length of the first segmented ray. Sometimes it is difficult to find the anterior-most supernumerary spine. In the soft dorsal only, the first of three spines is always short (usually a stub) and sometimes does not reach through the skin, while the sec- ond usually is moderately well developed and may be greater than half the length of the first segmented ray. Supernumerary spines occur in the second dorsal fin more often than in the anal fin. In samples from 3 successive years (in each, n = 30), the percentages of the fins with extra spines are: second dorsal = 40, 13, and 33; anal = 10, 3, and 3. These data are from direct obser- vations. Photographs with X-rays revealed the pres- ence of additional spines beneath the skin and of their basal elements. In one collection (n = 30, from 1955, but a different sample than the one mentioned in the preceding paragraph), 53% of the specimens have two spines in the soft dorsal and 10% have three (total =z 63%). In the same sample, 13% of the fish have two spines in the anal fin, but none have three. The situation is further complicated in that the number of spines does not always agree with the number of basals. Three specimens with only one spine in the fin have two basal elements. There is an additional basal anterior to the basal of the first, and only, spine. In two specimens, on the other hand, two spines share a single basal. In these cases, the first spine is merely a stub. In the specimens with three spines, three basals are found. Generally, ele- ments are added anteriorly onto the anal fin, and especially the soft dorsal fin. This is done either by adding spines, basal elements, or both. Axial deformities are conspicuous in fish from the Salton Sea (Fig. 13). Mostly they are situated anterior to the origin of the second dorsal fin. These are predominantly direct dorsal flexures, but may tend to right or left. Posterior to the first dorsal fin, the flexures are divided about equally between dorsal, lateral, and ven- tral. A few specimens have severe lateral flexures in the hypural plate (Fig. 13). More than one Gillichthys mirabilis- -Barlow 65 flexure per fish is not uncommon. Photographs with X-rays indicate that the structure of the vertebrae has broken down at the apex of the more extreme curvatures. Whether there is a disintegration and/or fusion of the vertebrae is not clear from the photographs. The occurrence of axial deformities increases during the hottest months of the year (Table 8). During July, the incidence of twisted speci- mens increases from around the usual 25% to 30'%. The maximum occurs in the period of October to December. (The very high value for October is based on a small sample and therefore is subject to greater errors of random sampling. There is also a serious problem in obtaining a truly random sample, because of the sudden appearance of deformities in the young fish.) Axial aberrations usually are manifest for the first time in subadults during August, the peak of the hot season. There is no relationship be- tween size and incidence of deformities in these young fish. The deformities appear during the same period of time but not at a given size. Axial deformities are also apparent in the sciaenid fishes that have been introduced into the Salton Sea. Here, too, anterior kyphosis seems to be prevalent. Other abnormalities are seen in mirabilis but are not as common as the axial terata. These include forward projection of the lower jaw (Fig. 13), dermal flaps on the upper jaw, and pronounced elongation of individual rays in the median fins. DISCUSSION Body Form Differences in size and shape are apparent in comparisons between various kinds of fishes from waters of different temperatures. Almost invariably in the Northern Hemisphere, the more northern representatives of a species or of a genus are larger than those to the south (Hubbs, 1926: 60; Vladykov, 1934: 120). The changes in body proportions have not been as thoroughly investigated, but some tentative gen- eralizations can be made. Northern, slowly grow- ing races of a species usually have smaller heads, eyes, maxillas, and fins than do their southern counterparts, although opposite effects are not uncommon (Hubbs, 1926: 62; Vladykov, 1934: 118; Martin, 1949: 23; and included refer- ences ) . Morphological differences in Gillichthys mira- bilis, as indicated by lines fitted to part lengths versus standard length, are apparent between the aggregates of populations from the Gulf of California and the Pacific Coast, and between the populations within these regions. There is no pattern among the regressions (length of parts on standard length) which might provide a due to the nature of the differences. There is only the slightest tendency for the height of the anal fin to be greater in more southern popula- tions of the Pacific Coast region. The Gulf populations differ, for the most part, from those on the outer coast in the way one would predict for faster developing fish. Fig. 13. Gillichthys mirabilis, female, standard length 101 mm, from the Salton Sea, California, and typical of extreme terata. Fold below the pectoral fin is from an incision. 66 PACIFIC SCIENCE, Vol. XVII, January 1963 TABLE 8 Axial Deformities in Gillichthys mirabilis FROM S ALTON SEA MONTH TOTAL FISH DEFORMED FISH PERCENT DEFORMED 1955 April 19 5 26 May 98 26 27 June 124 34 27 July 37 11 30 August 45 14 31 October 15 8 53 December 98 36 37 1956 January 136 33 24 February 32 8 25 The Gulf fish have higher fins and longer jaws, the head is more depressed, and they tend to be more completely scaled verier ally and have slightly larger scales. The head is shorter, how- ever. Except for the slightly shorter head length, the form of mirabilis from the Gulf is consistent with a hypothesis of rapid development. As seen in the fish from the Salton Sea, shorter head length is correlated with lower fins, shorter jaws, and reduced scalation, and presumably this has resulted from retarded development. If, as sug- gested, the Gulf fish differ morphologically be- cause of faster development, then they should have longer, not shorter, heads. This contradic- tion probably stems from genetic differences. No satisfactory explanation can be given for the distinctly greater distance between dorsal fins of the specimens of mirabilis from the up- per Gulf as compared with the other popula- tions. The form of the fish from the upper Gulf, however, resembles that of fish from the lower Gulf, except for the distance between dorsal fins. Environmentally induced changes usually manifest themselves in several ways, instead of being expressed in one character only. The in- creased space between dorsal fins, therefore, most likely has a genetic basis. Countable characters One of the commonly accepted generaliza- tions in ichthyology is that the number of ele- ments in serially repeated characters tends to be greater at higher latitudes (Vladykov, 1934: 102). Hubbs (1922, 1926) argued that this relationship depends chiefly on the tem- perature of the water during early development; lower temperatures were thought to result in increasing numbers of elements, and higher temperatures in lower numbers. Local temper- ature conditions have been observed to be asso- ciated with the anticipated changes in counts (reviewed in Barlow, 1961$). Moreover, any factor of the environment affecting rate of de- • velopment, such as salinity or oxygen tension, seems to have effects similar to those thought to be produced by the thermal regime (Hubbs, 1926; Taning, 1952; Seymour, 1956). There are exceptions, however, where the change in counts goes contrary to- expectations. In certain instances (Schmidt, 1919: 148; Hubbs, 1921: 150; 1924; Schultz, 1927: 420) the counts from one fin only showed an increase associated with warmer temperatures. The other fins manifested the normal negative correlation between number of elements and water tem- perature. In spite of laboratory findings that the rela- tionship between meristic structures and devel- opmental temperatures is exceedingly complex, the pattern in natural populations of the same species is usually straightf orward : higher tem- peratures during development normally bring - about lower counts (Barlow, 1961$). At first glance the changes in countable char- acters over the geographic range of mirabilis seem to indicate a gradient resulting from direct modification by the environment. Presumably, factors which cause a decrease in the dorsal and anal fin counts conversely bring about an in- crease in the number of pectoral fin rays. Geographically, then, the number of median fin rays is negatively correlated with the num- ber of pectoral fin rays. Johnsen (1936: 8-9) reported the same relationship between fins for various populations of Gobius flavescens and Gobius minutus .. These changes could be con- strued as the result of dissimilar water tempera- tures or some related factors. It is more probable, nonetheless, that the cline in meristic characters results primarily from different genotypes, not exclusively from environmental modification. This hypothesis can Gillichthys mirabilis— BAMLOW be supported by more than one line of evidence. The gradient of meristic characters in mira- bilis follows the temperature gradient of the late summer period. The number of rays in the dorsal and anal fins increases from north to south in the Gulf of California, whereas over the same latitude on the Pacific Coast the oppo- site relation holds, and the counts tend to de- crease from north to south. The same reversal of trends between the Gulf and outer coast oc- curs in the pectoral fin. The fin indices decrease from north to south on the Pacific Coast, but continue to decrease in the Gulf from south to north. Thus a con- tinuous dine in the meristic characters exists from. San Francisco south, on the Pacific Coast, and then north up into the Gulf of California. It might be argued that the late summer tem- perature gradient in the Gulf is opposite to that of the outer coast, and that this environmenal difference directly accounts for the reversal of the fin-index gradient. This cannot be the case, because the counts are determined while the eggs and larvae develop during late winter through early summer; then the temperature gradients of the Gulf and Pacific Coast are parallel, warmer to the south. If any correlation exists between environmental temperatures and number of countable elements, it is with the gradient of extremely warm temperatures found late in the summer, not those occurring during early development. This dine, however, might reflect nothing more than the progressive physical separation between populations. The fin index for seta in the upper Gulf of California continues the trend noted in mira- bilis , that is, for lower indices toward the north in the Gulf. The average value for seta there is about 82 as compared to about 95 for adja- cent populations of mirabilis. The temperature situation that prevails in the habitat of seta could be regarded as an extension of the gra- dient seen in the habitats of mirabilis from south to north in the Gulf. The habitat of seta is located on the coast in the high intertidal zone among black lava flows, boulders, and streams of seepage water (Barlow, 196L*). Cli- matically this environment is even more rigor- ous, and surely experiences higher summer 67 temperatures than do the nearby sloughs where mirabilis occurs. Returning to the fin indices of mirabilis, specimens from the Salton Sea, compared to their parental stock, have much higher counts in their dorsal and anal fins, and in their pec- toral fins as well The differences probably can be attributed to retardation of developmental rate. If the interaction between these fins be- haved as expected from observations on the geographical dine, the pectoral fin counts should have decreased while the median fin counts in- creased. As shown by the fin index, the rela- tionship between median and paired fins, re- markably, is unchanged in these fish. Fluctuations in the mean number of seg- mented rays from year to year in a given popu- lation are in the same direction for the median fins and the pectoral fins. In other words, the changes in the number of elements in paired and median fins are positively correlated in their response to environmental changes. Colder years induce the formation of more rays in all of these fins. In each fish, however, the number of pectoral fin rays appears to be inherited independently from the number of median fin rays. In a given year class, the fish with more numerous rays in the median fins have, on the average, the same number of pectoral fin rays as do fish with less median fin rays. Thus the median fin and the pectoral fin elements respond to the environ- ment in the same way, but independently of each other. From the foregoing evidence it seems reason- able to conclude that the change in relationship between the number of elements in the median fins and in the pectoral fin, the fin index, over the geographic range of mirabilis indicates ge- netic divergence. Analogous differences between inter- and i ntra- population al variation have been observed in other gobies (Johnsen, 1936: 8-9) . In dif- ferent populations of Gobius flavescens and G. minutus , the vertebral counts (median fin counts follow vertebral counts) are negatively correlated with pectoral fin counts. In contrast, these meris- tic characters are positively correlated within a given population; larger fish have more verte- brae and average more pectoral fin rays as well. 68 PACIFIC SCIENCE, VoL XVII, January 1963 Johnsen indiscriminately attributed both kinds of variation to direct modification by the en- vironment. A further similarity between the results of Johnsen s investigation and the present study lies in the course of divergence of the fin index (calculated from his data). There is a geo- graphic cline in the index which is minimal in the Baltic Sea and increases to the west and then independently south and north in the North Sea, but is maximal to the north. In Gillie bt by s mirabilis , a genetic cline is also suggested by comparisons of population means of combined fin ray counts. Variation between year classes and a comparison of Salton Sea fish with their parental stock indicate that slower development is correlated with an in- crease in the total number of fin elements. If the developmental rates within the populations are altered by climatic temperatures, and this is a reasonable assumption, then the combined counts should decrease from north to- south. But the combined counts are about the same for all populations, with a few exceptions, and the ex- ceptions are obvious deviants. One could argue for a cline of combined counts for the seven populations in California. But the counts increase, not decrease, steadily from San Francisco south to San Diego. Such differences from north to south could be at- tributed to increasingly saline water to- the south, or to later spawning seasons to the north. Yet the northernmost habitat, the Alviso salt ponds, is highly saline, and spawning commences at about the same time throughout the range of mirabilis . There seems to be homeostasis of the com- bined counts in spite of changes in latitude, since over the rest of the range of mirabilis the means are about the same from population to population. This suggests genetic compensation, or relatively complete acclimation of develop- mental rates. The most obvious deviant, the population from just north of San Felipe, is also atypical in body form. The variance of the countable characters in each population is even more stable than the mean of the characters. It does not appear pos- sible to estimate the respective roles of heredity and environment in producing this constancy. Even though the means of the counts were found to shift within limits from year class to year class, the variances showed no statistical differences. The environmental changes respon- j si hie for the changes in mean counts, most likely temperature, apparently do not influence the variances. This would seem to indicate an in- trinsic homeostasis where variability is con- cerned. Photoperiod might determine variability. Day j length is constant from year to year on the same day at a given location. Since the change in photoperiod over the spawning season in- creases faster at higher latitudes, one can demon- || strate a positive, although perhaps false, correla- tion between degree of change in day length and I variance. Interestingly, estimates of the variances ) of the different fins in populations of a different but closely related species, Gillichthys seta ( Bar- low, 1961*), fall directly in the midst of the corresponding data (Fig. 10) for mirabilis . \ Somewhat further aside, in two different popula- tioes of seta the measurements of four body j parts, not counts, having different means had the same variances (Barlow, 1961*). If the variance of each character is largely genetically determined, then the data indicate two different types of dines in mirabilis. Based j on the variances, there would be a simple linear relationship, north to south, irrespective of whether the populations are situated on the Pacific Coast or on the east or the west coasts of the Gulf of California. The V-like arrangement deduced from the distribution of the means of the counts, the fin indices, and measurements of body parts have already been described and jj will be summarized below. The possible relationship between the dif- jj ferential of the photoperiod and the variance j cannot be considered as evidence for or against |j modification or genetic determination. Even ■ though the environment originally may have induced the variance in question, the phenotype easily could have become genetically reinforced. The kind and pattern of the differences found . between the populations of mirabilis seem too *1 complex to be accounted for on the assumption that the dissimilarities result chiefly from direct environmental modification. The most obvious j part of the explanation is that the divergences i Gillkhthys mirabilis-— Barlow 69 reflect the degree of physical separation between populations. The less the exchange of genetic material, the greater the opportunity for dis- similarities to arise and persist. If we are to understand why the variation is in a particular direction and not random, then we must first seek correlations between environmental factors and the observed geo- graphic cline. Temperature and illumination are the only obvious parameters suitable to such an inquiry. The only evidence available relates to temperature differences, so illumination will not be treated. This raises the problem, inherent in such an approach, of formulating a self- realizing assumption. I have already pointed out the crude correla- tion between the dines of meristic characters reported on and the gradient of temperatures found late in the summer. In another article (Barlow, 1961b) , I have argued that relatively stable differences in the means of counts be- tween populations probably reflect physiological differences, most likely temperature responses, that are genetically determined. Changes in counts between populations, as suggested by Hubbs (1928), might constitute examples of the well-known Baldwin effect (Baldwin, 1896): phenotypic modification of the counts (resulting from physiological adjustment) be- comes genetically augmented and characterizes the population. A similar argument might hold for the differences in variances that apparently are associated with different photoperiods. Peculiarities of G. mirabilis from the Salton Sea The Salton Sea is a saline lake situated in the Salton Sink, a northward extension of the Gulf of California basin. According to Carpelan (1958), it has a surface area of about 340 square miles, but a maximum and highly vari- able depth of only 12 m. The salinity of the water is about 33 %©, although the Salton Sea is not of marine origin; it has been formed by the inflow and evaporation of Colorado River water. Relative to ocean water, sulfate (2.7 times more concentrated) and calcium (1.9) ions are especially abundant, whereas potassium (0.58) and magnesium (0.75) are appreciably less concentrated. The annual maximum and minimum surface water temperatures are usu- ally about 36C and 10C, respectively. Great daily fluctuation in temperature occurs, espe- cially in shallow water (Barlow, 1958). The specimens of mirabilis from the Salton Sea differ from those on the outer coast in just the ways one would predict for fish whose growth had been drastically retarded. All fin ray counts and the vertebral counts are high (but the relationship between the fin counts is un- changed ) , supernumerary spiny rays are formed, scalation is incomplete, the head and its related parts are small, the fins are low, and the com- pletion of the interorbital canal is delayed. Anomalies, such as a continuous slit for the anterior and posterior nates, an abbreviated postorbital canal, and flaps on the jaws, are holdovers of conditions usually seen only in postlarvae. The unusual combination of salts in the water of the Salton Sea probably interferes with the development of the fish. The water temperatures are not to be considered responsible, for they are moderate during that period of early develop- ment when the characters are determined. In- deed, the temperatures are comparable to those of habitats along the sea coasts. Other water conditions (oxygen tension, pH, and illumina- tion) also are similar to those encountered by naturally occurring populations. The Salton Sea fish at times are heavily infested with a mono- genetic trematode, but so are coastal populations. Besides, many of the characters are determined before hatching and so could not be modified by debilitation caused by parasites (Hubbs, 1927). The possibility exists, of course, that the mirabilis in the Salton Sea have diverged ge- netically, and that the structural differences are the result of such genetic change. The original stock consisted of merely 500 fish, of which only a fraction could have reproduced success- fully. By chance alone, certain alleles must have been lost, others fixed, even if all 500 fish are assumed to have reproduced (Wright, 1951). The population expanded rapidly to a large number of fish, as did later introductions of other species. In a large population, genetic drift would no longer be operative (Wright, 1951). During the 25 years after introduction, selection has been effected, for the most part, 70 through the rigorous environment and through intraspecific competition. If the gene pool of the Salton Sea fish has been altered, then the morphological changes might be due to such genetic differences. This cannot be known without rearing experiments. Experiments on the metabolic physiology of the adults suggest slight genetic divergence has taken place (Barlow, 1961c). My opinion is that the morphological pecu- liarities of the fish from the Salton Sea can be ascribed primarily to direct modification caused by the chemical composition of the water in which they develop; the same could be said of the San Felipe fish from a saline pool. Conclusions The investigation of the species Gillichthys mirabilis has revealed a complex situation in which genetic divergence is evidently inter- twined with phenotypic plasticity. Over the range of mirabilis, the adaptive norms seem to shift gradually, but with some plateaus. A comparison of fin indices and means of the fin counts indicates the following ar- rangement : ( 1 ) Populations of mirabilis are nearly the same genetically on the California coast, though the Salton Sea and San Francisco populations may be somewhat distinct. (2) A genetic cline may exist on the outer coast of Baja California. (3) The fish from the lower region of the Gulf of California are similar to, but slightly different from, those on the outer coast of southern Baja California. (4) The pop- ulations found in the northern part of the Gulf have diverged from those of the southern part to a greater degree than can be accounted for, in a consistent manner, by environmentally in- duced changes. Investigations of the body form of mirabilis support the conclusions based on countable characters. The body form of fish from Califor- nia is about the same in each of the populations, although local differences exist. Fish from the Gulf of California, as a group, differ from those from California; no adult specimens were avail- able from the outer coast of southern Baja Cali- fornia. Within the Gulf populations, fish in the upper northern region of the Gulf are separable from those in the southern region. PACIFIC SCIENCE, Vol. XVII, January 1963 Thus mirabilis is seen as one widely ranging species whose geographic subdivisions are con- sistent with known faunal regions. For example, the fish fauna of San Francisco Bay is isolated from its southern counterpart, and the degree of differentiation of the mirabilis population there reflects this separation. The broad transi- tion between faunas along the outer coast of Baja California also seems to be an area of tran- sition for mirabilis, though more samples are needed from this region. Likewise, the morpho- logical separation of the mirabilis of the Gulf ! from those of the outer coast is in keeping with the known distribution of this species. Finally, the division of mirabilis from the Gulf into northern and southern groups is in harmony with faunistic and oceanographic findings (Hubbs, I960; Walker, I960; Roden, 1938). Subspecific names are not proposed for the geographic variants. Until more collections are available it would only create confusion to name the geographic forms as they are known at pres- ent. Even if they were better known, I doubt that subspecific names would be of utility here. None of the evidence suggests that any group, j or groups, of populations is evolving into a dis- 1 tinct species. The findings indicate only that in | this widely distributed species appreciable dif- ferences exist; these are thought to reflect adap- tive changes. SUMMARY 1. Gillichthys mirabilis Cooper is found in | coastal sloughs from Central California south to ' near Cabo San Lucas, and then again in the cen- ; tral and northern parts of the Gulf of California. It has been introduced into the Salton Sea. 2. The fish spawn from about January to June and reach maturity within 1 year. The lar- vae have a brief pelagic phase, during which time the species is distributed to new habitats. 3. The head is shortest in specimens of mira- bilis from the Salton Sea and most depressed in those from the Gulf of California. The length of the upper jaw and the height of the median fins are greatest in samples from the Gulf and least in those from the Salton Sea. Specimens from the upper Gulf region have the dorsal fins separated more than in mirabilis from else- where. Gillichthys mirabilis — Barlow 71 4. The scales are larger and the ventral area more scaled in specimens from the Gulf than in those from the outer coast. Scalation is notice- ably reduced in fish from the Salton Sea. 5. Trends in means of meristic characters, from north to south along the Pacific Coast, and then from south to north up into the Gulf of California, are as follows: spiny dorsal fin — no change or slight increase; segmented rays of median fins — decrease; pectoral fin — increase; combined fin elements — irregular or no change; fin index (segmented dorsal plus anal rays, di- vided by pectoral rays, times 100) — decrease; pectoral fin asymmetry — -decrease; vertebrae — no change. 6. The variance of the number of elements in each fin is a function of latitude, greater vari- ance being associated with higher latitudes. 7. The means of the counts of the median fin are negatively correlated with the means for pectoral fin rays when the comparison is be- tween the various populations. In successive year classes within a given population, however, changes in the number of rays in the median fins are associated with commensurate changes of the same sign in the counts of the pectoral fin rays. But within each year class there is no correlation between the number of pectoral fin and median fin rays. 8. All meristic characters have higher mean values in specimens from the Salton Sea, but the mean of their fin index accords with the Pacific Coast population from which they were derived. The Salton Sea fish are aberrant in many ways, evidently as a result of living in a marginal environment. 9. The conclusions based on these findings are condensed on the last page of the foregoing article. REFERENCES Baldwin, J. M. 1896. A new factor in evolu- tion. Amer. Nat. 30: 441-475, 536-553. Barlow, G. W. 1958. Daily movements of desert pupfish, Cyprinodon macularius , in shore pools of the Salton Sea, California. Ecology 39: 580-587. — - — \%la. Gobies of the genus Gillichthys, with a comment on the sensory canals as a taxonomic tool. Copeia 1961: 423-437. 1961 A Causes and significance of mor- phological variation in fishes. Syst. Zool. 10: 105-117. — 196lc. Intra- and interspecific differ- ences in rate of oxygen consumption in go- biid fishes of the genus Gillichthys. Biol. Bull. 121: 209-229. Carpel AN, L. H. 1957. Hydrobiology of the Alviso salt ponds. Ecology 38: 375-390. 1958. The Salton Sea. Physical and chemical characteristics. Limnol. Oceanog. 3: 373-386. Garth, J. S. I960. The biogeography of Baja California and adjacent seas, Part II. Marine biotas. Distribution and affinities of the brachyuran Crustacea. Syst. Zool. 9: 105-123. Gilbert, C. H., and N. B. Scofield. 1898. Notes on a collection of fishes from the Colo- rado Basin in Arizona. Proc. U. S. Nat. Mus. (20): 487-499. Hoel, P. G. 1954. Introduction to Mathematical Statistics. 2nd ed. Wiley & Sons, New York. 331 pp. Hubbs, C. L. 1921. Geographic variation of Notemigonus crysoleucas — an American min- now. Trans. 111. Acad. Sci. 11: 147-151. 1922. Variations in the number of ver- tebrae and other meristic characters of fishes correlated with temperature of water during development. Amer. Nat. 56: 360-372. 1924. Studies on the fishes of the order Cyprinodontes, IV. The subspecies of Pseu- doxiphophorus bimaculatus and Priapichthys annectens. Misc. Publ. Mus. Zool. Univ. Mich. (13): 17-23. 1926. The structural consequences of modifications of the developmental rate in fishes, considered in reference to certain prob- lems of evolution. Amer. Nat. 60: 57-81. 72 PACIFIC SCIENCE, Vol. XVII, January 1963 1927. The related effects of a parasite on a fish. A retardation of early growth, the retention of larval characters and an increase in the number of scales. J. Parasit. 14: 73-84. 1928. An hypothesis on the origin of graded series of local races in fishes. Anat. Rec. 41: 49- (Abstract.) — 1948. Changes in the fish fauna of west- ern North America correlated with changes in ocean temperature. Sears Found. J. Mar. Res. 7: 459-482. I960. The biogeography of Baja Cali- fornia and adjacent seas, Part II. Marine bio- tas. The marine vertebrates of the outer coast. Syst. Zool. 9: 134-147. and C. Hu BBS. 1953. An improved graphical analysis and comparison of series of samples. Syst. Zool. 2: 49-57. — and L. C. Hubbs. 1945. Bilateral asym- metry and bilateral variation in fishes. Papers Mich. Acad. Sci. Arts and Letters 30 (1944): 229-310. and K. F. LAGLER. 1949. Fishes of the Great Lakes Region. Cranbrook Press, Bloom- field, Mich. 186 pp. Johnsen, S. 1936. On the variation of fishes in relation to environment. (Preliminary ac- count.) Bergens Mus. Aarb. (4): 1-26. MAcGlNlTlE, G. E. 1939. The natural history of the blind goby, Typhlogobius calif ornien- sis Steindachner. Amer. Midi. Nat. 21: 489- 505. Martin, W. R. 1949. The mechanics of en- vironmental control of body form in fishes. Univ. Toronto Stud. Biol. Ser. (58), Publ. Ontario Fish. Res. Lab. (70) : 1-91. Roden, G. I. 1958. Oceanographic and meteor- ological aspects of the Gulf of California. Pacif. Sci. 13: 21-45. and G. W. Groves. 1959- Recent oceanographic investigations in the Gulf of California. Sears Found. J. Mar. Res. 18: 10- 35. Schmidt, J. 1919. Racial studies in fishes, II. Experimental investigations with Lebistes reticulatus (Peters) Regan. J. Genet. 8: 147— 153. Schultz, L. P. 1927. Temperature controlled variation in the golden shiner, Notemigonus crysoleucas. Papers Mich. Acad. Sci. 7: 417— 432. Seymour, A. H. 1956. Effects of temperature upon young chinook salmon. Dissertation Abstr. 16: 2249. Starks, E. C., and E. L. Morris. 1907. The marine fishes of southern California. Univ. Calif. Publ. Zool. 3: 159-251. TAning, A. V. 1952. Experimental study of meristic characters in fishes. Biol. Rev. 27: 169-193. TA VOLGA, W. N. 1954. Reproductive behavior in the gobiid fish Bathygobius sop orator. Bull. Amer. Mus. Nat. Hist. 104: 431-459. Vladykov, V. D. 1934. Environmental and taxonomic characters of fishes. Trans. Roy. Canad. Inst. 20: 99-140. Walker, B. W. I960. The biogeography of Baja California and adjacent seas, Part II. Marine biotas. The distribution and affinities of the marine fish fauna of the Gulf of Cali- fornia. Syst. Zool. 9: 123-133. Weisel, G. F. 1947. Breeding behavior and early development of the mudsucker, a gobiid fish of California. Copeia 1947: 77-85. 1948. Relation of salinity to the activity of the spermatoza of Gillichthys, a marine teleost. Physiol. Zool. 21: 40-48. Wright, S. 1951. The genetical structure of populations. Ann. Eugen. 15: 323-354. Preliminary Notes on Molluscan Assemblages of the Submarine Banks Around the Izu Islands Takashi Okutani1 It is well known that there are several sub- marine banks along the submerged rise which extends southwestward from the southern tip of Izu Peninsula, central Honshu. Small islands such as Toshima, Niijima, Shikine, and Kozu, with several other islets, lie on this rise. These, together with a few other islands situated far- ther south, are called the Izu Islands. They are linked by a volcanic system, and there is a con- siderable number of such banks in the neighbor- hood. A few papers concerned with hydrograph- ical, bathymetrical, and faunistic characteristics of these submarine banks have been prepared by Suzuki and Sato (1944), Niino (1935, 1952, 1955), and Shirai (1958). On the basis of these works, together with information furnished by the present author, Horikoshi (1957) discussed the topographical peculiarity in relation to the general molluscan fauna on these banks. Another group of submarine banks is found around the O'sumi Islands, south of Kyushu. Presumably their hydrographical and bathymet- rical characters are similar to those banks men- tioned above, but no information about the molluscan fauna has been available until now. As a contribution to knowledge about mol- luscan fauna on the submarine banks and in- sular shelves around the Izu Islands, this paper deals with the general account of the molluscan assemblages of the area and their faunal simi- larity to another series of banks near the Osumi Islands in the Kuroshio area. It is based on biological dredge samples collected chiefly by research vessels during 1955-59. The present writer wishes to express his grati- tude to Dr. Z. Nakai, Tokai Regional Fisheries Research Laboratory, for the material. Thanks are extended to Mr. Y. Kurata, Tokyo Fisheries 1 Tokai Regional Fisheries Research Laboratory, Tokyo, Japan. Manuscript received September 11, 1961. Experimental Station; Dr. M. Horikoshi, Ocha- nomizu University; Dr. K. Sakurai and Mr. A. Teramachi, members of the Japan Malacologi- cal Society, for their facilities and advice ren- dered in the course of this study. Thanks are also due the crews of the research vessels for their cooperation in collecting the present materiai. TOPOGRAPHY OF SUBMARINE BANKS AROUND IZU ISLANDS The Izu Isiands extend from the mouth of Sagami Bay to the south. In the northern group are (Izu-)Oshima, Toshima, Niijima, Shikine, and Kozu. To the south there are Miyake, Mi- kura, Hachijo, Aogashima, and Torishima as the southern extremity (Figs. 1, 2). The banks are usually isolated from the series of these is- lands, with depressions deeper than 200 m lying between them. The tops of such types of banks are usually flat and about 80-120 m in depth. They are usually elongate-oval in shape with the axis in a northeast-southwest direction. The Hyotanse Bank, one of the representa- tives of this series of banks, located west of Kozu Island, has been described by Niino (1955) as follows: The slopes around the bank are steep and rocks are exposed there; gravels and coarse material cover the broad and flat plain on its top; andesite and basalt, which are very common in the bedrock, are found mingled with liparite gravels together with a number of manganese concretions from the bank; the litho- logical characters of these rocks are the same as those of the main islands in the Fuji Vol- canic Zone. According to gross observation of the present material, the sediments (gravels and shells) are heavily coated by calcareous algae. The bottom of Zenisu Bank, also studied by Niino (1935), reveals coarse sand and shells, 73 74 PACIFIC SCIENCE, Vol, XVII, January 1963 N 36° 34° 32° 30° 28° Fig. 1. Index map for the position of submarine banks and other localities, a, Boso Peninsula; b, Izu Penin- j sula; c, Kii Peninsula. A, (Izu-) Oshima Is.; B, Hachijo Is.; C, Torishima Is.; D, Goto Islands; E, Amami- Oshima Is.; F, Kikaijima Is.; G, Okinoshima Is. in the Bungo Straits, a, Omurodashi (bank); P, Kurose (bank); 7, Shinkurose (bank). (See Figs. 2 and 3 for details of areas I and II, respectively.) and several steep rocks are exposed above the sea surface. At Watarinose Bank, which lies between Kozu Island and Zenisu, the character of the bottom is supposedly generally the same as that at Zenisu. The neighboring waters of those banks are noted as excellent fishing grounds. Kurose and Shinkurose are the south- ern banks situated around Hachijo Island. The bottom characters of these two are known to be similar to that of Hyotanse. On the other hand, Omurodashi Bank off Izu-Oshima is said to be different from the others in having a sandy mud sediment at its top. The geographical positions of the submarine banks of Izu Islands under study are referred to in Figures 1 and 2. OCCURRENCE OF SPECIES BY AREA 1. Hyotanse Bank material: Dredged by the R.V. "Soyo- maru” on Nov. 20 and 23, 1935, at 7 stations from depths of 118, 135, 140, 145, 148, 153, and 170-230 m. EARLIER WORKS: Niino (1955) reported 11 Molluscan Assemblages — Okutani 75 species of Pelecypoda and 5 species of Gastro- poda. Shirai (1958) reported 5 pelecypods and 5 gastropods. SPECIES IN THE PRESENT MATERIAL: A car congenitum ( Smith ) ; Mimar carta aizoi Sakurai (ms), 137 m; Striarca fausta Habe, 137 m; Samacar pacifica (Nomura and Zimbo), 137 m; Barbatia tamikoae Sakurai (ms), 140 m; Pseu- do grammatodon obliquatus Yokoyama, var.; Nipponolimopsis decussata (A. Ad.), 145 m; Tucetona shinkurosensis Hatai, Niino and Ko- taka; Malleus irregularis ( Jousseaume) , var., 118 m; Chlamys mollita (Rve. ), 118 m; C. lem- niscata (Rve.), 118 m; C. vesiculosus ( Dkr. ) ; C. tissotii (Bernard), 153 m; Spondylus anacan- thus (Mawe); Limatula japonica (A. Ad.); Lima fujitai Oyama; Septifer gray ana (Dkr.); Cardita nodulosa (Lamarck), 153 m; Gians sa- gamiensis Kuroda and Habe; Chama argentata Kuroda and Habe; Erigidocardium eos (Ku- roda), 140 m; Meiocardia tetragona (Ad. and Rve.); Emarginula fragilis Yokoyama, 153 m; E. teramachii Habe; Microgaza sp. aff. sericata Kira, 153 m; Talopena lifuana (Fischer), 153 m; Galeoastraea guttata (A. Ad.), 153 m; Ten- agodus anguinus (L.), 140 m; Serpulorbis me- dusae Pilsbry, 140 m; Apollon hirasei Kuroda and Habe, 145 m; Phanozesta semitort a Kuroda and Habe, 135 m; Latiaxis pagodus (A. Ad.), 137 m; Bursa ranelloides (Rve.), var., 135 m; Mitrella sp. cf. lischkei (Smith); Conus gra- tacapy Pilsbry, 170 m; Conus sp., 153 m. OTHER SPECIES REPORTED BY NIINO: Euce- tona hanzawad (Nomura and Zimbo); Limop- sis tajimae (Yokoyama); Hawaiarca uwaensis (Yokoyama); Plicatula muricata (Sowerby); T richomus cuius coralliophaga (Gmelin); Lima basilanica (Ad. and Rve.), 260 m; Ctenoides annulata (Lamarck), 260 m; Pecten albicans (Schroter); Lucinoma spectabilis (Yokoyama); Perotrochus beyrichii ( Hilgendorf ) , 134, 160, 128 m. MOLLUSCS REPORTED ONLY BY SHIRAI: Li- mopsis obliqua A. Ad., 250 m; Septifer excisus (Wiegmann), 104 m (this may be S. gray ana ) ; Eragum loochooanum Kira, 250 m (this may be Gians sagamiensis ) ; Galeoastraea millegranosa Habe, 260 m (this may be G. guttata) ; Sil- iquaria cumin gii Morch, 260 m (this may be Tenagodus anguinus ). 2. Zenisu Bank MATERIAL: Dredged by the R.V. "Soyo- maru” on Nov. 20, 1955, at 2 stations from depths of 85 and 170 m. EARLIER WORKS: Niino (1935) reported on the bottom character only. SPECIES IN THE THE PRESENT MATERIAL: Acar congenitum (Smith) ; Hawaiarca uwaensis (Yokoyama), 170 m; Pseudo grammatodon ob- liquatus (Yokoyama); Pectunculina cernata (A. Ad.); Limopsis cumingii A. Ad.; E ucetona shinkurosensis Hatai, Niino and Kotaka, 170 m; Glycymeris rotunda (Dkr.); Polynemamus- sium intuscostatum (Yokoyama); Chlamys ve- siculosus Dkr.; C. tissotii (Bernardi); C. lem- niscata ( Rve. ) ; Plicatula muricata ( Sowerby ) ; Spondylus anacanthus (Mawe); Pecten albicans (Schroter); Lima zushiensis (Yokoyama); Vol- sella sp.; Pycnodonta musashiana (Yokoyama); Meiocardia tetragona (Ad. and Rve.); Frigido- cardium eos (Kuroda); Poromya flexuosa (Yo- N 34° 30' 34° 20 34° 10 34° 00’ 33° 50' Fig. 2. Detail for I in Fig. 1. Dots indicate the biological stations for the R.V. "Soyo-maru” in Nov., 1955. 76 koyama); V erticordia japonica A. Ad.; Emargi- nula incisura A. Ad.; Bursa ranelloides ( Rve. ) , var.; Distorsio sp. 3. W at arinose Bank material: Dredged by the R.Y. "Soyo-maru” on Nov. 20, 1955, at 2 stations from depths of 80-100 m and 140-220 m. earlier works: None. SPECIES IN THE PRESENT MATERIAL: A car congenitum (Smith); Barbatia tamikoae Saha- ra! (ms); Saniacar pacifica (Nomura and Zimbo); Tucetona shinkurosensis Hatai, Niino and Kotaka; Malleus irregularis ( Jousseaume) , var.; Chlamys mo Hit a (Rve.); C. lemniscata (Rve.); Lima fujitai Oyama; Astraea okamotoi Kuroda and Habe; Galeoastraea guttata (A, Ad.) ; Casmaria cervica (Sowerby) ?; Tenagodus anguinus ( L. ) ; Clavus sp. aff. rufovaricosa Ku- roda (MS). 4 . Toshimatai Bank and Niijima Niskitai Bank (T oramaguri) MATERIAL: None is available for the present study except Gians sagamiensis Kuroda and Habe and Conus kinoskitai Kuroda from Nii- jima-Nishitai. earlier works: Niino (1955) reported 19 pelecypods and 4 gastropods from Toramaguri and a single gastropod from Toshimatai. Shirai (1958) reported 9 bivalves and 4 univalves from Niijima-Nishitai Bank. SPECIES REPORTED BY NIINO AND SHIRAI: Toshimatai Bank— Tucetona shinkurosensis Ha- tai, Niino and Kotaka; Hiatella arctica orientalis Yokoyama; Perotrochus beyrichii (Hilgendorf), 88 m. Niijima-Nishitai Bank— -Bar batia plicata (Dillwyn) (probably A. congenitum ), 120 m; Tucetona shinkurosensis Hatai, Niino and Ko- taka, 104 rn; Chlamys pelseneeri Dautzenberg and Bavay, 120 m; C. vesiculosus Dkr., 78 m; Lima zushiensis Yokoyama, 73 m; Lima fujitai Oyama, 104 m; Limatula japonica A. Ad., 104 rn; Cr as satellites oblongatus (Yokoyama), 120 m; Venus toreuma Gould, 78 m; Aloides venusta (Gould), 120 m; Perotrochus beyrichii (Hil- gendorf), 104, 120 m; Emarginula fragilis Yo- PAGIFIC SCIENCE, Vol. XVII, January 1963 koyama, 104 m; E. sp., 104 m; Punctmella fasti - giata A. Ad., 78 m; Trivirostra oryza (Lamarck), 88 m. X Ombase Islet ( near Kozu Island) MATERIAL: A part of the specimens dredged by the R.V. "Tonan-maru” at depths of 32, 55, 60, and 100 m. These were examined by cour- tesy of Mr. Kurata. EARLIER WORKS: Igarashi and others (1956) ■ reported on the bottom character and upon pis- cifauna from the viewpoint of fishing ground investigation. SPECIES IN THE PRESENT MATERIAL: SdMacar pacifica (Nomura and Zimbo), 60" m; Barbatia sp.; Tucetona shinkurosensis Hatai, Niino and Kotaka, 32 m; Glycy metis amamiensis Kuroda, 32 m; Venus toreuma Gould; Galeoastraea gut- tata (A. Ad.), 55, 100 m. 6. Kurose and Shinkurose MATERIAL: Collected from Kurose Bank at a depth of ca. 200 m by Cor allium fishing net of the R.V. "Tonan-maru.” Examined by courtesy of Mr. Kurata. EARLIER WORKS: Niino (1952) surveyed around Shinkurose Bank and reported 13 gas- tropods, 20- pelecypods including 3 new forms, and 1 scaphopod. SPECIES IN THE PRESENT MATERIAL: KufOSC Bank- - -A car congenitum (Smith); Area mania takii Hatai, Niino and Kotaka; Barbatia sp.; Samacar pacifica (Nomura and Zimbo); Lima fujitai Oyama; L. quanto ensis Yokoyama; Cras- satellites oblongata Yokoyama; Perotrochus bey- richii (Hilgendorf); Emarginula sp.; Tenagodus anguinus ( L. ) ; Bursa ranelloides (Rve.) var.; Chicoreus laciniatus (Sowerby); Conus sp. Shinkurose (Niino)— Area sp., 290 m; Nip- ponolimopsis nipponica (Yokoyama), 290 m; Tucetona hanzawai (Nomura and Zimbo), 280 m; Tucetona shinkurosensis Hatai, Niino and Kotaka, 290 m; Venericardia ryukyuensis No- mura and Zimbo, 290 m; V asticardium sp., 280 m; Chione chlorotica , 280 m; Cadulus sp., 290 m; Collisella heroldi (Dkr.), 290 m (a littoral species, may be carried down to the deep by some means); Margarites cincereus (Couthouy), 290 Molluscan Assemblages — OKUTANI 77 m; Pseudoliotia micans (A. Ad.), 290 m; Clath- ofenella reticulata (A. Ad.), 290 m; Mucron- alia subulata (A. Ad.), 290 m; Tonna luteo- stoma Kiister, 280 rn; Coralliophaga euginiae (Bernard), 280 m; Bursa bufonia (Gmelin), 280 m; Conus sp. (identified as C. tone by an illustration in Niino’s paper). 7. Insular Shelf around Hachijo Island MATERIAL: None is available here. EARLIER WORKS: Niino (1952) reported 13 pelecypods and 3 gastropods from depths of 115 and 200 m. SPECIES REPORTED BY NIINO: Barbatia ha- chijojimensis Hatai, Niino and Kotaka, 115, 200 m; Area mauia takii Hatai, Niino and Kotaka, 200 m; Tucetona sp., 200 m; Spondylus cruen- tus Lischke (S. anacanthusl ) , 115 m; Spondylus sp.; Lima lima L. 115, 200 m; Pycnodonta mu- sashiana (Yokoyama), 200 m; Pseudochama sp., 200 m; V asticardium arenicolum ( Rve. ) ; Mere- trix sp., 115 m; Callista pilsbryi Habe, 200 m; Venus sp., 200 m; Phalium sp., 115 m; Ocenebra adunca (Sowerby), 115 m; Conus sp., 115 m. 8. Insular Shelf around Torishima Island material: Part of specimens collected by test fishing for Cor allium operated at a depth of 150-250 m. The material was examined by courtesy of Mr. Kurata. EARLIER WORKS: None. SPECIES IN THE present MATERIAL: Spon- dylus anacanthus (Mawe); Pycnodonta musa- shiana (Yokoyama); Plicatula muricata (Sow- erby); Notolimea sp. cf. tosana Oyama; Chama argent ata Kuroda and Habe; Tenagodus an- guinus (L.) ; Talopena lifuana (Pilsbry); Can- tharus sp. SIMILAR MOLLUSCAN FAUNA FROM SOME BANKS SOUTHWEST OF KYUSHU On the Pacific Ocean side of Japan, groups of submarine banks other than those mentioned above are scattered throughout the southwest- ern waters off Kyushu (Figs. 1, 3). These also are situated on the submarine rise of the vol- Fig. 3. Detail for II in Fig. 1. Dots indicate the biological stations for the R.V. "Soyo-maru” in Feb., 1959. canic system on which the Osumi Group lies; this situation is quite similar to that of the Izu Banks. Hereafter, this second series of banks is tentatively called the Osumi Banks. The tops of these banks are usually at about 100 m, and they are isolated from the neighboring islands by depressions of about 200 m in depth. Their general features are similar to the Izu Banks in that they have submarine sediments of coarse sand and gravel which are coated by calcareous algae. Some rocks exposed at the sea surface are also found (e.g., the Uji Islets). Oceanographically, the two series of banks are decidedly exposed to similar conditions with respect to the Kuroshio Current, for the usual main axis of the current passes through the Osumi Straits as well as across the Izu Sub- marine Ridge. The occurrence of molluscan species revealed by the present survey is as follows: 9. Off Takeshima Island At a depth of 210 m; dredged by the R.V. "Soyo-maru” on Mar. 5, 1956; sta. 114: 30° 50' N, 130° 28' E. Area mauia takii Hatai, Niino and Kotaka; 78 Glycymeris amamiensis Kuroda; Chlamys mol- lita ( Rve. ) ; C. vesiculosus ( Dkr. ) , var.; C. tissotii (Bernardi); Lima tomlini Prashad; Cte- noides japonicus (Dkr.); Plicatula muricata Sowerby; Septifer gray anus (Dkr.); Gians kyu- shuensis, n. sp.; Chama argentata Kuroda and Habe; Frigidocardium eos (Kuroda); Vasti- cardium sp.; Venus toreuma (Gould); Pitar sp.; Bursa sp.; Galeoastraea millegranosa Habe; G. tayloriana (Smith); C er at o stoma vespertilis Kira; Chicoreus laciniatus (Sowerby)?; Polyn- ices sp.; Conus sp. FIG. 4. Distribution of Perotrochus heyrichii (Hil- gendorf). (See section 12 for numerals on dots.) PACIFIC SCIENCE, Vol. XVII, January 1963 10. Uji Islets At 3 stations from depths of 126-140 m; dredged by the R.V. "Soyo-maru” on Feb. 8, 1959. A car congenitum (Smith), 126 m; Striarca fausta Habe, 126 m; Samacar pacifica (Nomura and Zimbo), 126 m; Mimarcaria aizoi Sakurai (MS), 126 m; Hawaiarca uwaensis (Yoko- yama), 126 m; Spondylus anacanthus (Mawe), 126 m; Chlamys lemniscata (Rve.), 140 m; Malleus irregularis ( Jousseaume) , 126 m; Chama argentata Kuroda and Habe, 126 m; Atrina penna Habe, 140 m; Penicillus giganteus (Sowerby), 126 m; Perotrochus salmiana I (Rolle), 126 m; Serpulorhis medusae Pilsbry, 126 m; Emarginula sp.; Malluvium otohimeae (Habe), 140 m; Bursa ranelloides (Rve.), var., 126, 140 m; Semicassis sp., 140 m. 11. Insular Shelf around Goto Islands Sakurai (1959, and personal communication) reported the following species which have been collected by Cor allium fishing nets operated off the Goto Islands; depths of operation may be about 100-200 m. Mimarcaria aizoi Sakurai ( ms ) ; Barbatia tamikoae Sakurai (MS); A car congenitum (Smith); Striarca fausta Habe; S. soyoae Habe; Chlamys lemniscata ( Rve. ) ; Dymia argentata Habe; Samacar pacifica (Nomura and Zimbo); Chama argentata Kuroda and Habe; Perotro- chus hirasei Pilsbry. TYPICAL SPECIES OR SPECIES-GROUPS FOR BANKS-ASSOCIATED MOLLUSCA Because the present data are not based on quantitative samplings, it is not possible to dis- cuss the matter from a quantitative point of view. However, several species-groups may be indicated as endemic ones (or semiendemic) for submarine banks or insular shelves, because of their frequency or abundance in occurrence. 12. Perotrochus beyrichii (Hilgendorf) Fig. 5 It is well known that this "living fossil” occurs on the lower shelf around Sagami Bay. The Molluscan Assemblages — -Okutani 79 FIG. 5. Perotrochus beyrichii (Hilgendorf) from Kurose, ca. 200 m, 72.7 mm X 70.0 mm (Mr. Kurata coll.) . known localities of the species are as follows (numbers for records refer to those of localities shown in Fig. 4) : 1) Off Iioka, Pacific side of Boso Peninsula; dead shells collected by commercial trawl- ers ( Sakurai, 1954). 2) Off Hachiman-Saki ("Soyo” sta. 40a; July 4, 1956; 35° 04.7' N, 140° 21.5' E), 250 m in depth; a fragment only. 3) West coast of Boso Peninsula, off Katsu- yama and Tomiura, about 200-300 m in depth (Wada, 1954). 4) Sagami Bay, west coast of Miura Penin- sula. Type locality is off Misaki. Also col- lected by the H. M. "The Emperor” near Hayama. 5 ) Okinoyama, a small bank located near the southwestern tip of Boso Peninsula. 6) Senzu, northwestern coast of Oshima Is- land; dead shell with hermit crab, col- lected by Mr. Kurata. 7) Around Omurodashi Bank (Shirai, 1958). 8) Hyotanse, 128, 134, 160 m (Niino, 1955). 9) Niijima-Nishitai, 104, 120 m; and Toshi- matai, 88 m (Niino, 1955, and Shirai, 1958). 10) Kurose, collected by Mr. Kurata. As is shown above, this pleurotomariid gas- tropod is characteristically distributed on the lower shelf and submarine banks around the area. Other species of Perotrochus, as ecological equivalents, are also found in similar habitats in other districts; Perotrochus hirasei Pilsbry is known to be distributed around the Pacific coast of Kii Peninsula and Tosa Bay, and as an in- habitant of the continental shelf around the main Japanese islands. According to Mr. Tera- machi ( personal communication ) , P. hirasei has been collected by him from depths of 60 fath- oms (ranging 20-180 fathoms) in such locali- ties. 1 ) Okezoko Deep located south of Okino- shima Islet in Bungo Straits. This is col- lected from thanatocoenose, mainly of Glycymeris rotunda, with Siphonalia fil- osa, etc. 2) Off Urado in Tosa Bay, 70-80 fathoms, sometimes with Erronea hirasei, Chicoreus dilectus, and some other common shelf dwellers. 3) Off Tanabe, southwestern coast of Kii Peninsula. 4) Somewhere in Hyuga-Nada, eastern waters off Kyushu. Recently, Kuroda and Habe (1953) have shown that this species is also distributed in the Fig. 6. Perotrochus salmiana (Rolle) from Uji Islet, 126 m, 65.0 mm X 62.7 mm. 80 PACIFIC SCIENCE, Vol. XVII, January 1963 N 36° 34° i Fig. 7. Distribution of three species of Japanese Perotrochus. Dots, P. beyrichii; horizontal lines, P. hirasei; cross lines, P. salmi ana. western waters off Kyushu and around the Goto Islands, where some banks-associated assem- blages are found (Sakurai, 1959, and cf 11). Perotrochus salmiana (Rolle) (Fig. 6) is also known from waters off Kii and Tosa provinces. In the present survey, a new locality for this rare species was discovered from the Uji Islets CSoyo” sta. 72a, Feb. 8, 1959, 31° 24.7' N, 129° 37.6' E, 126 m), in company with the similar banks-associated fauna (cf 10). The smallness of the specimen in the present ma- terial may coincide with the fact pointed out by Parker and Curray (1956: 2436), . . most of these bank forms were considerably smaller, although they appeared to be mature specimens. . . .” 13. Galeoastraea guttata (A. Adams) Fig. 8 According to Habe (1958), this fascinating species is distributed around Boso Peninsula and Sagami Bay. As is shown in the map (Fig. 20 a), this species is characteristically distrib- uted on the banks around the Izu Islands, i. e., Hyotanse, Ombase Islet, and Okinoyama. It is very rare from the insular shelf, despite the fact that there is an example from the Izu Penin- FlG. 8. Galeoastraea guttata (A. Adams) from Hyotanse, 153 m, 22.2 mm X 30.3 mm. Molluscan Assemblages — Okutani 81 Fig. 9. Bursa ranelloides (Reeve) var. from Uji Islet, 140 m, 48.6 mm X 30.3 mm. sula coast off Iro-Zaki ("Soyo” sta. 42, Apr. 2, 1958, 34° 34.2' N, 138° 50.7' E, 80 m). No record from Boso Peninsula is available for the present study. A possible ecological equivalent, G. millegranosa Habe, is distributed on the Osumi Banks as well as in its adjacent waters, where G. guttata is not found. It is also inter- esting to note that one of the related species, Astraea caelata Gmelin, is indicated by Parker and Curray ( 1956) as a typical banks-associated mollusca from calcareous banks on the conti- nental shelf off Texas, U.S.A. 1 4. Bursa ranelloides ( Reeve ) var. Fig. 9 A rather small form of the species is widely distributed on the banks mentioned above. It has been collected from Hyotanse, 135 m; Zenisu; Kurose, 200 m; off Takeshima, 210 m, and the Uji Islets, 126 m. 15. Small Arcid Pelecypods Figs. 10, 11 On the submarine banks under study, some small species of Arcidae are abundant and are found together with other members of the group. The occurrence of species in the present material is as follows: IZU BANKS OSUMI BANKS (KYUSHU) Hyotanse Zenisu Kurose Goto Uji Sta. 114 Acar congenitum (Smith) + + + + + — Area mania takii Hatai, Niino, Kotaka.... — — + — — + Mimarcaria aizoi Sakurai (MS) + — + + Hawaiarca uwaensis ( Y okoyama ) + + — + + — Samacar pacifica (Nomura, Zimbo) + — ? + + + — Striarca fansta Habe + — — + + ■ — 16. Small Pectinid Pelecypods Figs. 12-14 Chlamys vesiculosus (Dkr.) is known to be very abundant on the continental shelf border- ing southwestern Japan. Being a sandy bottom dweller, it is also distributed on the banks of the northern group at a depth of 32-118 m. A related species, C. tissotii ( Bernardi ) , often occurs in the same localities. They are not found Fig. 10. Area mania takii Hatai, Niino, and Ko- taka from Kurose ca. 200 m, 28.0 mm X 15.0 mm (Mr. Kurata coll.). 82 PACIFIC SCIENCE, Vol XVII, January 1963 on the southern half of Izu Banks but are dis- tributed on the Osumi Banks. C. lemniscata (Rve.) is similar in its distribution to the pre- ceding two species; it occurs rather oftener than C. mollita (Rve.) does on the banks un- der study. IZU BANKS OSUMI BANKS Hyotanse Zenisu Kurose Uji Sta. 114 Chlamys vesiculosus (Dkr.) + + — + + C. tissotii ( Bernardi ) + + — + + C. lemniscata ( Rve. ) + + — + — C. mollita ( Rve. ) + — — - + + 17. Tucetona hanzawai (Nomura and Zimbo); T. shinkurosensis Hatai, Niino, Kotaka Figs. 15, 16 These small glycymerid species occur on the banks under study. As the former, T. hanzawai, originally described from a fossil bed of Ryukyu ( Riu Kiu ) Limestone, was recorded from Hyo- tanse and Shinkurose (Niino, 1952, 1955). Hatai, Niino, and Kotaka (in Niino, 1952: 106) stated that "the occurrence of this species Fig. 11. Samacar pacifica (Nomura and Zimbo) from off Miyake Island, 170 m, 12.5 mm X 6.5 mm. FIG. 12. Chlamys tissotii (Bernardi) from Hyo- tanse, 140 m, 12.6 mm X 13.4 mm. in the marine fauna of the environs of Hachijo I Island is interesting, because it may suggest j the occurrence of a submerged (geological) formation corresponding in age to the Ryukyu Limestone.” The latter species has been found from Shinkurose (the type locality), Hyotanse, J Zenisu, Watarinose, Niijima-Nishitai, Toshi- matai, as well as Ombase Islet (Fig. 20 b) . According to Hatai, Niino, and Kotaka (in Niino, 1952), T. shinkurosensis (loc. cit. p. 109, figs. 11, 12) is distinguished from T. hanzawai (Nomura and Zimbo, 1934, p. 152(44), pi. 5(1), figs. 3a, b: as Glycymeris ) by the follow- ing points: The shell of T. shinkurosensis is longer than high, while that of T. hanzawai is higher than long; the radial ribs of the former are broad, flat, low, and are 27 in number, while Molluscan Assemblages— OYJJTANl 83 there are 25 rounded ribs in the latter species; the interspaces of shinkurosensis ribs are much narrower than the ribs, while the interspaces of the latter species are nearly equal to the ribs; the hinge teeth of the former are 5 in the an- terior half and 6 in the posterior half, instead of 8 and 9 on the anterior and posterior half of the hinge plate of the latter species. In the present survey, 40 odd valves from Izu Banks were examined. The size ranges of the material were 7.7-17.4 mm in length and 7.6-18.1 mm in height. Of these specimens, no valve was found which had rib interspaces as wide as the ribs themselves, despite the fact that meristic characters vary with individuals, i. e., height/length, ranges 0.915-1.090; num- ber of radial ribs, 24-33; anterior hinge teeth, 5-10; posterior teeth, 7-11. On these evi- dences, all of the specimens are identified as T. shinkurosensis. The measurements of two odd valves of topo- types of T. hanzawai are: HEIGHT HEIGHT/ LENGTH RADIAL RIBS TEETH Ant. Post. No. 1 (left valve) 11.1 mm 1.057 29 7 7 No. 2 (right valve) 12.0 mm 1.190 25 8 7 The interspaces of radial ribs are estimated to be about half as wide as the ribs. In compari- son with specimens of two species of similar size, T. hanzawai is provided with more distin- guishing features than are pointed out in the foregoing lines: it has a more prominent umbo, a deeper shell, and less angular shoulders, thus a V asicardium- like appearance, partly because of its shell which is longer than it is high. Nev- ertheless, the morphological similarity of these two requires further biometrical study in the future. 18. Chama argentata Kuroda and Habe Fig. 17 This is found on many banks and insular shelves under study. It was collected from the Niijima-Nishitai Banks, the insular shelf of Torishima, the Uji Islets, insular shelves of Takeshima (Sta. 114), and the Goto Islands (Fig. 20c). Other species than those enumerated above, Spondylus anacanthus (Mawe), Plicatula muri- cata ( Sowerby ) , and Lima fujitai Oyama ( Figs. 18, 19, and 20d), are usually found abundantly in the area. The occurrence of two or three of them may be a remarkable faunal characteristic of the banks. GEOGRAPHICAL DIFFERENCES AND VERTICAL LIMITS OF TYPICAL BANKS-ASSOCIATED ASSEMBLAGES Since the present material was obtained with different kinds of gear from a limited number of stations, a conclusive quantitative analysis can not be made at present. However, the abun- dance or probable dominancy of the several Fig. 13. Chlamys lemniscata (Reeve) from Uji Islet, 126 m, 22.0 mm X 20.0 mm. 84 PACIFIC SCIENCE, Vol. XVII, January 1963 species might be estimated. If in the future these areas are more thoroughly surveyed by quantitative methods, the faunistic differences or peculiarities will be more clearly demon- strated. For the same reason, present data are presumed to be insufficient for the following discussion and a supplemental report is to be expected. Geographical Differences of the Assemblages Within the Izu Banks, local differences in the assemblage are scarcely observed. The typical species-groups found in the area are roughly subtropical or warm-temperate forms from the lower part of the continental shelf bordering the main Japanese islands. On the other hand, probably because of the differences in latitude, a slight difference exists between the faunas of the northern half and those of the southern half of the Izu Banks. Geographically, the Izu Islands and the Osumi Group are distant from each other. Similar patches of banks are not found between the Izu Banks and the Osumi Banks. Therefore, the continuity or discontinuity of the fauna has not yet been studied. However, many species are found in both areas: Area mania takii Hatai, Niino, and Kotaka; A car congenitum (Smith); Samacar pacifica (Nomura and Zimbo); Haavaiarca uwaensis (Yokoyama); Mimarcaria aizoi Sakurai ( MS ) ; Striarca fausta Fig. 14. Chlamys mollita (Reeve) from Hyotanse, 118 m, 18.5 mm X 17.0 mm. Habe; Chlamys vesiculosus ( Dkr. ) ; C. tissotii ( Bernardi ) ; C. lemniscata ( Rve. ) ; C. mollita (Rve.); Spondylus anacanthus (Mawe); Pli- catula muricata (Sowerby); Frigido cardium eos (Kuroda); Chama argentata Kuroda and Habe; Tenagodns anguinus (L.) ; Bursa ranelloides (Rve.), var.; and Conus sp. Of the species that are not common to the two areas, some forms are apparently ecological equivalents of banks-associated forms. Several examples are shown here: IZU ISLANDS AREA OSUMI STRAITS AREA Perotrochus beyrichii (Hilgendorf) P. salmiana (Rolle) Galeoastraea guttata ( A. Ad. ) G, millegranosa Habe or G. tayloriana (Smith) Tucetona shinkurosensis Hatai, Niino, and Kotaka Lima fujitai Oyama L. tomlini Prashad? Gians sagamiensis Kuroda and Habe G. kyushensis, sp. nov. Possible Vertical Limits of the Banks-Associated Fauna The distribution of the benthic mollusca is governed much more forcefully by the sub- stratum of the depths than by other environ- mental factors. At equal depths in similar lati- tudinal position, or under the same ocean climate, the differences in representation are due to such bottom factors as mud, sand, gravel, and rock. The banks-associated molluscan as- semblages shows mainly hard-bottom facies mingling with some sandy bottom dwellers of the depths. This is closely associated with the insular shelf fauna, since the surrounding areas of almost all the small islands in the region are Molluscan Assemblages — Okutani 85 almost entirely rocky. On the other hand, this is not always similar to the shelf fauna of the main islands because the shelf bordering the main islands does not always present a hard bottom. For instance, a molluscan assemblage from the soft bottom within Sagami Bay at a similar depth ("Soyo” sta. T2 6, Nov. 15, 1958, 35° 15.4' N, 139° 27.9' E, 102 m) is made up of: Glycymeris rotunda (Dkr.) ; Delecto- pecten macro chrili cola Habe (believed to be found attached to the carapace of the giant spider crab, but recently recovered as free living individuals ) ; Venus faveolatus Sowerby; Onus- tus exutus (Reeve); Granulifusus niponicus (Smith). Moreover, collections from a certain area in Sagami Bay sometimes contain species similar to those from the banks under study (cf 12 and 13), while a different assemblage is found from another hard bottom in the Bay at the same level. Therefore, the most remark- able character of a banks-associated assemblage is the constant occurrence and constant domi- nancy of the typical species of the area. The typical banks-associated assemblage ap- pears at depths ranging from 32 m as the shal- lowest to 290 m as the deepest, centering around 100-250 m. In waters shallower than this, a certain upper-shelf assemblage is observed, which is frequently collected by a lobster-net or in Gelinidium collecting. Examples of hard- bottom dwellers in the shallower zones of the Oshima Islands are: Turbo cornutus (Solander); F usinus nicobaricus (Lamarck); Fasciolaria gla- bra (Dkr.); F. trapezium audouini (Jonus), Fig. 15. Tucetona shinkurosensis Hatai, Niino, and Kotaka from Niijima-Nishitai, 14 mm in height. Fig. 1 6. Tucetona hanzawai (Nomura and Zimbo) from Kikaijima (fossil). etc. The sandy bottom facies of the upper shelf in Okada, Oshima Island, is reported by the dominance of the following species: Terebra nebulosa Sowerby; Conus tes sulatus Born; Chlamys vesiculosus (Dkr.); Callista pilsbryi Habe; Dosinia iwakawai Oyama and Habe; An- tigona lamellaris Schumacher; Sunetta concinna (Dkr.), var.; Paphia amabilis (Philippi). Judging from these observations, the ecotone between a shallow insular shelf fauna and a lower insular shelf fauna, which is almost equi- valent to banks-associated fauna, is presumably at a depth of 50 m or so. The deeper limit may be about 300 m. For example, at a station near Okinoyama, about 300 m deep, the following species were collected from a gravel bottom ("Soyo” sta. T20’C, Aug. 11, 1958, 34° 59.0' N, 139° 32.3' E) despite the fact that the similar banks-associated assemblage is observed on the upper part of this bank: Chlamys sp. aff. mollita (Rve.); Delectopecten macrocheilicola Habe; Keenaea sakuraii Habe; Pandora sp.; Turcicula crumpii Pilsbry; Trophonopsis echinus (Dali); Japonacteon archibenthicola Habe. Gravels from such depths are not covered by calcareous algae. This archibenthal assemblage almost agrees with that found from similar depths near Omu- rodashi reported by Igarashi and Kurata (1956). As has been stated, the molluscan community on the islands near the banks under study is allied to those from neighboring banks; while, out of biological dredge samples from stations lower than that, at a depth of 470-480 m, near Hachijo Island, a different archibenthal mollus- can assemblage is found ("Soyo” sta. B3, March 86 PACIFIC SCIENCE, Vol. XVII, January 1963 20, 1958, 33° 07.5' N, 140° 03.2' E, 480 m and do. Nov. 17, 1958, 33° 10.0' N, 140° 02.7' E, 470 m): Fusinus? sp.; Trophonopsis echinus (Dali); Nep tunea constricta (Dali), var.; Ben- thovoluta sp.; Conus sp. nov. The bottom is found to be of volcanic gravel which is not coated by algal matter. SUMMARY AND CONCLUSION As the result of examination of biological dredge samples taken from submarine banks around the Izu Islands, it became clear that pos- sible characteristic banks-associated molluscan assemblages are found in this area. Since they are closely associated with the bottom nature and the depth of the banks, the assemblages are composed mostly of the hard-bottom lower shelf fauna found on the southwestern Pacific coast of the Japanese main islands. The present material was collected at random with various kinds of gear, so that a quantita- tive analysis of the fauna was not possible. How- ever, constant occurrence and relative abundance of certain species were assumed as indicators of the fauna of the areas. The typical banks- associated molluscan assemblages from the Izu Islands area are possibly represented by such gastropods as Perotrochus beyrichii, Galeoastraea guttata, and Tenagodus anguinus, together with several pelecypods such as A car congenitum, Area mauia takii, Hawaiarca uwaensis, Samacar pacific a, Mimarcaria aizoi, Tucetona shinku- rosensis, Chlamys vesiculosus, C. tissotii, C. lem- niscata, C. mollita, Spondylus anacanthus, Plica- tula muricata, and Lima fujitai. Fig. 17. Chama argentata Kuroda and Habe from insular shelf of Takeshima Is., 11.6 mm X 10-1 mm. Fig. 18. Spondylus anacanthus (Mawe) from Uji Islet, 126 m, 40.2 mm X 37.5 mm. The geographical difference of the assemblage within the area is not so obvious, but it is clear that the insular shelf fauna is substantially the same as that under discussion. It seems that the species found in the area are those distributed also in warmer districts. This impression may be corroborated by facts such as these: 1 ) Similar banks-associated assemblages are found in southern waters off Kyushu which are much more subtropic in average ' climate. 2) Tucetona hanzawai and Samacar pacifica, which are found from the Izu Islands area, were originally described by Nomura and Zimbo (1934) from the Ryukyu Lime- j stone of Kikaijima (or Kikai-ga-shima) , an island located far to the southwest of ■ the Osumi Group. Concerning the fossil fauna found from this Ryukyu Limestone, j Nomura and Zimbo cited Yabe and Han- zawa’s opinion that the Ryukyu Lime- j stone was deposited in waters warmer than are found there at the present time, and that fossils found there are very sim- ilar to the recent fauna found around the j Molluscan Assemblages — Okutani 87 Fig. 19. Lima fujitai Oyama from Hyotanse, 118 m, 15.5 mm X 12.0 mm. Ryukyu Archipelago, which is situated farther south than Kikaijima Island. 3) S triarc a fausta is another example de- scribed from a fossil bed (at Moeshima in Kagoshima Bay, Kyushu) containing warmer water species. 4) Some elements of tropical origin are found in the banks fauna, e.g., Chicoreus superbus, which has been known from the Bonin (or Ogasawara) Islands, and Area mauia takii, which is presumably an en- demic form of a Hawaiian species. These two species are found at Kurose, which is the most southern bank in the area. An assumption that the molluscan assem- blage of the Izu Banks is an isolated shelf fauna representing species derived from warmer wa- ter corresponds to that of Parker and Curray (1956: 2438) for the Gulf of Mexico. They concluded from their studies that the banks- associated molluscan assemblages in waters off Texas represented a population now isolated from the main centers of abundance ranging from southeast Florida to the West Indies and different from the surrounding level-bottom communities. According to Teramachi (personal communi- cation), there are some distributional gaps of upper-shelf mollusca even within the warm Kuroshio area, though they are far less con- spicuous than those existing at about 36° N on the Pacific coast of Honshu. One such type of gap presumably is found around Ashizuri-Saki, for instance. The distributional difference of certain species of Fusinus, Siphonalia, Ancilla, and Fulgoraria off the east and west coast of the cape (about 34° 42' N, 133° E) may be found there. Regardless of this fact, the typical Fig. 20. Distribution of bank-associated mollusca. a, Galeoastraea guttata; b, Tucetona shinkurosensis; c, Chama argentata; d, Spondylus anacant bus -Lima fujitai-Plicatula muricata (or any two of these). 88 PACIFIC SCIENCE, Vol. XVII, January 1963 Fig. 21. Gians kyushuensis, sp. nov. (holotype), 16.0 mm X 13.0 mm X 5.5 mm. banks-associated molluscan assemblages are found from the discontinuous biotopes, i. e., southwestern Kyushu (about 129° 30'' E) and Izu Islands areas (about 140° E). Latitudinally, they are distributed from the southern tip of Izu Peninsula (34° 34' N) southward to Tori- shima Island (30° 30' N). Above all, this study shows how marine mol- lusca on the same level under similar ocean climate are governed in their distribution by the substratum. Moreover, a particular mollus- can assemblage (or simply a combination of the species) can serve to indicate the sedimen- tary characteristics of the banks in the Kuroshio area. REFERENCES Habe, T. 1958. On the radulae of Japanese ma- rine gastropods (4). Venus: Tap. Tour. Make 20(1): 43-60, 2 pis. Horikoshi, M. 1957. Note on the molluscan fauna of Sagami Bay and its adjacent waters. Sci. Rept. Yokohama Nat. Univ., sec. 11, (6): 37-64, 13 tfs., 1 pi, 2 maps. Igarashi, M.. and Y. KURATA. 1956. Fish and mollusks collected by trawl net. Survey Rept. Tokyo Fish. Exp. Sta. 4 (publ. 94): 32-40. Mimeo. [In Japanese.] Kuroda, T . and T. Habe. 1952. New locality of Perotrochus hirasei . Venus: Jap. Jour. Malac. 17(1): 5. [In Japanese.} Niino, H. 1935. On the bottom nature of the banks of Zenisu, Izu Islands. Chigaku Zasshi 47(562): 32-37, 1 pi, 2 tfs. [In Japanese.] 1952. The bottom characters of the in- sular shelf around Hachijo Island and the neighboring banks. Jour. Tokyo Univ. Fisher. 39(1): 101-110, 12 tfs. — — — 1955. On a manganese nodule and Pero- j trochus beyrichii dredged from the banks j near the Izu Islands, Japan. Rec. Oceanog. Wks. Japan. 2(2): 1-5, 1 pi, 1 tf. Nomura, S., and N. Zimbo. 1934. Marine mol- j lusca from the "Ryukyu Limestone” of Kikai- jima, Ryukyu Group. Sci. Rep. Tohoku Imp. ; Univ., 2nd ser. (Geol.) 16(2): 109-164. Parker, R. EL, and J. R. Curray. 1956. Fauna ; and bathymetry of banks on continental shelf, I Northwest Gulf of Mexico. Bull Amer. Assoc. Petrol Geol. 40(10): 2428-2439, 6 j tfs., 1 pi Sakurai, K. 1954. [A note.] Yumehamaguri 76: 22. Mimeo. [Journal in Japanese.] — — — 1959. [A note.} Yumehamaguri 97: 68. Mimeo. [Journal in Japanese.] Shirai, S. 1958. On some species of mollusca collected on the banks of Izu-Shichito. Venus: Jap. Jour. Malac. 20(1): 87-96, 4 tfs. [In Japanese.} Suzuki, K., and A. Sato. 1944. On the bottom nature in the waters from the southern ex- tremity of the Izu Peninsula to Kozushima and Toshima. Jour. Oceanogr. Soc. Japan. 3(4): 193-206. [In Japanese.] Wada, R. 1954. [A note.] Yumehamaguri 75: 170-171. Mimeo. [Journal in Japanese.} APPENDIX Gians kyushuensis Okutani, sp. nov. (Carditi- dae) Shell equivalve, dosed, stout, gibbous in Molluscan Assemblages — Okutani 89 shape; quadrangular in outline; beak anteriorly oblique; surface rough, uniformly pink in color, sometimes with delicate wavy pattern of darker color on posterior region; radial ribs running from umbo to distal margin, 20 or 21 in num- ber, as wide as shallow interspacial grooves, granulated on anteroventral ones, while sparsely spinose on rest; longest spinose scale as long as V5 of shell length; escutcheon narrow; lunule deeply impressed, cordate in outline. Interior smooth, with shallow radial grooves; ashy white with pinkish tint; on right valve, anterior cardinal teeth vestigial, while posterior ones elongated; on left valve, anterior cardinal teeth short, small but prominent, posterior ones elongated, plate-like in shape; ventral margin crenulated; mantle scar smooth. locality: 35° 50' N, 130° 28.5' E, 210 m in depth, gravels and shell. TYPES: Left odd valve: 16 mm long, 13 mm high, 5.5 mm thick (holotype). Left odd valve: 13 mm long, 11.5 mm high, 5 mm thick (para- type). Right odd valve: 18 mm long, 16 mm high, 7 mm thick (paratype). REMARKS: This new species is closely allied to G. hirasei (Dali),2 but the latter has more radial ribs which are closely scaly. This new species is distinguished from G. millegranosa (Nomura and Zimbo), 3 which has a strongly inflated shell and granulated radial ribs. DISTRIBUTION: The type locality and off Kii Peninsula (collected by Mr. Teramachi; con- joined valve measures 24 X 20 X 18 mm). 2 Dali, 1918: Proc. U. S. Nat. Mus. 54(2234) : 234. 3 Nomura and Zimbo, 1934: Sci. Rept. Tohoku Imp. Univ., 2nd ser., 16(2): 154(45), pi. v(i), figs. 13a, b, 14a, b. Also Okutani, 1958: Venus 20: 220, tfs. 2, 3. Notes on the Osteology and Systematic Position of Hypoptychus dybowskii Steindachner and Other Elongate Perciform Fishes1 William A. Gosline2 Longer ago that can gracefully be admitted, Dr. Paul Kahsbauer of the Vienna Naturhis- historische Museum was kind enough to send me a specimen of Hypoptychus dybowskii from Steindachner’s (1880) original series taken off "Northern Japan.” Steindachner placed this fish alongside the Ammodytidae, and there has been a division of opinion ever since as to whether it should be included in or excluded from that family (cf, Regan, 1913; Jordan, 1923; Duncker and Mohr, 1939; Berg, 1940). In order to in- vestigate its relationships, the Vienna specimen has been stained and dissected, and its oste- ology compared with that of the ammodytids Bleekeria gilli (Fig. la) and Ammodytes tobi- anus. The specimen of Bleekeria is Hawaiian and was retrieved from tuna spewings. Am- modytes is represented by two series, sent to me from the U. S. National Museum and the Museum of Comparative Zoology through the courtesy of Dr. L. P. Schultz and Dr. G. W. Mead, respectively. That Hypoptychus belongs to the superfamily Ammodytoidae seems certain. The relationships of the superfamily Ammodytoidae are more obscure. A second objective of the present in- vestigation has been to look into this matter, and a preliminary discussion of certain of the problems involved here will serve as an intro- duction to the paper. In a typical percoid fish, e.g., Epinephelus, there are 24 vertebrae, and the dorsal fin is composed of an anterior spinous portion and a posterior soft portion. In such a fish the ma- jority of the basal supporting elements, i.e., pterygiophores, of the spinous dorsal have a one-to-one relationship with the vertebrae be- low them; the soft dorsal rays and their ptery- 1 Contribution No. 1176 of the Hawaii Marine Laboratory and of the University of Hawaii Depart- ment of Zoology. Submitted July 25, 1961. 2 Department of Zoology, University of Hawaii. giophores, in contrast, are more closely spaced so that there is more than one ray and pterygio- phore to each vertebra. Time and again, how- ever, the percoids and their derivatives have become elongate. This change in shape is fre- quently accompanied by a whole series of other alterations. Thus, the cranial crests become low or disappear, the number of vertebrae increases, the distinction between dorsal spines and rays becomes reduced, both types of dorsal (and anal) rays develop an approximately one-to-one relationship with the vertebrae, and the caudal fin becomes rounded and its principal rays re- duced in number. All of these changes are to be found among the percoids, e.g., the Cepolidae, trachinoids, ammodytoids, blennioids, schindler- ioids, and most gobioids. Indeed, it seems that all of these modifica- tions occur together in the majority of elongate percoid derivatives and that those forms, such as the ophidioids, where there is more than one dorsal and anal ray per vertebra are the excep- tion rather than the rule. On the other hand, the author is aware of no prepercoid teleost with a one-to-one relationship between soft dor- sal and anal rays and the vertebrae. If what has just been said is correct, it follows that any fish with such a relationship is a percoid deriva- tive, but that the unit correspondence between soft rays and vertebrae is of little use in dis- tinguishing the various lineages of percoid der- ivation. Here, the schindlerioids and gobioids will be dismissed from further consideration, as each of these groups has peculiarities by which it may easily be defined. However, Hawaiian specimens of the trachinoids Parapercis schauinslandi (Par- apercidae, Fig. 1 c) and Crystallodytes cookei (Trichonotidae, Fig. lb), and of the blennioid Tripterygion atriceps ( Tripterygiidae, Fig. Id) have been stained and dissected. These speci- mens will be used both for purposes of com- 90 Elongate Perciform Fishes— Gosline 91 Fig. 1. Sketches of a, Bleekeria gilli, from Gosline and Brock, after Fowler, based on a specimen 3 inches in total length; b, Crystallodytes cookei, based on a specimen 2 inches long; c, Parapercis schauinslandi, from a 3 -inch specimen; and d, T riptery gion atriceps, from a 1-inch fish. parison with the three ammodytoids and in an attempt to obtain some understanding of the lineages to which each of them belongs. With regard to these six fishes that have been investigated in some detail, it may be stated in advance that the author has not had any great success in discovering significant cranial differences between them. It is not so much that such differences do not exist, as that they appear to have rather slight systematic value. Though it may be that the author has simply overlooked significant differences, it would ap- pear that the percoids and their immediate derivatives have a rather standardized skull pat- tern and that the major morphological differen- tiation of percoid groups has occurred in other features. lacrimal (preorbital) followed by about five sepa- rately movable, canal-bearing ossicles (cf, Ka- tayama, 1959: figs. 2-5). Above the fifth, the infraorbital lateral line canal joins the supraorbital canal. Of the five ossicles the upper- most is particularly variable and is sometimes fused to and sometimes free from the sphenotic. In Parapercis (Fig. 2a) the infraorbital canal is complete, passing through a lacrimal and six separate circumorbital bones. The upper- most of these is firmly attached to the cranium in Parapercis. Because six circumorbitals ap- peared to be a high number, the opposite side of the same specimen and of a larger specimen of Parapercis schauinslandi were checked; no variation was found. The circumorbital struc- tures of T riptery gion differ from those of Para- percis in having three instead of six circum- orbital bones and in the failure of the bone to close over the sensory canal externally. In Crys- tallodytes the circumorbital canal is still com- plete but there are only a lacrimal and two circumorbitals. The lacrimal and second circum- orbital are large and laminar, but the anterior circumorbital is quite small. In Ammodytes (Fig. 2b) there is a large lacrimal, followed immediately by a moderate- sized first infraorbital; then there is a broad gap followed by two small circumorbitals, the HEAD SKELETON CIRCUMORBITAL BONES: In the typical per- coid the circumorbital series is made up of the FIG. 2. Lacrimal and circumorbital bones of a, Para- percis schauinslandi, and b, Ammodytes tobianus. There are no circumorbital bones bordering the por- tion of the orbit indicated by the dashed line in Ammodytes. co, Circumorbital bones; and la, lacrimal. 92 PACIFIC SCIENCE, Vol. XVII, January 1963 upper of which articulates with the skull. In Bleekeria a similar break in the circumorbital ring occurs, but it is shorter than in Ammo- dytes and the posterior series seems to contain three or four small elements instead of two. Hypoptychus has the same two anterior ele- ments followed by a broad gap; posteriorly, however, there is only a single ossicle, and it is fused to the sphenotic. None of the six fishes have any subocular shelf from the circumorbitals. JAWS: In sand-diving fishes like Ammodytes the mouth is usually not terminal; either it is withdrawn below an overhanging snout, as in Crystallodytes, or protected by a prognathous chin, as in Ammodytes. In Ammodytes the leading, lower jaw is firmly attached, but the upper has developed excessive powers of pro- trusion when the mouth is opened (van Dob- ben, 1935: 34-36). The great protrusibility of the upper jaw in Ammodytes is accompanied by a weakening of the bony elements, and it is probably in relation to this that Ammodytes and Bleekeria have edentulous premaxillaries. So far as jaw structure is concerned, Hypopty- chus is intermediate between the normal per- coid type and the specializations found in Am- modytes and Bleekeria. The premaxillary of Ammodytes (Fig. 3 a) consists of a long pedicel movably articulating at its base with the remaining portion of the premaxillary. The distal half of the maxillary tapers gradually to a point (Fig. 3 a). The upper jaw of Bleekeria is essentially similar to that of Ammodytes except that a number of small ossicles are to be found in the ligamentous tissues connected with the jaw apparatus. Thus, there is an ossicle above the more lateral of the two pedicels of the premaxillary, another at the distal end of the premaxillary, and a whole series in the ligamentous tissue that runs be- tween the upper and lower jaws. The upper jaw of Hypoptychus (Fig. 3 b) differs from those of Bleekeria and Ammodytes in the following features: the premaxillary bears a row of teeth (there are about 14 conical teeth in a single row on each side, not shown in Fig. 3 h)\ the premaxillary is fused to its pedicel; and the tip of the maxillary is expanded distally. A movable articulation between the pre- Fig. 3. Head skeletons of ammodytoids: superior views, with the premaxillary somewhat protruded, of a, Ammodytes tobianus, and b, Hypoptychus dybow- skii; lateral view, c, of cranium of Ammodytes tobi- anus. ba, Basisphenoid; bo, basioccipital; ca, cartilage; co, circumorbital; eo, exoccipital; ep, epiotic; fr, fron- tal; ha, anterior hyomandibular socket; hp, posterior hyomandibular socket; in, intercalar; la, lacrimal; le, lateral ethmoid; me, mesethmoid; mx, maxillary; na, nasal; pa, palatine; pc, prootic; pi, pleurosphenoid; po, posttemporal; pr, parietal; ps, parasphenoid; pt, ptero- tic; px, premaxillary; so, supraoccipital; sp, sphenotic; tb, tabular; tf, trigemino-facial foramen; and vo, vomer. In the superior view of Ammodytes, a, the epiotic is covered by the posttemporal. maxillary and its pedicel is also found in Crys- tallodytes, as well as in blennioids such as Cirripectus and Istiblennius. In Parapercis and Tripterygion, which have strong premaxillary teeth, the pedicel is stout and fused to the toothed portion. gill covers: In Ammodytes (Fig. 4b) and Crystallodytes (Fig. 4c) the subopercles are ex- panded, presumably to protect the throat region. Indeed, the lower border of the articular in Crystallodytes is greatly expanded below as well (Fig. 4c). SUSPENSORIUM: The suspensorium of Am- modytes is specialized in a number of regards (Fig. 4b). Most notable among these is the elongated palatine strut. The whole length of this strut from its forward tip to its articulation with the quadrate is made up of the palatine Elongate Perciform Fishes — -Gosline 93 bone; the ectopterygoid is a minute triangular splint at the very base. In Bleekeria the struc- ture of the suspensorium is essentially similar except that the ectopterygoid is somewhat larger so that the palatine does not meet the quadrate. The suspensorium of Hypoptychus (Fig. 4a) is a quite different structure. The palatine and ectopterygoid are about equal in size and are united to one another by a digitate suture. The metapterygoid is a small splint and the meso- pterygoid appears to be absent. Undoubtedly the greatest specialization in the suspensorium is that found in Crystallo- FlG. 4. Right gill covers, suspeosoria, and lower jaws, external view, of a, Hypoptychus dybowskii; b, Ammodytes tobianus; c, Crystallodytes cookei; and d, Tripterygion atriceps (with the lower jaw dislocated). an, Angular; ar, articular; de, dentary; ec, ecto- pterygoid; hy, hyomandibular; io, interopercle; mt, metapterygoid; mx, mesopterygoid; op, opercle; pa, palatine; pp. preopercie; qu, quadrate; sb, subopercle; and sy, symplectic. dytes (Fig. 4c). Here the suspensorium is divided into two well-developed and strong portions with the ectopterygoid forming a long, delicate strut between them. In the anterior portion a large, firm mesopterygoid forms a shelf under a large part of the eyeball; it is firmly attached to the strong palatine anteriorly, but only by membrane to the ectopterygoid. SKULL: In all of these fishes, there are no frontal-parietal crests, and the minute supra- occipital crest does not reach above the skull surface. Tripterygion , however, has a fringed, backwardly slanted crest running across the rear of the skull. This crest lies just behind the tabular ossicle on each side which bears the apparently incomplete supratemporal sensory canal commissure. ( In the related Entomacrodus the supratemporal commissure is almost com- pletely enclosed in the skull. Laterally, the com- missure passes through a tabular ossicle that is fused to the cranium and thence medially through the rear of the parietals, leaving a large opening on the middorsal line.) In Parapercis and Tripterygion the crania are somewhat more highly arched over the orbit than in the others. Probably in association with this, the wings of the parasphenoid extend farther up the sides of the postorbital bar than in the remaining four fishes. In all, however, the prootic extends over the top of the para- sphenoid wings to the edge of the orbit. (In Istiblennius, related to Tripterygion, the para- sphenoid wings meet the frontals in the usual blennioid fashion.) In Ammodytes and Bleekeria the two exoc- cipital condyles lie adjacent to one another and form the upper portion of the facet for the articulation of the convex head of the first vertebra. In the other forms, including Hypo- ptychus, the exoccipital condyles lie at either side of the basioccipital articulation; the two exoccipital bones do not meet below the fora- men magnum; and there is no rounded articu- lar head on the first vertebra. GILL ARCH SYSTEM: In Ammodytes the bran- chiostegal ray count is 8-7; in Crystallodytes, 7; in Bleekeria , 7; in Parapercis and Tripterygion, 6; and in Hypoptychus, 4-4. In all the fishes under consideration the lower pharyngeals are separate. Ammodytes and 94 Bleekeria appear to be the only forms with 3 distinct upper pharnygeals on each side; Hypo- ptychus has separate upper pharyngeals on arches 2 and 3, but appears to have none on arch 4. Parapercis and Crystallodytes also have two pairs of upper pharyngeals, but the posterior pair seems to represent a combination of pharyngobranchials 3 and 4. Tripterygion ap- pears to have only a single set of upper pharyn- geals. FINS, FIN SUPPORTS, AND AXIAL SKELETON ANAL FIN: In the six fishes under considera- tion there is never more than a single unseg- mented ray at the front of the anal fin, and even this is lacking in Crystallodytes and Ammodytes. All of the remaining anal rays are branched in Parapercis, some in Hypoptychus, only the last in Tripterygion, and none in Crystallodytes, Bleekeria, and Ammodytes. Unlike the other three fishes, there is in the three ammodytoids a well-developed caudal peduncle behind the base of the last anal (and dorsal) ray; this is supported by about five vertebrae with bladelike neural and hemal arches. DORSAL FIN: Parapercis and T rip ter y gion are the only fishes under consideration that have spinous dorsals. Furthermore, in these two the dorsal fins commence farther forward (over the 3rd vertebra in Parapercis, Fig. 3 a, the rear of the skull in Tripterygion, Fig. 5b) than in the others (over the 5th vertebra in Bleekeria, and still farther back in the remaining forms). Structurally the spinous dorsal fin differs con- siderably in Parapercis and Tripterygion. In Parapercis it appears that the spinous dorsal has undergone some condensation, possibly as a result of forward movement of the soft dorsal, for the pterygiophores of the five spines in- terdigitate between neural arches 2 and 5 (Fig. 5a) \ one supraneural remains (rather than the three usually found in the lower percoids). In Tripterygion there are two spinous dorsals, the first of 3 spines and the second of 14; it appears very much as if the anterior 3 have appropriated the usual percoid supraneurals as their support- ing bases. In the structure of the pterygiophores supporting the dorsal spines, Parapercis is con- siderably more generalized than Tripterygion. PACIFIC SCIENCE, Vol. XVII, January 1963 Fig. 5. Anterior vertebrae, ribs, dorsal rays and their supports in a, Parapercis schauinslandi, and b, Tripterygion atriceps. dr, Dorsal soft ray; ds, dorsal spine; ep, epipleural rib; na, neural arch; ns, neural spine; pb, pleural rib; pg, pterygiophore; sh, supra- neural; so, supraoccipital. In Parapercis the pterygiophores of the spines (except that of the first 2) retain their basic bisegmental structure (Fig. 5a) \ whereas in Tripterygion each pterygiophore is a fused monolithic unit (Fig. 5b). In the soft dorsal, as in the anal, all the rays are branched in Parapercis, some in Hypopty- chus, only the last in Tripterygion, and none in Crystallodytes, Bleekeria, and Ammodytes. In soft dorsal structure, there are again certain dif- ferences between Parapercis and Tripterygion on the one hand, and Crystallodytes and the ammodytoids on the other. In the first place, Parapercis and Tripterygion have the last dorsal (and anal) ray cleft to the base; Crystallodytes and the ammodytoids do not. Second, the ptery- giophore of each soft dorsal ray in Parapercis and Tripterygion interdigitates deeply between a pair of neural spines (Fig. 5a), and there is an exact correspondence between vertebrae and soft dorsal rays. In Crystallodytes and the ammodytoids the pterygiophores of the soft dorsal ( and anal ) rays are short, weak structures Elongate Perciform Fishes— Gosline 95 that interdigitate little if at all between the tips of the neural spines, and there is a rough but inexact correspondence between soft dorsal (and anal) rays and vertebrae. CAUDAL FIN: The tails of the ammodytoids are somewhat forked, those of the other fishes under investigation more or less rounded. In all, there is a reduction in the caudal ray num- ber from the typical percoid count of 17 prin- cipal rays, 15 branched. In Parapercis there are 15 branched rays, but no outer principal un- branched rays. In Ammodytes and Bleekeria, there are 15 principal rays, 13 branched. In Hypoptychus, there are 13 principal rays; ap- parently 11 of these were branched, but since the fin rays of the available specimen are broken the branched ray count cannot be definitely established (the same is true of the dorsal, anal, and pectoral fins). Tripterygion and Crys- tallodytes have 10 principal rays, 8 branched. With regard to the caudal skeleton, Paraper- cis (Fig. 6a) is quite typically percoid (Gos- line, 1961). There are six separate hypurals, one uroneural, and three epurals; none of these elements are fused to the urostyle. In the caudal skeleton of the other five fishes, considerably more fusion and/ or reduction has occurred. Hy- purals 4 and 5 are always fused with the urostyle, and, in Crystallodytes (Gosline, 1955: fig. 7d) and Hypoptychus (Fig. 6d ), two or three of the lower hypurals as well. ( Fig. 6d must be viewed with some reservation, as the specimen from which it was drawn may have been aberrant in having the last two vertebrae fused.) There are two epurals in Tripterygion (Fig. 6b) and the three ammodytoids, and only one in Crys- tallodytes. PECTORAL FIN: The total number of pec- toral rays in the fishes investigated is 15 in Tripterygion and Parapercis, 13 in Ammodytes and Bleekeria, 12 in Crystallodytes, and 9 in Hypoptychus. Of these, all are segmented in Tripterygion and the ammodytoids; however, there is a small, unsegmented, splintlike upper- Fig. 6. Caudal skeletons of a, Parapercis schauins- landi; b, Tripterygion atriceps; c, Bleekeria gilli; and d, Hypoptychus dybowskii. ce, Centrum; ep, epural; hr, hemal arch; hs, hemal spine; hy, hypural; na, neural arch; ns, neural spine; un, uroneural; and ur, urostyle. 96 PACIFIC SCIENCE, Vol. XVII, January 1963 most ray in Parapercis and Crystallodytes. Judg- ing from the material available, it appears that the uppermost pectoral ray is homologous whether splintlike or segmented, for it has the same very peculiar basal structure. Like other soft rays it consists of two halves. However, in the uppermost pectoral ray the two halves are usually not mirror images of one another. Rather, the scapula articulates with a facet that lies entirely on the inner "half” of the ray (except, among the fishes investigated, in Hy- poptychus and Trip ter ygion ) , An inquiry into this peculiarity has shown that the scopeliform genus Sy nodus has a small bony nodule that lies between two equal halves, but is attached to the inner. As Starks (1930: 238) noted, Fig. 7. Right pectoral girdles, external view, of a, Tripterygion atriceps; b, Bleekeria gilli; c, Hypo- ptychus dybowskii; and d, Ammodytes tobianus. Only the uppermost and lowermost pectoral rays are shown in a, b and d; the top of the pectoral girdle is not drawn in b; and the lower postcleithmm is not in- dicated in a. ac, Actinost; cl, cleithrum; co, coracoid; pc, postcleithrum; pe, pelvic girdle; pm, postcleithrum; py, pectoral ray; sc, scapula; and scl, supracleithmm. this nodule probably represents a modified ac- tinost, which in many higher teleosts became incorporated into the inner, articular "half” of the uppermost pectoral ray (Fig. lb, d). The pectoral girdles of Tripterygion and of the ammodytoids are shown in Fig. 7. Those of the ammodytoids are peculiar in having the j supracleithra and cleithra more or less vertically aligned. There are two postcleithra in Parapercis, Tripterygion, Ammodytes, and Bleekeria, and apparently not any in Crystallodytes and Hy- poptychus, PELVIC FIN: Pelvic fins are lacking in the three ammodytoids studied here. However, two splintlike pelvic girdle elements are to be found J! in Bleekeria (Fig. 7b), and a small pelvic fin of a spine and three rays, located somewhat ahead of the pelvic bases, occurs in the related ammodytid genus Embolichthys (Jordan, 1902). The three other fishes studied here also have the pelvics originating ahead of the pectoral bases. Parapercis has a pelvic spine and five branched rays, the fourth considerably the long- est. Crystallodytes has a short pelvic consisting of a spine and five unbranched but segmented rays. In Tripterygion there are two well- developed unbranched, but segmented, rays; there is no spine. In all three fishes the pelvic girdle articulates anteriorly with the cleithra.' In Crystallodytes (Fig. 8 d), the two halves of the pelvic girdle are rather widely separated an- teriorly; in Parapercis (Fig. 8b) they are united for nearly their entire length; in Tripterygion (Fig. 8c) the two halves are not only united, but anteriorly they seem to have completely fused. Futhermore, Tripterygion has the pelvic girdle firmly wedged between the cleithra. VETEBRAL COLUMN AND RIBS: Parapercis has 30 ( 10 -j- 20) vertebrae (including the uro- style); Tripterygion , 34 (10 + 24); Crystal- lodytes, 55 (29 + 26); Ammodytes lanceo- latns, according to Regan (1913: 137), 69 (40 + 29); Bleekeria gilli, according to Duncker and Mohr (1939: 13), 57 (32 + 25); and Hypoptycbus , 55 (31 + 24). In Parapercis, Tripterygion, Ammodytes , and Bleekeria there are two sets of ribs. The lower, or pleural, ribs start from the 3rd vertebra; the upper, or epipieural, ribs start from the 1st, articulating with the upper surface of the pleu- ral ribs from the 3rd vertebra on (Fig. 4a, b) . In Crystallodytes there is only one set of ribs which starts with the 1st vertebra; this set would appear to represent the epipleurals. In Hypoptychus, there is also only one series of ribs, but it starts from the 3rd vertebra and would appear to represent the pleural series. DISCUSSION On the basis of osteological characters, Par- apercis, Crystallodytes , Tripterygion, and the three ammodytoids could be grouped in a num- ber of ways. One such system would separate the more elongate forms, i.e., Crystallodytes and the ammodytoids, from the shorter, stockier Parapercis and Tripterygion. Such a division could be expressed osteologically as follows. Parapercis and Tripterygion. Premaxillary pedicels stout, firmly fused to the toothed por- tions. Wings of the parasphenoid expanded, forming the lower portion of the postorbital bar. The abdominal portion of the vertebral column shorter than the caudal portion, consist- ing of 10 vertebrae. Dorsal with an anterior spinous portion which commences over or ahead of the 3rd vertebra. Last dorsal and anal rays cleft to their bases. Pterygiophores of the dor- sal and anal rays deeply interdigitating between successive neural and hemal spines, respectively. Crystallodytes and the ammodytoids. Pre- maxillary pedicels long and/or movably articu- lated with their lateral portions. Wings of the parasphenoid little developed, not extending up as a portion of the postorbital bar (Fig. 3c). The abdominal portion of the vertebral column longer than the caudal portion, of more than 10 vertebrae. No spinous dorsal, the soft dorsal commencing over or behind the 5th vertebra. Last dorsal and anal rays undivided. Pterygio- phores of the dorsal and anal rays interdigitat- ing little or not at all between the neural and hemal spines, respectively. To group the fishes under consideration in Fig. 8. Pelvic girdles, from above, of a, Caranx ignobilis; b, Parapercis schauinslandi; c, Tripterygion atriceps; and d, Crystallodytes cookei. The pelvic rays are not indicated in a. cl, Cleithrum; pg, pelvic girdle; and py, pelvic ray. 98 PACIFIC SCIENCE, Vol. XVII, January 1963 the above fashion is to separate those nearer the generalized percoid type from those that are more specialized. Such a division may merely represent levels of structural organiza- tion rather than relationships. Parapercis is much the least differentiated from the typical percoids of any of the six fishes dealt with in this paper. All of the re- maining five fishes ( Tripterygion , Crystallo- dytes, Ammodytes , Bleekeria, and Hypoptychus ) have in common the following specializations over and beyond those found in Parapercis : circumorbital series of bones incomplete or complete and consisting of a lacrimal and only 2 or 3 circumorbitals; no pungent dorsal spines; caudal with 13 or fewer principal rays; uro- style fused to the upper hypurals; pelvics re- duced (i.e., without branched rays) or absent. Most of the minor specializations that Para- percis does exhibit seem to be associated with fin structure. Thus the pelvics are advanced in position and have the inner (actually the 4th) soft rays the longest, but there appears to be no great specialization of the pelvic girdle (Fig. 8b). The caudal fin lacks the usual principal unbranched rays, though the caudal skeleton is typically percoid (Fig. 6a). The spinous por- tions of the dorsal and anal have been reduced. In the anal there is a single small unsegmented splint at the front of the fin. The spinous dorsal seems to have been pushed forward and con- centrated as well as reduced in size, for there is only one supraneural and the pterygiophores for all five spines extend between neural arches 2 and 5. The soft dorsal and anal retain a typical percoid condition. However, their pterygio- phores interdigitate deeply between successive neural and hemal spines and bear a one-to-one relationship with the vertebrae. Now in typical percoid families, though not apparently in the Cepolidae, the soft dorsal and anal rays usually do not interdigitate deeply between the neural and hemal spines, and there are usually about two pterygiophores per vertebra. This is true of even fairly elongate forms like the goatfish, Mulloidichthys samoensis. To change this ar- rangement to a one-to-one relationship between pterygiophores and vertebrae requires either an increase in the spacing between soft dorsal and anal rays, or an increase in the number of verte- brae, or both. The cigar-shaped labrid Cheilio inermis seems to be an instance where a one- to-one ratio has been brought about by in- creased spacing between rays, for this fish re- tains 24 vertebrae. In most elongate fishes, how- ever, an increase in vertebrae has also taken place. On the basis of Parapercis alone it is impos- sible to evaluate Regan’s (1913) percoid "Di- vision Trachiniformes” (equals superfamily Trachinoidae) . Suffice it here to remark that there does appear to be a somewhat extensive | group of usually rather deep-water bottom fishes having essentially the fin characters described above for Parapercis. Whether these fishes are really related is impossible to say at this point. The relationship of Crystallodytes to this group is also doubtful. Certainly Crystallodytes repre- sents a much higher degree of differentiation from the typical percoids than Parapercis. (For certain characteristics of Crytallodytes , see be- low.) Some knowledge of less specialized (or at least of other) forms of the Crystallodytes lineage should provide far better indications of its relationships than are available from a study of this form alone. (Throughout this paper, Crystallodytes has been considered a trichonotid, but that this is a correct family allocation is dubious, cf, Schultz, I960: 273.) Tripterygion is generally agreed to be a mem- ber of the perciform suborder Blennioidei. How- ever, this suborder, since it was defined and later restricted by Regan (1912, 1929), has under- gone considerable disintegration and rearrange- ment (Starks, 1923; Smith, 1952; Hubbs, 1952; Gosline, 1955; and Makushok, 1958). Even after certain nonblennioid groups have been re- moved, Hubbs and Makushok feel that most or all of the remaining families may be divided into a northern (cold water) group and a southern (warm water) group which may have had independent origins among the percoid families. Tripterygion is, in many respects, one of the more generalized, i.e., percoid, members of the whole warm water group. This group differs radically from the remaining fishes under con- sideration here in its mode of life. The southern blennies characteristically use their thickened pelvic rays to prop the anterior end of the body Elongate Perciform Fishes — Gosline 99 away from the hard substratum on which they feed. The more elongate forms at least charac- teristically rest with the tail bent, and, when disturbed, retreat into holes in the rock and coral by means of sinuous movements of the body. A number of the characteristic external features of the southern blennies are probably associated with this mode of life. For example, the pelvic rays though reduced in number are stout, and are attached to a short pelvic girdle that is firmly wedged between the wings of the cleithra. The dorsal fin extends far forward, in the extreme case of Xiphasia to above the eye. There is often a pair of tentacles or a transverse fringe of them, e.g., Cirripectes, on the nape; in Tripterygion there is a low transverse fringe across the nape in exactly the same position as in Cirripectes , but it is made up of bony flaps extending upward from the skull. A few other characters of Tripterygion atri- ceps may be mentioned because of their bearing on blennioid classification. The lateral line canal of the lacrimal and three circumorbitals is not covered by bone externally (Hubbs, 1952: 48, 50). A basisphenoid is present (Makushok, 1958: 58). The lowermost actinost in the pectoral girdle is not greatly longer than deep (Makushok, 1958: 58; compare Fig. la of the present paper with Makushok’s fig. 25). Finally, in the caudal skeleton of Tripterygion (Fig. 6b) the three lower hypurals have fused to one another. Hypurals 4 and 5 (above the axis ) have also fused to one another and to the urostyle and uroneurals. Hypural 6 is a small separate ossicle (the "minimum hypural" of Makushok, 1958), and there are two broad epurals. In the northern blennies, by contrast, the upper or epaxial hypurals are usually, though not always, separate from the urostyle (Maku- shok, 1958: 38, and fig. 22). Also, the northern blennioids usually have three epurals, rather than the two of Tripterygion. (In 1955: fig. 7f, I provided a sketch of the caudal skeleton of Istiblennius gibbifrons. I have not been able to relocate the specimen from which the draw- ing was made, but judging from specimens of Entomacrodus marmoratus and Istiblennius ze- bra, which have caudal skeletons very like that of Tripterygion, the figure is incorrect in showing a fusion of the lower hypurals and the epurals with the rest of the urostylar vertebra.) The investigation of Tripterygion reported on here would support, in its small way, Hubbs’ (1952) and Makushok’s (1958) thesis that the "northern” and the "southern” blennies are diphyletic. But whether they are diphyletic in the sense that the two groups have wholly dif- ferent origins or in the sense that they have diverged in two different directions from the same or from two closely related stocks would seem to remain an open question. That Trip- terygion has little relationship to the congrog- adids (cf, Smith, 1952) also seems clear. Among the more elongate fishes investigated, i.e., Crystallodytes and the ammodytoids, Crys- tallodytes and Arnmodytes at least are sand div- ers. The pointed heads, long bodies, low vertical fins, small or absent pelvics, modified scales, and peculiarities of the lateral line of the body are probably associated with this mode of life. The habits of the other two ammodytoids, Bleekeria and Hypoptychus, are unknown. That there is any phylogenetic relationship between Crystallodytes and the ammodytoids seems extremely doubtful. Crystallodytes differs immediately from the ammodytoids in the fol- lowing characters. The lower jaw is included and the upper appears to be nonprotrusile. The circumorbital series of bones is complete. The cleithra are very oblique with the lower ends far forward of the upper. The dorsal and anal extend far back, and there is no well-demarcated caudal peduncle. Both the principal and procur- rent fin rays are reduced in number. The lateral line runs low on the body posteriorly, just above the anal base. Finally there is a series of specializations related to the eye and suspen- sorium of Crystallodytes. In the first place, the eye itself is very characteristic, for Crystallo- dytes is a pop-eyed fish (Fig. 1 b) with the small pupil protruding notably beyond the rest of the eyeball. This eyeball is supported below by a large and firm subocular shelf composed of the greatly expanded mesopterygoid. This bone, together with the palatine, forms a nearly sepa- rate portion of the suspensorium only weakly attached to the rest of this structure by the long, weak ectopterygoid (Fig. 4c). Neither the sus- pensorium nor the peculiar eyeball shows any relationship whatever to those structures in 100 Parapercis, which is typically percoid in these respects. Nor does Crystallodytes show any re- lationship to any of the other fishes investigated in these structures. Indeed, the only fishes that would seem to have a suspensorium anything like that of Crystallodytes are the congrogadids (cf, Regan, 1912: fig. 2b, and Smith, 1952: pi. 6B). The three ammodytoid fishes may be defined as follows: elongate fishes with premaxillati.es highly protrusile. Circumorbital bones incom- plete, the lacrimal and first circumorbital sepa- rated from the rest of the series. Fins without spines or unsegmented rays except for the pro- current rays of the caudal (pelvic rays of Em- bolichthys?) . Caudal forked or emarginate, pre- ceded by a well-marked caudal peduncle which is supported by five or more vertebrae with bladelike neural and hemal spines. P'elvics ab- sent (of a spine and three rays in Embolicbtbys, Jordan, 1902). Vertebrae 55 to 69, the abdom- inal vertebrae more numerous than the caudal, in approximately a one-to-one relationship with the dorsal and anal rays above and below them. Cleithra and supracleithra almost verti- cally aligned. The analysis of Hypoptycbus indicates that it is widely separated from Ammodytes and Bleekeria, Though the three genera hold a num- ber of features in common it would seem that Hypoptycbus has evolved in quite a different direction from the other two. Thus, while Hy- poptycbus remains more percoid in jaw struc- ture and skull-vertebral column articulation, it has become more specialized (degenerate?) in almost every other feature: the bones are thin; the branchiostegal rays and fin rays are reduced in number; the scales have been completely lost; etc. Hypoptycbus well warrants the sepa- rate family Hypoptychidae apparently first as- signed to it by Jordan (1923: 230). The families Hypoptychidae and Ammody- tidae may be contrasted as follows. Hypoptychidae. Scales entirely lacking. Jaws subequal, the premaxillary with teeth and with its long pedicel firmly attached. Branchiostegal rays 4. Articular facets of the exoccipitals widely separate. Dorsal and anal fins equal in length, of about 20 rays. Caudal with 13 principal rays; pectorals with 9. PACIFIC SCIENCE, Vol. XVII, January 1963 Ammodytidae. At least some scales present. Lower jaw protruding, sharptipped. Premaxil- lary without teeth and with its pedicel movably attached to its lateral portion. Branchiostegal rays 7 or 8. Articular facets of exoccipitals con- tiguous. Dorsal extending far forward of the anal. Caudal with 15 principal rays; pectorals with 13. The problem of ammodytoid origin remains obscure. The majority of features point to a percoid origin of some sort, but none of the percoid families known to the author would seem to provide a suitable ancestor. The ammodytoids in turn would appear to have led to nothing with the exception of one highly speculative possibility. If the terminal vertebra of Hypoptycbus is not merely the re- sult of fusion in an aberrant specimen, then a progressive evolution along many of the lines already apparent in that fish might end in a neotenic form very like Schindleria (cf, Gos- line, 1959). Whatever the ancestors and derivatives of the ammodytoids may be, they remain, so far as known, sufficiently isolated and characterized to warrant fully the superfamily status among the Percoidei that has generally (cf, Regan, 1913) been assigned to them. REFERENCES Berg, L. S. 1940. Classification of fishes, both recent and fossil Trav. Inst. Zool. Acad. Sci. U.R.S.S., Leningrad, 5: 87-517, 190 figs. VAN Dobben, W. H. 1935. liber den Kiefer- mechanismus der Knochenfische. Arch. Neer- landaises de Zool. 2: 1—71, 50 figs. L> unck.ee, G , and E. Mohr. 1939’. Revision der Ammodytidae. Mitt. Zool. Mus. Berlin 24: 8-31, 4 figs. Goslime, W. A. 1955. The osteology and re- lationships of certain gobioid fishes, with particular reference to the genera Kraemeria and Microdesmus. Pacific Sci. 9(2): 158- 170, 7 figs. — — — - 1959. Four new species, a new genus, and a new suborder of Hawaiian fishes. Pa- cific Sci. 13(1): 67-77, 6 figs. Elongate Perciform Fishes— Gosline 101 - — 1961. The perciform caudal skeleton. Copeia, 1961: 265-270, 3 figs. Hubbs, Clark. 1952. A contribution to the classification of the blennioid fishes of the family Clinidae, with a partial revision of the Eastern Pacific forms. Stanford Ichth. Bull. 4: 41-165, 64 figs. Jordan, D. S. 1902. Supplementary note on Bleekefia mitsukurii, and on certain Japa- nese fishes. Proc. U.S. Nat. Mus. 25: 693— 696, pi. 30, 1 fig. — — 1923. A classification of fishes includ- ing families and genera as far as known. Stanford Univ. Publ, Univ. Sen, Biol. Sci. 3: 79-243. KATAYAMA, M 1959. Studies on the serranid fishes of Japan ( 1 ) . Bull. Faculty Educ, Ya- maguchi Univ. 8: 103-180, 39 figs. Makushok, V. M. 1958. The morphology and classification of the northern blennioid fishes ( Stichaeoidae, Blennioidei, Pisces ) . Trudy Zool. Inst. Akad. Nauk S.S.S.R. 25: 3-129, 83 figs. (In Russian; English translation mimeographed. ) REGAN, C. T. 1912. The classification of the blennioid fishes. Ann. Mag. Nat. Hist., ser. 8, 10: 265-280, 4 figs. — _ — __ 1913. The classification of the percoid fishes. Ann. Mag. Nat. Hist., ser. 8, 12: 111-145. — — - — - 1929. Fishes. In: Encyclopaedia Britan- nica, 14th ed., pp. 305-329. Schultz, L, P, et al. I960. Fishes of the Mar- shall and Marianas Islands. Vol. 2. U. S. Nat. Mus. Bull. 202: 1-438, pis. 75-123, 132 figs. Smith, J. L. B, 1952. The fishes of the family Haliophidae. Ann. Mag. Nat. Hist., ser. 12, 5: 85-101, pi. 6, 2 figs. Starks, E. C. 1923. The osteology and relation- ships of the uranoscopoid fishes. Stanford Univ. Publ., Univ. Ser., Biol. Sci. 3: 259-290, 5 pis. — — 1930. The primary shoulder girdle of the bony fishes. Stanford Univ. Publ., Univ. Ser., Biol. Sci. 6: 149-239, 38 figs. Steindachner, F. 1880. Ichthyologische Beit- rage (IX) . Sitzungsberichte k. Akad. Wiss., Wien, math.-naturwiss. Classe 82: 238-266, pis. 1-6. Observations and Experiments on the Food Habits of California Sea Hares of the Genus Aplysia Lindsay R. Winkler1 and E. Yale Dawson2 In A PREVIOUS paper the senior author (Wink- ler, 1959*2) presented the results of fecal pellet analyses to determine the influence of diet on the color of Aplysia calif ornica Cooper. Only the dominant plants were mentioned in con- nection with that immediate problem. The pres- ent paper contains more detailed information from that study and from subsequent observa- tions and reviews of references to Aplysia food plants. The food of A. calif ornica has received only limited consideration in connection with other studies. This is also true, however, of Aplysia species in general. Even in the large monograph of Mazzarelli ( 1893) the only indication of the nature of the food is in a discussion of the method of feeding, where we find the phrase . . algae (ordinarily Ulva lactuca ) which serves as nutriment.” In the extensive study of European Aplysia punctata by Eales (1921), about half a page is concerned with food. She notes an apparent change in food habits with age. The young, she says, live on red seaweed like Deles seria [probably D. sanguined] and then slowly migrate through the Laminaria zone, where they feed on Laminaria, into shal- low water, where the fully grown adults live on Fucus and Ulva. However, since she also states that they are found in the greatest numbers in the Zoster a beds, it may be that like Aplysia calif ornica they also feed on Zoster a. As to the feeding habits of A. calif ornica, MacGinitie (1935) stated that the animal eats Zostera in Elkhorn Slough. In a later work (MacGinitie, 1949) he stated that they eat "sea- weeds.” Winkler (1959^) referred to three red 1 Department of Pharmacology, School of Medicine, Loma Linda University, Loma Linda, California. Pres- ent address: College of the Desert Palm, Desert, California. Manuscript received November 6, 1961. 2 Beaudette Foundation for Biological Research, Santa Ynez, California. Present address: Hancock Foundation, University of Southern California. seaweeds used as food by Aplysia calif ornica and to their effect on the color pattern. Another paper (Winkler, 1959&) mentions the effect of diet, both qualitatively and quantitatively, on the purple secretion. Winkler (1961) mentions diet in connection with two special habitats which are reported here in more detail. Marcus (I96I) stated that A. calif ornica eats seaweeds, "especially Zostera and C odium.” MATERIALS AND METHODS Young specimens of Aplysia calif ornica 4-5 inches long were captured on the central shore area of Lunada Bay, Palos Verdes, California, and each was placed in an individual pint fruit jar until fecal pellets were passed. The pellets were then removed to small screw-capped vials and the animal was released at sufficient dis- tance to preclude resampling. Twenty-five speci- mens from this area were sampled in this man- ner. Pellets were also collected from larger specimens wandering in the deeper waters of the bay and from 1 mile north of the north side of Palos Verdes Point. The pellets were kept under refrigeration until examined and identified by the junior author. Since the effi- ciency of sea hare digestion appears to be in indirect proportion to the rate and quantity of food ingestion, identification of the algae of- fered no great problem since these sea hares were well fed. Observations on specimens taken from Do- heny Beach, California, and from a small lagoon at Lunada Bay, possessing a different flora than the former, are also reported. The food plants determined from crop contents were made dur- ing other studies involving the sacrifice of the animals. RESULTS Results from 25 young specimens taken from Lunada Bay in 1955 are presented in Table 1. In this area the following seaweeds were noted 102 Food Habits of Aplysia — Winkler and Dawson 103 TABLE 1 Diet of Young Aplysia calif omica at Lunada Bay KINDS OF PLANTS* NO. SPECIMENS IN WHICH FOUND NO. SPECIMENS IN WHICH DOMINANT Acrosorium uncinatum 1 1 Centroceras clavulatum 8 4 Ceramium eatonianum 21 14 Corallina vancouveriensis 12 1 Gelidium coulteri 11 2 G. crinale (?) 7 1 Gelidium sp. 1 0 Gigartina canaliculata 14 7 Hypnea valentiae 4 2 Lophosiphonia sp. 1 0 Phyllospadix sp. 6 0 Plocamium pacifcum 12 2 Pterocladia sp. 4 0 Pterosiphonia sp. 2 0 S per moth amnion snyderae 1 0 Sphacelaria sp. 2 0 Ulva sp. 2 0 * Although marked differences in the relative abundances of the various algae in the habitat would seem to negate the validity of this table in indicating selectivity of food by the sea hares, there is some interesting supporting evidence. Al- though Gigartina canaliculata was the dominant intertidal rock-cover alga throughout the area, and was eaten, it appeared far less dominant in the pellets than the small, relatively in- conspicuous epiphytic alga Ceramium. Similarly, although Ulva and Plocamium were so sparse in the areas as generally to go unnoticed, these, especially the latter, appeared in the pellets. to be common, though never found in the fecal pellets: Laurencia splendens, Chondria calif orn- ica, Gigartina leptorhynchos, Prionitis lanceo- lata, Laurencia subopposita, Cystoseira osmun- dacea, Gymnogongrus leptophyllus, Bossea or- bigniana, Herposiphonia sp., Rhodymenia sp., and Colpomenia sinuosa. All specimens taken from deeper water in Lunada Bay and off Palos Verdes Point were similar in the algal content of their pellets. In every case the dominant plant was Plocamium pacificum, but with some admixture. These ad- mixtures included Phyllospadix sp., Lophosiph- onia sp., Gelidium (several species), and Cer- amium eatonianum . In May 1961 the larger red-striped A. cali- fornica from Lunada Bay were found feeding exclusively on the red alga Laurencia pacifica, which turned their blood and digestive glands a deep red. Observations of the crop contents of several hundred Aplysia specimens were made during the course of work in 1959 at the Lunada Bay lagoon. Occasional spot checks were made in which the contents were determined as to species, but no single alga was recognized as dominant. The samples contained: Chondria calif omica, Lithothrix aspergillum, Ceramium eatonianum, Centroceras clavulatum, Plocamium pacifcum, Corallina gracillus, and Ecto carpus sp., intermixed with fragments of coralline al- gae. The flora of this small lagoon contained more coralline algae and much less Gigartina canaliculata than the afore-mentioned 1955 col- lection site located Vy mile to the south on Lunada Bay. The collecting area at Doheny Beach was located at the extreme north end of the state park, where a small lagoon of uniformly shal- low water is protected on the north by a strati- fied rocky peninsula submerged at higher tides, and on the south by a wave-formed reef of small boulders and gravel. The flora was com- paratively limited, consisting predominately of various jointed coralline algae with intertwined Hypnea valentiae. Algal growth improved as one moved to the seaweed area of the lagoon. In this habitat 86 specimens were collected, but because of apparently grossly similar crop con- tents only 8 specimens were analyzed for pre- sentation in Table 2. Collection dates were March 24, 1959, for numbers D-02 to D-08, and April 29, 1959, for the remainder. It will be noted that the latter collection shows a greater specialization of diet than the former and may reflect the concentration of sea hares farther out in the lagoon than at the former collection date. It may, on the other hand, sug- gest that the sea hare tends to become more selective in its food as it becomes older. This has generally been noted from other areas. The algal contents of crops of Aplysia vac- caria have also been examined. Animals col- lected in their usual haunts in the Egregia beds have contained only large amounts of Egregia in their crops. On the other hand, when the animals are inshore in breeding aggregations the crops are usually empty or contain sand or other material of little significance. On one oc- casion only, several such animals were found 104 PACIFIC SCIENCE, VoL XVII, January 1963 to have ingested a quantity of Pelvetia fastigi- ata, a food plant that was consistently refused by aquarium-confined specimens (Winkler, 1955 ) . The diet of the young of this species has not been determined. DISCUSSION AND CONCLUSIONS In northern California Aplysia californica is known principally from protected bays and es- tuaries such as Elkhorn Slough, Bodega Bay, and Tomales Bay. In these areas, beds of Ulva and Enteromorpha often occur abundantly, and J. J. Gonor has reported to us in a personal communication that A. californica lives in these beds and apparently uses both plants for food. Marcus (1961), reporting upon specimens from Bodega Bay and Tomales Bay, indicated that A. californica feeds on sea plants, especially Zoster a and C odium. MacGinitie (1935) stated that sea hares in Elkhorn Slough fed on Zoster a, but in subsequent personal correspondence has indicated that they also fed on Enteromorpha, which is sometimes the most prevalent plant in the slough. However, he also observed that at certain seasons the alga was absent from the area while Zoster a remained. He concluded, then, that Zoster a was their principal staple food. In southern California the eel grass, Zoster a, does not occur intertidally on the open coast and is uncommon in bays, and although the similar surf grass, Phyllospadix, grows in pro- fusion at low water levels it is rarely ingested by sea hares. Likewise, neither Ulva nor Entero- morpha is abundant in open coast habitats of southern California, and, although captive sea hares showed preference to Ulva or Enteromor- pha over any other food offered them in the laboratory, in nature they resort to feeding on red algae, which are almost their exclusive food. Young southern California sea hares seem to eat a wide variety of seaweeds depending upon the flora in their immediate surroundings, although some species are apparently habitually refused. As they grow older and wander farther J they seem to become more selective in their food habits and develop a behavior pattern that is difficult to break under changed feeding con- ; ditions in the laboratory. It has been generally observed that Aplysia californica in northern California is of larger ; average size than in southern California, and the senior author considers that this larger size, as well as the difference in color (Winkler 1959^), results from dietary differences. Ex- perimental evidence for the variation in size similar to that for color variation is needed to j confirm the homogeneity of the two forms. I Two distinctive dietary factors offer themselves for study: (1) the comparative nutritive value of a high Ulva-Enteromorpha component in the diet during spring and summer; (2) the effect of almost exclusive feeding on the marine grass, TABLE 2 Presence and Relative Abundance of Several Algae in Crops OF Aplysia californica FROM DOHENY Beach, CALIFORNIA COLLECTION NO. Hypnea valentiae Jania tenella Corallina gracilis Corallina pinnatifolia Corallina vancouveriensis Sphacelaria sp. Lauren cia sp. Lithothrix aspergillum D-02 XXX X D-04 XX XX D-06 X XXX D-08 X iiii X X X X X D-20 xxxx D-21 xxxx D-27 xxxx D-28 XXX 1 x Food Habits of Aplysia — Winkler and Dawson 105 Zostera marina, during winter. Further studies are also needed on the diet of adult Aplysia calif ornica which, according to Limbaugh (per- sonal communication ) are most numerous at depths of 30-40 ft and have been observed as deep as 100 ft, depending upon concentra- tions of plants. He mentions them eating the "thinner red and green seaweeds.” From evidence now available it would appear that Aplysia calif ornica and A. vaccaria are not competitive for food, and that the former uses red and green algae as well as marine phanero- grams for food, while the diet of the latter is essentially restricted to the brown alga Egregia, at least in the tidal zone. The nutritive value of Egregia seems to favor growth to large size, for Limbaugh has reported observations of speci- mens of A. vaccaria in the subtidal zone measur- ing up to 30 inches and an estimated 35 lb, making them the giants of the gastropoda. SUMMARY Young Aplysia calif ornica on the open coast of southern California subsist on a diet mainly of fleshy red algae but show a preference for Ulva or Enteromorpha under artificial feeding. In the northern end of its range, where it ap- pears to become largely estuarine in habitat, all reports indicate that it feeds almost exclusively on green algae and eel grass, Zostera marina. It is suggested that the dietary differences may in- fluence its size as well as its color, and that an adquate winter food supply is an important factor. A. vaccaria appears to feed exclusively on the brown alga, Egregia. REFERENCES Eales, Nellie B. 1921. Aplysia. L. M. B. C. Memoirs, Proc. Trans. Liverpool Biol. Soc. 35: 183-266. MacGinitie, G. E. 1935. Ecological aspect of a California marine estuary. Amer. Midi. Nat. 16: 737. 1949. Natural history of marine ani- mals. McGraw-Hill, New York. 473 pp. Marcus, Ernst. 1961. Opisthobranch mollusks from California. Veliger 3 ( Suppl. ) : 9. Mazzarelli, G. 1893* Monographia delle Ap- lysiidae del Golfo di Napoli. Mem. Soc. Ital. Sci. Modena, ser 3, 9: 1-205. Winkler, Lindsay R. 1955. A new species of Aplysia on the southern California coast. So. Calif. Acad. Sci. Bull. 54: 5. 1959^. A mechanism of color variation operating in the West Coast sea hare, Aplysia calif ornica Cooper. Pacific Sci. 13(1): 63- 66. 1959 b. Intraspecific variation in the purple secretion of the California sea hare, Aplysia calif ornica Cooper. Pacific Sci. 13 (4): 357-361. 1961. Preliminary tests of the toxin ex- tracted from California sea hares of the genus Aplysia. Pacific Sci. 15(2): 211-214. Littoral Sedimentary Processes on Kauai, A Subtropical High Island1 D. L. Inman,2 W. R. Gayman,2 and D. C. Cox3 ABSTRACT: Beach and shallow water sand samples from the island of Kauai, Hawaiian Islands, were studied to ascertain the effects of climate on the supply of sediment, and of wave action on the dispersal and transport of sand along the shores of this circular island. The littoral sediments of the island are made up of two components: biogenous material, such as shell, coral, and foraminiferal sands, formed near the shoreline; and terrigenous material, consisting mostly of volcanic mineral and rock fragments brought to the beach by rivers. The northeast trade winds play a dominant part in the climatic and oceanographic processes affecting the island. On the windward side of Kauai, the annual rainfall ranges from 30 inches near the coast to over 460 inches at higher elevations; on the leeward side, annual rainfall is less than 20 inches. This large variation in rainfall results in a climatic range from tropic-humid to semi-arid. Rivers draining the wet, windward portion of the island carry relatively little sand, and the shoreline on this side is characterized by intermittent fringing reefs and beaches of calcareous sand. A detailed study of several reefs on the windward side of the island shows that each pair of fringing reefs, divided by a relatively deep inlet, constitute sep- arate cells for the circulation of water and distribution of sediments. The highest concentrations of volcanic sand occur on the lee side of the island, where streams draining the semi-arid region enter the coast. The concentration of volcanic material in the beach sands decreases logarithmically with distance along the coast from the rivers. KAUAI is the next to westernmost of the wind- ward islands of the Hawaiian Archipelago ( Fig. 1). It is roughly circular in shape, with a di- ameter of about 25 nautical miles (Fig. 2). The island is volcanic and has a central plateau whose eastern rim exceeds 5,000 ft in elevation. This mile-high rim intercepts the trade winds, causing moderate to extreme precipitation on the windward slope and producing a rain- shadow desert along portions of the leeward coast. 1 Published as Contribution No. 24 of the Hawaii Institute of Geophysics and, with the approval of the Director, as Paper Number 88 in the Journal Series of the Experiment Station, Hawaiian Sugar Planters’ Association. Manuscript received November 22, 1961. 2 Scripps Institution of Oceanography of the Uni- versity of California, La Jolla, California. 3 Hawaii Institute of Geophysics, University of Ha- waii; formerly Experiment Station, Hawaiian Sugar Planters’ Association. The contrast between windward and leeward sides of the island in terms of climate and source of sediment suggested that this island was an ideal locality to study the relation be- tween nearshore sedimentation and the source and type of sediment. It was found that the windward streams, although high in runoff, con- tributed very little sand-size material to the nearshore areas. On this side, the coastal en- vironment is characterized by fringing coral reefs, with protected beaches composed of cal- careous sand derived from shells, reef detritus, and foraminifera. Kapaa Reef had been investi- gated by Kohn and Helfrich (1957) and, since it seemed typical of the windward reefs, it was selected for more detailed study. Consistently high concentrations of terrige- nous material were found on only a few of the leeward beaches. These deposits were particu- larly conspicuous near the mouth of the Wai- 106 Littoral Sedimentary Processes — Inman, Gayman and Cox 107 Fig. 1. Index chart and exposure to waves and winds. Heavy arrows indicate directions of approach of the five major wave types which affect the island. mea River, which drains the major canyon on the island and is in an area of relatively light rainfall. The most typical of the lee beaches in- vestigated lay along the shores of the Mana Coastal Plain. During the study, it was observed that beach and dune rock is especially well developed along portions of the Kauai coast. This was the sub- ject of a special investigation by Emery and Cox (1956). Some additional observations are discussed under the section on sediments. PHYSIOGRAPHY Kauai is one of a chain of volcanic islands in the Hawaiian Archipelago. It formed as a shield volcano built up from the deep floor of the Pa- cific Ocean by thin flows of basaltic lava. The original shield collapsed to form a broad cal- dera, which subsequently filled with thicker lava flows and talus deposits. The caldera is now al- most obscured by faulting, extensive erosion, and stream cutting. Ninety % of the flows are composed of normal olivine basalt, while the remaining rock is a basalt either very rich (pic- rite-basalt) or very poor in olivine. A compre- hensive and up-to-date review of the geology of Kauai is given by Macdonald, Davis, and Cox (I960). The ocean floor around Kauai is characterized by a number of sloping terraces which are ter- minated on their seaward side by a marked in- crease in slope. Although the depths of the breaks in slopes at the outer edges of the ter- 108 PACIFIC SCIENCE, Vol. XVII, January 1963 races range on various sides of the island from 30 ft to 500 ft, their grouping suggests that terracing occurred at about four separate levels. The shoalest and widest terrace commonly ex- tends to depths of 40 to 90 ft; others extend from about 150 ft to between 240 to 350 ft; and for the deepest, to about 450 ft. The latter depth presumably marks the outer edge of a wave-cut terrace formed during a still stand near the lowest Wisconsin Sea Level. Generally, the 450-ft contour ranges from 1 to 2 miles from shore, although it extends to 4 miles off the west coast. The most gently sloping part of the insular shelf, that part shoaler than 90 ft, is frequently characterized by many apparently closed depres- sions, generally ranging from 10 to 30 ft deep. These are most common on the north coast, although a very large depression occurs on the west coast off Makaha Point ( Fig. 2 ) . The con- figuration of the submarine ridge surrounding this depression suggests that it may be a relic coastal sand dune, formed at a lower stand of the sea, or possibly a relic barrier beach or reef.4 4 Observations made while diving on this reef dur- ing September 1962 subsequent to submission of this manuscript show the reef to consist of a framework of coral. It is believed that it represents a barrier reef formed during a previous lower stand of sea, and the name "Kaheko” is proposed for the reef because of its presumed great age. Fig. 2. Topography, bathymetry, and sample locations. Numbers along coastline refer to sample designations listed in text and in Table 2. Intermittent jagged line along shore shows location of major reefs. Littoral Sedimentary Processes— Inman, Gayman and Cox 109 * SYMBOL a NUMBER INDICATE GAGING STATIONS SLANTED NUMBERS ARE VALUE OF ANNUAL RAINFALL IN INCHES LETTERS DESIGNATE DRAINAGE AREAS FIG. 3= Rainfall and drainage basins on Kauai. Dotted lines indicate outline of drainage areas. Numbers and letters, designating stream gaging stations and drainage areas, respectively, are referred to in Table 1. Note area of lowest rainfall along western coast. Isohytel lines from the Meteorology Department, Pineapple Research Institute, and the Experiment Station, Hawaiian Sugar Planters’ Association, 1948. There are at least five well-developed sea valleys which cut the shallower portions of the insular shelf. All of these, except the one off Hanalei Bay, appear to terminate above the 450-ft con- tour. CLIMATE The annual precipitation varies from an av- erage of about 50 inches, on the windward (northeastern) shore, to over 450 inches, at the summit of Mt. Waialeale, and decreases to less than 20 inches in the rain-shadow produced on the extreme lee side of the island. Vegetation varies from tropical rain forest on the windward side, to arid on the southwestern slopes. Figure 3 shows the drainage pattern, the outlines of the major drainage basins, and the contour lines of mean annual rainfall, or isohyets, for the island, as well as the location of several stream- flow gaging stations. Langbein and Schumm (1958) have shown that the sediment yield from a drainage basin varies as a function of climate, and that the maximum yield occurs for an effective precipi- tation of about 12 inches. The effective precipi- tation5 is obtained from the actual precipitation by correcting for evapotranspiration losses to a standard temperature of 50 F. Sediment yield is less for precipitation either greater or less than 5 Langbein and Schumm (1958) define "effective precipitation” as the amount of precipitation that would be required at a mean temperature of 50 F to produce the actual annual runoff from the basin. 110 this optimum value of about 12 inches (Fig. 4). Although sediment yield is influenced by vege- tation, temperature, rainfall characteristics, and topography, a rough approximation for sedi- ment yield can be obtained from the effective precipitation. Data indicative of the differences between the climates of the various drainage basins on Kauai and of the supplies of sediments in those basins are given in Table 1. These data include: ( 1 ) a rough calculation of the mean annual rainfall for the various drainage areas; (2) ef- fective precipitation calculated from rainfall; ( 3 ) runoff data for the upper portion of some of the drainage basins, obtained from gaging stations; and (4) the percentage of carbonate in the beach sand sample taken nearest the stream mouths. In general, the table confirms that the great- est precipitation and the most runoff occur on the windward side of the island where the least amount of terrigenous sediment is found in the beach sand. The Langbein and Schumm curve (Fig. 4) shows that the potential sediment yield is appreciably higher when the effective pre- cipitation does not exceed about 35 inches, with a maximum yield corresponding to a value of about 12 inches. The equivalent values in terms of annual rainfall on Kauai6 give the band of high sediment yield as occurring where the mean annual rainfall does not exceed 46 inches, and 0 After correcting for evapotranspiration, following Langbein et al. (1949: fig. 2). FIG. 4. Relation between effective precipitation and annual sediment yield (taken from Langbein and Schumm, 1958: fig. 3). PACIFIC SCIENCE, Vol. XVII, January 1963 a maximum sediment yield corresponding to a mean annual rainfall of 22 inches. Figure 3 shows that a relatively high percentage of the drainage basins on the lee side of the island fall within the range of high sediment yield, whereas the rainfall in the windward drainage basins is appreciably greater and thus produces a somewhat lower yield of sediment. The mean annual rainfall for each drainage basin and the corresponding value for the mean effective pre- cipitation are listed in Table 1, columns 4 and 5, respectively. The Waimea River drainage basin (Fig. 3, B), which is the principal basin on the lee side of the island, has mean values of annual rainfall and effective precipitation of 70 inches and 53 inches, respectively, which are much nearer the value of maximum sediment yield than are those for the principal windward drainage basins (Fig. 3, E, F, G) . In fact, much of the Waimea drainage basin falls within the rainfall range of potentially high sediment yield. Also, the sediment yield in the Waimea basin is accentuated by the occurrence of ex- treme topographic relief in the low rainfall area. This combination of factors, tending to increase sediment yield, does not occur extensively in the windward basins. The high carbonate content of the beach sand samples on the windward coast tends to con- firm the difference in sediment yield between the lee and windward drainage basins (Table 1, last column). In addition, cursory inspection of river banks and river mouths indicated that not only is the yield greater on the leeward side but also that there is a significant difference in the size of the sediment: the leeward streams carry large amounts of sand-size material, whereas the windward streams carry silt and clay-size material. A predominance of silt and clay-size sediment indicates chemical weather- ing, a process which is enhanced in the wind- ward basins by greater vegetation density and slower rate of removal of disintegration prod- ucts. EXPOSURE TO WAVES Although no complete statistical wave data for spectrums and directions of waves are avail- able for the island of Kauai, a general descrip- tion of the prevailing wave conditions can be Littoral Sedimentary Processes — Inman, Gayman and Cox 111 TABLE 1 Rainfall, Runoff, and Sediment Composition for Drainage Basins on Kauai BEACH sand SAMPLE NEAREST DRAINAGE BASIN river basin above gaging station RIVER Desig- nation* Area (sq statute miles) Mean Annual Temp.f (°F) Mean Annual Rainfall (inches) Effective Precipi- tation! (inches) Desig- nation* River Drainage Area (sq statute miles) Runofff (inches) Maximum Inst. Flow (cubic ft/ sec) No. % Car- bonate A 86 70 30 18 B 96 69 70 53 5 Waimea 45.0 45.0 372 ' 24 1.3 C 64 69 76 55 4 Hanapepe 20.5 52.1 487 30 4.2 D 27 70 67 50 E 144 67 128 100 3 S. Wail u a 22.4 81.0 265 47 80.0 2 N. Wailua 14.6 99.8 2045 47 80.0 1 Anahola 5.5 53.5 1010 65 90.0 F 78 69 134 101 10 Hanalei 19.2 196.0 1130 71 62.0 9 Waioli 1.6 277.0 930 74 76.0 G 40 67 169 130 8 Lumahai 7.1 230.0 1090 76 62.0 7 Wainiha 14.7 91.9 211 78 55.0 H 14 67 93 70 6 Hanakapiai 2.6 102.0 1600 85 85.0 * Refers to Figure 3. t Adjusted for mean elevation of basin, assuming 3 F. decrease per 1000 ft altitude. Station temperature from Macdonald, Davis, and Cox (I960: 114). | Precipitation at a mean temperature of 50 F. required to produce the actual runoff. Runoff obtained from mean annual rainfall (extrapolation from Langbein et al., 1949: fig. 2); effective precipitation obtained from annual runoff (Langbein and Schumm, 1958: fig. 1). § Annual river discharge divided by the drainage area. River discharge from Territorial Planning Board (1939). obtained by generalizing from the wave sum- mary given by the Corps of Engineers (1955: appendix B) and from specific storm descrip- tions. such as that of Arthur ( 1948) . Review of these data suggests that the wave and wind re- gime can conveniently be classed in the follow- ing manner (Fig, 1): 1. Northeast Trades: Swell generated by the prevailing northeasterly winds. These waves are present all year, but are largest from April through November. Waves typically have pe- riods of 5-8 sec, are 4-12 ft in height, and approach from the east-northeast. 2. Kona Storms: Waves generated by intense winds, associated with local fronts or Hawaiian lows of extra-tropical origin. The wind may vary in direction from south through west to north, and usually blows hardest from the south- west. These storms are not frequent, but occur most commonly from November through March. Waves accompanying these storms typi- cally have periods of 8-10 sec, are 10—15 ft high, and are most intense from the southwest. 3. Winter Swell from Aleutian and Mid-lati- tude Lows : Waves from this source are inter- mittent, and can occur throughout the year, but are most intense from October through May. Typical waves have periods of 10-17 sec, are 8-14 ft high, and may approach from the north- west, north, and northeast. 4. Southern Swell: Long, low waves generated by storms in the southern hemisphere, and most prevalent from June through September. Typical waves have periods of 14-22 sec, are 1-4 ft high, and approach from the southeast through the southwest. 112 PACIFIC SCIENCE, Vol. XVII, January 1963 5. Hurricane: Waves from tropical storms, usually approaching from the southeast through southwest. They are most likely to occur in Au- gust and September, although occurrence may be between June and December. Of the five classes of waves, those generated by the northeast trades are by far the most per- sistent source of wave energy for the east and northeast shores of the island. Undoubtedly these waves account for the greater abundance of cal- careous reefs on the windward side of the island. Waves from Aleutian lows are probably the dominant agent responsible for recent erosion along the Napali Cliffs on the northwest coast of the island. These waves produce a southwest- erly drift of littoral sediment from the Napali Cliffs towards Nohili Point and have been a major influence in the formation of the straight coast line near Barking Sands (between sample stations 1 to 7, Fig. 11). The meteorological conditions giving rise to unusually high waves of this type in January 1947 are described by Arthur ( 1948). Waves of this type, with break- ers estimated to be 20 ft high and periods of 14-16 sec, were observed on the north coast of Kauai on 30 November 1935. Southern swell is the most common type of wave on the south and lee sides of the island. Although these waves are low, they are almost continuous during the summer months and pro- duce a westerly transport of sand from the Wai- mea River toward the Mana Coastal Plain. Hur- ricanes, although infrequent, are important, and the local residents attribute some recent erosion to the combined effects of the hurricane of 1957 and the tsunami of the same year. Kauai is ex- posed to tsunami waves accompanying seismic activity in the vicinity of the Aleutian and Ku- rile Trenches in the northern hemisphere, and, to a lesser degree, from tsunami waves generated off the coast of Chile in the southern hemis- phere. These catastrophic waves cause considera- ble coastal erosion and, at times, transport large blocks of reef material inland (Shepard, Mac- donald, and Cox, 1950). SEDIMENTS Eighty-three sand samples and five beach and dune rock samples were collected from the near- shore zone along the coasts of the island. Exami- nation of the sediment showed that it could easily be divided on the basis of origin into: ( 1 ) biogenous material derived from the skeletons of molluscs, coral, calcareous algae, and fora- minifera; and (2) terrigenous material derived from erosion of the island’s volcanic formations. These two sediment types were characterized by important differences in color, density, and solu- bility in dilute hydrochloric acid. In general, the biogenous material appeared to constitute the predominant sediment in the nearshore region of the island. On the windward coast, the per- centage of biogenous material was so high that the terrigenous fraction was rarely visible to the unaided eye. Most samples were composed of medium and coarse grain sand containing 80% to 95% cal- cium carbonate of biogenous origin. Analysis by X-ray diffraction showed that, mineralogically, the biogenous material was composed of three components: aragonite, magnesium-rich calcite, ! and magnesium-poor calcite. Terrigenous sedi- ments were common only on the lee coasts of the island, especially in the vicinity of and down current from river mouths. The terrigenous ma- terial consisted mainly of olivine grains and sand-size, partially-weathered feldspathic rock fragments, plus a small percentage of opaque iron oxides. A brief description of the size distributions of the sands follows, together with their biological, chemical, and mineralogical analyses. Detailed data on each sample are listed in Table 2. Size Distribution The size distributions of all the sediments ; were determined by the Emery settling tube (Poole et al., 1951), and some, especially the coarsest, were also analyzed by sieving. The me- dian diameter of the particle size distribution is expressed in microns, and measures of the standard deviation (sorting) and skewness are given in the graphic phi notation of Inman i (1952). Sieving was necessary in order to obtain the parameters for the coarser samples. Since sieve and settling tube analyses are based on different properties of the sediment, the para- meters obtained by the two methods are only roughly comparable. For example, the partly hollow tests of foraminifera are retained in a TABLE 2 Location, Type, Size Distribution, and Composition of Samples Littoral Sedimentary Processes — Inman, Gayman and Cox 113 OLIVINE to FELDSPAR PEAK HEIGHT RATIO § ITS i m i CARBONATE FRACTION § % Mg-Poor Calcite trace trace ITN 00 trace I % Mg-Rich Calcite ON WN c XT I x ssf 1—1 CN CN I x r- o CN CN 1 1 NO CN CN i-h i x 1-H 00 CCN O i x (00‘0) £Z‘— b .31 (.310) ir\ ON 00 m m .48 (.465) 00 o l rs l/"N 00 'T no beach rock .61 (.48) .45 (.42) .56 (.72) .38 (.46) q ws .56 (.64) H \T\ ITS ITS .48 (.49) Median Microns 253 (227) 309 (297) 451 (435) 4 73 (451) 651 (637) | 707 (646) 779 (747) 245 (243) 243 (230) 281 (330) 257 (262) cn no i r\ CN CN 297 (337) BEACH SLOPE, TANGENT .067 o .085 cCN .101 .096 m TYPE SAMPLE* PQ PQ Q PQ PQ BR PQ PQ PQ PQ PQ PQ PQ PQ LOCATION Barking Sands Barking Sands Barking Sands Barking Sands Barking Sands Barking Sands Nohili Point Mana Mana Mana ! Kokole Kekaha Kekaha Kekaha SAMPLE NUMBER - NO rs- 00 CN o - CN NT TABLE 2 (continued) Location, Type, Size Distribution, and Composition of Samples 114 PACIFIC SCIENCE, Vol. XVII, January 1963 olivine to FELDSPAR PEAK HEIGHT RATIO § i r\ O V \r\ O V o V ! CV vq CARBONATE FRACTIONS % Mg-Poor Calcite trace % Mg-Rich Calcite m 00 00 ICV % . Aragonite GV CN r^- CN NT [ % CARBONATE* 52.9 34.9 32.0 00 NT nt 00 00 7.7 14.7 4.4 6.9 rCV 00.5 0.7 84.6 86.0 8”06 SIZE distribution! a (TV o 71 \r\ \r\ o o f + ON \T\ cn M VO o m M i-i CARBONATE FRACTION § I % 1 Mg-Poor Calcite 18.6 16.7 p 00 On NO 23.1 16.1 18.5 23.2 VOZ 00 NT 12.4 p 1 trace o d o trace % Mg-Rich Calcite 52.1 57.5 54.9 56.5 53.8 CN |T I/O 53.8 54.5 rCN nt wo wo wo 60.4 62.3 00 54 48 1 1 Aragonite 29.3 25.8 p rT CN 26.6 23.1 26.7 L‘LZ 22.2 25.6 29.8 27.2 26.7 1 ON 46 CN i r\ ■pq «! ea < u 93.3 89.7 93.8 92.7 92.7 cO CN 87.9 90.7 95.0 93.3 90.3 88.2 88.5 89.7 ON 00 00 75.9 00 p 00 00 85.4 62.0 ON 00 NT SIZE DISTRIBUTIONt is. C3 or— —.06 © o NO GN o o r + CO NO m © n WO O 1' rO ON 7 1-1 NT 71 o d m oo ITN VAN ICN 56 r-. 00 wo ON wo 60 £ 62 63 64 wo NO 99 67 68 69 70 72 TABLE 2 (continued) Location, Type, Size Distribution, and Composition of Samples Littoral Sedimentary Processes— Inman, Gayman and Cox 117 OLIVINE TO FELDSPAR PEAK HEIGHT RATIO§ CN V CN 3.9 no o V 2.7 CARBONITE FRACTION § % Mg-Poor Calcite trace o % Mg-Rich Calcite r-' \rs r"~ .% . Aragonite ON CN l/A CN % carbonate! 74.2 76.2 XT NO 62.5 10.7 no nf i rs 11.7 92.5 90.7 ON 00 00 VO 64.4 53.0 85.3 23.8 \r\ © size distribution! o (N f xr o r 7 7 00 7 o x CN CN f NT + CN m + (—.09) —.18 (0.00) o NT r o US f ON o + CN 7 US us i/~\ NT i rs .71 \Q vr\ CN » rs us •43 .39 3 00 us rCN rCN CN NO | .66 00 r- _ ^ _ Median Microns 341 330 364 611 US r~- Nf m 369 L 717 871 819 (707) 369 (309) 344 295 620 242 (3732) BEACH SLOPE, TANGENT o m 00 so On On VO .20 .043 TYPE SAMPLE* PQ PQ pq PQ B, BC PQ B, BC PQ PQ PQ s q_i p4 RF PQ PQ CO LOCATION Hanalei Bay Hanalei Bay Hanalei Bay Lumahai Bay Lumahai Bay Wainiha Bay Wainiha Bay Wainiha Bay Haena Point Haena Ke’e Ke’e Ke’e Hanakapiai 1 Hanakapiai Hanakapiai | SAMPLE NUMBER rCN r~- i r\ i"- NO r- r"- 00 On r-. o 00 00 CN 00 m 00 84c NT 00 us 00 NO 00 r-~ 00 iil t d C j LONGSHORE ........ + CURRENT V | - A,, » © o 1 * Ax * OCEAN rr ri B; while the sediment Q, transported longshore consists of two components, a terrigenous frac- tion T, and a biogenous fraction B = 1 — T If complete mixing takes place within a coastal segment, having boundaries xi and X2, a distance Ax apart, then T is a function of the longshore distance x and can be written T ( x ) . The above assumptions permit the budget for the fraction of terrigenous material to be written: Q • T(Xl) = Q • T(x2) + F0T(xi +Lr) Ax ^ Q T(xi) T(x2) , , Ax, 0= x + F0T(xi. d — - ) Ax 2 but in the limit as Ax -> o ~ dT °=Q F„T(x) and integration gives lnT= — ^-x + k where In is the base of natural logarithms and k ■ is a constant of integration. The relation may be written in terms of B = 1 — T as: A small segment of shoreline (see figure) is ! assumed to be in a state of equilibrium, such i that the rate of longshore transport of material into the segment (the longshore influx) is con- stant and equal to the outflux, Qi = Q0- Also, the onshore influx and outflux are constant and equal, F| l- F0. The sediment influx from offi ; shore, F:> consists of 100% biogenous material, 9 Comparison of surveys of 1959 and 1962 indicate | a net erosion of beach immediately westward of the dredging area (approximately stations 51-57, Fig. 8) ! of about 2,500 cubic yd per year. I n B= -q- x + k REFERENCES Arthur, R. S. 1948. Forecasting Hawaiian swell from January 2 to 5, 1947. Bull. Amer. Me- teor. Soc. 29:395-400. BlEN, G. S. 1952. Chemical analysis methods. Scripps Inst. Oceanogr. Ref. 52-58. 130 Blackmon, P. D., and R. Todd. 1959. Mineral- ogy of some foraminifera as related to their classification and ecology. J. Paleo. 33 ( 1 ) : 1- 15. Bramlette, M. N. 1926. Some marine bottom samples from Pago Pago Harbor, Samoa. Car- negie Inst. Publ. 344. 35 pp. Chave, K. E. 1954^. Aspects of the biogeo- chemistry of magnesium 1, calcareous marine organisms. J. Geol. 62:266-283. 1954 A Aspects of the biogeochemistry of magnesium 2, calcareous sediment and rocks. J. Geol. 62:587-599. Corps of Engineers, San Francisco District. 1955. Beach erosion control report on coop- erative study of Waimea Beach in Hanapepe Bay, Island of Kauai, T. H. (Partially pub- lished as House Doc. 432, 84th Cong. 2nd Sess.) Emery, K. O., and D. C. Cox. 1956. Beach rock in the Hawaiian Islands. Pacific Sci. 10 (4) : 382-402. Gayman, W. R. In preparation. X-ray diffrac- tion calibration curves for ratios of aragonite and magnesium-rich and magnesium-poor cal- cite. Goldsmith, J. R., D. L. Graf, and O. I. Joen- SUU. 1955. The occurrence of magnesium cal- cites in nature. Geochim. Cosmochim. Acta. 7:212-230. Hinds, N. E. A. 1930. Geology of Kauai and Niihau. Bishop Museum Bull. 71. Inman, D. L. 1949. Sorting of sediments in the light of fluid mechanics. J. Sed. Petrol. 19:5 1— 70. 1952. Measures for describing the size distribution of sediments. J. Sed. Petrol. 22: 125-145. 1953. Areal and seasonal variations in beach and nearshore sediments at La Jolla, California. Beach Erosion Board, Corps of En- gineers, Tech. Memo. 39. 134 pp. Jamieson, J. C. 1953. Phase equilibrium in the system calcite: aragonite. J. Chem. Physics 21:1385-1390. PACIFIC SCIENCE, Vol. XVII, January 1963 Kohn, A. J., and P. Helfrich. 1957. Primary organic productivity of a Hawaiian coral reef. Limnology and Oceanography 2:241-251. Jelgersma, S., and A. J. Pannekoek. I960. Post-glacial rise of sea-level in the Nether- lands. Geol. Mijnbouw 39:201-207. Langbein, B. et al. 1949. Annual runoff in the United States. U. S. Geol. Surv. Circ. 52. 14 pp. and S. A. Schumm. 1958. Yield of sed- iment in relation to mean annual precipita- tion. Trans. Amer. Geoph. Union 39: 1076- 1084. Lowenstam, H. A. 1954. Factors affecting the aragonite-calcite ratios in carbonate secreting marine organisms. J. Geol. 62:284-322. Macdonald, G. A., D. A. Davis, and D. C. Cox. I960. Geology and ground water re- sources of the Island of Kauai, Hawaii. Haw. Div. Hydrog. Bull. 13. 212 pp. McF ARLAN, E., Jr. 1961. Radiocarbon dating of Late Quaternary deposits, South Louisiana. Geol. Soc. Amer. Bull. 72:129-158. Munk, W. H. and M. C. Sargent. 1954. Ad- justment of Bikini Atoll to ocean waves. U. S. Geol. Surv. Prof. Paper 260-C: 275-280. Poole, D. M., W. S. Butcher, and R. L. Fisher. 1951. The use and accuracy of the Emery settling tube for sand analysis. Beach Erosion Board, Corps of Engineers Tech. Memo. 23. 11 pp. Shepard, F. P. I960. Rise of sea level along northwest Gulf of Mexico. Recent Sediments, Northwest Gulf of Mexico, 1951-1958. Amer. Assoc. Petr. Geol. 1960:338-381. , G. A. Macdonald, and D. C. Cox. 1950. The tsunami of April 1, 1946. Bull. Scripps Inst. Oceanogr. 5:391-528. Territorial Planning Board, Territory of Hawaii. 1939. Surface water resources of the Territory of Hawaii, 1901-1938; Sum- mary of Records. Honolulu Star-Bulletin Lim- ited. 411 pp. NOTE Adoption of the Metric System and Celsius Scale The Intergovernmental Oceanographic Com- mission, Noting that international adoption of the Metric System and Celsius Scale should be one of the major keys of promoting marine sciences, Considering that the use of the Metric Sys- tem was recommended by the Working Group on Data Exchange held in Washington from 7 to 10 August, 1962, (Ref: NS/IOC/2-5(c)); Noting further that data are to be submitted to the World Data Centres in the Metric System and Celsius Scale. Bearing in mind that for purposes of inter- national exchanges of meteorological data, the World Meteorological Organization (WMO) has already adopted the said system at its 3rd Congress, Recommends that member countries of IOC take the necessary steps to encourage, as far as possible, the use of the Metric System and Cel- sius Scale in their oceanographic publications. 131 ILLUSTRATIVE MATTER Manuscript Form. Manuscripts should be typed on one side of standard-size, white bond paper and double-spaced throughout. Pages should be consecu- tively numbered in upper right-hand corner. Sheets should not be fastened together in any way, and should be mailed flat. Inserts should be either typed on separate sheets or pasted on proper page, and point of insertion should be clearly indicated. Original copy and one carbon copy of manuscript should be submitted. The author should retain a car- bon copy. Although due care will be taken, the editors cannot be responsible for loss of manuscripts. Introduction and Summary. It is desirable to state the purpose and scope of the paper in an introductory paragraph and to give a summary of results at the end of the paper. Dictionary Style. It is recommended that authors fol- low capitalization, spelling, compounding, abbrevia- tions, etc., given in Webster’s New International Dic- tionary (unabridged), second edition; or, if desired, the Oxford Dictionary. Abbreviations of titles of pub- lications should, if possible, follow those given in World List of Scientific Periodicals. Footnotes. Footnotes should be used sparingly and never for citing references (see later). When used, footnotes should be consecutively numbered by supe- rior figures throughout the body of the paper. Foot- notes should be typed in the body of the manuscript on a line immediately below the citation, and sepa- rated from the text by lines running across the page. Citations of Printed Sources. All references cited should be listed alphabetically by author at the end of the paper, typed double-spaced. References to books and to papers in periodicals should conform to the following models : Batzo, Roderick L., and J. K. Ripkin. 1849. A Treatise on Pacific Gastropods. Rice and Shipley, Boston, vii + 326 pp., 8 figs., 1 map. Crawford, David L. 1920 a. New or interesting Psyllidae of the Pacific Coast (Homop.). Proc. Hawaii. Ent. Soc. 4(1) : 12-14. 1920&. The sandalwoods of Hawaii. Proc. Hawaii. Ent. Soc. 4(2) : 374-375, 13 pis. In the text, sources should be referred to by author, date, and page, as follows: "It was noted (Rock, 1916: 18) that . . ” or "Rock (1916: 21-24) says . . .” Quotations. Quoted matter of fewer than five printed lines (about 200 characters) should be given in the text in the usual form, using double quote marks. Longer quotations should be set flush with left mar- gin. The author is responsible for the accuracy of quoted material. Numbers. Decimals, measurements, money, percent- ages, time; enumerations in which any figure is 10 or over; and isolated enumerations of 10 and over should be given in Arabic figures, rather than spelled out, except when the number begins a sentence. Only the minimum number of illustrations required to supplement the text will be accepted by the editors. Reproduction costs of illustrations in excess of the number allowed by the editors will be paid by the author. Artwork for illustrations should be SV2 x 11 inches or smaller, and it should accompany manuscript, on separate sheets. Often more valuable than a photo- graph is a good line drawing. Figures and Graphs. Copy for figures and graphs should always be drawn large enough to allow for at least one-third reduction by the engraver. Copy should consist of carefully prepared line drawings in one color only, drawn in India ink on plain white draw- ing paper or tracing cloth. Co-ordinate paper with lines in light blue (a color which will not show in a photograph) may be used; but co-ordinates which should appear in the finished graph must be drawn in India ink. If original figures may not be conven- iently submitted with manuscript, duplicate rough sketches or photographic prints may be furnished to aid the editors in their decisions. It is strongly urged that an indication of scale be incorporated as a part of all drawings in which mag- nification and size are critical considerations. Photographs. Photographs should be chosen for clarity in portraying essential information. They should be printed for contrast, on glossy paper, and should be sent unmounted. They should be identified with serial number written in soft pencil on the back to corre- spond with list of captions. Illustrations will be returned to the author. Tables. Tabular matter should be kept to a minimum. Each table, prepared to conform with Pacific Science style, should be typed on a separate page, and its posi- tion indicated on the manuscript. Mathematical Formulas. Complicated formulas cannot be set by the printers. Authors should submit them as illustrations. Captions. Readily identifiable captions for figures, graphs, photographs, and other illustrative matter should be supplied on a separate page. PROOF Proof should be corrected immediately and returned at once to Robert Sparks, assistant to the editors. Authors are reminded that the editors will allow only a minimum number of corrections on galley proof. Additions to the printed text and changes in style and content are not permitted. All queries on proof should be answered. If cor- rected proof is not received within four weeks after being sent to the author, author’s changes cannot be accepted. REPRINTS Reprints or separates should be ordered on the form provided and returned with author's proof. All correspondence concerning separates must be directed to the printer, Star-Bulletin Printing Company, 420 Ward Avenue, Honolulu 14, Hawaii. APRIL 1963 NO. 2 boo- / P//7 VOL. XVII PACIFIC SCIENCE A QUARTERLY DEVOTED TO THE BIOLOGICAL AND PHYSICAL SCIENCES OF THE PACIFIC REGION TADAO YOSHIDA, TAKE O SAWADA, and MASAHIRO HIGAKI Sargassum at Tsuyazaki, North Kyushu ALBERT TESTER Role of Olfaction in Shark Predation EDMUND S. HOBSON Feeding Behavior in Three Species of Sharks ALAN G. LEWIS Life History of Lepeophtheirus dissimulates MOHAMMED NIZAMUDDIN Studies on Udotea indica RITA D. SCHAFER Effects of Pollution on Mytilus edulis C. E. PEMBERTON Insect Pests of Sugar Cane UNIVERSITY OF HAWAII PRESS BOARD OF EDITORS O. A. BUSHNELL, Editor-in-Chief Department of Microbiology, University of Hawaii Robert Sparks, Assistant to the Editors Office of Publications and Information, University of Hawaii Thomas S. Austin Bureau of Commercial Fisheries, Hawaii Area (U. S. Fish and Wildlife Service) Honolulu, Hawaii L. H. Briggs Department of Chemistry University of Auckland Auckland, New Zealand Ai Kim Kiang Department of Chemistry University of Malaya, Singapore Gordon A. Macdonald Department of Geology University of Hawaii Donald C. Matthews Department of Zoology University of Hawaii Colin S. Ramage Department of Geology and Geophysics University of Hawaii Martin Sherman Department of Entomology University of Hawaii Donald W. Strasburg Bureau of Commercial Fisheries, Hawaii Area (U. S. Fish and Wildlife Service) Honolulu, Hawaii Albert L. Tester Department of Zoology and Entomology University of Hawaii Miklos F. Udvardy Department of Zoology University of British Columbia Vancouver, Canada Thomas Nickerson, Managing Editor Assistant to the University Provost INFORMATION FOR AUTHORS Contributions to Pacific biological and physical science will be welcomed from authors in all parts of the world. (The fields of anthropology, agriculture, engineering, and medicine are not included.) Manu- scripts may be addressed to the Editor-in-Chief, PACIFIC SCIENCE, University of Hawaii, Honolulu 14, Hawaii, or to individual members of the Board of Editors. Use of air mail is recommended for all communications. Manuscripts will be acknowledged when received and will be read promptly by members of the Board of Editors or other competent critics. Authors will be notified as soon as possible of the decision reached. Manuscripts of any length may be submitted, but it is suggested that authors inquire concerning possi- bilities of publication of papers of over 30 printed pages before sending their manuscripts. Authors should not overlook the need for good brief papers, presenting results of studies, notes and queries, com- munications to the editor, or other commentary. PREPARATION OF MANUSCRIPT It is requested that authors follow the style of Pacific Science described herein and exemplified in the journal. Authors should attempt to conform with the Style Manual for Biological Journals, Am. Inst. Biol. Sci. Washington. ( Continued on inside back cover) PACIFIC SCIENCE A QUARTERLY DEVOTED TO THE BIOLOGICAL AND PHYSICAL SCIENCES OF THE PACIFIC REGION VOL. XVII APRIL 1963 NO. 2 Previous issue published March 6, 1963 I _ CONTENTS PAGE Sargassum Vegetation Growing in the Sea around Tsuyazaki, North Kyushu, Japan. Tadao Yoshida, Takeo Sawada, and Masahiro Higaki 135 The Role of Olfaction in Shark Predation. Albert Tester 145 Feeding Behavior in Three Species of Sharks. Edmund S. Hobson 171 Life History of the Caligid Copepod Lepeophtheirus dissimulatus Wilson, 1905 ( Crustacea : Caligoida). Alan G. Lewis 195 Studies on the Green Alga, Udotea indica A. & E. S. Gepp, 1911. Mohammed Nizamuddin 243 Effects of Pollution on the Amino Acid Content of Mytilus edulis. Rita D. Schafer 246 Important Pacific Insect Pests of Sugar Cane. C. E. Pemberton 251 . Pacific Science is published quarterly by the University of Hawaii Press, in January, (April, July, and October. Subscription price is $4.00 a year; single copy, $1.23. Check or money order payable to University of Hawaii should be sent to University of Hawaii Press, Honolulu 14, Hawaii, U. S. A. Printed by Star-Bulletin Printing Company, 420 Ward Avenue, Honolulu 14, Hawaii. mmm mu-3 JIM 3 Sargassum Vegetation Growing in the Sea around Tsuyazaki, North Kyushu, Japan1 Tadao Yoshida,2 Takeo Sawada, and Masahiro Higaki3 Ecological studies on the subtidal marine vegetation, including the Sargassum community, have not progressed far because it is harder to make field surveys in subtidal vegetation than in terrestrial or intertidal vegetations. Some depth records for species of marine algae have been obtained using dredges and other instru- ments. From these data, we know that there is considerable vegetation developed at subtidal depths (cf Ueda and Okada, 1938, 1940). Walker (1947) carried out extensive studies of the Laminaria beds around Scotland. His at- tention has been centered chiefly on productiv- ity. Underwater observation of seaweeds has been done from time to time in the course of the development of diving apparatus. Waern ( 1952) published the results of his underwater survey of seaweeds. More recently Gilmartin (I960) carried out a survey of the deeper water algae of Eniwetok Atoll with the aid of a self-contained underwater breathing apparatus (SCUBA). Above all, a work published by Gislen (1930) is the most remarkable. He proposed a system of classification of life-forms for benthic plants and animals, and described many "as- sociations.” When these works are examined from a phytosociological point of view it is noticed that they are prescribed by the state of progress in terrestrial plant ecology at the par- ticular time of writing. For example, Gislen only recognized associations in the concept of Swedish school and based on data from but a few quadrats. Artificial stone beds have been laid down on shallow bottoms at many places around Japan for the propagation of useful algae, such as 1 Contribution from the Department of Fisheries,: Faculty of Agriculture, Kyushu University. Manuscript received November 23, 1961. 2Tohoku Regional Fisheries Research Laboratory, Shiogama, Japan. 3 Department of Fisheries, Faculty of Agriculture, Kyushu University, Fukuoka, Japan. Gelidium and Undaria. In order to evaluate the usefulness of these artificial stone beds many surveys have been carried out on the communi- ties of algae developed upon them. However, the data secured from them are too rough to provide satisfactory knowledge of the marine vegetation and succession in its populations (Katada, 1958). Many concepts and methods for the analysis of vegetational structure have been proposed recently in the course of the development of phytosociology. In contrast to the method of classifying the plant communities used by Rraun-Blanquet and others, Goodall ( 1953) in- troduced a new statistical method for classifica- tion using positive interspecific correlation be- tween the species as a criterion. On the other hand, Whittaker (1956) and others developed the concept of a vegetational continuum. These are applicable to the study of marine algal vege- tations. A Sargassum vegetation covers large areas of the upper subtidal region of the warmer waters around Japan. This vegetation affords a good habitat or spawning bed to many kinds of fish. Plants of Sargassum and its allied genera, float- ing at the sea surface after being detached from their substrata, are known as "floating seaweeds” or "nagaremo.” The floating seaweeds and many kinds of fish exist in a close relationship insofar as the fish spawning and growth of the larvae of the fish are concerned. The authors have been engaged in the survey of the floating seaweeds for several years, and feel acutely the necessity for knowledge concerning the Sargassum com- munity from which the floating seaweeds of Ja- pan are derived. The authors carried out a survey on the Sargassum community growing on the rocky shallow sea around Tsuyazaki, north Kyushu, Japan, and intend to analyse here the relationship of species populations to the gradient of environmental factors. 135 136 PACIFIC SCIENCE, Vol. XVII, April 1963 FIG. 1. Map of Kyushu, showing the position of Tsuyazaki. DESCRIPTION OF AREA SURVEYED The peninsula of Tsuyazaki (Fig. 1) lies at 130° 29' E, 33° 47' N, about 20 km north- northeast of Fukuoka City, and faces on the Genkai-nada. The peninsula protrudes from general coast line of sandy beach, and consists of hills more than 100 m high. Although these hills are for the most part composed of palae- ozoic sedimentary rock, there is a granite zone on the west side facing the sea, and a basalt layer lying above the 100 m contour line. Along the middle of the shore of the peninsula, there is a sandy beach called "koi-no-ura.” The rocky area, where the Sargassum community develops, occupies the northern and southern parts of the peninsula. An outline of the geological structure of the area concerned is shown in Figure 2. Although the nature of the bottom is rocky where the shore is exposed to strong wave ac- tion from the intertidal zone to a depth of about 10 m, or of boulders in more sheltered places, such hard bottoms give way to sand at the rela- tively shallow depth of 15 m or less. Sandy bottoms are met in deeper places all around the peninsula. On the coast of North Kyushu, tides are semi- diurnal, and the diurnal inequality of the tides is not remarkable. The spring tidal range is about 1.9 m, and 1.4 m at neap tide. According to the "Coast of Kyushu Pilot” (1947), in the Kuraraseto passage, lying to the north-northeast about 10 km from Tsuyazaki, the tidal current flows towards the north from 4 hr after low tide to 4 hr after high tide, and towards the south from 4 hr after high tide to 4 hr after low tide. The tidal current flowing north reaches a speed of 2.8 knots. Near the shore of the area surveyed, the tidal current runs even faster around Tsuzumi islet, lying to the north of the peninsula, and may reach a maximum speed of more than 1 knot. The current moves more slowly near Sone-no-hana and Kyodomari. Stronger waves and swells are generated with north or west winds, influenced by the topogra- phy of the neighbouring area. Hence, the north and west sides of the peninsula are more ex- FlG. 2. Geological map of the peninsula of Tsuya- zaki (after Takehara, 1937). Sargassum at Tsuyazaki— -Yoshida, Saw ADA, and Higaki 137 °c Fig. 3. Seasonal variation of surface temperature (solid line) and specific gravity (dotted line) of the sea water for I960, measured at the top of the break- water for the port of Tsuyazaki. posed to wave action. The eastern sides of Sone- no-hana and Kyodomari are more sheltered from wave action, although there is no method of determining the strength of wave action, judging from the results of direct observation of waves and from the characteristics of the in- tertidal algal zonation. The tidal current is faster where the shore is more exposed to wave action, and vice versa in the area surveyed. Surface temperatures rise to about 28 C in August and descend to about 10 C in January (Fig. 3). Transparency of the sea water is less than 15 m in the Secchi disc reading almost throughout the year. After a heavy rain, low transparency is sometimes recorded around Sone-no-hana. METHODS Field surveys were carried out during 7 days between July 23 and August 26, 1959, on 13 survey lines In the area (Fig. 4), as follows. A survey ship was anchored at the offshore end of each line. A rope was then strung from the ship to a rock on the shore. Survey spots were se- lected along the rope every 5 or 10 m depend- ing on the inclination of the bottom. One or two observers, a recorder, and a crew got into a small boat. The boat halted at each survey spot. The depth of water was measured with a sound- ing lead and then a quadrat was laid down. The quadrat used in this study was a 0.5 m square iron frame. Observations on the plants found in the quadrat were made by skin diving, using a face mask only. First, the percentage of vegeta- tional cover of the area around the quadrat was recorded. Then, the degree of cover by each species of the algae present in the quadrat was measured. The scale for the measurement of cover degree was as follows : + 0-10% 1 11- 20% 2 21- 40% 3 41- 60% 4 61- 80% 5 81-100% Number and height of the plants were recorded when this was possible. The species not present in the quadrat but found in the surrounding area were also noted. The number of quadrats placed along each Fig. 4. Map of Tsuyazaki peninsula, indicating the 13 survey lines. 138 survey line varied from 10 to 20, according to the inclination of the bottom; 197 sets of quadrat data were obtained. All data needed for analysis were recorded on hand-sort edge- punched cards to facilitate the compilation of the data. Readings of depth were adjusted to the low watermark of spring tide, and the quadrat of 0 m level indicates the quadrats placed above the low watermark. The depth represented as 1 m concerns the quadrat level laid down between low watermark and 1 m deep, and so on. RESULTS Stratal Structure of Sargassum Community Twelve species of Sargassum, one species of Cystophyllum, and one species of Hizikia oc- curred in the quadrats surveyed (Table 1). The fruiting period of these species is revealed by analysing the data obtained from surveys of the floating seaweeds (Segawa, Sawada, Higaki, and Yoshida, 1939) . According to these results and our observations made at the stations in- vestigated, all of the species growing there ex- cept Sargassum ringgoldianum had passed their fruiting period. Most of the individuals of each perennial species had lost their long branches bearing the receptacles, and most of the branches of the next year class had reached a length of 30-40 cm. Therefore, the vegetation was at its lowest height for the year. However, S . ring- goldianum was taller than the rest. Annual TABLE 1 Species of Sargassum and Related Genera Observed in Quadrats S. thunbergii (Mertens) O. Kuntze S, nigrifolium Yendo S. hemiphyllum C. Agardh S. confusum Agardh S. micracanthum (Kiitzing) Yendo S. ringgoldianum Harvey S, fulvellum Agardh S' patens C. Agardh S. tortile C. Agardh S. yendoi Okamura et Yamada ? S. piluliferum C. Agardh S' serratifolium J. Agardh Cystophyllum sisymbrioides J. Agardh Myagropsis my agr aides (Turner) Fensholt) Hizikia fusiforme (Harvey) Okamura PACIFIC SCIENCE, Vol. XVII, April 1963 species such as S. horneri were in a very early stage of growth. At the time of the year when the survey was done, therefore, all species of Sargassum were considered to contribute to- the same layer, and cover degree of the species was rather easy to determine. Stratal difference may be developed among the species of Sargassum during the period from winter to spring, be- cause of the difference in the growing and fruit- ing periods of the species concerned (cf Se- gawa, Sawada, Higaki, and Yoshida, I960). Species other than Sargassum occurred mainly at shallower places, and there was almost the total lack of a frondose lower layer under the well-developed layer of Sargassum . The crustose layer was represented by species of unarticu- lated corallines. In the area where the upper layer was open because of the absence of Sargas- sum, several species of frondose algae other than Sargassum formed a lower layer. Although the population of Sargassum became sparse in deeper places, the society of the lower frondose strata was not developed. This fact may be at- tributed partly to the survey time, when many kinds of algae were extremely depauperate in growth for the year. Vertical Distribution of Species The substratum profile and the vertical distri- bution of some of the major species of Sargas- sum are presented ( Fig. 5 ) for each station. At station l(Fig. 5), the rocky bottom is evenly sloped to a depth of more than 8 m, then it gives way to- sand. The dominant species of Sargassum changes with increasing depth: 5'. piluliferum is predominant at depths less than 3 m, then S. patens predominates over the other species on rock surfaces about 3 m deep, and finally S. serratifolium predominates in deeper places. On the other hand, at station 6 (Fig. 5), the bottom is of boulders for the most part and there is some solid rock bottom in places. The slope of the bottom is lower than at station 1. While three of the species of Sar- gassum just mentioned are also present here, their degree of cover and vertical range do not coincide with what was found at station 1. As stated above, the vertical ranges of the species are not the same among the stations surveyed. An explanation for these differences is desirable. Sargassum at Tsuyazaki — Yoshida, Saw ADA, and Higaki 139 Fig. 5. Profile of substratum and distribution of species: ri, S. ringgoldianum; pi, S. piluliferum; to, S. tortile; fu, S. julvellum; pa, S. patens; se, S. serrati folium. Vegetational Cover The percentage of vegetational cover was ap- proximated by eye for several square meters around the spot where the frame of the quadrat was laid down. For this measurement, crustose algae were not included. As shown in a later section, algae other than Sargassum were found mostly in the relatively shallow places and had only a slight effect on the degree of vegetational cover in the deeper places. Variation in average vegetational cover with increasing depth was determined using the whole quadrat samples. This mean value is in- dicated with a dotted line in Figure 6. From this figure it can be seen that the average vegeta- tional cover is nearly uniform from the low watermark to 5 m deep, and then decreases gradually in places beyond the 5 m depth. Here only depth of water is taken into consideration as an environmental factor. In this study, the 13 stations surveyed are placed in three groups according to the strength of wave action. For each of the three groups, then, the relationship between the depth of water and average vegetational cover is plotted in Figure 6 by the three solid lines. At the rather sheltered stations subject to weak wave action, the maximum value for average vegeta- tional cover is found at the 1 m depth. As deeper water is reached the values fall quickly. For the areas where wave action is moderate, vegetational cover reaches its maximum 2-3 m below low watermark and gradually decreases with increasing depth. An examination of the values from the stations exposed to the strong- est wave action shows the values for average vegetational cover reach 100% above the low watermark, then fall and again come to a peak at the depth of 4-6 m. Algae other than Sargas- sum contribute to the vegetational cover in shal- lower places exposed to strong wave action. Therefore, when the degree of cover for Sargas- sum only is considered, the depth at which the maximum value for vegetational cover is found becomes deeper with increasing strength of wave action. 140 PACIFIC SCIENCE, Vol. XVII, April 1963 Distribution of Species Populations along Environmental Gradient The distribution pattern of the populations of Sargassum may not be understood when only the depth of water is taken into consideration as an environmental factor. The effect of the wave action on it must be considered at the same time. The survey stations are classed in three groups according to the degree of wave action, as shown above. Average degree of cover of each species is plotted against depth. For five major species, the curves of average degree of cover were superimposed to obtain the graph represented by Figure 7. The curve for each in- dividual species overlaps the others and the peaks of the different curves occur at different depths. For the places moderately exposed to wave action (Fig. 7, lower graph), the curve for S. hemiphyllum has its peak at a depth of 1 m below low watermark, that of S. ringgold- ianum comes second, and next in order are S. piluliferum, S. patens, and S. serrati folium, from the shallowest to the deepest. In places subjected to stronger wave action (Fig. 7, upper graph), the order of appearance is the same as above, but the peaks of the curves for degree of cover occur deeper, except in the case of S. ringgold- ianum. The method of gradient analysis developed % Fig. 6. Variation of average vegetational cover in relation to depth of water. Average of whole quadrat data x X Sheltered O O Moderate wave action A A Exposed Fig. 7. Change of cover degree in relation to depth of water. (Abbreviation of species names as in Fig. 5.) by Whittaker (1956) is introduced here to make the distribution pattern of these Sargas- sum species more clear. The depth of water and exposure to wave action are considered as the principal environmental gradients. The popula- tion nomograms for six major species con- structed on this basis are shown in Figure 8. S. serratifolium, S. piluliferum, and 5". hemiphyl- lum have their peaks at the stations exposed to stronger wave action, but their peaks oc- curred at different depths, in the order given, with S. hemiphyllum at the shallower end of the series. The species which occurred more abun- dantly at the places subject to more moderate wave action are S. ringgoldianum and S. fulvel- lum. The former grows more abundantly at places of 2 m depths where the bottom is hard and wave action is moderate. It has a tendency to grow in shallower places with both increas- ing and decreasing strength of wave action. This bimodal distribution pattern is peculiar to this species only. Only young individuals of S. fulvellum, an annual species, were found when the investigation was carried out, and the degree of cover of this species was very low. Nevertheless a population nomogram of dis- tinctive pattern was obtained. S. patens and S. tortile grow abundantly on bottoms more shel- tered from wave action. However, the popula- tion nomogram for one species differs more or less from the others in the position of maximum value, even if the distribution patterns are similar to each other. Generally speaking, most species of Sargassum grow in deeper places as the strength of the wave action increases. Sargassum at Tsuyazaki — Yoshida, Saw ADA, and HiGAKi 141 Objective Classification of Sargassum Commu- nity Goodall ( 1953 ) proposed a method of divid- ing vegetation samples into statistically homo- geneous groups. This method involves prepar- ing a 2 X 2 table whereon the correlation be- tween every pair of species is analysed by com- puting a value from the number of quadrats in which one or both of two particular species are present or not. An objective classification can be arrived at when, with respect to all species growing in the sample quadrats, no interspecific correlations of significant level occur within each group of quadrats, so that the quadrats may be divided into some homogeneous groups. The significance of interspecific correlation is decided by x2 test, preparing the 2X2 table, as shown below: SPECIES B present absent Species A present absent a b C d a + c b + d a + b C+d n=a+b+c+d (ad-bc±n/2) 2 n (a+b) (c+d) (a+c) (b+d) Values of x2 are calculated between every pair of the six species of Sargassum contained fre- quently in quadrats, and are shown in Table 2. Only two combinations, i.e., the pairs of S. ring- goldianum-S. piluliferum and S. tortile-S. pa- tens, have significant correlation at the 5% level. According to the first procedure of classi- fication proposed by Goodall (1953: 46-50), as a result of the elimination of quadrats con- taining one species of high frequency in those four species mentioned above, no species was found which has a significant interspecific cor- relation with the others in the final two groups obtained. However, the two groups obtained from this procedure are not useful because quad- rats belonging to one group are considerably intermixed with the others. Distribution of Algae Other than Sargassum Although attention was not centered on the distribution of algae other than Sargassum, ob- servational data on those algae were recorded as often as possible. The species observed in this survey are listed in Figure 9. The frequency of quadrats containing algae other than Sargas- sum is calculated for each depth, and the graph in Figure 10 is the result. This graph illustrates that the species other than Sargassum occurred in all quadrats laid above the low watermark, but that the frequency of their occurrence in quadrats decreases sharply with increasing depth. The range of the species growing at dif- ferent depths is shown in Figure 9. Dictyopteris undulata was observed in the deepest quadrat. The marine vegetation was richest in number of species at the place exposed to the strongest wave action, and more abundant in number of individuals near the low watermark. At deeper places the density of algae is less and the vege- tational cover may be provided by the species of Sargassum alone. DISCUSSION With respect to the stratal structure of the marine algal community, Segawa considered that three layers are recognized, i.e., the upper, lower, and crustose layers, when a well-developed vege- tation is being considered (Katada, 1958). The upper layer is more than 10 cm in height, and the crustose layer is composed of algae less than 1 cm high. At the time of the year when TABLE 2 x2 — Values between Six Species of Sargassum Frequently Encountered S. hemiphyllum S. piluliferum S. ringgoldianum S. tortile S. patens S. serratifolium 6.093 2.652 4.330 2.923 1.661 S. patens 0.517 7.337 6.391 8.405 S. tortile 2.527 0.045 0.000 S. ringgoldianum 2.197 9-776 S. piluliferum 3.772 Depth In meters 142 PACIFIC SCIENCE, Vol. XVII, April 1963 ( V * \ *4\ U.l^ \3 1.2 * \ A ^ 1-0 V .7 ^.5 - \<4 .1 — .25 S .hemiphyllum a 1 \ Wave action Fig. 8. Population nomograms for six major species of Sargassum. this survey was made, all the species of Sargas- sum found in the subtidal region formed but a single upper layer, and there was no constituent in this layer other than Sargassum . Katada re- ported that Undaria pinnatifida may also be in- cluded in this layer, intermingled with Sargas- sum. As already pointed out, in reference to the population nomograms in Figure 8, the Sar- gassum community appears to form a contin- uum. This is confirmed by the application of Goodall’s objective method of classification. Hence it is desirable that our attention be cen- tered on the distribution of each species popu- Sargassum at Tsuyazaki — Yoshida, Saw ADA, and Higaki 143 lation in relation to the gradient of environ- mental factors, rather than centering it on classi- fying the communities. The junior author has reported that the verti- cal distribution of a species of Sargassum is hardly explained when only the depth of water is taken into consideration, and that, in addi- tion, presence or absence of a species at a certain station may partly be determined by the degree of exposure to wave action (Yoshida, 1961). At present, our knowledge is insufficient to elucidate the detailed effect of many environ- mental factors on the distribution of the species. For marine plants, increasing the depth of water has an important effect on growth in connection with the diminishing intensity and variation in the spectrum of the light penetrating into the water. On the other hand, we must take into account the other environmental factors, such as wave action or the turbulence of water. There is also a resultant complex of several factors. According to Gessner ( 1955: 198-209), the movement of water may have a close re- lationship with the assimilation and respiration rates of marine algae. It has already been re- (Codium fragile) Dictyota dichotoma Dictyopteris undulata Pachydictyon coriaceum Padina crassa P. japonlca P. arborescens Zonaria rtiesingiana Stypopodium zonale Undaria pinnatifida Eisenia bicyclis (Ecklonia kurome) Pterocladia tenuis Chondrococcus hornemanni Jania sp. Amphiroa ephedra ea A. erassi ssirna A. dilatata Cora Hina pilullfera Lithophyllum okamural Laureneia okamurai L. undulata (Plocamiua telfairiae) Fig. 9. Vertical ranges of distribution of algae other than Sargassum. Species in parenthesis did not occur in quadrats. % Fig. 10. Frequency in occurence of algae other than Sargassum (cf Fig. 9) in relation to water depth. ported that the distribution pattern of inter- tidal organisms is greatly affected by the turbu- lence of water (Southward and Orton, 1954). In this present study the distribution pattern of the species of Sargassum is clearly shown by the two-dimensional gradient analysis improved by Whittaker (1956). Two principal environ- mental gradients considered here are (a) the depth of water, and (b) the exposure to wave action. In the area surveyed the tidal current is faster where the shore is exposed to the stronger wave action, as stated above. It is hardly possible to treat the two factors, namely, exposure to wave action and speed of tidal current, independently. Determining the lower limit of the marine vegetation was prevented by the diving ability of the observer. It is observed around Tsuzumi islet that the species of Sargassum grow at places deeper than 10 m, where hard bottoms are found. At most cf the stations, the hard bottom gives way to sand before what would seem to be the lower potential limit for Sargas- sum growth is reached. SUMMARY The Sargassum community found in the upper part of the subtidal region in the sea around Tsuyazaki, North Kyushu, Japan was studied phytosociologically in July and August, 1959. It seems that this community can not be divided successfully into smaller groups by ap- plying Goodall’s objective method of using posi- tive interspecific correlation. The gradient analy- sis method developed by Whittaker (1956) 144 PACIFIC SCIENCE, Vol. XVII, April 1963 was applied here to the major species of Sar- gassum. Two principal environmental gradients considered here are depth of water and exposure to wave action. The stratal structure of this community is also noted. ACKNOWLEDGMENTS The authors wish to express their sincere appreciation for the guidance of the late Pro- fessor S. Segawa throughout this study. Their heartiest gratitude is due Professor T. Hoso- kawa, of the Faculty of Science, Kyushu Uni- versity, and to Professor Maxwell S. Doty, of the University of Hawaii, for their kind advice and criticism of the manuscript. The authors’ thanks are also due Mr. T. Furukawa in the Fishery Research Laboratory, Kyushu Univer- sity, who assisted them in the course of the held survey. REFERENCES Gessner, F. 195 5. Hydrobotanik, Bd. I. Deutscher Verlag, Berlin. 517 pp., 291 figs. Gilmartin, M. I960. The ecological distribu- tion of the deep water algae of Eniwetok Atoll Ecology 41(1): 210-221. Gislen, T. 1930- Epibioses of the Gullmar Fjord II. Skrifts. K. Svensk. Vetensk. Kris- tenbergs Zoo. Stat. 4: 1-380. Goodall, D. W. 1953. Objective methods for the classification of vegetation, I. The use of positive interspecific correlation. Austral. J. Bot. 1(1): 39-63. Katada, M. 1958. Phytosociological considera- tion on the artificial stone beds for Undaria pinnatifida. Aquiculture 5(3): 30-39. [In Japanese.} Maritime Safety Board. 1947. Coast of Kyu- shu Pilot. Tokyo. 338 pp. Segawa, S., T. Sawada, M. Higaki, and T. Yoshida. 1959. Studies on the floating sea- weeds, III. Some considerations on the forma- tion of the floating seaweeds. Sci. Bull. Fac. Agr., Kyushu Univ. 17(3): 299-305. [In Japanese.} I960. Studies on the floating seaweeds, IV. Growth of some sargassaceous algae based on the material secured from floating seaweeds. Sci. Bull. Fac. Agr., Kyushu Univ. 17(4): 429-435. [In Japanese.} Southward, A. J., and J. H. Orton. 1954. The effect of wave-action on the distribution and number of the commoner plants and animals living on the Plymouth breakwater. J. Mar. Biol. Ass., U. K. 33(1): 1-19. Takehara, H. 1937. Geology of the Tsuyazaki area, Fukuoka Prefecture. Mem. Fac. Tech., Kyushu Univ. 12(5): 263-276. [In Japa- nese.} UEDA, S., and Y. Okada. 1938. Studies on the vegetation of the marine algae in the seas of Japan, with special reference to the depth of the growing zone. Bull. Jap. Soc. Sci. Fish. 7(4): 229-236. [In Japanese.} 1940. Studies on the vegetation of the marine algae in the seas of Japan, with special reference to the depth of the growing zone (II). Bull. Jap. Soc. Sci. Fish. 8(5): 244-246. [In Japanese.} WAERN, M. 1952. Rocky-shore algae in the Oregrund Archipelago. Acta Phytogeogr. Suecica. 30: 1-298. Walker, F. T. 1947. Sublittoral seaweeds sur- vey. J. Ecol. 35 (1-2): 166-185. Whittaker, R. H. 1956. Vegetation of the Great Smoky Mountains. Ecol. Monogr. 26: 1-80. Yoshida, T. 1961. A brief study on the Sargas- sum vegetation around Ushibuka, west Kyu- shu, Japan. Jap. J. Ecol. 11 (5): 191-194. [In Japanese.} The Role of Olfaction in Shark Predation* 1 Albert L„ Tester2 It IS RECOGNIZED that most if not all species of sharks possess a keen sense of smell which is used in detecting dead and wounded prey or other edible material during their well-known scavenging operations. The early experiments of Parker ( 1910) , Sheldon ( 1911 ) , and Parker and Sheldon (1913) established the role of the paired nasal organs as olfactory receptors. Parker (1914) demonstrated directional response in the smooth dogfish (Must elm canis) and pro- vided a plausible explanation of how this was accomplished; he postulated that the two sep- arated nostrils have the ability to detect small differences in the concentration of odorous ma- terials enabling the shark to orient in the direc- tion of equal stimulation and to head "up- stream” to the source. This tracking ability is well recognized by skin divers and fishermen who have involuntarily attracted sharks by re- taining speared fish or by discarding trash fish and offal from their boats. It seems unlikely that any shark species could maintain itself entirely by scavenging opera- tions, except perhaps in areas where man pro- vides forage such as bait, fish offal, or other forms of edible garbage. Certainly the larger species are recognized as active predators which attack uninjured living prey, including man. Doubtless, vision plays a predominating role in locating and tracking normal living prey, and possibly other senses such as hearing are also involved. In considering olfaction, attention has Presented at the Tenth Pacific Science Congress of the Pacific Science Association, held at the University of Hawaii, Honolulu, Hawaii, August 21 to September 6, 1961, and sponsored by the National Academy of Sciences, Bernice P. Bishop Museum, and the Uni- versity of Hawaii. 1 Contribution No. 179 of the Hawaii Marine Lab- oratory, University of Hawaii, Honolulu 14, Hawaii. 2 Department of Zoology, University of Hawaii, Honolulu 14, Hawaii. Manuscript received November 23, 1961. focused largely on feeding activity induced by the release of attractive substances such as blood or other body fluids from the wounds of in- jured prey. The possible part played by olfaction in the detection and tracking of uninjured liv- ing prey has been largely overlooked. In this paper I will present the results of experiments on the olfactory response of cap- tive sharks to extracts of natural foods, to hu- man materials, and to uninjured living fish in the hope of clarifying the role of olfaction in shark feeding activity. The results form part of an investigation of factors affecting the be- havior of sharks sponsored by the Office of Naval Research (Contract Nonr 2736(00), Project nr 104503) over the period 1959-61. The work was undertaken at the Eniwetok Ma- rine Biological Laboratory, Eniwetok Atoll, Mar- shall Islands, and at the Hawaii Marine Labora- tory, Oahu, Hawaii. I am grateful to the respec- tive directors, Dr. R. W. Hiatt and Dr. A. H. Banner, for laboratory facilities. I am indebted to personnel of the Atomic Energy Commission and the Pacific Missile Range Facility for logis- tic and other help. I am particularly indebted to those graduate students who have assisted in phases of the project: Edmond S. Hobson, Su- sumu Kato, Taylor A. Pryor, and Bori L. Olla. FACILITIES, MATERIALS, AND METHODS Eniwetok Marine Biological Station At Eniwetok, small (18-36 inches) blacktip sharks (Carcharhinus melanopterus ) and small (20-36 inches) grey sharks ( C . menisorrah) were readily caught in shallow water by hook and line and established in captivity. Holding facilities consisted of two large concrete tanks housed in a building and illuminated by over- head fluorescent lighting (Fig. 1). The tanks were supplied with running sea water pumped 145 146 PACIFIC SCIENCE, VoL XVII, April 1963 Fig. 1. View of shark tanks at Eniwetok Marine Biological Laboratory. (Photograph by Atomic Energy Commission.) from the lagoon at a maximum rate of flow of about 10 gal per min. Lying side by side with a common middle wall, the tanks were 50 ft long, 4 ft wide and 3 ft deep with turning basins 6 ft in diameter at both ends (Fig. 2). The tanks could be divided into 5 -ft sections by gates which slid in notches in the walls. Thus the sharks could be confined in a compartment consisting of one or several sections. Observa- tion booths located midway along the tanks on both sides or blinds erected elsewhere effectively concealed the observer from the sharks. Olfaction tests were conducted on both "nor- mal” and blinded sharks. The sharks were blinded after anesthesia in a 1/1000 solution of MS 222- Sandoz (cf Gilbert and Wood, 1957) by coagulating the proteins of the aqueous hu- mor with the diode probe of a "Hyfrecator” in- serted through the cornea. Proof of blindness was lack of response to a hand waved close to the surface as they swam by or lack of response to the beam of a flashlight directed at their eyes. Within 1 hr after recovery from anesthesia the sharks circled their compartment, guided by the tip of the outstretched pectoral fin which touched the wall. Within a day they were able to circle the compartment without this tactile aide. They soon fed avidly on pieces of fish, squid, or other food which settled to the bottom before it was eaten. The sharks would detect the odor while swimming in mid-water and would spiral down, converging on the food by swimming in a figure-8 pattern on the bottom. Our attempts at blinding sharks with contact occluders (Mishkin, Gunkel, and Rosvold, 1959) were unsuccessful, perhaps because of faulty technique in molding plastic "lenses” and fitting them to the eyes. In general, the response of Olfaction and Sharks — -Tester 147 the blinded sharks to olfactory substances was considerably less variable than that of sharks with normal vision. Except during starvation experiments, the sharks were fed sparingly about once or twice a week; uneaten food was removed to avoid putre- faction and an unnecessary increase in olfactory level. After investigating several different tech- niques during the early summer of 1959, a stan- dard method of testing materials which could be dissolved or suspended in water was adopted. With natural foods such as fish flesh, usually 5 gm of material was macerated in a Waring blendor for 5 min with 250 ml of fresh water; sea water was not used because essential oils tended to accumulate in the froth. The material was then filtered and held in a refrigerator until used. Prior to an experiment a small quantity (usually 3.0 ml) of the clear solution was di- luted with sea water to 25 ml in a test tube and then further diluted to 100 ml with sea water in a funnel. Substances other than natural foods were diluted to various concentrations before being tested. Prior to testing, a glass funnel and tube lead- ing from the observation booth to a point just below the surface in the center of a test area (e.g., Fig. 2, Tank 1, D) was filled with sea water to remove air bubbles; the contents were maintained by closing a pinchcock. The flow of sea water to the tanks was shut off at the inflow valve. Usually, five 2 -min control periods were then run, during each of which the time (sec- onds) spent by one or more of the sharks in the test areas (e.g., Tank 1, C and D) was re- corded by an electric timer activated by a foot switch. At the same time, observations were re- corded of the behavior of the sharks and some- times of the number of passes or turns in the test areas. The test material was then introduced silently while the sharks were at the far end of the compartment, and the activity of the sharks was again noted and recorded during five (or more) 2 -min test periods. The nature of the response was then categorized as attraction, re- pulsion, etc. on the basis of the graphed data and the notes. A similar technique was used in I960 except that timing was abandoned in favor of counting the number of turns. Each of the two test areas was divided by an imaginary line into halves; turns in the four half-areas were given weights of 1 to 4, with the weights increasing toward the half-area of introduction (Fig. 2). The graphed "count index” of activity seemed to re- flect our subjective impression of a response SEAWATER Fig. 2. Diagram of shark tanks at Eniwetok Marine Biological Laboratory. 148 PACIFIC SCIENCE, Vol. XVII, April 1963 more realistically than the "time index.” Special techniques used in studying the response of the sharks to living fish will be discussed later. Haivaii Marine Laboratory At the Hawaii laboratory several grey sharks of two species {Car char hinus spp. ) , hammer- head sharks {Sphyrna lewini ) , and tiger sharks {Galeocerdo cuvier ), all 5 to 7 ft in length, were readily caught by set line in the adjacent waters of Kaneohe Bay. They were established in large seminatural ponds (Fig. 3), screened by gates and flushed slowly by tidal action. Three grey sharks and one tiger shark were success- fully maintained in captivity for 3 years and were still alive and healthy at the time of writ- ing (October 1961). Hammerheads were suc- cessfully maintained in captivity for periods from 3 months to 1 year. It is suspected that their death was due either to injury caused by the other sharks or lack of food. They were un- able to compete successfully with the fast, ag- gressive grey sharks; often our efforts at pre- ferential hand-feeding failed when the food was taken persistently by the other species. The sharks were fed sparingly about once or twice a week on cut or whole fish. Several experiments were undertaken on a tiger shark and a grey shark following their re- spective establishment in Ponds 2 and 3, both of which were about 100 ft long, 60 ft wide and 3-4 ft in maximum depth. Observations were conducted from a 16-ft tower between the ponds. By means of a pump and hose a continu- ous flow of water was taken from one pond, led to the top of the tower, and thence led into a test area of the other pond. Following a series of 3 -min control periods, during which quanti- tative data were collected on the activity of the shark, notes were made of overt responses and the path of the shark was diagrammed. The ma- terial to be tested was then introduced into the stream of salt water after dilution in a suction funnel on top of the tower, and the observations were repeated during a series of 3 -min test periods. During the winter of 1959-60 both the grey and the tiger shark were transferred to Pond 5 (Fig. 4), a much larger enclosure about 360 ft long and 66 ft wide. Other grey and hammer- head sharks were added to this pond; eventually Olfaction and Sharks— Tester 149 they were confined in two-thirds of its length by a fence. Observations were conducted on the shark population from the 16-ft tower which had been moved to a central location along one side. Usually the sharks would swim back and forth along the length of the enclosure in a channel which averaged about 9 ft in depth. Oc- casionally the tiger shark and the hammerheads ( but rarely the greys ) would traverse the length of the pond in shallows 1—4 ft in depth along the side of the pond opposite to the tower. Two testing techniques were employed which are henceforth referred to as "point” and "curtain- funnel” or "curtain-drum” introduction. In both, activity was recorded during the usual control and test periods in a test area 50 ft in length and extending across the width of the pond. The area, centrally located in front of the tower, was marked off by cords which stretched across the pond and were several inches above the water surface at high tide. In "point” introduction the material was con- tained in a 5 -gal funnel on top of the tower and was introduced at a point either just below the surface or at a depth by means of a rubber tube suspended from a boom (Fig. 5). In "curtain- funnel” introduction the material passed from the funnel to a perforated hose running trans- versely across the bottom of the pond at the center of the test area and extending part way into the shallows. In a modification, used in the spring and summer of 1961 and called "curtain- drum” introduction, a continuous stream of salt water was pumped into the hose before and during control conditions; the stream was then, switched to a 50' gal drum containing about 40- gal of sea water together with the test material After introduction, which usually consumed about three 3 -min test periods, the flow was again switched to salt water. With both methods care was taken to prevent the generation of air bubbles in the curtain for they produced a vari- able visual response. By using dye it was found that the curtain was fairly uniform and rela- FlG. 4. View of Pond E. S. Hobson.) 5 at Hawaii Marine Laboratory showing the observation tower. (Photograph by 150 PACIFIC SCIENCE, Vol. XVII, April 1963 Fig. 5. Diagram of the arrangement of test apparatus in Pond 5 at Hawaii Marine Laboratory. tively stable during periods at or near slack water. When there was a tidal current the cur- tain lacked uniformity in concentration and was irregular in shape; it slowly spread to one end or the other of the test area; sometimes it spread in one direction at the surface and in the op- posite direction at or near the bottom. Normally the sharks would encounter the curtain of ma- terial during their passage along the deep chan- nel. They could avoid it by swimming in the shallows. Interpretation of a response Based partly on quantitative data averaged as an index of activity or plotted in graphic form, partly on notes of overt responses, and, in the case of pond experiments, partly on diagrams of the swimming paths of the sharks in, out, or through the test area, the result of each experi- ment was classified as no noticeable response (O), sensing only (S), weak attraction (A), strong attraction (AA), weak repulsion (R), or strong repulsion (RR). Occasionally these were supplemented by other descriptions such as "startled reaction,” "alarm reaction,” "agita- tion,” etc. The category "no noticeable response” needs no further explanation. The category "sensing” was reserved for a response which consisted of a sudden start or turn on the part of the shark on first encountering the test material but with no other noticeable component suggestive of either attraction or repulsion. Sensing responses were obtained with a variety of materials in- cluding weak acids, bases, and salts. The re- sponse was interpreted merely as an awareness of any change in the composition of an other- wise uniform environment. An "attraction” re- sponse included an initial sensing followed by a more or less prolonged hunting response, usually with rapid swimming, circling at or near the surface, and occasionally with a "gulping” or flexing of the jaws as when feeding. The shark would usually make several excited circles in the test area on encountering the material, and would then dash down the tank, returning to the test area for further circling. Almost in- variably the average index of activity during test conditions was higher than during control conditions. A "repulsion” response included an initial sensing, but this was followed usually by rapid departure from the test area, a slowing of swimming speed, and a "cautious” re-ap- proach to the test area. Often in subsequent passes the sharks would turn short of the test area. A strong repulsion was often accompanied by shaking of the head or flexing of the gill slits. Almost invariably the average index of activity during test conditions was lower than during control conditions. Olfaction and Sharks — Tester 151 Unfortunately the response to a given ma- terial varied considerably in repeated testing and was often difficult to classify. The problems of bioassay should not be minimized. The re- sponsiveness of the sharks both at Eniwetok and Hawaii seemed to vary from day to day for unknown reasons despite our attempts to maintain standard conditions of testing and feeding. Erratic behavior, frequently encoun- tered during both control and test conditions, in some cases could be traced to obvious sources of disturbance, such as noise, but in other cases could not be explained. Particularly exasperating was an occasionally exhibited tendency to circle at one or the other end of the pond or tank for long periods of time so that tests could not be conducted. Even though an attempt was made to conduct tests only after some reasonable uni- formity in swim pattern persisted throughout control periods, there was always the question of whether or not a subtle change in behavior pattern was related to the material being tested. In classifying a response, greater reliance was placed on overt signs such as sudden turns, cir- cling, gills flexing, and head shaking than on the quantitative data. Unfortunately the location and concentration of the material being tested was not known precisely during tests and could only be estimated from the use of dyes after an ex- periment had been completed. Thus even overt responses could be related to the test material only by inference. Our caution in interpretation is reflected in the large number of responses relegated to doubtful categories in the results which follow and the numerous tests which were conducted on materials of particular interest. RESPONSE TO EXTRACTS OF NATURAL FOODS Experiments were conducted on the response of the tiger shark to extracts of tuna flesh and decayed shark flesh and on the response of the blacktip and grey sharks to a wide variety of potential foods including tuna, eel, grouper, snapper, parrot fish, jack, giant clam, octopus, squid, lobster, fresh shark flesh and skin, and decomposed shark flesh and skin. In general, the extracts of all food substances tested could be classed as attractants, although because of variability in the response of the sharks some- times several tests of the same substance were necessary to establish this conclusion. Because of this variability it was not possible to make rigorous comparisons between the attractive- ness of extracts from equivalent quantities of the various foods. However, it seemed certain that extracts from moist- or oily-fleshed fish such as grouper, tuna, and eel generally resulted in greater activity than those from dry-fleshed fish such as snappers. In attempting to determine the response to nonfood substances, frequently standard extracts of grouper, tuna, or eel were used either before or after tests of the other substances to appraise the sharks’ responsive- ness. From Table 1, summarizing the results of 169 tests, it will be noted that the response was classed as a strong attraction in 59, as a weak or doubtful attraction in 62, as nil or merely a sensing in 39, and as a weak or a doubtful re- pulsion in 9. The last, comprising 5% of the tests, warrants further comment. The five instances of apparent repulsion in the 1959 tests at Eniwetok involved extracts of little tunny ( Euthynnus yaito ) , yellowfin tuna ( Neothunnus macropterus) , and giant clam (Tridacna) , and occurred early in the summer when testing techniques were being developed. Without doubt the response was related to either incipient pollution of the tanks or decomposi- tion of the test materials. Excessive quantities of extracts were being used and excess food was not being removed from the tanks; one or both of these factors resulted in the death of several sharks in one compartment before the condition was rectified. Our notes state that the tunny ex- tract, which had been kept for 9 days, smelt foul. The four instances of apparent repulsion in the I960 tests, involving standard extracts of eel, again took place early in the summer and involved not pollution of the tanks but decom- position of the extract even though it was held at ice-box temperature. The extract was pre- pared on July 7, I960. On that and the follow- ing day tests of both greys and blacktips showed strong attraction (Figure 6A). On July 13, in seven tests the responses were indicated as weak or questionable attraction, sensing only, or nil. On July 16, the material produced erratic re- 152 PACIFIC SCIENCE, VoL XVII, April 1963 TABLE 1 Response of Sharks to Extracts of Natural Food at Hawaii Marine Laboratory (hml) and Eniwetok Marine Laboratory (embl) SHARKS, LABORATORY, AND RESPONSE* YEAR MATERIAL RR R-R? O-S A-A? AA Total 1959 Tiger, HML Fresh tuna extract 3 2 2 7 Aged shark extract - 1 1 2 1959 Blacktips, EMBL Various extracts ! 5 21 32 32 90 Aged shark extract - - 7 8 3 18 1960 Blacktips and greys, EMBL Fresh extracts 12 18 30 Aged eel extract - 4 8 6 2 20 Aged shark extract - - - 1 1 2 Total - 9 39 62 59 169 * RR, strong repulsion; R-R?, weak or doubtful repulsion; O-S, no apparent response or sensing; A-A?, weak or doubtful attraction; AA, strong attraction. sponses, some of which were classed as doubtful repulsion. The notes indicated that the material smelt foul. Similar results were obtained with the same material on July 18. The sharks were tested with freshly prepared standard eel ex- tract on July 20 and both species showed a strong attraction response. In direct contrast to the above results are those with extracts of decayed shark flesh, which after a week in the hot sun smelt particularly foul. Our material consisted of extract of decayed hammerhead and of decayed tiger shark tested on the tiger shark at the Hawaii Laboratory (two tests) and of extracts of decayed blacktip shark flesh and skin tested on blacktip sharks at Eni- wetok ( five tests ) . In addition, we tested black- tips at Eniwetok on an alleged shark repellent, supplied by a fisherman, which contained ex- tract of decayed shark flesh as the principle component ( six tests ) . We also tested fractions of extract of decomposed shark flesh which were supplied by Dr. M. A. Steinberg, Bureau of Commercial Fisheries Technological Laboratory, Gloucester, Mass. (11 tests ) . No repellent ef- fects were noted in any of the tests. On the con- trary, the majority yielded responses which were classed as either weak or strong attraction. Our results with the fractions of extract were in agreement with those reported by Steinberg (I960) when his material was later tested on the lemon shark ( Negaprion hrevirostris) , the reef shark ( Carcharhinus falciformis ) , and the bull shark ( Carcharhinus leucas ) at the Lerner Marine Laboratory, Bimini, Bahamas, B. W. I. Our results with extracts of decomposed shark flesh seem to be at variance with those of Springer (1955), who found that the feeding of the dogshark ( Mustelus canis ) was consist- ently inhibited by the presence of decayed shark flesh. Although several hypotheses might be formulated to account for the difference in re- sults, no convincing explanation can be made at the present time, particularly in view of the apparent repulsion noted with decomposed eel and other extracts noted in preceding para- graphs. BEHAVIOR OF STARVED SHARKS In considering shark predation, the questions arise as to how long a shark can exist without food and whether its olfactory response is modi- fied by starvation. Some information on these points was obtained for small sharks at the Eni- wetok laboratory. In 1959, following the summer’s work, AEC personnel at Eniwetok volunteered to keep track of the fate of four small blacktips under starva- tion conditions. Three of the sharks died after about 2 months in captivity. One survived for 3 months but it was not known to what extent it had maintained itself by feeding on the sharks which had died. ACTIVITY INDEX CONTROLS TEST Fig. 6. Activity index during successive 2-min periods, illustrating (A) response of blacktip and grey sharks to standard eel extract, and (B) response of starved (normal) and fed (blinded) blacktip sharks to a 1/1000 dilution of standard grouper extract. 154 In I960, four blacktip and four grey sharks were starved under close supervision. Of the blacktips, a 20-inch male died after 36 days, a 28-inch female died after 40 days, a 20-inch female died after 40 days and a 27-inch female survived for 43 days and was then fed. Of the greys, a 28-inch female died after 32 days and a 36-inch male died after 40 days. Two 30-inch females were starved respectively for 34 and 46 days and were then fed. During starvation, the sharks became very thin. Those which died had difficulty in maintaining their equilibrium for several hours before death; they could not be revived by forced feeding. These experiments show that small blacktip and grey sharks can survive for more than a month without food but that some will die after 5 or 6 weeks. Using standard extract, tests were run at in- tervals to measure the response activity of both the starved blacktips and greys and, for com- parison, that of four blinded blacktips which were fed two or three times a week. Activity data are given in Table 2. The activity index for the fed sharks fluc- tuated from day to day but showed no trend. In contrast, the index for the starved sharks showed a more or less steady increase during both con- trol and test conditions, particularly in the case of the blacktips. For the latter, the decrease in index on the last day is due to the moribund condition of two of the four sharks. As starva- tion proceeded, the sharks seemed to become increasingly restless and to respond with in- creasing vigor to the standard extract. Tests on the greys were discontinued after August 5, 1961 because of the death of one shark and the moribund condition of another (starvation be- PACIFIC SCIENCE, Vol. XVII, April 1963 gan about 2 weeks earlier than with the black- tips). Normally the blinded blacktips were more sensitive to odorous substances than the black- tips with normal vision. This situation was re- versed when the latter sharks were starved. This is illustrated by one experiment (August 20, I960) when both the blind, fed and the normal, starved sharks were tested with a 1/1000 dilu- tion of standard extract (Fig. 6B). Using 0.003 ml (rather than the usual 3.0 ml) the fed sharks showed a weak attraction response which did not differ greatly from mere sensing. The starved blacktips on the other hand gave a strong at- traction response which included the usual ex- cited circling and hunting activity. There is no doubt that hungry sharks are much more responsive than fed sharks to minute traces of odorous substances. RESPONSE TO HUMAN MATERIALS In this section are reported the results of tests on the response of normal and blinded blacktip sharks and normal grey sharks to human urine, blood and sweat, and to L-serine, a presumed component of human sweat. Other materials such as faeces and vomit were not investigated. Urine At Eniwetok, in both 1959 (eight tests) and I960 (three tests) blacktip sharks were pre- sented with human urine in quantities ranging from 3 to 80 ml of whole material. The urine was sensed, as indicated by a swirl or turn on encountering it, but there was no other con- sistent response. TABLE 2 Activity Index of Fed (Blind) and Starved (Normal) Sharks to Standard Extract at Eniwetok Marine Biological Laboratory, I960 DATE FED BLACKTIPS STARVED blacktips STARVED GREYS Controls Test Controls Test Controls Test 7/20 29.2 41.2 31.0 40.4 33.8 60.4 7/23 21.2 32.0 31.8 44.2 23.8 38.6 7/29 34.8 37.6 57.4 76.8 52.6 74.2 8/5 29.6 54.0 50.6 71.2 86.3 96.8 8/18 27.2 36.2 56.0 97.0 — — 8/25 23.4 32.2 36.4 49.2 — — Olfaction and Sharks— -Tester 155 TABLE 3 Response of Sharks at Eniwetok Marine Biological Laboratory to Human Blood RESPONSE* YEAR MATERIAL AND SHARKS RR R-R? o-s A-A? AA Total 1959 Fresh or aged blood Normal blacktips 4 4 1 9 Blinded blacktips 1 4 2 2 - 9 Total 1 8 6 3 - 18 1960 Aged blood (4-6 days) Normal blacktips 1 2 1 4 Blinded blacktips - 2 3 - - 5 Normal greys - - 1 4 - 5 Total 1 4 5 4 - 14 1960 Fresh blood (1-2 days) Normal blacktips 2 2 2 6 Blinded blacktips - - 2 2 3 7 Normal greys - - 1 4 - 5 Total - - 5 8 5 1 18 * RR, strong repulsion; R-R?, weak or doubtful repulsion; O-S, no apparent response or sensing; A-A?, weak or doubtful attraction; AA, strong attraction. Blood Most authors agree that blood in the water excites sharks. For example, Whitely (1940) notes that small blacktip sharks on the Great Barrier Reef would follow persons who had scratched their legs on coral and would dog their footsteps through slightly bloodied water. Bige- low and Schroeder (1948) remark that if per- sons in the water are bleeding from injuries the danger from shark attack may be imminent and the results may prove fatal. Moreover, they state that the more voracious of the larger sharks are excited by blood in the water to such a degree that they will make ferocious attacks whether the object be fish, whales, or man, dead or alive. In contrast, based on experience with releasing turtle and sheep blood while fishing for sharks, Wright (1948) concluded on ad- mittedly weak evidence that blood alone, with- out the presence of some moving object, did not release the attack pattern. Steinberg ( 1961 ) reports that a captive lemon shark was not at- tracted by solution of dried beef blood. The results of experiments with human blood in 1959 are included in Table 3. The responses were much more variable and erratic than those with food extract. A sensing was at times fol- lowed by an attraction response and at other times by an apparent flight reaction and a tend- ency to avoid the test area. The erratic behavior was unlikely related to a visual stimulus as it oc- curred in both the blinded sharks and those with normal vision. Moreover the quantities used, even when the techniques were being de- veloped, were not sufficient to produce notice- able coloration in the water. It was suspected that the variability in response was related to the freshness of the blood. In I960, 32 experiments were conducted with human blood. The results are summarized in Table 3 and are given in detail in Table 7. Quantities ranged from 0.03 to 6.0 ml of a sus- pension of 5 ml of whole blood in 250 ml of sea water. With fresh blood tested within 1 or 2 days after collection, 3.0 ml of the suspension usually produced a moderate or strong attraction response with the usual behavior components: excited circling, swirling and hunting. An attrac- tion response was obtained with 0.3 ml of the fresh suspension on several occasions and with as little as 0.03 ml in one test. By the use of dye it was estimated that the shark first encoun- tered the material when it had mixed with Va to Yi of the volume of the test compartment. If this dilution is assumed, it may be estimated 156 PACIFIC SCIENCE, Vol XVII, April 1963 that the sharks were attracted to human blood at a concentration of about 0.1 to 0.01 parts per million of sea water. A blood suspension held under refrigeration for 4 days or longer usually underwent hemoly- sis and acquired a faint to strong putrid odor. The aged blood produced erratic results with blacktip sharks as had been suspected in the 1959 tests. At times there was only a sensing of the material, an avoidance of the area, or possibly a slight attraction. At other times there seemed to be a "startled” or 'alarm” reaction with speeding from the area such as had been noticed the previous year. This was classed as repulsion. With grey sharks, on the other hand, the hemolyzed blood seemed more consistently attractive. Our results prove that fresh blood excites blacktip and grey sharks and promotes a strong hunting response. They suggest that decomposed human blood contains a component which is repellent to blacktips. Sweat A large number of experiments were con- ducted at both the Eniwetok and the Hawaii laboratories on the response of sharks to human sweat. They were stimulated by the observation of Brett and McKinnon (1954) that water in which human hands had been rinsed retarded the upstream migration of salmon and induced an "alarm” response. Sweat was collected initially by sponging the body and wringing the sponge in 500 ml of sea water. Later, at Eniwetok it was collected directly from the body as it ran down arms, chest and abdomen in the hot, humid atmos- phere of the shark house, and at Hawaii it was collected in the same way by exercising and subjecting the body to heat lamps. The material was tested according to the standard procedures already described. Each sweat test was usually followed or preceded by standard extract to ap- praise the sharks’ responsiveness to a known attractant. The results of 29 tests conducted on normal and blinded blacktip sharks at Eniwetok are included in summary form in Table 4 and are given in detail in Table 8. The majority of the tests yielded results which were classed as re- pulsion. A weak repellent effect (R) was com- prised of an initial sensing, followed by a slow- ing of swimming speed, an apparent wariness, TABLE 4 Response of Sharks to Human Sweat at Eniwetok Marine Biological Laboratory (embl) and Hawaii Marine Laboratory (hml) RESPONSE* YEAR LABORATORY AND SHARKS RR R-R? o-s A-A? AA Total 1959 EMBL, Normal blacktips 1 6 3 1 — 11 Blinded blacktips 8 9 1 - - 18 Total 9 15 4 1 - 29 I960 EMBL, Normal blacktips 5 7 1 1 _ 14 Blinded blacktips 2 4 1 - - 7 EMBL, Normal greys - 9 6 - - 15 Total 7 20 8 1 - 36 1959-60 HML, tiger, Pond 2 HML, tiger, grey, 1 3 2 - - 6 hammerhead, Pond 5 - 8 7 - - 15 Total 1 11 9 - - 21 1960-61 HML, tiger, greys, Pond 5 - 3 2 - - 5 Total 17 49 23 2 - 91 * RR, strong repulsion; R-R?, weak or doubtful repulsion; O-S, no apparent response or sensing; A-A?, weak or doubtful attraction; AA, strong attraction. ACTIVITY INDEX CONTROLS 0.3ML SQUID EXTRACT J I I— J I I I L TEST TEST BLINDED ' BLACKTIPS 8-12-59 j t—i i I i CONTROLS TEST TEST Fig. 7. Activity index during successive 2 -mm periods in two experiments, illustrating the difference in response of blinded blacktip sharks to human sweat and to food extract. 158 PACIFIC SCIENCE, Vol. XVII, April 1963 TABLE 5 Index of Activity in Successive Rings (No. 1, Center) of a Target Area for a Hammerhead Shark During Successive Tests of "Tilapia Water’’ and Human Sweat in Pond 5, Hawaii Marine Laboratory, June 16, 1960 RING NUMBER TEST 1 2 3 4 AVERAGE Controls, sea water 4.5 5.5 14.8 20.7 11.4 Tilapia water 27.3 29-3 22.0 11.8 22.6 Sweat 5.4 8.3 13.1 12.3 9.8 Tilapia water 28.4 36.8 29.4 20.6 28.8 Controls, sea water 15.0 19.1 16.9 20.0 17.7 Tilapia water plus sweat 15.1 30.0 21.2 16.9 20.8 and a tendency to avoid the test area. A strong repellent effect (RR) included in addition a rapid exit from the area following sensing, sometimes accompanied by head shaking. When no noticeable response occurred it was found usually that this was also the case with the known attractant. In the one case classed as doubtful attraction, the first sweat test which was conducted, it is likely that initial sensing was interpreted as attraction. The quantitative data of Table 8, illustrated for two experiments in Figure 7, give convincing evidence of a de- pression of shark activity following the intro- duction of sweat. In each of the 29 tests, the time spent in the test areas during test condi- tions was less than during control conditions; this is in striking contrast to the results with standard extract where the reverse is almost invariably encountered. The apparent aversion to sweat was exhibited by both normal and blinded blacktips, possibly to a greater extent in the latter. The sweat of one donor (ALT) seemed to be effective at a roughly-calculated concentration of about 1 part per million. The sweat of a second donor (TAP) produced no obvious repellent effects in the two tests which were made. Because of the possibility that the sharks in 1959 had become conditioned to associate sweat with punishment, e.g., from handling, the ex- periments were repeated during I960, taking precautions against sweat dripping accidentally into the water and using fresh sharks, some of which had not been touched by hand. The re- sults in I960 were similar to those in 1959 (Tables 4, 9). With the blacktips the majority of the responses were classed as repulsion; in the one case of apparent attraction, again the first test of the season, the sharks had just been introduced and still exhibited erratic behavior. There was no noticeable difference in response between the blinded and normal blacktips. With the grey sharks an aversion to human sweat was present in the majority of the tests but it seemed less pronounced than with the blacktips. There were no obvious differences in response between the sweat of three donors. The sweat did not decrease in potency with aging at room tem- perature for several days; rather, its repellent properties seemed to increase but this could not be established with certainty. In tests conducted during the winter of 1959— 60 on the tiger, hammerhead, and grey sharks at the Hawaii laboratory, larger quantities of sweat (10-40 ml) were used because of the greater volume of the ponds compared with the Eniwetok tanks. In most cases the results, sum- marized in Table 4, showed vague repellent ef- fects following the introduction of sweat. For the most part, the sharks displayed only a tend- ency to avoid the test area (R? ), occasionally with a sharp veering from the presumed loca- tion of the material and rapid swimming through or away from the test area (R). Apart from these latter overt responses which were displayed on occasion by all three species, the only strong repulsion concerned the tiger shark in one test and consisted of obvious "agitation” and head shaking. Repellent effects were noted with the sweat of all three donors ( ALT, RJ, and ESH) but more consistently with that of the first two than the last. Olfaction and Sharks — -TESTER 159 A different technique was employed in tests with the hammerhead which was particularly responsive to attractants. The introduction hose running from the funnel on top of the tower was submerged just below the surface at the center of a bullseye target area 32 ft in diameter, the boundaries of which were judged by eye from reference points on the bottom. During control conditions, sea water was introduced from the funnel. During test conditions the fol- lowing materials were used in varied sequence: ( 1 ) an attractant consisting of water from the funnel in which fish ( Tilapia ) were swimming, (2) 50 ml of sweat mixed with sea water in the funnel, and ( 3 ) a mixture of the attractant and sweat in sea water. Activity data on one test are given in Table 5. Based on records of the time spent and the path followed by the shark in the target area, activity was calculated as the distance- swum in each of four rings of the target per unit of time. It is apparent that, in general, activity was greatest with the attrac- tant, intermediate with the mixture of sweat and attractant, and least (less than controls) with sweat alone. The sweat depressed but did not eliminate the response to the attractant. By tracing the spread of materials in the pond with the use of dye and calculating the volume of sea water involved, it was concluded that the actual sea water concentration of sweat in the pond experiments at the Hawaii laboratory were still considerably less than those used in the tank experiments at Eniwetok. Additional ex- periments were undertaken during the early summer of 1961 using much larger quantities of sweat (100-400 ml per test) and the im- proved "curtain-drum” method of introduction. The results are summarized in Table 4. Despite the larger quantities of sweat which were used no strong repulsion was noted. In the five tests, there was weak or doubtful repulsion in three and sensing only in two. In those tests indicat- ing repulsion, all three species of sharks, espe- cially the tiger, showed definite signs of aversion including veering from the curtain and gill flexing. The sweat of one donor (sk) seemed to be more active than that of the other (blo) . From the above experiments on blacktip, grey, tiger, and hammerhead sharks one cannot conclude' that human sweat, per se, is an active shark repellent. On the other hand, it is certain that human sweat does contain, at least at times, a component which is aversive to sharks. Oc- casionally this induces overt signs of repulsion such as head shaking, gill flexing, veering, and rapid retreat; more frequently it induces only a subtle wariness manifested chiefly by avoid- ance of the area of introduction. The response is highly variable. This is unfortunate but almost inevitable when one considers the uncontrolled environmental conditions and the many factors which could contribute to both the variability of shark behavior and variability of sweat com- position. Steinberg (1961) found no evidence of re- pellent properties in either human sweat or pure compounds forming constituents of human sweat in tests with a captive lemon shark at the Lerner Marine Laboratory. Unfortunately he gives no information on the concentrations of material used. Moreover, he reports that the lemon shark was not responsive to solutions of dried beef blood nor would it eat chunks of fresh shark liver which, at other times, had been particularly attractive to captive sharks. His negative results are understandable. It has been our experience that sharks which have not yet fed in captivity do not respond to either highly attractive substances such as eel extract or fresh human blood, nor do they respond to subtle repellent substances such as human sweat. In view of the results of this series of tests, it seems safe to assume that shark attack on hu- mans is not motivated by the smell of human sweat. L-serine Following the discovery by Brett and Mc- Kinnon ( 1954) that human hand rinse retarded the migration of salmon, Idler, Fagerlund, and Mayoh (1956) undertook an analysis of hand rinse to determine the repellent component ( s ) . By employing various fractionation techniques and testing the fractions on migrating salmon, they were able to identify the active fractions as amino acids of which serine was a major com- ponent. In further tests, the L-isomer of serine was found to induce the alarm response whereas D-isomer did not. They stated "L-serine defi- nitely elicited a typical alarm reaction but the effects were neither so dramatic nor so long a duration as the response obtained by hand 160 rinse.” As hand rinse doubtless contained sweat and as sweat produced a repellent effect on sharks, it was decided to investigate their re- sponse to L-serine. Three tests were conducted at the Hawaii laboratory on the tiger shark in Pond 2, during the winter of 1959-60, using 1.0 gm of L-serine per test. In the first, there was no overt response except an obvious sensing on encountering the material. In the second, the shark showed a sensing of the material, followed by rapid exits from the area and violent head shaking. In the third test, there was no noticeable response. In all three, however, the time spent in the test area during test conditions was less than during controls, as had also been the case with sweat. Three tests were conducted on blacktip sharks at the Eniwetok laboratory in I960, using 6 ml of a solution containing 1.0 gm of L-serine (i.e., 0.12 gm per test). No repellent effects were noted other than a "wariness” in one test. Again, however, the time spent in the test area during test conditions was less than during con- trols. As definite repellent results had been noted in one test with the tiger shark, it was decided to run a third series at the Hawaii laboratory during the summer of 1961, using much larger quantities of L-serine despite its high cost. Three tests were conducted on the tiger and grey sharks co-inhabiting Pond 5, using the "curtain- drum” technique. In all three tests the tiger shark displayed an aversion to the chemical but only after intro- duction of the material had been completed. In- troduction required about 10 min (three to four 3-min periods). It seemed either that the response occurred after a threshold concentra- tion of the material had been reached, or that there was a latent period between exposure to the material and response. The overt response was a violent head shaking either while In or while leaving the area of concentration. At times this took place at the surface and caused considerable splashing. However, in only the first test (25 gm L-serine) was there frequent rapid exit from the area on encountering the material. In the second test (50 gm L-serine) there was swerving and head shaking after en- countering the material, but no turning-back on initial encounter. The response was less pro- PACIFIC SCIENCE, Vol. XVII, April 1963 nounced than in the first test even though twice the quantity of material had been used. This may have been due to a higher tide and thus a larger volume of water and greater dilution of the material despite the larger quantity used. The shark frequently avoided the material by swimming in the shallows on the far side of the pond. In the third test (25 gm L-serine plus 10 ml of a 99% nicotine solution), the tiger shark again displayed agitation and head shak- ing. The response, however, occurred less fre- quently than in the other two experiments. The nicotine had been added in the hope of a syn- ergistic effect; it had been our impression that the sweat of smokers was more repellent to the sharks than that of nonsmokers. Possibly it tended to inhibit rather than Increase the ef- fect of L-serine. The grey sharks, in contrast to the tiger shark, were not obviously agitated by L-serine; no head shaking or gill flexing was observed. In the first test, two grey sharks of the same species veered sharply on first encountering the material and returned to the end of the pond. Thereafter all three grey sharks circled in the end zone for the duration of the experiment. In the second test no veering was noted but there was re- peated circling in the end zone. It was uncertain whether this could be interpreted as a repellent effect, for the same habit was noted occasionally during control periods. In the third test, all three grey sharks passed through the test area without signs of awareness, agitation, or repulsion. Although there is no doubt that the tiger shark was actively repelled by L-serine, the phys- iological mechanism producing the response is unknown. We can offer no satisfactory expla- nation of the difference in response of the tiger and the greys to L-serine. It may have involved species differences in physiological effect or differences in the concentration of materials to which they were subjected. The latter is possible even though the tests were conductd simultaneously on the tiger and the greys, for uneven curtains of material were formed by tidal currents in all three tests and the shallow- swimming tiger shark may have encountered different concentrations than the deep-swim- ming grey sharks. It may be added, however, that directly opposite results were obtained with a highly irritating lachrimator which is presently Olfaction and Sharks — Tester 161 Fig. 8. View of a tiger shark attempting to swallow a spiny puffer. (Photograph by E. S. Hobson.) being tested as a potential shark repellent. With this substance, the greys responded violently with gill flexing, head shaking, and definite avoidance of the curtain. The tiger shark, on the other hand, was not noticeably affected until the concentration of the material had been dou- bled. In the foregoing sections it has been shown that certain species of sharks have an aversion to and at times are repelled by aged food ex- tracts, aged human blood, fresh or aged human sweat, and finally L-serine. Serine, presumably the D-isomer, is a common amino acid in both foodstuffs and blood. It may by hypothesized that at least one of the repelling substances in all of the above materials is L-serine, which is presumably present as the L-isomer in human sweat, but which may be formed in foodstuffs and blood from the nonrepellent D-isomer dur- ing decomposition. It has not been demonstrated that L-serine has sufficiently active repellent properties to deter shark attack on prey, including man. How- ever its repellent properties, its presence in hu- man sweat, and its possible generation during the decomposition of foodstuffs and blood war- rant further investigation. RESPONSE TO LIVING FISH Although at times sharks may obtain a large portion of their food by scavenging dead ma- terials, they also feed on living prey. When the prey is wounded, the sharks are doubtless at- tracted by the odor of body juices as well as by visual and possibly by other stimuli. It is 162 PACIFIC SCIENCE, Vol. XVII, April 1963 reasonably certain that most species also attack- healthy, undamaged, living prey, although apart from attack on man, observations of feeding activity are singularly lacking in the literature. In the summer of 1959, a group of biologists from the Hawaii Marine Laboratory witnessed the persistent attack of a tiger shark on a spiny puffer which had inflated itself and was floating at the surface in Kaneohe Bay. The shark’s at- tempts to swallow the puffer (Fig. 8) lasted for about 10 min despite the presence of the observers who circled in an outboard motor boat. During the shark’s slow, awkward passes at the puffer the sound of its jaws clamping together as it missed the prey could be heard. Other species of sharks are capable of catch- ing fast moving prey. For example, Eibl-Eibes- feldt and Hass (1959) observed both the grey shark ( Carcharhinus menisorrah ) and the black- tip (C. melanopterus ) actively feeding on healthy fish in the Indian Ocean, and even herd- ing them against the shoreline to facilitate cap- ture. Although vision is doubtless the predominat- ing sense which is used by sharks on converging on living undamaged prey, it is possible that olfaction may also be involved. I have found only one observation in the literature which supports this possibility, that reported by Shel- don (1911) and again by Parker and Sheldon (1913), who found that the dogshark ( Mustelus canis) was able to locate undamaged living crabs concealed in a wrapping of eelgrass. The response of sharks to living, presumably un- damaged fish was investigated at both the Eni- wetok and the Hawaii laboratories. Results In one series of experiments at Eniwetok in 1959, an empty wire cage (about 6 X 6 X 12 inches) was silently lowered to the bottom of a test area at the upstream end of a compart- ment containing four blinded blacktip sharks. After the usual series of control periods during which activity was recorded, the cage was re- moved, a living fish was added, and it was again lowered into the test area when the sharks were at the far end of the compartment. Activity was again observed during a series of test periods. The water flow was maintained during both control and test conditions. The results are included in Table 6. In most of the experiments attraction responses were obtained with a 12 -inch grouper ( Epinephalus fuscoguttatus ) , an 8-inch squirrel fish (Holo- centridae), and an 8-inch stone fish ( Synancaja verrucosa) . Although probably excited by con- finement in the cage, the fish did not move about much after the cage had been lowered. There was often a delay of several test periods before the sharks showed any response. Then, in most of the experiments, one or more sharks suddenly TABLE 6 Response of Sharks at Eniwetok Marine Biological Laboratory to Living Fish, 1959 RESPONSE* SHARKS AND FISH RR R-R? O-S A-A? AA Total Blinded blacktips Caged grouper - - 1 3 1 5 Caged squirrel fish - - 1 1 2 4 Caged stonefish - - - 2 - 2 Total - - 2 6 3 11 Blinded blacktips Grouper water, grouper present - - 1 2 2 5 Grouper water, grouper absent - - 2 - 1 3 Eel water, eel present - - - 1 - 1 Blacktip water, blacktip absent - - 2 - - 2 Total - - 5 3 3 11 * RR, strong repulsion; R-R?, weak or doubtful repulsion; O-S, no apparent response or sensing; A-A?, weak or doubtful attraction; AA, strong attraction. Olfaction and Sharks — Tester 163 TABLE 7 Response of Normal Blacktip Sharks (nb), Blinded Blacktip Sharks (bb), and Normal Grey Sharks (ng) to Human Blood at Eniwetok Laboratory, I960 DATE (1960) TIME TANK SHARKS BLOOD (ML) DONOR, DATE ACTIVITY INDEX RESPONSE * Controls Test 7/12 0954 I (F-I) 2BB 3.0 JK 7/8 26.0 9.0 R 1040 II (A-C) 2NG 3.0 JK 7/8 13.0 20.2 A 1108 I (A-D) 2NB 3.0 JK 7/8 57.8 39.0 R 1445 I (F-I) 2BB 6.0 JK 7/8 21.6 6.6 R 1515 II (A-C) 2NG 6.0 JK 7/8 19-8 24.6 A 1540 I (A-D) 2NB 6.0 JK 7/8 15.6 13.2 R? 7/18 1529 I (F-I) 2BB 3.0 JK 7/8 21.6 24.6 O? 1552 I (A-D) 2NB 3.0 JK 7/8 37.2 17.8 RR 1645 II (A-C) 2NG 3.0 JK 7/8 11.0 10.0 O? 7/21 1015 I (F-I) 4BB 3.0 ALT 7/20 22.8 35.6 AA 1045 I ( A-D ) | 4NB 3.0 ALT 7/20 27.7 26.4 O? 1135 II (A-C) i 4NG 3.0 ALT 7/20 37.4 49.8 A 1350 I (F-I) 4BB 3.0 ALT 7/20 30.4 39.2 AA 1415 I (A-D) 4NB 3.0 ALT 7/20 21.8 28.0 A 1450 II (A-C) 4NG 3.0 ALT 7/20 33.7 44.2 A 7/22 1015 I (F-I) 4BB 0.3 ALT 7/20 26.4 41.6 AA 1045 I (A-D) | 4NB 0.3 ALT 7/20 31.6 40.0 AA 1135 II (A-C) 4NG 0.3 ALT 7/20 24.6 35.2 A 1405 I (F-I) 4BB 0.03 ALT 7/20 30.6 35.0 A 7/26 0910 I (F-I) 4BB 3.0 ALT 7/20 22.8 25.0 O? 1000 I (A-D) 4NB 3.0 ALT 7/20 51.0 46.4 O? 1135 II (A-C) j 4NG 3.0 ALT 7/20 28.9 40.8 A 2035 I (F-I) j 4BB 3.0 ALT 7/20 34.6 36.8 O 2150 II (A-C) 4NG 3.0 ALT 7/20 31.0 36.8 A 7/27 0925 I (F-I) 4BB 0.3 SK 7/26 23.6 27.6 O 1.0 SK 7/26 — 22.6 O 3.0 SK 7/26 — 31.8 A 1045 I (A-D) 4NB 0.3 SK 7/26 45.6 46.2 O 1.0 SK 7/26 — 56.0 A 3.0 SK 7/26 — 57.2 AA 1500 II (A-C) 4NG 1.0 SK 7/27 46.8 53.2 O 3.0 SK 7/27 — 52.2 A * RR, strong repulsion; R, weak repulsion; O, no apparent response except sensing; A, weak attraction; AA, strong attraction. became excited and engaged in the typical hunt- ing response. It is assumed that they were stim- ulated by odors emanating from the fish. In view of the delayed response, it seems unlikely that the blinded sharks were attracted by vibra- tions or sounds that may have been made by the fish although this possibility could not be ruled out in these rather crude experiments. In another series of experiments at Eniwetok in 1959, an uninjured fish was held in a bucket of saltwater for 15 to 20 min prior to an ex- periment. The bucket was slapped or agitated Fig. 9. Diagram of apparatus used at Eniwetok Ma- rine Biological Laboratory for investigating response of sharks to living fish. 164 PACIFIC SCIENCE, Vol. XVII, April 1963 to keep the fish in an excited state. During con- trol conditions, water from another bucket of sea water was siphoned into the test area of the compartment containing the four blinded black- tip sharks. The siphon was then switched to the bucket of water which either contained the fish or from which the fish had been removed. The usual observations were made during a series of control and test periods. The results are included in Table 6. The blinded blacktip sharks showed an attraction re- sponse in most of the experiments with the "grouper water” and in the one experiment with "eel water” ( Gymnothorax ). It was concluded that the water in which these fish had been con- fined contained some substance which was at- tractive to the sharks. The blinded blacktips did not show a noticeable attraction response to "blacktip water.” During the winter of 1959-60 similar ex- TABLE 8 Response of Normal Blacktip Sharks (nb) and Blinded Blacktip Sharks (bb) to Human Sweat at Eniwetok Laboratory, 1959 DATE (1959) TIME TANK SHARKS SWEAT (ML) DONOR, DATE ACTIVITY INDEX RESPONSE* Controls Test 8/8 1905 I (A-D) 3NB 0.3 ALT 8/8t 97.7 65.9 A? 1930 I (A-D) 3NB 0.3 ALT 8/8+ — 73.7 R 2115 I (F-I) 3BB 0.3 ALT 8/8+ 67.6 52.7 R 8/9 1800 I (F-I) 2BB 3.0 ALT 8/8+ 29.4 14.1 RR 1945 I (A-D) 2NB i 3.0 ALT 8/8+ 54.4 41.7 | R 8/10 0845 I (A-D) 2NB 3.0 ALT 8/8+ 54.0 52.2 O 0925 I (F-I) 2BB 3.0 ALT 8/8+ 45.0 26.8 R 8/11 2322 I (F-I) 2BB 1 2-3 TAP 8/11 39.4 25.0 O? 8/12 0035 I (A-D) 2NB 1 2-3 TAP 8/11 70.0 j 58.8 O? 1910 I (F-I) 2BB 3.0 ALT 8/8+ 29.5 9.3 RR 2010 I (A-D) 2NB 3.0 ALT 8/8+ 73.6 61.9 R 8/13 1553 I (F-I) 2BB 3.0 ALT 8/8+ 33.0 5.9 RR 1710 ! I (A-D) 2NB 3.0 ALT 8/8+ 68.2 58.5 R 8/14 1905 I (F-I) 2BB 3.0 ALT 8/14 54.0 | 35.3 ! RR I 2025 I (A-D) 2NB 3.0 ALT 8/14 67.0 57.7 R 8/16 1900 I (A-D) 2NB 0.0 6 ALT 8/14 80.8 77.8 O 1930 I (A-D) 2NB 3.0 ALT 8/14 — 74.2 R? 2010 I (F-I) 2BB 3.0 ALT 8/14 9.2 5.7 R 8/18 1355 ! I (A-D) | 2NB 50.0 ALT 8/8+ 34.8 17.5 RR 8/24 2010 | II (G-I) 4BB 10.0 ALT 8/8+ 85.7 56.3 RR 8/25 1515 j II (G-I) IBB 10.0 ALT 8/8+ 24.2 1.9 RR 8/31 1930 I (F-I) 2BB 3.0 ALT 8/31 26.9 12.2 RR 9/1 2155 I (F-I) 2BB 3.0 ALT 8/31 37.0 32.0 R 9/3 1450 j I (F-I) 3BB 3.0 ALT 8/8+ 38.6 13.5 RR 1900 I (F-I) 3BB 3.0 ALT 8/8+ 53.0 36.0 R 2145 I (F-I) 3BB 3.0 ALT 8/8+ 49.6 35.2 R 9/4 1930 ! I (F-I) 2BB 3.0 ALT 9/4 j 37.0 24.2 R 2145 I (F-I) 2BB 3.0 ALT 9/4 j 38.0 22.6 R 9/5 1400 I (F-I) 2BB 3.0 ALT 8/8+ | 21.6 12.2 R * RR, strong repulsion; R, weak repulsion; O, no apparent response except sensing; A, attraction. f Sweat collected by sponge. weak attraction; A A, strong Olfaction and Sharks — Tester 165 TABLE 9 Response of Normal Blacktip Sharks (nb), Blinded Blacktip Sharks (bb), and Normal Sharks (ng) to Human Sweat at Eniwetok Laboratory, I960 DATE (1960) TIME TANK SHARKS SWEAT (ML) DONOR, DATE ACTIVITY INDEX RESPONSE * Controls Test 6/30 1250 I (A-D) 2NG 3.0 ALT 6/29 14.6 13.2 R 1340 I (A-D) 2NG 3.0 ALT 6/29 12.2 8.0 R 1350 I (A-D) 2NG 5.0 ALT 6/30 — 7.4 R 1530 I (A-D) 2NG 3.0 ALT 6/29 13.8 10.0 R 1555 I (A-D) 2NG 6.0 ALT 6/29 8.7 8.6 o 7/1 1007 I (A-D) 2NG 3.0 ALT 6/29 17.0 18.0 j O 1035 I (A-D) 2NG 6.0 ALT 6/29 17.8 17.6 O 1105 I (F-J) 2NG 3.0 ALT 7/1 13.2 7.0 R 1130 I (F-J) 2NG 6.0 ALT 7/1 7.4 8.3 o 1400 I (A-D) 2NG 6.0 ALT 7/1 13.6 8.8 R 1422 I (F-I) 2NG 6.0 ALT 7/1 14.0 12.2 R 1942 I (A-D) 2NG 6.0 ALT 7/1 22.4 17.8 R 7/2 0903 I (A-D) 2NG 6.0 ALT 7/1 18.4 18.6 O 1140 I (F-I) 2NB 3.0 ALT 7/2 50.5 92.8 A7 1430 I (F-I) 2NB 6.0 ALT 7/2 68.5 71.1 O 1445 I (A-D) 2NG 9.0 ALT 7/2 18.4 15.6 R 7/6 1300 I (A-D) 2NB 5.0 SK-ALT 7/2 52.2 23.2 RR 1400 I (F-I) 2BB 6.0 JK 7/6 31.4 31.0 R? 1430 I (A-D) 2NB 6.0 SK 7/6 47.0 31.4 R 7/7 | 0905 I (A-D) 2NB { 6.0 JK 7/6 40.8 33.8 R? 7/9 1010 I (A-D) 2NB 3.0 SK 7/6 ; 20.0 7.0 RR 1500 I (A-D) 2NB 3.0 JK 7/6 22.6 5.4 RR 7/11 0839 I (A-D) 2NB 3.0 SK 7/6 33.0 29.0 R 1549 I (A-D) 2NB ! 3.0 JK 7/6 31.4 10.4 RR 7/19 1055 I (F-I) 2BB 3.0 SK 7/6 24.6 14.2 RR 1120 I (A-D) 2NB 3.0 SK 7/6 31.4 31.6 R? 1145 II (A-C) 2NG 3.0 SK 7/6 7.8 10.0 O 7/20 1505 ! I (F-I) 4BB i 3'° JK 7/6 25.4 18.6 RR | 1530 | I (A-D) 4NB 3.0 JK 7/6 15.4 15.2 R 7/23 | 1243 I (F-I) 4BB 3.0 SK 7/11 20.4 24.8 O 7/24 ! 1450 ! I (F-I) | 4BB 3.0 ALT 7/23 23.8 22.6 R | 1655 ; i (A-D) ! 4NB 3.0 ALT 7/23 28.8 13.2 RR 7/25 i 0940 I (F-I) 4BB 3.0 ALT 7/23 25.2 18.4 R | 1020 ! I (A-D) 4NB ; 3.0 ALT 7/23 33.8 25.4 i R 8/25 1230 I (F-I) 4BB 3.0 ALT 8/25 28.0 26.2 R j 1300 1 I (A-D) 4NB 3.0 ALT 8/25 34.0 32.2 ! R * RR, strong repulsion; R, weak repulsion; O, no apparent response except sensing; A, weak attraction; AA, strong attraction. periments were conducted at the Hawaii labora- tory on the population of sharks in Pond 5, which at that time consisted of a tiger, a ham- merhead, and a grey shark. The method of "point introduction" was used. During tests, three or four uninjured fish ( Tilapia mosam- bique ) were confined in the 5 -gal funnel of sea water on top of the tower. During tests, the water from the funnel containing the fish was introduced through the hose. This produced a strong hunting response in the hammerhead. The response of the tiger shark and the grey ACTIVITY INDEX 166 PACIFIC SCIENCE, Vol. XVII, April 1963 STARVED BLACKTIPS QUIESCENT A6ITATE0 GROUPER GROUPER WATER WATER I I I i ! L I I 1—1 L 0900 0912 TIME - CONTROL AND TEST PERIODS Fig. 10. Activity index during successive 2-min periods, illustrating response of starved grey sharks to "'quiescent grouper water” and of starved blackdp sharks to "agitated grouper water.” shark was positive but less intense. "Tilapia water” was the attractant used in the experi- ment involving human sweat which was de- scribed earlier. Several, more definitive experiments were conducted at Eniwetok during the summer of I960 using an experimental arrangement illus- trated in Figure 9. Living fish were placed in a 2 5 -gal plastic container into which sea water was flowing. The sea water could be siphoned from the container into the test area of either Tank 1 or Tank 2, or it could be spilled to the ground. The compartment of Tank 1 contained four starved blacktip sharks and that of Tank 2 held four starved grey sharks, all with normal vision. One observer, manipulating the siphons and living "prey” fish, was concealed from the sharks by a blind; a second observer, recording data on shark behavior, was concealed in the observation booth. Only two of several experiments will be de- scribed in detail. In one (Fig. 10), four group- ers ( Epinephelus merra ) had been placed in the container the previous night, with the water siphoning into the blacktip compartment. In the morning, following a series of control peri- ods which started at 0830 (timed on a 24-hr clock ) , the "quiescent grouper water” was siph- oned into the grey shark compartment. In the first test period (at 0842) the grey sharks ACTIVITY INDEX Olfaction and Sharks — Tester 167 showed obvious awareness and mild attraction, with one shark biting the siphon tube, but the response quickly subsided. Siphoning into the blacktip compartment was then resumed (at 0900) with no noticeable response from the sharks; this was anticipated as the water had been siphoning into this compartment all night. The groupers were then frightened and excited by threatening them with a moving stick. There followed a noticeable hunting reaction by the blacktips (at 0912), stimulated by the "agi- tated” grouper water. Similar results were ob- tained with both starved blacktip and grey sharks using quiescent and agitated surgeon fish and mullet in place of the groupers. In the second experiment to be described in detail ( Fig. 11), the grey sharks showed a nor- mal behavior pattern during control periods which started at 1600. When "quiescent grouper water” was introduced (at 16 12) they responded, as above, with a mild hunting reaction; one bit the tube and others milled near it. In the mean- time, a small grouper had been removed from the aquarium and held in a dip net in air for 30 min, at which time it was still alive and ap- parently undamaged. Wth the water from the quiescent groupers still flowing into the grey shark compartment, the "distressed” grouper was quietly lowered into the container (at 1622) by a string tied around its body; it was removed after the third test period (1628). The sharks displayed a violent hunting reaction with cir- cling and biting of the tube. The "quiescent grouper water” continued to siphon into the compartment for about 1 hr, at which time (1730) the sharks exhibited normal activity during control conditions. The small grouper which had been returned to the dip net and was still alive after 74 min, was again lowered into the container (at 1740) for three test periods. Fig. 11. Activity index during successive 2-min periods, illustrating response of starved grey sharks to "quiescent grouper water,” to "distressed grouper water,” and to "dead grouper water.” 168 PACIFIC SCIENCE, Vol. XVII, April 1963 Fig. 12. Activity index during successive 2-min periods, illustrating response of starved blacktip sharks to "quiescent grouper water,” to "distressed grouper water,” and to "dead grouper water.” The sharks again responded with a violent hunt- ing response and tube biting. The procedure was repeated after an additional 4 hrs (at 2200) during which time the quiescent grouper water had been flowing into the compartment and the sharks were responding normally. The small grouper in the dip net, however, had died. After the dead fish was lowered into the container (at 2210), there was a spectacular hunting re- action by the sharks. Only the string was re- covered from the container at the end of the third period: the small dead grouper had been swallowed by one of the larger "quiescent” groupers. The water in the container was clear indicating that the fish had been engulfed whole without the escape of body juices. The above experiment was repeated on the starved blacktip sharks, using the same four "quiescent” groupers and another "distressed” grouper. In this case the "distressed” grouper was not eaten after it had died. The results (Fig. 12) were almost identical to those obtained with the starved grey sharks. A similar test using four "quiescent” and one "distressed” mullet gave similar results (Fig. 13). The sharks even responded to a small "distressed” blacktip shark which was held in a dip net for a few minutes and then, still alive, was lowered into the con- tainer, in this case in the absence of any "quies- cent” fish. Discussion These experiments show that "quiescent” prey give off an odor which can be detected by sharks when it is first introduced into their environ- ment but to which they soon become habituated. There is still the question, of course, as to whether the "quiescent” fish were still under stress because of the artificial environment of the plastic container. Regardless of this, the experiments demonstrate that when the prey becomes frightened and excited it gives off an additional or a new odor which again stimulates the habituated sharks, provoking the typical Olfaction and Sharks — Tester 169 Fig. 13. Activity index during successive 2-min periods, illustrating response of starved blacktip sharks to "quiescent mullet water,’’ "excited mullet water,” "distressed mullet water,” and "dead mullet water.” hunting response. Moreover it seems that shark activity, and thus presumably the amount of odorous material released, increases with in- creased agitation of the prey. It seems unlikely that the odorous material is associated with body juices released by direct injury to the prey on the part of the observer. The fish used were healthy aquarium specimens which in some cases, e.g., groupers, were used over and over again and yet suffered no obvious ill-effects from being repeatedly "agitated.” It is unlikely they would be damaged by rubbing against the sides of the smooth plastic container. They could, of course, rub against each other when excited. This may possibly have removed part of their mucous coating and enabled body juices to escape through the skin. That the results were not induced by the arti- ficial environment of the shark tanks was dem- onstrated in follow-up experiments with sharks in the natural environment of Eniwetok lagoon by Hobson (1963). Water siphoned into the lagoon from a plastic container in which large, living, agitated but apparently undamaged groupers had been placed, attracted both white- tip ( Triaenodon obesus ) and grey (C. menisor- rah ) sharks. They detected the "grouper water” from a distance and followed its path upstream to the source — a concealed plastic tube. If the substance which attracts the sharks is released by some subtle damage to the skin of the prey it might be similar to that demonstrated by von Frisch (1941) in the injured skin of the minnow ( Phoxinus laevis ) . As with von Frisch’s material, identified as a purine- or pterin-like substance by Hiittel (1941), it might produce an alarm reaction among the prey but still be attractive to the sharks. On the other 170 PACIFIC SCIENCE, VoL XVII, April 1963 hand it is tempting to postulate that the sub- stance is some metabolite which is released from gill, vent, or skin by excitement rather than by injury of the prey. Whatever may be the source and nature of the attractant, we have presented evidence that olfaction is involved in the predation of sharks on normal, healthy fish. It is suggested that in the natural environment, fish give off odors to which the sharks are conditioned. It is further suggested that when the fish become frightened or excited, and certainly if they rub against each other or against a coral head, they give off ad- ditional or new odors which stimulate the hunt- ing response in sharks. This hypothesis is con- sistent not only with our experimental data but also with our observations of the behavior of the sharks in their natural environment. For the most part they display a complete disregard for the myriad of normal, healthy fish which surround them. However they are able to track down and converge on a distressed fish (such as a live fish suspended from a hook through the jawbone but otherwise uninjured) with un- canny speed and accuracy. REFERENCES Bigelow, H., and W. C. Schroeder. 1948. Sharks. In, Fishes of the western north At- lantic. Sears Found. Mar. Res. Mem. 1, Pt. 1:59-546. Brett, J. R., and D. McKinnon. 1954. Some aspects of olfactory perception in migrating adult coho and spring salmon. J. Fish. Res. Bd, Canada 11 ( 3 ): 310-318. Eibl-Eibesfeldt, I., and H. Hass. 1959. Erfah- rungen mit Haien. Zeit. Tierpsych. 16(6): 733-746. Gilbert, P. W., and F. G. Wood, Jr. 1957. Method of anesthetizing large sharks and rays safely and rapidly. Science 126:212-213. Hobson, E. S., Jr. 1963. Feeding behavior in three species of sharks. Pacif. Sci. 17(2). Huttel, Rudolf. 1941. Die chemische Unter- suchung des Schreckstoffes aus Elritzenhaut. Naturwiss. 29:333-334. Idler, D. R., U. H. M. Fagerlund, and Helen Mayoh. 1956. Olfactory perception in mi- grating salmon, 1. L-serine, a salmon repellent in mammalian skin. J. Gen. Physiol. 39(6): 889-892. Mishkin, M., R. D. Gunkel, and H. E. Ros- vold. 1959. Contact occluders: a method for restricting vision in animals. Science 129: 1220-1221. Parker, G. H. 1910. Olfactory reactions in fishes. J. Exper. Zool. 8 : 535—542. 1914. The directive influence of the sense of smell in the dogfish. Bull. U. S. Bur. Fish. 33:61-68. and R. E. Sheldon. 1913. The sense of smell in fishes. Bull. U. S. Bur. Fish. 32:33— 46. Sheldon, R. E. 1911. The sense of smell in selachians. J. Exper. Zool. 10:51-62. Springer, Stewart. 1955. Laboratory experi- ments with shark repellents. Proc. Gulf, Car- ibbean Fisheries Inst. 1954: 159-163. Steinberg, M. A. I960. Report on progress on contract NR 104-525, chemical analysis of shark repelling substances. (Mimeo.) — 1961. Final report on contract NR 104- 525, chemical analysis of shark repelling sub- stances. (Mimeo.) von Frisch, K. 1941. Ueber eine Schreckstoff der Fischhaut und seine biologische Bedeu- tung. Zeit. vergl. Physiol. 29:46-145. Whitely, G. P. 1940. The Fishes of Australia: Part 1, Sharks. Royal Zoological Society of New South Wales. Wright, B. S. 1948. Releasers of attack be- havior pattern in shark and barracuda. J. Wildlife Mgt. 12(2): 117-123. Feeding Behavior in Three Species of Sharks1 Edmund S. Hobson2 This report concerns a study of the feeding behavior in three species of sharks: Car char - hinus menisorrah Muller and Henle, the grey shark (Fig. 1), Car char hinus melanopterus Quoy and Gaimard, the blacktip shark ( Fig. 2 ) , both of the family Carcharhinidae; and Triae- nodon ohesus Ruppell, the whitetip shark (Fig. 3 ) , of the family Triakidae. The study was con- ducted in the lagoon at Eniwetok Atoll, Mar- shall Islands, during the summers of 1939 and I960. It was a segment of a broad program of investigation of shark behavior in which labora- tory and field work were coordinated whenever possible. The overall program, conducted at both the Eniwetok Marine Biological Labora- tory and the Hawaii Marine Laboratory, Coco- nut Island, Hawaii, was under the direction of Dr. Albert L. Tester, with financial support from the Office of Naval Research (Contract Nonr 2756(00), Project NR 104503). Observations of sharks in their natural en- vironment have been the basis for most of the shark literature which is available today and yet comparatively little has been offered toward a realistic understanding of shark behavior. Most of this material has been written for popu- lar consumption and is therefore oriented to- ward the sensational rather than the scientific. We do find scientifically oriented accounts in the literature (as for example, Eibl-Eibesfeldt and Hass, 1959; Limbaugh, 1958; and Wright, 1948) but these observations are limited largely to incidental encounters. A critical study of shark behavior, undertaken with planned experiments in the sharks’ natural 1 Prepared as partial requirement for the Master of Science Degree, University of Hawaii. Contribution No. 180, Hawaii Marine Laboratory, University of Hawaii, Honolulu, Hawaii. Manuscript received No- vember 23, 1961. 2 Department of Zoology, University of Hawaii. Presently at the Department of Zoology, University of California, Los Angeles 24, California. environment — the aim of this investigation — has been almost completely neglected. In this study, a comparison is made of the feeding behavior of the three species. Behavior is further related to habitat and to distribution within the lagoon. Experiments designed to il- lustrate the roles played by the major sensory modalities are presented. This last portion of the investigation, dealing primarily with grey sharks, includes a consideration of the specific stimuli involved in releasing feeding behavior. Consideration is also given to food preferences and to factors which may possibly inhibit feed- ing. GENERAL DESCRIPTION OF STUDY AREAS The study was centered about two primary locations: (1) the lee of Engebi Island during 1959, and (2) the lee of Bogen Island, adjacent to Deep Channel, during I960. The Engebi Island site was in 1 5 ft of water over a relatively shallow sand and coral rubble flat extending out from shore for approximately 400 yd before dropping suddenly into the deeper regions of the lagoon. Currents are weak in this area and the water is generally turbid, with underwater visibility commonly less than 20 ft. At the Bogen Island site strong tidal currents are pres- ent and underwater visibility often exceeds 100 ft. The edge of Deep Channel at this point drops abruptly from a depth of approximately 10 ft at the rim to 110 ft at the bottom. At these locations, observations were made from the following vantage points: 1. An underwater chamber was fitted to a vessel moored in the lagoon. This chamber was a metal cylinder, 14 ft long and 2V2 ft in di- ameter, open at the top and closed at the bot- tom, with viewing ports on three sides near the bottom. An observer in the chamber was situ- ated approximately 8 ft below the surface. 171 172 PACIFIC SCIENCE, Vol XVII, April 1963 Fig. 1. The grey shark, Carcharhinus menisorrah. (Photo by E. Hobson.) 2. A wire cage was suspended from a raft anchored on the slope at the edge of Deep Channel. An observer in this cage was located immediately below the surface. 3. A canvas screen, anchored on the bottom, concealed an observer lying prone on the bottom wearing SCUBA equipment. 4. A 16-ft glass-bottom boat, having the added advantage of mobility, provided for ob- servations of activity directly below. 5. In many cases, the observations involved incidental encounters with sharks during the course of such routine underwater activity as installing apparatus, etc. Experimental procedures and apparatus will be further described as they pertain to the re- port. SPECIES STUDIED Identification of the sharks is based on Schultz et al (1953). All three species, Carcharhinus menisorrah , C. melanopterus , and Triaenodon obesus, reportedly have a wide Indo-Pacific dis- tribution. They are reported from the Red Sea as well as the Maidive Islands by Klausewitz (1958, 1959). Although positive identification of the two carcharinids awaits a revision of the family on a world-wide basis, all three species appear to be prominent in the shark popula- tions of most Pacific atolls. Harry (1953) re- ports them from the Tuamotus, as does Randall (1955) from the Gilbert group. The author found them abundant at Palmyra and they were the only species of sharks consistently seen in the lagoon at Eniwetok during the present in- vestigation. GENERAL OBSERVATIONS OF BEHAVIOR Often when we were engaged in various types of activity in the lagoon the resulting commotion, which commonly involved splash- ing on the surface or striking metal tools on hard objects underwater, was immediately fol- lowed by the rapid approach of an obviously alerted blacktip or grey shark. In spite of this initial attraction to many stim- uli, both species exhibited varying degrees of caution when encountering unfamiliar situa- tions. This was particularly apparent in the blacktip, a species which seemed especially sen- sitive to potential danger. When work was be- gun at Engebi, blacktips often circled at the limit of visibility in the baited area for as long as 2 or 3 hr before approaching the bait. Then, when an approach was made, it seemed to be done reluctantly and was often cut short by some stimulus, unnoted by the observer, which startled the shark as it neared the bait and caused it to swim rapidly away. Usually the shark repeated the approach, but in these early tests it was not until several sharks had become active in the area that a blacktip finally took the bait. We subsequently found that we were dealing with the same blacktips day after day at Engebi and that as the work progressed, their initial caution steadily declined. This was pre- sumably due to a growing familiarity with the situation. Nevertheless, a month of testing passed before the blacktips took the bait with- out hesitation. Once released, the attack of the blacktip was very fast and aggressive. The grey was a notably bolder species. Al- Feeding Behavior of Sharks— HOBSON 173 though a black dp often approached rapidly in response to splashing at the surface, it then gen- erally reversed direction and retreated just as rapidly if the investigation of the splashing brought it into an encounter with a human. In the same situation, the grey ordinarily reacted to the encounter with a human by continuing the approach to approximately 5 yd from the human, at which point it would veer aside and circle with an apparent cautious interest. If no further stimuli were introduced, the grey moved on. Black tips and greys showed a marked increase in excitement when feeding in numbers. This phenomenon, generally referred to in extreme cases as the feeding frenzy, has been observed in many species of sharks. In the blacktips and greys, the presence of more than one shark ap- peared to lower the threshold for the release of feeding behavior. Whitetips did not show this group effect. Even when feeding in numbers, members of this species responded individually and without a notable increase in excitement. The whitetip seemed to be relatively unresponsive to many of the stimulus situations which elicited a sharp reaction in the grey and blacktip. There was little overt response seen in this shark when en- countering a human in the water for the first time. In this situation we did not see the curi- osity frequently exhibited by the greys, nor the start and rapid flight usually exhibited by the blacktip. On several occasions whitetips appeared and took fish from the spears of divers before the fish could be removed from the water. Even in these instances, the slow, deliberate actions of this shark did not give the impression of being a threat to the diver. This impression may have been an illusion based on the sluggish behavior of the animal. On one such occasion a 6-ft whitetip bit the fish in half and then made a slow pass at the diver holding the other end of the spear. This appeared to be a warning pass at a potential competitor rather than an active- attack. Such warning passes were noted on sev- eral occasions directed at grouper or snappers which approached a bait that was under attack by a whitetip. Similar behavior was noted in- greys and blacktips. Tester (personal communi- cation) observed a small blacktip pursue a grouper of comparable size from a bait which had been placed on the bottom in shallow water near Aniyaani Island. In this case, the blacktip showed considerably more than a warning pass, as it actively pursued the grouper among a con- centration of small coral heads. We did not ob- serve these sharks exhibiting aggressive behav- ior toward members of their own species. One observation made on numerous occasions might at first glance seem to oppose this view. When a bait which was too large to be immediately swallowed was presented to a group of feeding greys or blacktips, the shark which succeeded in taking the bait would invariably swim rapidly away from the area, shaking its head vigorously from side to side presumably in attempts at cutting up and swallowing the bait. As it fled, Fig. 2. The blacktip shark, Carcharhmus melanopterm. (Photo by E. Hobson.) 174 PACIFIC SCIENCE, Vol. XVII, April 1963 Fig. 3- The whitetip shark, Triaenodon obesus. (Photo by E. Hobson.) the other sharks always followed in close pur- suit. Although the flight of prey was commonly noted to release attack in the grey, the aggres- sive behavior of these other sharks in this case was believed to have been directed against the bait rather than the fleeing shark. There was never any mistaking an alerted grey or blacktip from one engaged in normal patrolling activity. Thus we had no difficulty distinguishing a grey or blacktip entering the test area in response to a stimulus situation we had presented from one incidentally passing through the area. The movements of the alerted greys and blacktips were markedly accelerated and the grey in particular seemed tense and highly responsive to subsequent stimulation. Movements of the grey immediately before at- tack were markedly abrupt; its body often ap- peared stiff, with back slightly arched and head extending straight out and slightly upward. The pectoral fins were characteristically pointed no- ticeably downward. Attack was prefaced in many cases by such anticipatory movements as a lateral shaking of the head (noted also by Eibl-Eibesfeldt and Hass, 1959) and a move- ment of the jaws as in biting (Hobson et al, 1961). Although it was not always so easy to make this distinction with the seemingly unrespon- sive whitetip, these sharks were undoubtedly more responsive than they appeared to be. It is likely that their reactions were simply more subtle than those of the more excitable greys and blacktips. Whitetips usually appeared in an area shortly after divers had undertaken various types of underwater activity. However, after their appearance the whitetips would swim about without any apparent interest in the pro- ceedings. Nevertheless, the consistency of these appearances indicated that the sharks sensed ac- tivity and were interested. DISTRIBUTION OF SPECIES WITHIN LAGOON Blacktips were the most commonly observed shark over the sand and coral rubble flats lying under approximately 1-40 ft of water at the perimeter of the lagoon. These flats extend out from shore for distances ranging from approxi- mately 50 yd to several miles before the bottom falls off sharply into the depths. Coral growth in this area is generally restricted to large iso- lated heads which, in many cases, reach the surface of the water at low tide. Although black- tips exceeding 6 ft in length were seen, speci- mens of more than 4 ft were not common. Small blacktips were very common on the seaward reef flats when the sea covered these flats at high tide. The seaward flats were largely exposed at low tide. The whitetip also frequented the shallow waters of the lagoon, although unlike the black- tip, which foraged widely over the flats, the whitetip centered its activity among the coral heads and about the coral-rock ledges which border the seaward passages. Unlike the other two species, the whitetip was commonly seen resting motionless on the bottom, often under Feeding Behavior of Sharks — Hobson 175 ledges and in caves. The whitetip was common to a length of 6 ft, with individuals of 7 ft seen on occasion. We seldom encountered the blacktip when we moved down the slopes from the shallow flats into the deeper waters of the lagoon or seaward passages, but here we found the grey shark in abundance. A census of the shark popu- lation of the entire lagoon would probably show the grey shark to be the most numerous. This shark was commonly seen up to 7 ft in length. All three species were generally observed swimming close to the bottom unless drawn to- ward the surface to feed. Figure 5 shows shark sightings by species during a period of 30 days in the vicinity of Fig. 4. The test area off Bogen Island, showing the relative position of Deep Channel (top), raft (at edge of channel), and the barge. The island is out of the picture to the left. (Photo by R. A. Boolootian.) 176 PACIFIC SCIENCE, VoL XVII, April 1963 Fig. 5. Area in the vicinity of Parry Island, showing the sharks sighted during the period 26 July to 26 August, I960, and indicating species involved and depth of water where sighting occurred. Parry Island and the depth of water in which these sightings occurred. Two basic rules were observed in making this count: (1) no more than one count was made in any one area on any one day; (2) if there was any question of whether or not a particular shark had already been counted on any one day, then this shark was disregarded. This survey was not intended to show shark abundance, but rather to illus- trate the areas and depths in which each of the three species was normally seen. The distributional picture which emerges is consistent with the observations made through- out the program. For example, during 2 months of work in 15 ft of water off Engebi, only two grey sharks, both approximately 2 ft in length, were seen. On the other hand, the experiments involved many blacktips and whitetips. In con- trast, when experiments were conducted along the edge of the dropoff into Deep Channel, both whitetips and greys were in abundance, while blacktips were only occasionally seen. Feeding Behavior of Sharks — Hobson 177 In December 1959, while fishing outside the seaward reef slope at Palmyra, we saw only grey sharks, finding these in considerable abundance. At the same time, only blacktips were common on the reef flat a hundred yards away. We saw comparatively few whitetips at Palmyra, these in shallow water over the reef. Klausewitz (1959) found these same species occupying similar habitats in the Red Sea. Exceptions to this general distribution pat- tern were noted. Large grey sharks appeared at poison stations in water scarcely 6 ft deep, while whitetips were seen swimming over open bottom, far from the nearest coral head or rock. Furthermore, Strasburg (1958) reports the catching of two blacktips at sea in the Mar- quesas. BEHAVIOR AND HABITAT The sluggish behavior noted in the whitetip is consistent with a life in and about the caverns and crevices of the coral reef. This species was noted as being clumsy and ineffective in at- tempts at taking baits which were suspended in midwater. However, this same shark was re- markably effective in tracking down and cap- turing prey which had taken shelter deep in one of the many holes or crevices typical of a coral reef, thus making available to it prey which are beyond the reach of both greys and black- tips. Considerable time was spent placing wounded fish far back into small holes in the reef and then watching as a whitetip appeared, nosed about tentatively for the correct hole, and then swam in and captured its prey. Large whitetips were seen disappearing into small holes from which they presently emerged, al- ways head first. The experiment described be- low involved whitetips and greys and illustrates the division of the food source between these two species. Experiment 1 Three small wounded fish, each essentially identical, were presented simultaneously at three positions below the raft at the edge of Deep Channel, where water depth was approximately 35 ft: (1) Suspended mid-way between surface and bottom; (2) on the bottom in an exposed position; (3) concealed in a hole beneath a large rock on the bottom. Bait 2 was lowered to the bottom at the end of a weighted line. Baits 1 and 3 were both secured to a single line which ran from the raft, down under one side of the rock, through the hole, out the other side and back up to the raft. By alternately hauling in one end of the line or the other, both baits could be simultaneously hauled aboard the raft or lowered into position. The experiment was repeated 16 times when both greys and whitetips were in the vicinity. Although the two species seemed equally adept at taking the exposed bait on the bottom, the suspended bait in all but one instance was taken by a grey, while the whitetips completely mo- nopolized the bait concealed in the hole. When grey sharks encountered humans in shallow water they often started and fled in much the same manner described for the black- tip, rather than exhibiting their usual relatively bold inquisitive approach. Possibly this apparent change in behavior was a result of their being out of their usual habitat. This observation might offer an insight into the characteristically timid behavior of the blacktip. Perhaps this ap- parent timidity is an adaption of these relatively large animals, which must remain in motion, to a shallow water habitat. The shark, unable to take shelter and without room to maneuver, may find flight the alternative. In the blacktips, this characteristic timidity is more apparent in the larger individuals. The suggestion that this behavior might have some survival value im- plies the existence of a natural predator. The only evidence we found of such a predator was the presence of an 18-inch blacktip in the stomach of an 80-lb grouper. DETECTION AND CAPTURE OF FOOD Sharks are well known to feed avidly on dead fish, meat, and many other food materials dumped as garbage or used as bait. It is also well known that they will consume living fish impaled on a hook or spear. These, however, are unnatural situations. Undoubtedly, under nat- ural conditions these sharks will feed on such prey as may have been killed or weakened by disease or injury. The sharks in the Eniwetok lagoon, as elsewhere, are highly responsive to stimulus situations which suggest injured and/or distressed, as well as dead or moribund 178 PACIFIC SCIENCE, Vol XVII, April 1963 prey. Nevertheless, it seems unlikely that this source of food alone is sufficient to support such a large shark population. It seems probable, then, that they act not only as opportunistic scavengers, but also as predators on healthy free-moving animals. Eibl-Eibesfeldt and Hass (1959) report observing both C. menisorrah and C. melanopterus in the Indian Ocean herd- ing schools of mullet against the shoreline and actively feeding on these fishes. Similarly, Stras- burg (1958) observed pelagic whitetips (Pterol- amiops longimanus ) herding squid under a night light. Strasburg also cites other evidence of pelagic sharks apparently capturing what would appear to be highly motile elusive prey. We observed no such activity among the three species in the lagoon at Eniwetok. These sharks seemed oblivious to the presence of the numer- ous reef fishes which were continually present during the shark’s patrolling activities. This be- havior might be expected, however, as only un- der such conditions would the reef fishes allow sharks to move in amongst them without ex- hibiting immediate alarm and taking shelter. This same apparent oblivion to what appears to be potential prey is also standard behavior seen in many other reef predators, for example the groupers, snappers, and moray eels. It seems likely that this behavior on the part of the predators is advantageous in allowing them to catch their prey unawares with a frequency which, while sufficient to maintain life, does not destroy the illusion of their non-aggressive- ness. It is also probable that the threshold for the release of feeding on healthy prey fluctuates with the relative availability of more readily obtainable food items, such as dead or disabled fish. The present study is confined to feeding be- havior with respect to dead, damaged, and dis- tressed prey. An effort is made to determine which of several sensory modalities are involved and which are dominant in the sequence of events between initial stimulation and the act of consuming the prey. RESPONSE TO OLFACTORY STIMULI A number of experiments were conducted which elucidated the role of olfaction in detect- ing dead and living prey. Only three (II, III, and IV) will be reported in detail. Experiment II This experiment was designed to study the response of these sharks to an uninjured fish struggling on a line, which might thus produce visual, mechanical, auditory, and perhaps olfac- tory cues. The glass-bottom boat was anchored in 40 ft of water on the steep slope of Deep Channel where the current ran in one of two directions, depending on the tide. Ten trials were con- ducted, each at a time of strong flood or ebb current when visibility was good. As a precau- tion against the sharks becoming conditioned to feeding at this location, the trials were spaced over a period of several weeks, with only one trial on any one day. Each trial involved one fish, either a grouper ( Serranidae ) , snapper (Lutjanidae) , or mullet (Mugillidae) , 2 to 3 lb in weight, secured to a line by a piece of soft, light cord which passed through the mem- brane behind the maxillary. The fish had been caught by barbless hook and kept in tanks at the laboratory until needed; they appeared to be healthy and uninjured. Before the fish was lowered to a point 5-10 ft above the bottom, a 5 -min observation period was conducted to insure that no sharks were in the area. If sharks were seen the test was de- layed until at least 5 min after they had disap- peared. Observations included species of shark, the time each took to locate and take the bait, its general behavior, and particularly the nature and direction of its approach. It was presumed that if the sharks approached directly and con- sistently from downstream the attracting stim- ulus had been initially olfactory, inasmuch as only the olfactory stimulus was affected by the current. If the approach was from random di- rections, then other sensory cues, such as visual, were likely to be involved in the initial attrac- tion. The results are summarized in Table 1. In 9 of the 10 trials the sharks appeared from down- stream, swimming rapidly and directly toward the bait, thus indicating they were responding to olfactory cues carried by the current. The Feeding Behavior of Sharks— -HOBSON 179 TABLE 1 Response of Sharks to Fish Which Are Struggling on a Line (Experiment II) TRIAL BAIT TIDAL CURRENT NO. AND LENGTH OF SHARKS INVOLVED TIME TO APPEAR DIRECTION FROM 1 mullet flood, moderate 1 4-ft grey 17 min downstream 2 mullet flood, strong 1 4-ft grey 10 min downstream 3 grouper flood, moderate 2 6- ft greys 15 min downstream 4 snapper flood, moderate 2 4-ft greys 20 min downstream 5 grouper flood, moderate 1 6-ft grey 18 min downstream 6 mullet flood, moderate 1 4-ft grey 1 6-ft grey 5 min downstream 7 mullet ebb, moderate 2 4-ft greys 10 min downstream 8 grouper flood, moderate 1 4-ft grey 1 3 -ft whitetip 16 min downstream 9 grouper ebb, moderate 2 6-ft greys 1 6-ft whitetip 14 min downstream 10 grouper ebb, moderate 2 4-ft greys 25 min upstream one test which might at first glance appear to deviate from this pattern deserves quotation from the field notebook: Introduction was made at 1035 . . . bait positioned 10 ft off the bottom, being carried 20-30 yd astern by the current where it struggled vigorously until 1050 when noticeably tired. By 1055 bait no longer strug- gled, but simply maintained position, looking quite natural. At 1100, two 4-ft grey sharks appeared from upstream, cruising slowly along the bottom toward the test area and giving no indication of having sensed bait. When still about 20 yd upstream of the boat (and 40 ft down, on the bottom) they both became notice- ably alerted. At this point apparently they had become aware of the boat on the surface, as they veered up- ward and swam at an accelerated rate directly to within 5 yd of the boat, turned aside at this distance and circled twice. Then, seeming to lose interest, they re- turned slowly to the bottom and continued at their leisurely pace downstream, passing within 10 yd of the bait without apparent notice (as stated, the bait was not struggling, but merely maintaining position in the current). The sharks continued downstream ap- proximately 20 yd below the bait, at which point they both obviously and simultaneously became alerted, turned around, and with increasing speed raced back upstream straight toward the bait, with one of them taking it. The first response in the above observation was apparently one of vision to the boat on the surface. The second response, in which the sharks returned upstream to the bait, was ob- viously one of olfaction. In this experiment, the sharks appeared to be following an olfactory cue in a direct line to an uninjured fish. However, there was no as- surance that other stimuli were not also in- volved in the detection and approach to the bait. The question arises whether these sharks can follow an olfactory cue directly to its source in the absence of other cues. The classical experiments of Parker (1914), in which he observed the approach of dogfish to bait when both nostrils were free and when one was occluded, has been offered as proof that directional response to olfactory cues is possible by virtue of the ability of each nostril to detect minute differences in the concentration of odor- ous material. In standing water in the ponds at Coconut Island we have observed the hammer- head ( Sphyrna lew ini) describing the typical figure 8 pattern, described by Parker, in which the shark is apparently locating the source of stimulation by continually turning toward the nostril exposed to the greater concentration of the material. Tester ( 1963 ) also describes how blinded blacktips spiral down from above in converging on bait on the floor of the tanks at the Eniwetok Marine Biological Laboratory. It seems likely that sharks are capable of following an olfactory trail in running water, particularly when the current is strong and the trail narrow, thus forming what would essentially be an ol- factory corridor. Under such conditions they could be expected to make a direct-line ap- proach by taking advantage of the normal lat- eral movements of the head, which are part of the swimming motion, in keeping themselves oriented in the stimulus trail. 180 PACIFIC SCIENCE, Vol. XVII, April 1963 Experiment III The purpose of this experiment was to de- termine if these sharks are capable of orienting themselves in a current and following an olfac- tory corridor in a direct line to its source in the absence of other stimuli. A location on the edge of Deep Channel sim- ilar to that used in Experiment II was chosen. Essentially clear, colorless fish extracts were used as the olfactory cues. These were prepared from several species of grouper but always of the same concentration (600 gm of fish flesh or skin macerated in a Waring blendor with 6 liters of fresh water and diluted in a large plas- tic container with 64 liters of sea water). The material was introduced from the glass-bottom boat by siphoning from the container through a clear plastic 34-inch hose which ran down to the top of the reef. From here it continued down the slope through a series of holes to a small cave at the base of a large rock. The hose was thus effectively concealed for a distance of ap- proximately 30 yd from the cave. Eight tests were conducted under various ve- locities of tidal current over a 15 -day period with no more than one trial on any one day. Each test was preceded by a 30-min control period during which any unusual behavior in sharks sighted was noted. The duration of the introduction varied according to the response of the sharks. Observations included the species and general behavior of the sharks and particu- larly the direction and nature of approach. The results are given in Table 2. A consistent response was exhibited by both grey and blacktip sharks during this experiment. In the presence of a current they appeared from downstream, swimming at an accelerated rate directly to the mouth of the cave where, in most cases, they briefly stuck their snouts inside. They then turned in very small circles here for a short period of time before returning slowly, in a random manner, downstream. After an initial approach by any one shark, this same shark often reappeared and repeated the above de- scribed pattern several times. However, upon the third or fourth approach by the same shark, this shark often reversed direction 5-10 yd short of the hole and returned downstream without the characteristic circling. TABLE 2 Response of Sharks to Extracts of Fish Flesh or Skin (Experiment III) (All Sharks Came from a Downstream Direction) TRIAL EXTRACT TIDAL CURRENT NO. AND LENGTH OF SHARKS RESPONDING TIME TO INITIAL APPEARANCE TOTAL INTRODUCTION TIME 1 frozen flesh ebb, moderate 1 6-ft grey 1 4-ft grey 1 4-ft whitetip 10 min 1 hr 2 frozen flesh flood, moderate 1 4-ft grey 1 5 -ft grey 2 4-ft whitetips 12 min 1 hr 3 frozen flesh ebb, moderate 1 4-ft grey 1 5 -ft grey 25 min 30 min 4 frozen flesh ebb, weak 1 4-ft grey 1 4-ft whitetip 35 min 1 hr 5 frozen flesh ebb, weak to moderate 1 4-ft grey 1 4-ft blacktip 1 5 -ft whitetip 16 min 45 min 6 frozen flesh flood, moderate to strong 3 4-ft greys 1 4-ft whitetip 9 min 45 min 7 fresh flesh flood, strong 2 6-ft greys 1 4-ft grey 1 4-ft whitetip 1 5 -ft whitetip 7 min 30 min 8 fresh skin flood, moderate to strong 3 6-ft greys 1 4-ft grey 6 min 30 min Feeding Behavior of Sharks — Hobson 181 It might be suggested that the olfactory stim- ulus had not itself been directional, but that the sharks, having been alerted by the olfactory cue, had simply turned upstream, and oriented to the current. However, while a rheotaxic re- sponse may have influenced the shark in its ini- tial decision as to which way to go in the cor- ridor, this observer does not feel that the re- sponse observed could have been directed by such a cue. Among other considerations, In an approach directed solely by current the shark would have at least briefly overshot the source of the olfactory stimulus. This would have re- sulted in a brief but certainly noticeable period of uncertainty as the shark turned back to pick up the olfactory cue once again. Nothing of this sort was seen. Furthermore, the following of an olfactory corridor was observed in the absence of current and will be described shortly. Occasionally the extract elicited a response from other fish in the area which could have provided an approaching shark with supplemen- mentary cues. However, in only 2 of 27 ob- served approaches might this source of error have affected the results. When the current subsided, the approach of the sharks immediately became less direct — illustrating the importance of the current in maintaining the definition of the corridor. No new sharks appeared during slack water and those already present milled about continuously within 30 yd of the cave. Approaches to the cave during slack water were made in a random manner and from all directions. At this time it was apparent that the material was diffusing out in all directions from its source and was at the same time being retained in the immediate area. The whitetip did not seem to be as respon- sive to the introductions as were the blacktip and the grey. On several occasions whitetips swam directly to the cave in the manner noted in the grey and blacktip. However, whitetips just as often swam past the hole without any noticeable response to the extract. As apparent unresponsiveness has been noted as characteris- tic of the whitetip, it is difficult to draw any direct comparisons between this species and the grey and blacktip from the observed behavior. Whitetips do have the ability to follow an olfactory corridor. The following quotation from the field notebook describes an incident which clearly illustrates this and also the formation of a corridor in the absence of current: A large parrot fish (Scaridae, 10 lb.) was speared in about 20 ft. of water. The fish tore itself from the spear and took shelter in a large coral head. Within 1 min. a 5 ft. whitetip appeared. It became obvious that the shark had sensed the presence of the wounded fish as it poked about the holes of the coral head and then swam into one of them. The chase which followed was witnessed from the surface. The two fish could periodically be seen through one or another of the many holes which honeycombed the coral head. First the parrot fish would flash by and then the whitetip in pursuit. The coral head contained an extensive net- work of caves and the chase seemed to take advantage of most of them. Suddenly the parrot fish emerged from a hole. Apparently it had temporarily eluded the whitetip because there was no immediate sign of the shark. The parrot fish swam off rapidly on a straight course for about 30 yd. where it made a 90° turn and continued on the new course, in a straight line, until it was out of sight. The whitetip emerged from the hole just seconds after the parrot fish, but already its prey was out of sight. The whitetip circled briefly, then started out along the same path taken by the par- rot fish. When it arrived at the point of the 90° turn it continued on a few yards, but quickly slowed and turned around. After another brief period of circling the shark picked up the second leg of the trail and followed it straight out of sight. In this instance, the corridor was formed by olfactory substances given off by the moving wounded fish. There was no noticeable current at the time. The sharks involved in Experiment II had apparently been initially alerted by an olfactory stimulus emitted by a fish which, while in dis- tress, was uninjured. The ability of these sharks to detect the pres- ence of an unwounded fish in a state of stress through an olfactory cue was noted by Tester (1963) early in the program. This point has been neglected since the observations of Shel- don (1911), in which a dogfish was noted to locate an undamaged crab wrapped in eelgrass. However, in Sheldon’s experiment the attract- ing stimulus (or stimuli) may have been a movement or sound made by the crab instead of, or in addition to, an olfactory stimulus. Any conclusions derived from Experiment II are open to the same criticism. Another source of criticism might be the assumption that a fish with a line passing through its maxillary mem- brane may be considered uninjured. The ex- 182 PACIFIC SCIENCE, VoL XVII, April 1963 TABLE 3 Response of Sharks to Water Containing a Grouper Under Stress (Experiment IV) (All Sharks Came from a Downstream Direction) TRIAL TIDAL CURRENT NO. AND LENGTH OF SHARKS RESPONDING TIME TO INITIAL APPEARANCE TOTAL INTRODUCTION TIME 1 flood, moderate 2 5 -ft greys 15 min 30 min 2 ebb, strong to slack 3 4-ft whitetips 5 min 30 min 3 flood, moderate 1 4-ft grey 1 3 -ft whitetip 10 min 1 hr periment described below was designed to mini- mize these sources of error. Experiment IV This experiment was conducted to determine if these sharks are capable of detecting and tracking down, exclusively by olfaction, an un- injured fish under stress. The experiment was carried out using the procedure described for Experiment III, except that, in place of the extract solution, the large plastic container was filled with sea water and contained an uninjured grouper (1-10 lb) which was presumably under stress. In each case, the fish was caught by line with a barbless hook. During the experiment the fish was agitated intermittently with a pole, using care not to in- flict any damage, while the water was running from the container to the bottom as described for Experiment III. Three trials were conducted over a period of 10 days, with at least 3 days between each trial. The results are given in Table 3. The response was essentially the same as that to the extracts in Experiment III. Greys and whitetips were involved in the experiment; no blacktips were seen. The captive grouper obviously emitted an ol- factory stimulus which attracted the sharks up- stream to the cave. Critics of this experiment may question whether a recently hooked fish can be considered uninjured. There is justification for pointing out the hook wound as well as the skin rubbed and mucus dislodged during han- dling. These factors offer a possible source of error. However, the results were consistent with findings in the laboratory tanks where it was possible to maintain considerably more control over experimental conditions. In the latter tests (Tester, 1963) the strength of the stimulating olfactory component varied with the degree of distress of the fish being used. It appeared that a maximum level was reached shortly after the death of the fish. It has been shown that the sharks were able to track down a distressed but apparently un- injured fish by olfaction alone in Experiment IV. Although they could also have made exclu- sive use of the olfactory sense in tracking down the hooked fish in Experiment II, it is highly unlikely that they did so. We shall see that, in all probability, other sensory modalities not only contributed to, but in fact dominated, certain phases of the approach. RESPONSE TO COMPRESSION WAVES Compression waves are regarded by many to have an important effect on the behavior of sharks, both as an attractant and as a repellent. Wright (1948) claims that attack patterns in sharks are released by sounds. Many investiga- tors (e.g., Eibl-Eibesfeldt and Hass, 1959) re- port that the vibrations and sounds made by a wounded fish will attract sharks. On the other hand, some sounds have been reported to have a repelling effect, as for example the underwater shouts of divers (Hass, 1951)- These reports are based on incidental encounters with sharks and are generally complicated somewhat by the presence of stimulating factors other than com- pression waves which might themselves have been influential in eliciting the observed re- sponse. For example, Wrights conclusion is based to a considerable degree on observations of sharks appearing just after an underwater explosion to feed on the dead and stunned fish. Feeding Behavior of Sharks — Hobson 183 The presence of the disabled fish is mentioned almost incidentally, although in such a situation it is impossible to dismiss the olfactory and visual cues emitted by these fish. Similarly, the presence of olfactory and visual cues also complicate reports of sharks being attracted to a struggling fish in response to compression waves. In regard to the repelling effect of the shouts of divers, we did not see any such re- sponse in the species studied in Eniwetok. How- ever, sudden movements and/or a sudden burst of bubbles from the aqua-lung, such as might readily accompany a shout, often startled the blacktip and put it to flight. It is important to consider the ability of sharks not only to sense these stimuli but also to locate their sources. Parker (1912), investi- gating sound as a directing influence on the movements of some teleosts, noted that there was a directing effect only during the duration of the sound. These results cast some doubt on the ability of such noncontinuous stimuli as a single underwater explosion to attract sharks. Two experiments on the perception of com- pression waves by sharks were conducted in the field at Eniwetok. Experiment V The aim of this experiment was to investi- gate the effects of compression waves of a variety of sonic frequencies on these sharks. The fol- lowing were presented: A. Continuous pure tones, covering a fre- quency range in steps from 100-1,000 cycles per sec. B. Sounds of mixed frequencies recorded on a circular tape, including: 1. wood struck against wood, 1/ sec and 6/ sec; 2. rock struck against rock, 2/sec and 6/sec; 3. rasping effect, continuous; 4. rattling of nuts and bolts in a bottle, continuous. C. Shouts by the observer, directly projected through the apparatus. The experiment was conducted at the Engebi location, using a tape recorder equipped with an underwater speaker lowered 6-8 ft below the surface. Each sound was presented during 20- min trials, with each trial immediately follow- ing a 20 -min control period and conducted under two sets of conditions: (1) following a different experiment in which sharks had been drawn into the experimental area and were still present, and (2) upon arrival at the site, with no sharks present. All sounds were audible to a submerged human at distances greater than 150 ft. Although we watched closely for signs of any sort of response, for example curiosity, we saw nothing in the behavior of the sharks which suggested that they were able to perceive the sounds. Experiment VI This experiment was designed to investigate the effects of various subsonic compression waves on these sharks. A metal hoop, having a diameter of 36 inches, was fitted with a rubber diaphragm held in posi- tion by a series of surgical tubing lacings. This piece of apparatus was installed midway through a natural tunnel, approximately 20 ft long, in the coral ridge bordering Deep Channel beneath the raft. A line was secured to the center of the diaphragm on that side facing away from the channel, and this line was run out of the tunnel and up to the surface where a small skiff rode at anchor. The diaphragm could thus be vibrated by jerking on the line. Two trials were conducted in which the dia- phragm was vibrated irregularly for a period of 20 min immediately following another experi- ment in which a number of grey sharks had been drawn into the area and were still present. Throughout both trials there were grey sharks swimming about calmly in the area. At times they passed within 5 yd of the tunnel’s entrance. In observations from the raft we looked for any type of response in the sharks which might indicate that they were aware of the vibrating diaphragm. We saw no such response. The negative results in these two experiments do not mean that the sharks could not perceive these stimuli. If releasers of feeding activity are emitted by struggling fish in the form of com- pression waves, they probably are of a specific nature or involve a characteristic pattern. It would have been a rare stroke of luck if we had duplicated a specific releaser with our crude experiments. The problem of delimiting the 184 sensory capacity for stimuli of this sort is prob- ably best approached with conditioned response experiments on captive sharks under controlled conditions before attempting to establish their role in feeding behavior under natural con- ditions. Although the work of Parker (1903, 1911), Kritzler and Wood (1961), and others has demonstrated that at least some sharks are sensi- tive to compression waves of a relatively wide range of frequencies, it has yet to be proven experimentally that such stimuli normally re- lease feeding behavior. Nevertheless, incidental observations at Eniwetok indicated they do play an integral role in feeding activity. These ob- servations are of a nature similar to those men- tioned at the beginning of this section and are subject to the same reservations. One such ob- servation is quoted from the field notebook: ... A 30 lb. grouper was speared on the slope border- ing Deep Channel . . . the grouper fled, dragging the spear, into a small cave. . . . Within seconds, 5 greys 4—7 ft. in length swam excitedly into the area from downstream. As soon as they were in the area, how- ever, their excited state diminished. They swam about in the area for several minutes, appearing to steadily lose interest, before slowly drifting off downstream and out of sight. After a few minutes ... 5 greys made another approach which seemed as highly motivated as the preceding one. Immediately I looked below to see a small puff of sediment emerge from the cave — the grouper was obviously thrashing about inside. Almost as soon as the sharks were in the area, all be- came quiet in the cave, whereupon the sharks imme- diately lost their excited state and settled down to cruising about as before. Two 4-ft. whitetips joined them before they all again drifted off downstream. After a few minutes . . . another small puff of sedi- ment was visible at the mouth of the cave — the grouper was struggling again. I quickly looked down- stream to see the greys on their way in as before. The pattern previously described was repeated, but this time there were 4 whitetips (3—5 ft.) in addition to the 5 greys. Several minutes after the sharks disap- peared downstream for the third time the grouper scrambled out of the hole and, dragging the spear, struggled along the bottom toward the bigger, more protective caves farther up the slope. The 5 greys charged into the area from downstream before it had gone 20 ft. At this point, the grouper stopped, and ceased all movement. Although the grouper was in a completely exposed position, the excited state of the onrushing greys diminished almost as soon as the grouper stopped moving. Instead of attacking the com- pletely vulnerable grouper as I expected, the greys settled down to swimming slowly and randomly about the area. Often they came within inches of the mo PACIFIC SCIENCE, Vol. XVII, April 1963 tionless grouper, and on at least one occasion grazed it. The 4 whitetips rejoined the group, along with 2 blacktips (3 and 4 ft.) making a total of 12 sharks, all obviously interested, but offering no show of ex- citement or indication that they regarded the wounded grouper as prey. Nevertheless, the sharks did not gradually drift off downstream as they had previously, but continued to swim about in the area. Thirty min- utes later, when we had to leave, the situation was unchanged — the grouper was still sitting motionless in the same spot, while the 12 sharks swam slowly about. Most of this random swimming about the area, both at this point and earlier, had occurred downstream of the grouper. It was significant that all approaches and de- partures, as well as most of the random swim- ming in the area, occurred downstream of the grouper. This indicated that olfaction was in- volved throughout the incident. Although the excited approach of the greys during this inci- dent appeared to coincide with the periods when the grouper was struggling, it is possible that they had first been conditioned by an olfactory cue. In this case, the olfactory cue may have had a threshold lowering effect for the postulated compression wave stimuli. At least one author (Wright, 1948) doubts that olfaction alone will release attack patterns in sharks. In this latter regard, blinded sharks in the tanks fed avidly on chunks of fish flesh which were lying on the bottom, thus offering little other than olfactory stimuli before the sharks came into contact with them. The question here is whether or not the behavior of these blinded sharks can be con- sidered normal. While the sharks in the latter phase of the incident did not attack the wounded grouper, they did not drift off downstream as they had done when this same fish was concealed in the cave. We are probably safe in assuming that the behavior pattern demonstrated in swimming slowly about in the area was appetitive in nature. This suggests that an additional stimulus was required to release the actual attack. This is to say that a combination of stimuli, e.g., an olfac- tory element and erratic motion might have been necessary to release the attack in this case. The olfactory cue may have simply released an ex- ploratory behavior pattern which drew the sharks to the area. Once there, the release of the attack may then have required a specific stimulus not presented by the quiet, natural looking grouper, the coloration of which blended Feeding Behavior of Sharks — Hobson 185 in very well with the bottom on which it was resting. The cessation of movement by this grouper upon the approach of the sharks may have been a well-established protective behavior pattern. We will consider this incident further in the next section. During a later period of the study, a grouper of the same species and ap- proximately the same size was caught at the raft where the underwater action was witnessed from the observation cage. The grouper wrapped the line about a piece of coral and by the time the line was freed and the struggling fish hauled toward the surface, a single 5 -ft grey came rush- ing into the area from downstream. The shark went directly for the struggling fish, took the tail section into its mouth, and with a few vigorous lateral shakes of its head came away with the after portion of the fish. There is no doubt that rapid erratic move- ments are a prime releaser of attack patterns in these sharks, particularly in the presence of spe- cific olfactory cues. Many authors have noted an excited state in sharks resulting from the sensing of movement (for example Limbaugh, 1958). As both a mechanical disturbance and a visual stimulus are usually produced by a moving ob- ject underwater, it is very often difficult to say with certainty which is more significant as a releaser of attack in a given situation. On one occasion, while fishing for sharks, a 3 -ft grey was hooked in the presence of five larger greys. As the shark struggled vigorously on the line, the other five sharks became highly active and appeared to be chasing the hooked animal. Initially we supposed that an attempt was being made to attack the hooked shark, but upon landing this individual after several min- utes of activity we noted no evidence of injury. If the larger sharks had been attempting to feed on the smaller hooked shark there is no doubt that they could have done so. It is probable that the activity of the hooked shark excited the others which were then simply following the focal point of this activity. Inasmuch as other species of fish, similarly hooked and struggling, were immediately taken by the grey shark there is a suggestion here that this species of shark is inhibited in some way from attacking members of its own species. This possibility is further supported from observations made while fishing for snapper ( Lutjanus bohar) at Palmyra. Grey sharks drawn to the fishing area showed no in- terest in the many free-swimming snapper, but would immediately attack a snapper which be- came hooked and began to struggle on the line. Contrariwise, although the sharks would swim rapidly about one of their own species which was hooked and struggling, no attacks were observed. RESPONSE TO VISUAL STIMULI Contrary to many reports on the subject (for example Halstead, 1958) vision was found to play a major role in the feeding activity of the sharks in the Eniwetok lagoon. We briefly men- tioned vision above in regard to instances in- volving movement. The following experiments were designed to further clarify the role of vision in cases where there was little or no movement of the bait. Experiment VII The object of this experiment was to deter- mine the role of vision in these sharks when they are approaching a motionless bait up an olfactory corridor. Two baits of similar appearance were pre- pared for each trial, one of which was a 4-inch cube of grouper flesh while the other was a 4-inch cube of wood. After we were confident that no sharks were visible in the area, the two baits were presented together in a strong cur- rent, suspended 3 ft apart, midway between the surface and the bottom. Observations were made from the cage, and a record was kept of which bait was hit first, as well as of the nature and direction of approach of the sharks. Experiment III had shown that these sharks are capable of orienting on an olfactory stimulus in a current and swimming directly to its source. In every case during the present experiment the sharks appeared from downstream, swimming at an accelerated rate, presumably following an olfactory trail emitted by the grouper flesh. If the sharks had continued to orient exclusively on the olfactory stimulus all the way to the bait, then the fish would presumably have been taken in all trials. This, however, did not occur. In 20 trials conducted over a period of 2 days, the fish was struck first 11 times while the block of wood was hit first 9 times. As no preference was 186 shown, it seems that at least the final phase of the approach was visually directed. When the shark selected the wood, the ob- ject was either simply bumped or briefly taken into the mouth and then rejected. No teeth marks were found on the wood following the test. In all cases the fish-baits were carried away. Only grey sharks were involved. If we accept as fact that these sharks orient visually during the final phase of their approach to a motionless prey, we must then question the nature of the stimulating visual image, bringing us to the subject of visual acuity. Most elasmobranchs reportedly possess an all- rod retina with a high ratio of visual cells to ganglion cells, which provides for low visual acuity. While this would presumably result in an inability to see objects in detail, the sharks would be able to utilize their visual sense under conditions of relatively little light (Gilbert, 1961). Kato (1962) found no evidence of cones in the retina of blacktips or whitetips. Perhaps even more significant, he found no specialized area, corresponding to the human fovea, which is generally assumed to be associated with visual acuity. At Engebi, we often noted that sharks at- tracted by a dead fish suspended on a line often struck objects in the immediate vicinity of the bait before taking the bait itself. A rock about IV2 ft square, which was situated in the middle of a sandy area below the baits, as well as float- ing seaweed and a lead weight on the line a foot above the bait were often tested with a nudge or by actually being taken momentarily into the mouth. These observations suggest one of two alternatives: (1) these sharks lack the visual acuity necessary to distinguish the visual appearance, i.e., form, markings, etc. of fish from inedible objects in the immediate vicinity; or ( 2 ) although these sharks do possess the visual acuity necessary for such a distinction, it was not utilized by them in this situation. Tester and Kato (ms) showed that small blacktips and greys have the ability to discrimi- nate between a number of objects of different shape, e.g., squares, triangles, and rectangles, and that they make this discrimination from a distance of at least 6 to 12 ft. These species then possess at least the degree of visual acuity re- quired to discriminate between these forms. PACIFIC SCIENCE, Vol. XVII, April 1963 The following experiment was an attempt to detect evidence of the utilization of a high de- gree of form discrimination ability in the feed- ing behavior of these sharks. Experiment VIII This experiment was designed to determine whether or not the visual cues offered by a fish through its characteristic morphological features are significant attractive visual stimuli to these sharks. Two fresh dead fish were selected for each trial, both being of the same species, but one a little larger than the other. Three species were used: goatfish (Mullidae), 8 trials; surgeon fish ( Acanthuridae), 3 trials; and squirrel fish (Ho- locentridae), 10 trials. The larger fish was then decharacterized by removing the head and all the fins, reducing it in size to approximately that of the smaller, still natural-appearing fish. A long, deep incision was then made along the ventral side of the natural-appearing fish to in- sure that olfactory stimuli emitted by both baits were essentially the same. The baits were both secured to a single line, one spaced approxi- mately 18 inches above the other, with their respective position being reversed on alternate trials. After assurance that no sharks were vis- ible, the baits were suspended motionless at a point midway between surface and bottom in a strong current. Observations were made from the cage and a record was kept which bait in the pair was taken first, along with the species of shark involved and the nature and direction of approach. It was assumed that if the characteristic mor- phological features which had been removed from one of the pair were significant as attrac- tive visual stimuli, then the natural-appearing fish would be favored in the initial choice made by the sharks. This in turn would demonstrate a high level of form discrimination ability in these sharks. However, no apparent preference was shown to either of the two bait types in 21 trials. The decharacterized fish were hit first 11 times, while the natural-appearing fish were hit first 9 times. Once both baits were hit simul- taneously. The relative positions of the baits on the line had no apparent effect, with the top bait being hit first 9 times and the bottom bait Feeding Behavior of Sharks — Hobson 187 1 1 times. As in Experiment VII, only grey sharks were involved and these approached in every case at an accelerated rate from downstream. By positioning the bait in mid-water we favored this species. A number of whitetips were seen about the bottom at various times during the trials. The baits introduced during this experiment were the first food given by us to the sharks in over a month, with the exception of a few un- injured living fish. This is an important con- sideration in this experiment, as any condition- ing of the shark population to feeding on our presentations would have resulted in a tendency to take our offerings indiscriminately. However, no conditioning was noted at this stage of experimentation. The failure of these sharks to show a pref- erence for the natural-appearing fish over the one which had been reduced to a simple object suggests that the visual appearance of a fish per se is not itself significant to these sharks when feeding. When the results of this experi- ment are considered along with the observations at Engebi already mentioned, in which the sharks struck at inedible objects in the imme- diate area of the baits, it appears that the sig- nificant visual cue in these cases was simply that of an object within the immediate area of the source of the olfactory stimulus. This evidence suggests that these sharks do not utilize a high degree of visual, acuity in their feeding activity. It follows, then, that if the visual stimulus is not moving, it must otherwise contrast with the background against which it is viewed in order to provide an effective visual cue. The following experiment involves the rela- tive effectiveness of two objects in attracting the attention of sharks, when these objects differ from one another in degree of contrast in bright- ness with their background. Experiment IX This experiment was designed to determine which of two baits, one white and the other black, would be taken first when both are pre- sented to these sharks together at the surface. Paired 3 -inch cubes of fish flesh (parrot fish, snapper, and grouper) were used as bait in this experiment, with members of each pair identical except that one retained its natural white color, while the other was dyed black with nigrosine dye. It had been previously determined that nigrosine dye was not sensed by blinded sharks in the tanks. The experiment was conducted at the raft over 40 ft of water when a current was run- ning. A wounded grouper was placed in the cage until a number of sharks had been drawn about the raft. The fish was then removed from the water, allowing the source-point of the olfac- tory stimulus to be carried off downstream and taking the sharks with it. When the sharks were about 20 yd downstream from the raft, the two baits were dropped into the water, spaced ap- proximately 10 ft apart. Movement of both baits was negligible as they slowly settled. At times the response of the sharks immediately followed the entry of the baits into the water, indicating that the splash had been sensed; at other times the response was noted after the baits began to sink, indicating a visual response. These re- sponses followed the introduction too closely to have been those of olfaction. In either case, the sharks came racing back upstream and in all cases the baits were taken before they had fallen 10 ft. The bait taken first in each case was recorded, although the alternate bait was always taken almost simultaneously. As it was antici- pated that the amount of incident light present would influence the results, trials were con- ducted under different light conditions, with from 10 to 56 trials held each day for 4 days. The results under each of the different sets of conditions prevailing were as follows: ( 1 ) day- light with clear sky, in 96 trials black was taken 79 times, or 82%; (2) daylight with overcast and drizzle, in 21 trials black was taken 15 times, or 71%; (3) daylight with overcast and heavy rain, in 14 trials black was taken 8 times, or 57%; (4) after sunset, with clear sky but almost dark, in 41 trials black was taken 32 times, or 78%. The total: for 172 trials under all conditions, black was taken 124 times or 72%. As the sharks, all greys, raced back toward the test area they were viewing the baits against light surface water. The black bait, then, con- trasted with its background to a greater degree than did the white bait. This was as noticeable 188 to the observer sitting in the cage as it obviously was to the sharks. Recalling some of the observations made earlier, we might consider again the speared grouper which rested motionless on the bottom while 12 sharks swam slowly about in its im- mediate area. Although this grouper was in a completely exposed position, its coloration blended in well with the bottom on which it was resting. In this position the grouper was apparently at least temporarily safe from attack by the 12 sharks, in spite of the fact that it was wounded and still emitting the olfactory stim- ulus which presumably had released the ap- petitive behavior pattern which these sharks were at that time demonstrating. At Engebi, the sharks struck at the rock which was sitting in the middle of the sandy area, floating seaweed, and the lead weight — all inanimate objects in the immediate area of the bait, but all of which sharply contrasted in brightness with the back- grounds against which they were viewed by the sharks. These results are consistent with the con- clusions drawn by Gilbert (1961) from ana- tomical studies of the shark eye. Although there is little doubt that vision is the predominant directing sense within the visual field, the effective distance involved here will be highly variable. Such external factors as water clarity, incident light, and whether or not the prey is under cover, no doubt determine the effectiveness of vision in any given situation. RESPONSE TO GUSTATORY AND TACTILE STIMULI While lack of visual discrimination appar- ently led to selection of the wood as often as it did the fish during Experiment VIII, the shark was quick to learn its error. Initially the wood was actually taken into the mouth, but after one or two successive trials a nudge was gen- erally sufficient to dismiss the inedible object. An appraisal of the bait by the visual or olfac- tory sense may have been the basis for the nudge, but when the bait was taken into the mouth, other senses, e.g., gustation and/or tac- tile sense, undoubtedly came into play. A review of the gustatory sense of sharks is presented by Tester (ms). In this review, he points out that while the receptors of the gus- PACIFIC SCIENCE, Vol. XVII, April 1963 tatory sense (e.g., taste buds or terminal buds) occur in the skin of the body, fins, and barbels of many species of fishes, in the elasmobranchs they appear to be associated with papillae which are confined to the epithelial lining of the mouth and pharnyx. He also calls attention to the claim of Budker (1938) that the "pit or- gans" located on the body of elasmobranchs have a gustatory function. An effort was made to design an experiment which would illustrate the respective roles played by both gustation and the tactile sense in accepting or rejecting food taken into the mouth. As the block of wood in Experiment VIII differed from the fresh fish in tactile cues as well as those of olfaction and gustation, it was impossible to say which of the two might have been more influential in the rejection of the wood. An attempt was made to synthesize a bait which possessed the visual and tactile properties of acceptable bait, but which lacked the olfac- tory and gustatory properties. Tester et al. ( 1955) concluded that in the flesh of many fishes there is present a substance or substances which, when extracted with alcohol or water, can be perceived by a fish through its sense of smell or taste and which promotes the urge to feed. Tester commented (personal communica- tion) that extraction by alcohol was effective in removing this substance from a piece of flesh. He observed that squid prepared this way was not detected by blinded blacktips and although taken into the mouth by normal blacktips, they were subsequently rejected. He attributed this rejection to the lack of acceptable gustatory stimulation, in as much as the texture of the prepared squid seemed to be similar to untreated squid. The following experiment was an attempt to duplicate Tester’s results in the field. Experiment X This experiment was designed to determine whether or not a bait offering the visual and tactile stimuli of food, but which lacks the ol- factory or gustatory stimuli, will be acceptable as food by these sharks when they are actively feeding. A number of 3 -inch cubes of grouper flesh Feeding Behavior of Sharks — Hobson 189 were prepared. Half of these baits were soaked for 4 days in each of the following concentra- tions of ethyl alcohol: 35%, 75% and 95%, in that order. After the alcohol treatment the baits were soaked in sea water for 3 hr. It was hoped that the product of this process would be a bait which felt and looked like food, but did not taste or smell like food. In actuality, how- ever, the texture of the product was not identi- cal to that of normal flesh, being notably dry and leathery. The experiment was nevertheless continued, with each of the treated baits being paired with a normal bait. It was assumed that the normal bait would be taken. If the treated bait was also taken this would indicate that the wood in Experiment VIII had been rejected due to its tactile properties rather than its lack of acceptable olfactory or gustatory stimuli. If, however, the treated bait was not taken, Testers results would be confirmed and it would appear that gustation (and perhaps the tactile sense too) had been involved in the rejection of the wood. The test was carried out at the raft, in a mod- erate current, using essentially the same tech- nique described for Experiment IX, except that in this case all 15 trials of the experiment were conducted during a 30-min period of one after- noon and the response to both baits was noted. Only grey sharks were involved in the results. The treated bait was hit first 6 times, while the natural bait was hit first 9 times, indicating that the sharks did not make a distinction between the visual appearance of the two baits. All the baits were taken into the mouth, but although in all trials the natural bait was swallowed im- mediately, the treated bait was rejected within seconds. As this rejection took place after the bait had been taken into the mouth, it appeared to be based on gustatory stimulation (if we are correct in assuming that the bait offered ac- ceptable tactile stimulation). A snapper was seen taking, and apparently retaining, one of the treated baits rejected by the grey shark. Another effort to clarify the roles played by gustation and the tactile sense in feeding in- volved the use of sponges. It was thought that perhaps a sponge which had been soaked in an extract of fish flesh would be accepted by the sharks as food. If this were the case, then by pairing the soaked sponge with a normal sponge it would have been possible to present a choice of baits which posed the same problem to the sharks as did Experiment X. In order for the experiment to have been a success, however, it was necessary for the sharks to have accepted the soaked sponge as food, and this proved to be only temporarily true. Initially, the soaked sponge was carried off in the manner seen with a chunk of fish, while the plain sponge was re- jected as was the prepared bait in Experiment X. However, after 3 trials the sharks began to lose interest in the soaked sponges until they either dismissed them after an unenthusiastic nudge, or ignored them completely. Further- more, as the sharks learned that the soaked sponges were inedible, the extract no longer excited them. In this latter regard it is difficult to say whether the sharks actually learned that the extract did not indicate food or whether olfactory fatigue was the major factor. A definite change in general behavior of the grey sharks was first noted during Experiment IX and became very evident during Experiment X. As the sharks became conditioned to feeding about the raft they responded instantly to any object which was presented. In addition, there was a sharp increase in the number of sharks responding to our test situations. Where ini- tially a maximum of 5 or 6 grey sharks were in- volved in our experiments, the number steadily increased during this period of increased food supply until more than 20 were being drawn about the raft. The initial population had been relatively stable and through successive observa- tions involving experiments which utilized min- imum feeding, it had become possible to recog- nize a certain group of about 6 individuals that appeared day after day. These same individuals were seen after the population numbers in- creased, but apparently many other sharks which normally frequented other areas were drawn to this point of concentrated feeding. This change involved only grey sharks, which is understand- able as this species was completely monopoliz- ing the additional food. The increase in num- bers of sharks resulted in a markedly increased element of competition. When we witnessed the simultaneous rush at the baits by up to 10’ sharks, it was understandable that all baits were taken into the mouth before any discrimina- tions were made as the slightest hesitation by 190 a shark in this situation immediately eliminated it from a chance at a bait. Fortunately the pres- ent tests, which required concentrated feeding, were concerned with probing questions which did not suffer from these effects. This change in behavior, along with the increase in numbers of the local sharks, was therefore viewed with interest rather than alarm. FOOD PREFERENCES Sharks have been popularly described in a general way as creatures with an exceptionally voracious appetite, feeding on such unlikely ob- jects as tin cans, bottles, and other trash (Linea- weaver, I960). Although considerable effort was brought to bear on the problem of food pref- erences during this study, the experiment de- scribed below was the only one, of many experi- ments conducted, which clearly indicated a pref- erence for one of two food materials presented. Experiment XI This experiment was an effort to determine whether or not these sharks show any preference between grouper flesh and the flesh of various species of mollusks, e. g., Tridacna sp., (10 trials); Spondylus sp ., (5 trials); and Cassus sp., (3 trials). Mollusk flesh in 2 -inch cubes was paired with chunks of grouper flesh of the same size, with both baits thus offering an essentially identical visual appearance. Prior to the test, large pieces of grouper and mollusk flesh of the species to be tested were placed in the cage on the raft to attract sharks. When the introductions began, there were 10 grey sharks, 3 to 6 ft in length, present in the area. The method of introduction was as used in Experiments VIII, IX, and X, ex- cept that all trials were conducted during one session. Observations were made to detect any evidence of a preference which might be shown between the two baits. The fish baits were hit first 8 times, while the mollusks were hit first 9 times, indicating that there was no visual preference. However, al- though both the fish and mollusks were taken into the mouth with equal vigor, the fish were presumably swallowed while the mollusks were rejected almost immediately. The results were comparable to those of Experiment X, with the PACIFIC SCIENCE, Vol. XVII, April 1963 response toward the mollusks similar to that shown toward the treated bait. On one occasion a single shark took the Tridacna and then the grouper. After a short period, with both baits in its mouth, the shark rejected one, presumably the Tridacna. These same rejected baits were subsequently taken by groupers and snappers which waited below the feeding sharks. Al- though the mollusks tested were unacceptable as food by these sharks, it is known that they will feed readily on squid. It was possible, therefore, to observe a pref- erence when one of the bait-choices offered was unacceptable to the sharks. However, when both bait-choices presented were acceptable, it be- came difficult to make this distinction, even though one might have been significantly more attractive than the other. Thus, most of our ex- periments concerning food preferences yielded inconclusive results. For example, a test might have been conducted to determine the compara- tive attractiveness of two baits, A and B, both motionless and presented on the lines within 10 ft of each other. As indicated in Experiment VIII, although these sharks might have been drawn in by an olfactory stimulus produced by bait A, they would then have been quite likely to have hit B inasmuch as the final phase of the approach to the bait would have been visually directed. Once having taken bait B, this bait would be retained as long as it was not actually unacceptable. It has been noted (Tester, 1963) that some types of fish flesh appear to be more attractive to sharks than do others. In the ponds at Coco- nut Island the notably dry-fleshed snapper, Lu- tianus gib bus, appeared to be far less desirable to captive sharks as food than did the much juicier tuna, Katsuwonus pel amis. Springer (1958) also noted this preference for tuna. Is it possible that this apparent preference is ac- tually due to a higher concentration of some basic attractant which is common to the flesh of both fish? It has become increasingly appar- ent that some substance (or substances) in fish flesh is perceived by the sense of taste and/or smell of these sharks which is a powerful ele- ment in the release of a highly motivated feed- ing pattern. Furthermore, it was indicated in Experiment X and also in the work of Tester et al. (1955) that this substance could be ex- Feeding Behavior of Sharks — Hobson 191 tracted from the flesh, whereupon the flesh it- self was left undesirable to the sharks. Tester et al. (1955) expended considerable effort in attempts at purification, fractionation, and iden- tification of the attractant extracted from fish flesh which released a response in captive tuna. While the precise identity of the substance eluded the investigators, many of its chemical properties were determined and presented by the authors. Tester et al. (1954) suggest that the so-called attractant which is present in the body juices of fish, squid, shrimp, and other forms may be a substance (or substances) which is common to all these forms. FEEDING DETERRENTS The recent increased interest in the habits and behavior of sharks has stemmed largely from a growing awareness of the need to de- velop a more effective means of protecting hu- mans from shark attack. A number of tests of proposed repellents were conducted, including a test of the repellent now in general use by the armed services. The re- sults of some of these tests illustrate many of the problems involved in producing an effective repellent. Experiment XII This experiment tested the effectiveness of the standard shark repellent (copper acetate- nigrosine dye) in protecting both dead and wounded fish from attack by these sharks. The experiment was conducted at two loca- tions: (1) on the bottom at the edge of Deep Channel, and (2) from the raft. In the tests on the bottom, involving whitetips, three baits were tied to packages of repellent and anchored on the bottom, while seven baits, without re- pellent, were anchored nearby. In the tests from the raft, which involved grey sharks, all the baits were tied to packages of repellent and lowered halfway to the bottom. No attempt was made to attract sharks before beginning these tests. The whitetips took all of the unprotected baits but did not take any of those tied to the packages of repellent. On the other hand, al- though the first grey to appear during the tests at the raft occasionally exhibited a slight hesi- tation when making its approach, it was never long before several greys were in the area and the bait and repellent package were quickly taken together. The repellent packages were shortly rejected, but there was no indication of a subdued appetite. One grey was seen swim- ming away trailing a black cloud of shark re- pellent from its gills with no show of discom- fort. The relative effectiveness of the repellent in protecting the baits from the whitetips may be related, at least in part, to the absence of the group feeding effect in this species. This effect, which was discussed earlier, was noted to lower the threshold for the release of feeding patterns in grey sharks. Experiment Kill In this experiment the standard shark repel- lent was presented as a large cloud, both by it- self and as a protective screen for various at- tractants. Trials were conducted at the raft in which 30 gm of repellent were dissolved in 64 gal of sea water in a plastic container. This solu- tion was presented alone as well as mixed with each of the following attractant materials: (1) two small macerated goatfish, (2) 500 ml fish extract, (3) 1,000 ml of fish extract, and (4) 2,000 ml of fish extract. The extracts were pre- pared by using 50 gm of grouper flesh per 500 ml of fresh water. The presentation was made by simply pour- ing the contents of the container into the water after a number of sharks had been drawn in about the raft and noting the results. In each case, a cloud approximately 6 ft deep and 12 ft wide formed at the surface next to the raft and slowly drifted downstream. When the plain repellent cloud was intro- duced, the sharks rapidly converged on the cloud to a distance of approximately 5 yd where they circled slowly, following the cloud as it drifted downstream. After a few minutes, however, the sharks lost interest in the cloud and returned to the raft. When the repellent cloud containing the macerated goatfish was introduced, the fish fragments sank beneath the cloud where they were quickly taken by the sharks. The sharks, then excited, swam unhesitatingly up into the cloud where they circled vigorously. Sixteen 192 greys, 3-6 ft long, were involved in the first two phases of this experiment. The response to the repellent clouds contain- ing the extract solutions was essentially the same in each case. After rapidly approaching the cloud, the sharks circled briefly about its edges and then moved slowly inside without notable excitement. Within 5 min of introduction, all sharks had lost interest in the introduction. Eight greys, 3-6 ft in length, were involved in these tests. It seems noteworthy that the extract elicited an unexcited pattern of exploratory behavior rather than the highly excited aggressive be- havior seen following the introduction of mac- erated goatfish. In any case, the most that can be said for the repelling qualities of the material used in this test is that the sharks did not swim into the cloud without incentive to do so. It may theoretically have served to conceal an otherwise attractive visual stimulus which might have been inside. Experiment XIV This experiment tested the effectiveness of an underwater light suspended on a line in protecting dead fish. The apparatus was designed so that the bait (dead goatfish) was suspended 2 ft below the light in a position where both could be viewed from the underwater chamber. Observations were then made of the response of the sharks to the bait, both in the presence and in the ab- sence of light. Tests were conducted during four different nights using both 110- and 300-watt bulbs. A steady light was used, as well as a light which flashed at rates of approximately 15, 35, and 50 flashes per min. Periods of light, 20 min long, were alternated with 20-min periods of darkness during both of which the baits were presented. Sharks had been drawn to the area by bait placed in a wire basket which was sus- pended in the water and then raised prior to each test. A strong current was flowing toward Deep Channel during each test. Initially the steady light was effective in pro- tecting the bait from the sharks for the entire 20-min period, with the bait being taken soon after the light was extinguished. During the early periods of light the sharks milled about PACIFIC SCIENCE, Vol. XVII, April 1963 at the edge of the field, consistently downstream of the bait and light. Upon continued contact with the light the sharks became progressively bolder until despite the light, flashing or steady, they showed little hesitation in taking the bait. This boldness seemed to increase notably with the number of sharks present. There were usually between 5 and 10 grey sharks present during these tests; no other species was seen. This test proved to be a good illustration of progressive loss of apprehension through grow- ing familiarity with an initially strange situa- tion, as well as the mutually stimulating effect of some species of sharks on one another in certain situations. CONCLUSIONS 1. Three species of sharks are common in Eniwetok lagoon: the grey shark, Car char hinus menisorrah; the blacktip shark, C. melanopterus; and the whitetip shark, Triaenodon obesus. 2. Each species inhabits a rather characteris- tic habitat within the lagoon: (a) the blacktip in relatively shallow water ( 1-40 ft approx. ) over sand and coral rubble flats which extend out from shore for distances ranging from a few yards to several miles; (b) the grey, along the outer slope of these flats, in deeper water and in the passages to the sea; and ( c ) the whitetip about rock-ledges and coral heads. 3. The feeding behavior of these sharks shows differences which in many cases appears to be associated with their characteristic habitats. 4. Blacktip and grey sharks show a marked increase in excitement when feeding in num- bers. This phenomenon, often referred to as a "feeding frenzy” in extreme cases, does not seem to occur in the whitetip. 5. All three species are highly sensitive to stimuli emanating from or suggesting injured and/or distressed, as well as dead or moribund prey. 6. These sharks can detect, by olfaction, both injured fish and uninjured fish in a state of stress. 7. Olfaction is the most effective sense in detecting prey at a distance, providing the ol- factory stimulus has had sufficient time and means for effective dispersal. Feeding Behavior of Sharks — Hobson 193 8. These sharks are able to follow an olfac- tory stimulus quickly and directly to its source without benefit of other orienting stimuli when the olfactory material has been drawn out as a narrow trail by a strong current or, in standing water, when the source of the olfactory stimulus is moving. 9. Generally, olfactory stimuli are effective in releasing a highly motivated pattern of ex- ploratory behavior in grey sharks, which is ap- petitive in nature. Consummation of the feeding act then requires a subsequent specific stimulus, as for example one of vision. 10. The grey shark is often highly excited by, and attracted to, the source of rapid and/or er- ratic movement. The effect is intensified in the presence of certain olfactory stimuli but is not dependent on their presence. 11. Despite displaying varying degrees of ini- tial curiosity toward many unusual stimulus situ- ations arising in their environment, the sharks exhibit caution when encountering a situation which is unfamiliar. This caution will steadily subside, however, with an increasing familiarity with the situation. 12. The final phase of approach to a motion- less prey by the grey shark is normally directed by vision, even though the feeding pattern may have been initially released by some other sen- sory modality, for example, olfaction. 13. Although vision is an important sense in the feeding activity of these sharks, a high de- gree of acuity and form discrimination is not demonstrated. Rather, the significant visual cues seem to involve the detection of movement or contrasting brightness, or both. 14. These sharks may attack any object which they sense in a high concentration of an olfac- tory material. 15. The grey shark rejects, from the mouth, food which does not permit acceptable gustatory stimulation. 16. The standard shark repellent (copper acetate-nigrosine dye) now in use by the armed forces is ineffective in preventing the grey shark from swimming into a cloud of this material when the species is present in numbers and is motivated by food within and adjacent to the material. ACKNOWLEDGMENTS The author is indebted to Dr. Albert L. Tester for many helpful suggestions during the course of the investigation and for his constructive criticism of the manuscript. The study would not have been possible without assistance in the field from Dr. Tester and Mr. Francois Mau- tin, as well as from Mr. John Kay, Mr. Bryant Sather, Mr. Susumu Kato, Mr. Taylor Pryor, and Mr. Albert Smith, graduate students, and Mr. Henry Yoshihara, undergraduate student, all of the University of Hawaii. The facilities at Eniwetok were made available by Dr. Robert W. Hiatt, Director of the Eniwetok Biological Laboratory, as well as by the Atomic Energy Commission, the Pacific Missile Range Facility of the United States Navy, and by the Holmes and Narver Construction Company. The author also wishes to thank Dr. William A. Gosline and Dr. Ernst S. Reese of the Department of Zoology and Entomology, University of Hawaii, and Dr. Donald W. Strasburg of the Honolulu Biological Laboratory, United States Fish and Wildlife Service, for critically reviewing the manuscript. REFERENCES Budker, P. 1938. Les cryptes sensorielles et les denticles cutanes des plagiostomes. Ann. Inst. Oceanog. 18:207-288. Eibl-eibesfeldt, I., and H. Hass. 1959. Er- fahrungen mit Haien. Zeit. Tierpsych. 16 (6): 733-746. Gilbert, P. S. 1961. The visual apparatus of sharks and its probable role in predation. Ab- stracts of Symposium Papers, Tenth Pacific Science Congress, Honolulu, Hawaii, pp. 176— 177. Halstead, B. 1958. The relationship of trau- magenous fishes to man. A.I.B.S. Shark Con- ference, New Orleans, Apr. 8-11, 1958 (MS). Harry, R. R. 1953. Ichthyological field data of Raroia Atoll, Tuamotu Archipelago. Atoll Res. Bull. 18:1-190. 194 Hass, H. 1951. Diving to adventure. Doubleday and Co., Garden City. 280 pp. Hobson, E. S., F. Mautin, and E. S. Reese. 1961. Two shark incidents at Eniwetok Atoll, Marshall Islands. Pacific Sci. 15 (4): 605-609. Kato, Susumu. 1962. Histology of the retinas of the Pacific sharks, Carcharhinus melanop - terus and Triaenodon obesus. M. S. Thesis, University of Hawaii, Honolulu, Hawaii. Klausewitz, W. 1958. Die Atoll-Riffe der Malediven. Natiir und Volk 88 (11): 380 — 390. 1959. Fische aus dem Rotenmeer. I. Khorpel Fische. Senck. biol. 40:43-50. Kritzler, H., and L. Wood. 1961. Provisional audiogram for the shark Carcharhinus leucas. Science 133 (3463) : 1480-1482. Limbaugh, C. 1958. Abstracts from field notes on sharks. A.I.B.S. Shark Conference, New Orleans, Apr. 8-11, 1958 (ms). Lineaweaver, T. I960. Shark! Sports Illus- trated, Feb.: 54-63. Parker, G. H. 1903. Hearing and allied senses in fishes. Bull. U. S. Fish. Comm. 1902:45-64. 1911. Influence of the eyes, ears, and other allied sense organs on the movements of the dogfish, Mustelus canis. Bull. U. S. Bur. Fish. 29:43-57. 1912. Sound as a directing influence in the movements of fishes. Bull. U. S. Bur. Fish. 30: ( 1910) :97-104. 1914. The directive influence of the sense of smell in the dogfish. Bull. U. S. Bur. Fish. 33:61-68. PACIFIC SCIENCE, Vol. XVII, April 1963 Randall, J. E. 1955. Fishes of the Gilbert Is- lands. Atoll Res. Bull 47:1-243. Sheldon, R. E. 1911. The sense of smell in selachians. J. Exper. Zook 10:51-62. Schultz, L. P., E. S. Herald, E. A. Lachner, A. D. Welander, and L. P. Woods. 1953. Fishes of the Marshall and Marianas Islands. Bull. U. S. Nat. Mus. 202(1): 1-685. Springer, S. 1958. Field observations on large sharks. A.I.B.S. Shark Conference, New Or- leans, Apr.: 8-11 (ms). Strasburg, D. W. 1958. Distribution, abun- dance, and habits of pelagic sharks in the central Pacific Ocean. Fishery Bull. U. S. Fish Wildlife Serv. 58 ( 138) : 335-361. Tester, A. L. MS. The olfactory and gustatory apparatus in sharks. 1963. The role of olfaction in shark predation. Pacific Sci. 17(2). , and S. KATO. ms. Form and color dis- crimination by blacktip and grey sharks ( Car- charhinidae). , H. Yuen, and M. Takata. 1954- The reaction of tuna to stimuli 1953. U. S. Fish Wildlife Serv. Spec. Sci. Rpt. Fish. 134:1-33. , P. B. Van Weel, and J. J. Naughton. 1955. Reaction of tuna to stimuli 1952-53. Part 1. U. S. Fish Wildlife Serv. Spec. Sci. Rpt. Fish. 130:1-61. Wright, B. S. 1948. Release of attack behavior patterns in shark and barracuda. J. Wildlife Mgt. 12 (2): 117-123. Life History of the Caligid Copepod Lepeophtheirus dissimulatus Wilson, 1905 (Crustacea: Caligoida)1 Alan G. Lewis2 Copepods of the family Caligidae are found as external parasites on both fresh and salt water fishes throughout the world. Even with their wide distribution, relatively little is known about their life history. The presentation of the life history of Lepeophtheirus dissimulatus , a caligid found on salt water fishes, should pro- vide more information on the life history and general biology of this group of copepods. Although there are many references to im- mature and juvenile or larval stages (reviewed by Heegaard, 1947 : 2 1-3 1 ) , only a few workers have had the material to describe the complete life history of a caligid. Through methods de- scribed below, the author was able to obtain a series of Hawaiian specimens representing all of the stages of development of L. dissimulatus. Lepeophtheirus dissimulatus is one of the more common parasitic copepods of Hawaiian acanthurid (surgeon) fishes. The species is not limited only to Hawaiian fishes but is known from a wide variety of localities and a wide variety of hosts (Lewis, in press). Even with its wide distribution, in both the Atlantic and Pacific oceans, larval stages have been described only once, by Shiino (1959:309), from fishes of the Revilla Gigedo Islands. Attempts to rear the copepod from the egg to the adult in culture were only partially success- ful. Ovigerous adult females were placed in clean finger bowls partially filled with cotton-filtered or unfiltered sea water. Circulation of the water was accomplished by means of a fine jet of air from a slender glass tube immersed in the medium. The water was changed at least once 1 A portion of a thesis submitted to the University of Hawaii in partial fulfillment of the requirements for the degree of Doctor of Philosophy. Contribution number 181, Hawaii Marine Laboratory. Manuscript received November 23, 1961. 2 Department of Zoology, University of New Hamp- shire, Durham, New Hampshire. every 8 hr and usually more frequently. At the time the eggs, carried by the female, hatched, a portion of the first nauplii, the first larval stage, was removed with a wide-mouthed pipette and placed in small stender dishes. The remain- ing nauplii were left in the finger bowl and the adult female was removed and placed in a separate bowl. Water in the finger bowls was circulated in the same manner as above and changed at the same time intervals. Water in the stender dishes was not circulated but instead was changed at more frequent intervals, every 3-4 hr. As moulting occurred into the second naupliar stage, the newly moulted second nau- plii were transferred to new finger bowls or stender dishes except for three to four indi- viduals which were preserved in either 10% buffered formalin or 95% ethyl alcohol. Shortly after the second naupliar stage moulted into the succeeding stage, the copepodite, a piece of loosely woven nylon cloth was placed in the dishes to offer a substrate for attachment. (It was found that the copepodite stage is the stage which attaches to the host.) In addition, pieces of host tissue were placed in some of the dishes. The host tissue was left in the dishes for 1 hr and then removed and the animals transferred to a new dish in order to reduce contamination and bacterial action. The mortality rate between the first and sec- ond naupliar stages was high and appeared to be partially due to the inability of the nauplius to break out of the old cuticle. The rate from the second naupliar stage to the copepodite was relatively low. Out of the 30 attempts that were made to rear the copepod, each with at least one female carrying from 20-60 eggs, one copep- odite was obtained that attached to the nylon cloth placed in the dish. The remaining copep- odites died without attaching and the single copepodite that attached died immediately thereafter. None of the copepodites attached to the host tissue placed in the dishes. 195 196 PACIFIC SCIENCE, Vol. XVII, April 1963 endo. ~exa Fig. 1 . Hypothetical thoracic leg of Lepeophtheirus dissimulatus, showing the various armament compo- nents. C, Heavy setae; c, fine setae; endo, endopodite; exo, exopodite; f, membrane; H, well-developed spine; H', laterally projecting spine appearing as a continua- tion of segment; h, poorly developed spine; Hm, mem- brane margined spine; P, large plumose setae; P', large naked setae; p, small plumose setae; p’ , small naked seta; prot, protopodite; Q, seta (or spine) plumose on one side and membrane-margined on other; rh, spinule; s, hairlike seta; sp, sternal plate. Because of the failure of the culture material and in order to obtain a complete timed series of developmental stages, attached copepodites were taken from host material and placed in in- dividual stender dishes. After these copepodites moulted into the first chalimus (the following stage ) , the time was noted until the succeeding moult, into the second chalimus. To ascertain the time passed in each stage after the first chalimus, specimens of the previous stage were taken from the host and treated as above. Due to the mortality rate under laboratory condi- tions, many specimens of each stage had to be collected so that one would pass through the succeeding stage. Because of the inability to get any number of specimens to pass through each stage and due to the necessity of submitting the larvae to laboratory conditions, the time indi- cated for each stage in Figures 2-7 should be regarded as indicative, not conclusive. The descriptions of the body, appendages, and processes of all of the developmental stages and the adult is followed by a discussion of the general behavior of the various developmental stages. The graphical figures included in the text (Figs. 2-7) give the means and ranges of various measurements of the larval forms. Larval specimens other than nauplii were killed and preserved in 95% ethyl alcohol. Speci- mens to be dissected were placed in 85% lactic acid to soften and clear them, stained with methyl blue dissolved in 85% lactic acid, and dissected in clear 85% lactic acid. Drawings of the total animal were made from specimens placed in 85% lactic acid and cov- ered with a 9 mm cover slip. The cover slip was raised slightly above the younger specimens by thin pieces of glass so that the specimen was not distended. A camera lucida and a Bausch and Lomb arc projector were used to make the drawings of the total animal. Drawings of the appendages and processes were made, with the use of a camera lucida, from appendages and processes either in situ on the whole mount or dissected off and mounted in Hoyer’s mounting medium. Measurements were made with an ocu- lar micrometer. The terminology used, with some exceptions which are given in the text, is that used in Lewis (in press). To facilitate the use of the thoracic leg tables, a hypothetical thoracic leg is shown in Figure 1, giving all of the component parts of the armament of the thoracic legs dis- cussed in tabular form in this publication. Fur- ther, a table of the hypothetical thoractic leg is given below in Table 1. 3.5 3.0 2.5 £ .§2.0 .c CD £ 1.5 1.0 0.5 6ocy<. Fig. 27. Crystallophrisson rubrum. Characteristic spicules from body regions b, c, and d. Fig. 28. Crystallophrisson rubrum. Sta. 2150. between the mouth opening and the brain, then semicircular in the brain region, and some- what widened and quadrangular in the region of the radula. Numerous buccal glands are vis- ible behind the sphincter and all three kinds of salivary glands are present. The radula is typical and lacks denticles. The mid-gut is wide and has a cuticular region near its opening to the liver. The rectum is slightly winding. The pericardium is spacious, the ventricle slender and not very muscular, with two atrio- ventricular openings. The dorsal vessel begins with a bulb. Gills are conspicuously small and have only a few gill lamellae. The epithelium of the gonoducts is not continued on the wall of the cloacal chamber. DISTRIBUTION: Two specimens come from the San Pedro area, southern California, in 310-386 fm, one from clay sediments, the other from mud with glass sponge. 280 PACIFIC SCIENCE, Vol. XVII, July 1963 Cry st alio phr is son rubrum n. sp. Figs. 27, 28 COLLECTION : 2150 (1) in 310 fm, with C. hartmani and C. riedli. DIAGNOSIS: This species is readily distin- guished from others by its outer purple color and its plump body ( Fig. 28 ) . The single speci- men measured 17 mm long and 2.5 mm at its greatest width. The front end was withdrawn and the exact outline of the buccal plate is thus not known, but it is certainly pierced by the mouth opening. The spines (Fig. 27) are large and stout and have a distinct keel; the abdominal needle-like spines attain a length of 800 n. A glandular zone in front of the opening of the cloacal chamber is present and "giant cells" are frequently encountered in the epidermis of the entire body. The animal is highly muscular; there are two pairs of dorsal, one pair of lateral, and one pair of ventral retractors of the front region, all of them very strong. These retractors split off from the longi- tudinal muscle layer which is located dorsally in the prothorax, dividing only into a few rather large bundles. The six pairs of gill-re- tractor muscles are typical; the anterior dorsal one has a double origin. The diaphragms are very strong, and an anterior vertical one can be distinguished. The nervous system shows three pairs of small compact precerebral ganglia, a small brain, separated from the dorsal wall of the fore-gut by a large sinus, and a cerebrobuccal connective split off the common trunk at some distance from the latter’s origin. The fore-gut is provided with strong longitudinal muscle fibres and with a stout sphincter, with buccal glands anterior to the last and with all three kinds of salivary glands, of which the dorsal ones open into an unusually large pouch of the fore-gut. The strong radula consists of the typical parts in addition to a pair of small pointed, curved denticles. The reproductive organs are typical; the gonoducts are simple without sacculations; their dorsal ciliated parts are long and slender tubes. The epithelium of their outlets contin- ues on the wall of the cloacal chamber. The pericardium is very large and together with the posterior part of the atrium surrounds the suprarectal ganglion. The muscular ventricle, connected with the atrium by two atrioventricu- lar openings, is attached at its front end to the dorsal and ventral pericardial wall. The dorsal vessel begins as a large bulb. distribution: This species is known through a single specimen from Redondo cafi- yon, in the San Pedro area, southern California, in 310 fm, in mud. ACKNOWLEDGMENTS The author is indebted to all those who made these collections possible. The operations of the "Velero IV" were supported by funds provided partly by the Allan Hancock Foundation, the California State Water Pollution Control Board, and the National Science Foundation. The sort- ing of solenogasters was done by staff members of the Hancock Foundation; the anatomical preparations were prepared in the laboratory of the Department of Zoology, University of Vi- enna. The author is also indebted to the Aus- trian Department of Education and the Depart- ment of Zoology of the Free University of West Berlin. APPENDIX Additional Station Data Physical data are published for most of the station numbers listed above. Those not sep- arately listed below are to be found in Hart- man (1955) and Hartman and Barnard (1958, I960). The following are newly listed; all localities are in southern California. 2457-53. Oct. 17, 1953. 4.0 miles SE of Point Fermin lighthouse, 33° 39' 19" N. lat., 118° 14' 20" W. long. In 21.5 fm. A snapper sampler took a very small sample of sand. 2884-54. July 27, 1954. 5.6 miles SW of end of Newport pier. 33° 31' 58" N., 118° 00' 00" W. In 190 fm. The orange-peel-grab took 1.93 cu ft of gray green mud. Solenogaster Mollusks — SCHWABL 281 2888-54. July 27, 1954. 14.7 miles EWE of East end, Santa Catalina Island. 33° 21' 58" N., 118° 02' 00" W. In 390 fm. OPG took 2.67 cu ft of gray green mud. 2895-54. July 28, 1954. 9.9 miles SSW of end of Newport pier. 33° 27' 57" N., 118° 01' 58" W. In 265 fm. OPG took 2.58 cu ft of gray green mud. 2899-54. July 28, 1954. 8.25 miles SW of end of Newport pier. 33° 30' 00" N., 118° 02' 02" W. In 266 fm. OPG took 2.64 cu ft of gray green mud. 2901-54. Aug. 11, 1954. 12.8 miles SW of East end, Santa Catalina Island. 33° 26' 00" N., 118° 06' 06" W. In 312 fm. OPG took 3.02 cu ft gray green mud. 3037-55. May 8, 1955. 3.75 miles from North light, Santa Barbara Island. 33° 31' 17" N, 118° 58' 00" W. In 360 fm. OPG took 2.58 cu ft of gray green sandy mud. 3204- 55. July 7, 1955. Santa Monica Bay. 33° 58' 10" N., 118° 39' 10" W. In 115 fm. OPG took 2.08 cu ft of green silty mud. 3205- 55. July 7, 1955. Santa Monica Bay. 33° 58' 00" N, 118° 37' 00" W. In 88 fm. OPG took 2.08 cu ft of green silty mud. 4756-56. Dec. 8, 1956. 4.1 miles from Scripps pier, La Jolla. 32° 51' 55" N., 117° 20' 30" W. In 102 fm. OPG took 2.39 cu ft of green mud and some broken shells; temperature of sediment 9.8 C. 4776-56. Dec. 10, 1956. 2.2 miles from Dana Point. 33° 26' 00" N., 117° 43' 30" W. In 87 fm. OPG took 3.46 cu ft of green mud; temperature of sediment 10.2 C. 4778-56. Dec. 10, 1956. 5.2 miles from Dana Point. 33° 30' 20" N., 117° 47' 45" W. In 100 fm. OPG took 2.96 cu ft of green mud. Temperature of sediment 9.7 C. 4835-57. Feb. 6, 1957. 11 miles from Port Hueneme Light. 34° 02' 15" N., 119° 01' 45" W. In 77 fm. OPG took 1.38 cu ft of olive green sandy silt. REFERENCES Hartman, Olga. 1955. Quantitative survey of the benthos of San Pedro basin, southern California, Part 1. Preliminary results. Allan Hancock Pacific Exped. 19 ( 1 ) : 1-185, 9 pis. — and J. Laurens Barnard. 1958, I960. The benthic fauna of the deep basins off southern California. Allan Hancock Pacific Exped. 22 (1, 2): 1-297, 19 pis., map. Heath, Harold. 1911. Reports on the scien- tific results of the expedition to the tropical Pacific, in charge of Alexander Agassiz, by the US Fish Commission Steamer "Alba- tross,” from August, 1899, to June, 1900. The Solenogasters. Mem. Mus. Comp. Zool. Harvard 45: 1-182, 40 pis. 1918. Solenogasters from the eastern coast of North America. Mem. Mus. Comp. Zool. Harvard 45:183-263, 14 pis. SCHWABL, Mathilde. l%la. Crystallophrisson ( = Chaetoderma ) hartmani, nov. spec., eine neue Aplacophore aus dem Ostpazifik. Zool. Anz, Bd. 166, Heft 7-8:258-277, 10 figs. — — — 1961 A Plathyucenia branchiosa, nov. gen., nov. spec., ein neuer Vertreterder Neo- meniidae aus dem Ostpazifik. Zool. Anz., Bd. 167, Heft 3-4:100-115. Thiele, J. 1932. Die Solenogastres des arkti- schen Gebietes. Fauna Arctica, Hena, Bd. 6, Lief. 4:379-382. Studies in Fitchia (Compositae) : Novelties from the Society Islands; Anatomical Studies Sherwin Carlquist and Martin L. Grant1 The genus Fitchia has attracted interest be- cause of its arborescent habit, its distribution in southeastern Polynesia, the endemism of its species, and its phylogenetic relationships. In a monograph of this genus, six species occur- ring on live islands were recognized ( Carlquist, 1957). Following the appearance of that mono- graph, Grant notified Carlquist of his collec- tion, while on a Bishop Museum Fellowship in 1931, of specimens of Fitchia from two ad- ditional islands (Bora Bora and Tahaa) in the Society Group. The genus had not been pre- viously reported from these islands, and manu- script names, as a new species and subspecies, respectively, had been applied to the taxa. On Grant’s invitation, Carlquist made the anatomi- cal studies presented here which, it is felt, puts the recognition of the two new entities on a much firmer basis than gross morphology alone could have provided. In addition, descriptions are given of one additional specimen each of two previously known species of Fitchia, one from Tahiti, and one from Rapa. The occur- rence of the two novelties is of more than or- dinary interest because of the nature of this genus, and permits more extended remarks on the nature of speciation in Fitchia. Flowering material of F. rapensis, not available for the monograph, has been studied here. This paper, therefore, may be considered an addendum to the monograph. The taxonomic descriptions be- low have been prepared by Grant. The anatomi- cal account which follows is the work of Carl- quist, and has been prepared to conform with the types of data offered in the monograph. 1 Claremont Graduate School, Claremont, California; and Iowa State Teachers College, Cedar Falls, Iowa. Manuscript received December 28, 1961. TAXONOMIC DESCRIPTIONS 1. Fitchia cor data M. L. Grant et S. Carlquist, sp. nov. Fig. 31 Frutex arboreus 2 m altus glaber, caule basi- liter 4 cm crasso. Folia simplicia petiolata petio- lis 25-65 mm longis, longitudine media (N = 42) 47 mm. Laminae late obovatae vel subor- biculares luteovirides 40-72 mm longae 36-65 mm latae, magnitudine media foliarum grandi- orum circa 58 X 54 mm, basi cordata vel ro- tundata, apice obtuso acutove vel parum acu- minate, margine crenulato vel subintegro, venu- lis lateralibus 5-10 (numero medio 8.5) pro unico latere. Pedunculi terminales 20-40 mm longi, longitudine media ( N = 7 ) 33 mm. Capi- tula solitaria circa 40-florata. Involucra 25 mm longa 25 mm lata, bracteis 3-4 seriatis viride- lutei coriaceis marginibus- scariosis, exterioribus late triangulatis 3 mm longis 9 mm latis, medi- oribus semiorbicularibus, interioribus lanceo- i latis 13-17 mm longis 5-7 mm latis. Bracteae receptaculares 12-17 mm longae 2-3 (-5) mm latae. Corollae 15-20 mm longae, tubis circa 10 mm longis, sinibus subaequalis. Antherae saccis 5 mm longis, apicibus staminum cuneato-lance- olatis 1-1.5 mm longis. Ovaria ad anthesin 7-8 mm longa, stylis 20-22 mm longis. Achaenia matura 13-16 mm longa, aristis 8-10 mm longis plerumque triangulatis in sectione transversa subdeciduis. Tree-like shrub, 2 m tall, 4 cm in diameter \ at the base, glabrous. Bark 1 mm thick, brown, almost smooth, with fine longitudinal ridges 0.5 mm deep. One-year twigs on flowering shoots greenish, 2.5-3 mm in diameter; two- 282 Studies in Fitchia — CARLQUIST and GRANT 283 FIGS. 1—6. Secondary xylem. Figs. 1-4, Fitchia cuneata ssp. tahaaensis. Figs. 5-6, F. cordata. Figs. 1, 2: Tran sections; note tyloses in vessels; Fig. 2, a sclerified tylosis in which the lumen is nearly occluded. Figs. 3, 4: Tangential sections; Fig. 3 is a portion showing long, thick-walled fibers; Fig. 4 shows storied apotracheal parenchyma cells. Fig. 5: Transection; note band of apotracheal parenchyma above center of photograph. Fig. 6: Tangential section. Fig. 2, X 300; others, X 117. year twigs pale brown, 4-5 mm in diameter; internodes 4-12 mm long, on vegetative shoots to 20 mm long. Leaves simple. Petioles 25-65 mm long, the average being 47 mm ( N = 42 ) ; the stipular sheaths 0.4-0.6 mm high. Blades broadly ovate to suborbicular, yellowish-green, with a narrow (0.3 mm) stramineous margin, and with a dark callous at the apex; 40-72 mm long, 36-65 mm wide, the average ( N = 8 ) size of the largest blade on the flowering shoots 284 PACIFIC SCIENCE, Vol. XVII, July 1963 being 62 X 55 mm, the second largest 58 X 54 mm, and the smallest of the fully-developed leaves 64 X 43 mm; the base cordate to a depth of 1-2 (-3) mm or rounded; the apex obtuse to acute, with the margins near the tip slightly rounded to straight, or, if barely acu- minate, departing at most 1 mm frorn^a straight line; the margin crenulate, with 2-3 indenta- tions (up to 0.5 mm deep) per cm, to suben- tire, the base entire for 20 mm adjacent to the petiole; with 5-10 (average 8.5) pairs of major lateral veins, usually curving somewhat towards the tip of the leaf. Peduncles terminal, 20-40 mm long, the average (N = 7) 33 mm, often arcuate and reflexed. Head solitary, with about 40 flowers. Involucres 25 mm long and 25 mm wide; the bracts in 3-4 series, greenish-yellow, coriaceous, scarious-margined, the outermost broadly deltoid (3 mm long, 9 mm wide), the middle ones semicircular, and the inner ones lanceolate (13-17 mm long, 5-7 mm wide). Receptacular bracts 12-17 mm long, 2-3 (-5) mm wide. Corollas 15-20 mm long, the tube about 10 mm long, the shallower sinuses nearly equalling the ventral cleft, and the lobes with terminal hairs 0. 1-0.6 mm long. Anther sacs 5 mm long, the stamen tips cuneate-lanceolate, 1-1.5 mm long. Ovary at anthesis 7-8 mm long; the styles 20-22 mm long. Ripe achenes 13-16 mm long, with awns 8-10 mm long; awns gen- erally triangular in transection and somewhat deciduous. DISTRIBUTION: Society Islands, endemic to Bora Bora. SPECIMEN EXAMINED: Bora Bora, in dwarf rain-forest on the summit of Mt. Tarapaia, alti- tude 645 m, Jan. 3, 1931, M. L. Grant 4968 (bish, type; RSA, ISTC, isotypes). This species was found in several places on Mt. Tarapaia, none less than 25 m in altitude below the summit, but only four inflorescences were discovered, two in bud, one at anthesis, and one in fruit. Associated woody plants were Alstonia, Metrosideros, Freycinetia, and Glochi- dion. Native name: anei. Mt. Tarapaia appears on U. S. Hydrographic maps as Mt. Pahia, elevation 2165 ft (— 660 m), the name and elevation being taken from French maps. It has been recorded as "Tarao- pai’a.” The only higher point on Bora Bora is Otemanu (Mt. Temanu of the maps), eleva- tion 725 m, a peak which apparently has never been climbed by anyone other than Polynesians. 2. Fitchia cuneata J. W. Moore, Bishop Mus. Bull. 102:48. 1933 ssp. tahaaensis M. L. Grant et S. Carlquist, subsp. nov. Fig. 32 Frutex arboreus 2.5 m altus glaber, caule basaliter 2.5 cm crasso. Folia simplicia petiolata petiolis 10-37 mm longis, longitudine media (N=30) 20 mm. Laminae plerumque ellipti- cae vel parum ovatae obovataeque vel rhom- boidales subatrovirides 40-110 mm longae 25- 58 mm latae, magnitudine media foliarum grandiorum circa acuminato, margine crenu- lato serratove vel subintegro, venulis lateribus 5-10 ( numero medio 7.5 ) pro unico latere. Pedunculi terminales 25-50 mm longi, longi- tudine media (N= 10) 33 mm. Capitula soli- taria circa 75-florata. Involucra 20-40 mm longa, 15-25 mm lata, bracteis 3-4 seriatis, viride- lutei coriaceis marginibus . chartaceis, exteriori- bus breviter ovatis reniformibusque 2-5 mm longis 6-15 mm latis, medioribus semiorbiculari- bus 8-12 mm longis 9-13 mm latis, interioribus cuneatis 12-20 mm longis, 4-8 mm latis. Brac- teae receptaculares lanceolatae 12-17 mm longae 2-4 mm latae. Corollae ad anthesin 15-22 mm longae, tubis 4-10 mm longis, dentibus 5-7 mm longis. Antherae saccis, 4.5-5 mm longis, apici- bus staminum cuneato-lanceolatis 1.5 mm longis. Ovaria ad anthesin 8-10 mm longa, 3-4 mm lata, stylis 24-30 mm longis. Achaenia matura 10-20 mm longa, 4-6 mm lata, aristis 10-13 mm longis plerumque teretibus sub- persistentibus. Treelike shrub, 2.5 m tall, 2.5 cm in diameter at the base, glabrous. Bark 1 mm thick, brown, roughened with irregular scaly ridges. One-year- old twigs on flowering shoots greenish, 2 mm in diameter; two-year twigs pale brown, 3 mm in diameter; internodes 4-8 mm long, on vege- tative shoots to 12 mm long. Leaves simple. Petioles 10-37 mm long, the average (N = 30) 20 mm; the stipular sheaths 0. 5-0.8 mm high. Blades mostly elliptical, varying to slightly ovate, obovate, or rhomboid, dull green, with a narrow stramineous margin, and with a dark Studies in Fitchia — Carlquist and Grant 285 callous at the apex; 40-110 mm long, 25-58 mm wide, the average size (N=10) of the largest leaf on flowering shoots 84 X 45 mm, the second largest 75 X 42 mm, and the small- est of the fully developed leaves 59 X 30 mm; the base cuneate (usually) to barely rounded; the apex acute, occasionally slightly acuminate; the margin crenulate to toothed or subentire, with 2-4 indentations (up to 1 mm deep) per cm; with 5-10 (average 7.5) pairs of major lateral veins, mostly curving slightly toward the tip of the leaf. Peduncles terminal, 25-50 mm long, the average (N=10) 39 mm long, al- ways reflexed and usually arcuate at anthesis. Heads solitary, with about 75 flowers. Involucres 15-25 mm long, 20-40 mm wide; the bracts in 3-4 series, greenish-yellow, coriaceous, charta- ceous-margined, the outermost short ovate or reniform (2-5 mm long, 6-15 mm wide), the middle ones semi-circular (8-12 mm long, 9-13 mm wide), and the inner cuneate (12-20 mm long, 4-8 mm wide ) . Receptacular bracts lanceo- late, 12-17 mm long, 2-4 mm wide. Corollas at anthesis 15-22 mm long, the tube 7-10 mm long, and the teeth 5-7 mm long, the shorter teeth about half the length of the limb, or eventually equalling it; lobes with hairs about 0.75-1 mm long. Anther sacs 4.5-5 mm long, the stamen tips cuneate-lanceolate, 1-1.5 mm long. Ovary at anthesis 8-10 mm long, 3-4 mm wide; styles 22-30 mm long, the stigmatic branches 0.8 mm long, barely separating at an- thesis. Mature achenes 10-20 mm long, 4-6 mm wide; the awns 10-13 mm long, generally rounded in transection, subpersistent. DISTRIBUTION: Society Islands, endemic to Tahaa. SPECIMEN EXAMINED: Tahaa, district of Ruutia, slopes of Mt. Ohiri, in rain-forest of Crossostylis, Alstonia, and Morinda, altitude 465 m, Jan. 25, 1931. M. L. Grant 5161 (bish, type; RSA, ISTC, isotypes). This subspecies was observed in three other nearby localities, all within an altitudinal range of 15 m, and about 70 m below the top of Ohiri, the highest point on the island. Other associated woody plants were Xylosma, Meryta, Metrosideros, Wikstroemia, F agar a, Astronia, and Hernandia. 3. Fitchia nutans Hook f., London Jour. Bot. 4:640, t.23-24. 1845. Since so few specimens of this species are available in the United States, and the several Kew and Paris sheets are without detailed habi- tat data, the citation of an additional collection may be of interest: Tahiti iti, district of Teahu- poo, on Mt. Ronui, in Metrosideros- Weinman- nia forest, altitude 890 m, July 2, 1930, M. L. Grant 3925 (BISH, istc). The following de- scription is from this collection alone. Treeshaped glabrous shrub, 2.5 m tall, with a basal diameter of 5 cm; bark 2 mm thick, brown, with longitudinal ridges about 1 mm deep; the wood very sweet-smelling. One-year twigs 2-3 mm in diameter; two-year twigs 3.5- 4.5 mm thick; internodes 7-10 mm long. Peti- oles 30-60 mm long; stipular sheaths 2-3 mm high. Blades ovate, yellowish-green, 60-130 mm long, 45-80 mm wide, the average of 12 leaves being 96 X 61 mm; truncate or cordate at the base, occasionally barely rounded, often somewhat oblique; the apex slightly acuminate; the margin irregularly and shallowly crenulate to entire; with 8-10 major lateral veins. Pedun- cles two in each of the four inflorescences pres- ent, 60-65 mm long, reflexed. Heads shattered with age, the involucral bracts and corollas hav- ing dropped. Receptacular bracts 13-14 mm long, 3-5 mm wide. Ripe achenes 9-10 mm long, 2-2.5 mm wide, with persistent awns 7-8 mm long. This collection fits the general description of the species (Carlquist, 1957: 63), except as follows, with Carlquist’s measurements and notes in parentheses: plant smaller (4. 5-7. 5 m tall), blades often crenulate (entire), maximum leaf size greater (115 X 70 mm), base often truncate to cordate (acute to obtuse), heads paired (solitary), receptacular bracts shorter (20-22 mm long), achenes much shorter (lb- 17 mm long at anthesis) and narrower (3-4 mm). These differences, however, do not seem significant enough to warrant separation, and the material matches Hooker’s plate closely. Although the type of the species was sup- posedly from "Elizabeth Island,’’ all the evi- dence (Carlquist, 1957: 63) suggests that it is confined to Tahiti. The present specimen comes FIGS. 7-12. Figs. 7, 8, 11, Fitchia cordata. Figs. 9, 10, 12, F. cuneata ssp. tahaaensis. Fig. 7: Leaf transec- tion, adaxial face above. Secretory canal is above vein at right. Fig. 8: Transection of awn from an achene, adaxial face at right. Note lignification of hypodermal layers, presence of a single vascular bundle. Fig. 9: Leaf transection, adaxial face above. In addition to secretory canals adjacent to vein, left, secretory cavities may be seen, center and right. Fig. 10: Transection of awn from an achene, adaxial face upper right. Tri- chomes in section, above and below. Fig. 11: Transection of pith. Secretory canal with adjacent fibrosclereids, above; inner margins of vascular cylinder, below; note thin-walled cells at top of figure. Fig. 12: Longitudinal section of pith. All cells have secondary walls, some such are thicker walled. Figs. 7, 9, X 145. Figs. 8, 10, X 255. Figs. 11, 12, X 135. Studies in Fitchia — CARLQUIST and GRANT from the smaller mountain (Tahiti-iti) of the two which make up the island. Hoffmann (1890: 353) reported F. nutans from Tabuai ( = Tubuai ) , as well as from Tahiti and Elizabeth (now Henderson) I., but as Hemsley (1885: 20) had earlier shown, this supposed locality is based on the type sheet (Cuming 1424) which Hooker had earlier at- tributed to Elizabeth Island. Erdtman (1952: 124) cites a specimen of F. nutans as "Fiji, Moseley, anno 1875.” Carl- quist (1957:63) suggested that this is in error, 287 because no other collections have been made of Fitchia in the Fiji group. Also, Hemsley ( 1885: 20 ) reports that there is at Kew a specimen collected by Moseley in Tahiti at an altitude of 4,000 ft, and this is probably the specimen cited by Erdtman. Papy (1955: 325) mentions F. nutans as growing on "Tahiti, Moorea, Raiatea, Rapa.” The record from Rapa is probably based on the report of Riley (1926: 55), which was made before F. rapensis had been described as a separate species (Brown, 1935: 366) , although FIGS. 13-16. Fig. 13: Fitchia cordata. Transection of involucral bract taken about midway along length of bract; adaxial face at left. Fig. 14: F. cordata. Transection receptacular bract, adaxial face at left. In Figs. 13, 14, tracheary elements shown in bold outline, fibers and sclereids stippled. Fig. 15: "Typical” leaf of F. cordata, showing main veins. Fig. 16: "Typical” leaf of F. cuneata ssp. tahaaensis. Figs. 13, 14, X 105. Figs. 15, 16, X 2/3. 288 PACIFIC SCIENCE, Vol. XVII, July 1963 Papy was aware of this. Neither Moore nor Grant found F. nutans on Raiatea, and Papy was aware of the description of F. cuneata from that island. Moorea is certainly a likely locality for F. nutans, but for none of these localities does Papy cite specimens. Thus, the geographic distribution offered by Carlquist (1957:2) must be broadened to in- clude the occurrence of F. nutans on Tahiti-iti, F. cordata on Bora Bora, and F. cuneata ssp. tahaaensis on Tahaa. 4. Fitchia rapensis F. Brown, Bishop Mus. Bull. 130:366. 1935. Since the ten previously recorded collections of this species are all without flowers (Carl- quist, 1957 : 62 ) , the description of a flowering specimen should be of interest: Rapa, moist zone, altitude 520 m, Jan. 3, 1922, W. B. Jones 372 ( Whitney Expedition) ( BKL ) . "Tree,” glabrous. Petioles 40-75 mm long. Blades ovate, the larger ones averaging 90 mm long and 84 mm wide, rounded or scarcely cuneate at the base, the apex acute; entire; with about 12 pairs of major lateral veins. The sin- gle inflorescence is terminal, with two heads, the peduncles 50 and 60 mm long, reflexed. Heads 45 mm long and 45 mm wide. Bracts in about 4 series, those of the second layer semi- circular, 10 mm long and 22 mm wide, and those of the inner layer 20 mm long and 12 mm wide. Receptacular bracts cuneate, 16-17 mm long, 3-4 mm wide. Mature corollas 24-26 mm long, the ventral cleft reaching to 6-8 mm of the base, the segments eventually separating to within 10 mm of the base; the tips of the lobes often densely covered with sclerified cells, which are up to 0.75 mm long. Anther sacs 5.5-6 mm long, caudate, the tail 0.3-0.4 mm long; the stamen tips cuneate, 1.5 mm long. Mature style ( broken off) at least 32 mm long. Achenes, at and just after anthesis, 10-12 mm long, 4 mm wide, the awns 12-15 mm long, strongly bent and twisted at the base. ANATOMICAL DESCRIPTIONS Anatomical studies were based on the her- barium specimens cited above. Liquid-preserved heads of F. cuneata ssp. tahaaensis and wood samples of both F. cuneata ssp. tahaaensis and F. cordata , for which the herbarium specimens serve as vouchers, were available. The methods of preparation of these materials are the same as those employed for such materials in the two papers (Carlquist, 1957, 1958) dealing with the anatomy of this genus. SECONDARY XYLEM: With respect to quali- tative features, the species of Fitchia other than F. speciosa are much alike in wood anatomy (Carlquist, 1958). Fitchia cuneata ssp. tahaaen- sis and F, cordata share the qualitative features of those species. The bands of apo tracheal parenchyma are clearly shown by the two new taxa ( Figs. 4, 5 ) . Thick-walled fibers are readily apparent In F. cuneata ssp. tahaaensis (Figs. 1, 3 ) ; libriform fibers are less thick-walled in F cordata (Figs. 5, 6). Sclerosed tyloses have been reported in F. nutans and F. speciosa (Carlquist, 1957, 1958); they are abundant in F. cuneata ssp. tahaaensis (Figs. 1, 2), although none were observed in F. cordata . Quantitative features of the two new taxa are as follows: F. cordata : vessels per group: 1.37; average vessel diameter: 61.5 /x; diameter of widest vessel: 88" "p; average length vessel elements: 325 ft; average length libriform fibers: 502 /a; average length apotracheal parenchyma cells: 356 g; average height multiseriate rays: 1.17 mm. Fitchia cuneata ssp. tahaaensis: ves- sels per group: 1.66; average vessel diameter: 62.8 /a; diameter of widest vessel: 92 /a; average length vessel elements: 326 /a; average length libriform fibers: 507 /a; average length apo- tracheal parenchyma cells: 380 /a; average height multiseriate rays: 1.51 mm. Quantitative as well as qualitative data point up a close similarity between F. cuneata ssp. tahaaensis and the typical F. cuneata. Quanti- tative and qualitative data for F cordata re- FIGS. 17-19. Fitchia cuneata ssp. tahaaensis. Fig. 17: Transection of involucral bract taken about midway along length; adaxial face below. Fig. 18: Transection of receptacular bract; ad axial face and adjacent achene at left. Fig. 19: Transection of style. At center of style, stigmatoid tissue. Exterior to each bundle is a secretory canal, except for bundle at upper left, which is flanked by a pair of canals. At right, anthers in transection. Note secretory canal in each of the two anthers. Fig. 17, X 100; Fig- 18, X 80; Fig. 19, X 120. 290 PACIFIC SCIENCE, Vol. XVII, July 1963 semble the figures given (Carlquist, 1958: 6-7) for F. cuneata and F. nutans. PITH: Pith anatomy proved important in dis- tinguishing species of Fitchia (Carlquist, 1957). Likewise, the two new taxa possess characteristics useful for taxonomic purposes. Pith of F. cordata (Fig. 11) consists of both thick-walled, lignified, and thin-walled, non- lignified, cells. Secretory canals are present; each is surrounded by an eccentric zone of fibro- sclereids. Although secretory canals surrounded by small lignified thick-walled cells occur in F. nutans and F. tahitensis, the presence of these combined with occurrence of thin-walled non- lignified cells in pith is characteristic only of F. cordata. Pith of F. cuneata ssp. tahaaensis (Fig. 12) lacks secretory canals and consists of thin-walled and thick-walled lignified cells. The only pith pattern which matches this is that described in the monograph for typical F. cuneata. LEAF: Leaf dimensions are mentioned in the taxonomic description above. However, aver- ages of dimensions of leaves from a collection were used in the monograph, and these may be compared with those of the new taxa. Mature leaves of the collection of F. cordata (Fig. 15) show an average lamina width of 52.5 mm, an average lamina length of 53 mm, and a petiole length of 49.7 mm. The only Fitchia which re- sembles F. cordata both in form and dimensions is the extreme population of F. rapensis from the summit of Rapa. Leaves of F. cordata have longer petioles than that collection, and there is certainly no close relationship between F. cordata and F. rapensis. Leaves of F. cuneata ssp. tahaaensis (Fig. 16) which could be termed mature show the following average dimensions: lamina width, 48.0 mm; lamina width, 87.5 mm; petiole length, 26 mm. When compared with the chart in the monograph (Carlquist, 1957: 50) the lamina dimensions fall in the vicinity of F . nutans and some collections of F. rapensis. The leaves are wider than those of F. tahitensis but longer than those of the typical F . cuneata. Thus, leaf shape distinguishes both F . cordata and F. cuneata ssp. tahaaensis. Histological features of leaves of the new taxa also distinguish them. Fitchia cordata pos- sesses fibers in bundle sheaths of many veins, but they are lacking in sheaths of finer veinlets. No secretory cavities are present in the meso- phyll, although secretory canals occur adaxially or abaxially to the veins, or both (Fig. 7). Mesophyll is about 10 layers thick. These char- acteristics are not matched by any other Fitchia collection. Presence of a few fibers would ally F. cordata to F. nutans or F. tahitensis, but these two species possess secretory cavities. Ab- sence of secretory cavities does characterize F. speciosa and the typical F. cuneata, but these taxa lack sclerenchyma in bundle sheaths. Leaves of F. cuneata ssp. tahaaensis (Fig. 9) lack sclerenchyma along all but the largest veins. In addition to the secretory canals above and below veins (Fig. 9, left), secretory cavi- ties are present in the mesophyll (Fig. 9, right) . The mesophyll is about 10 cells thick. This description agrees closely with that of typical F. cuneata except for the presence of secretory cavities. Although such a difference might con- ceivably arise from a difference in maturity of a plant, it seems more likely a valid difference, for specimens of both subspecies were of flow- ering age. INVOLUCRE: The heads of F. cordata are small, and the involucre of a living specimen would probably measure about 2 cm in diam- eter— the smallest involucre size in the genus except for F. tahitensis and F. cuneata. Involu- cral bracts of F. cordata (Fig. 13) have a thick- ness comparable to some bracts of F. nutans, but differ in their lack of the abundant scleren- chyma which characterizes bracts of F. nutans. The mature state of the bracts of F. cordata illustrated is certain, because they came from a head in fruit which was in the process of shattering. Occasional lignified thick -walled cells may be seen along the inner face of the bract and around the larger veins. In presence of sclerenchyma and other histological charac- teristics, the involucral bracts of F. cordata are midway between those of F. nutans and F. cu- neata. The receptacular bracts of F. cordata (Fig. 14) show similar characteristics. As might be expected, lignified thick-walled cells are more frequent than in the involucral bracts. Fitchia cuneata ssp. tahaaensis also has very narrow involucres, like those of the typical F. cuneata or F. tahitensis. Histologically, involu- Studies in Fitchia— CARLQUIST and Grant 291 FIGS. 20-24. Figs. 20, 23, Fitchia cordata . Figs. 21, 22, 24, F. cuneata ssp. tahaaensis. Figs. 20, 21: Tips of corolla lobes, showing exterior surface. Sclerenchymatous cells are shown in black. Figs. 22, 23 : Stamen tips, showing inner surface. Anther sacs shown in outline, below. Fig. 24: Base of anther and adjacent filament. Note caudate anther sac, right. All, X 65. 292 cral bracts of F. cuneata ssp. tahaaensis (Fig. 17) resemble closely the description given for typical F. cuneata in the monograph. This re- semblance is especially evident in the lack of sclerenchyma. Only an occasional lignified cell (usually near a vein) can be found. Collenchy- matic thickening may occur near outer or inner faces of the bract. In the section shown, more than a single series of bundles may be seen; some of these bundles have distorted orienta- tion of xylem and phloem, a condition men- tioned in the monograph for F. speciosa. This condition is probably occasional among larger bracts of several species. The thickness of the involucral bract, as well as its lack of scleren- chyma, identifies the two subspecies of F. cu- neata closely with each other. The receptacular bract (Fig. 18) of a head at anthesis certainly lacks sclereids, although a greater degree of lignification might be apparent in bracts of a head in fruit. ACHENES: Awn shape and anatomy proved useful characteristics in distinguishing species of Fitchia. In comparing illustrations of these ( Carlquist, 1957: plate 8) with these of the new taxa (Figs. 8, 10) this also proves to be true. In F. cordata (Fig. 8), the awns tend to be rounded in transectional shape, like those of F. nutans. However, they show a tendency to be wider than those of F. nutans. They have a single vascular bundle and, unlike at least some awns of F. nutans, they lack a secretory canal. Awns of F. cuneata ssp. tahaaensis (Fig. 10) are like those figured for the typical F. cuneata in all respects. Presence of three longitudinal zones of trichomes tends to make them triangular as seen in transectional view. Where such tri- chomes are absent, awns may have a more rounded shape. COROLLA: In the monograph, considerable attention was focused on the importance of venation patterns of Fitchia corollas, both for their phylogenetic importance within Compos i- tae and for their usefulness in distinguishing species. Corolla venation patterns are not illus- trated here for the new taxa because they con- tribute relatively few novel features and can be referred to patterns illustrated in the mono- graph. Although few flowers of F . cordata could be studied in this respect (on account of the PACIFIC SCIENCE, Vol. XVII, July 1963 ravages of the living plant by insects ) , the vena- tion pattern of this species is substantially like that figured for F. cuneata or F. tahitensis. There is some variation in the levels at which adja- cent lateral veins in each lobe join beneath sinuses, but differences in these levels are not nearly so extreme as the condition figured for F. nutans (Carlquist, 1957: 10). In their rela- tively short length (ca. 18 mm) the corollas of F. cordata approximate those of F. tahitensis also. Corollas of F. cuneata ssp. tahaaensis have similar size and venation features. Typically, the patterns are the same as mentioned for F. cordata. However, occasionally a pair of adjacent laterals do not join beneath a sinus, but con- tinue into the tube and achene without joining. This condition does occur occasionally in the other species, such as F. speciosa. A feature of interest in the corolla of F. cuneata ssp. taha- aensis is the presence of one or two supernu- merary veins in each lobe. In addition to the median and lateral veins, these fine supernu- merary veins may be present for a short distance near the base of the lobe. Such veins were not seen in the typical F. cuneata, for which, how- ever, mature corollas were not available. Super- numerary veins are occasional in corollas of F. speciosa, and characteristically abundant in those of F. mangarevensis. The occurrence of larger numbers of veins in corollas of F. mangarevensis is particularly interesting, since that species is close to F. rapensis. Corollas, now available, of F. rapensis confirm this relationship in the similarly elabo- rate venation pattern (Fig. 26). This similarity is shown in the presence of five veins per lobe, rather than three. Such a condition is basic in the construction of the F. rapensis corolla, al- though additional veins or vein fragments may be present. The outermost veins of each lobe fuse at the tips of many corolla lobes. The co- rolla venation of F. mangarevensis (Carlquist, 1957:14) is somewhat more complicated than this, because although the five-vein condition may be observed in some lobes in that species, additional veins are more frequent. The pres- ence of complex corolla venation in both F. rapensis and F. mangarevensis is interesting in that it raises the question of how complex Studies in Fitchia— CARLQUIST and GRANT 293 venation may have been present ancestrally in Fitchia . Is the more complex condition primitive or a specialization? The question cannot be answered in terms of the data available, and cytological information would be very desirable as a line of supporting evidence. Distribution of thin-walled and thick-walled trichomes and occurrence of sclereids on corolla- lobe tips have proved useful features in the systematics of Fitchia. In the two new taxa these features are also of interest. Fitchia cordata ( Fig. 21) possesses lobe tips not identical to those of any species previously figured. The frequency of sclereids at the lobe tip is notable. Long multiseriate trichomes, all the cells of which are sclereids, are present. No trichomes composed of nonlignified cells were observed. This condition is most closely matched in the genus only in the new subspecies of F. cuneata (Fig. 20). In F. cuneata ssp. tahaaensis sclereids are rare or absent at the lobe tips. The trichomes are long, multiseriate (rarely unisef. fate ) , and composed wholly of sclereids. This condition is not unlike the condition in typical F. cuneata, except that the trichomes in the Tahaa plants are few and long. The corolla-lobe tips of F. rapensis (Fig. 25) are, as might be expected, rather similar to those of F. mangarevensis (Carlquist, 1957: 27 ) . They are different in that in F. rapensis the sclerified trichomes are slightly longer, in gen- eral, and extend farther down the lobe. Oc- casional thin-walled hairs may be seen on the terminal portion of lobes of F. rapensis, but the extremely dense coating of thick-walled lig- nified trichomes clearly marks this pattern as closest to that of F. mangarevensis. STAMENS: Stamen tips likewise offer taxo- nomic criteria. The long stamen tips of F. cor- data (Fig. 23) exceed those of F . nutans in size, and are thus the longest of the Society Islands species of Fitchia. The stamen tips of F. cuneata ssp. tahaaensis (Fig. 22) are longer than those of the typical F. cuneata, but do not differ mark- edly from those of F. tahitensis or F. nutans. The stamen tips of F. rapensis (Fig. 27) are 1.5 mm long, thus matching those of F. man- garevensis. This is particularly interesting be- cause the two species are so much alike in this respect and because they have longer stamen tips than the other species of the genus, with the exception of F. speciosa. A feature of some interest not previously observed is the occurrence of somewhat caudate anther sacs in F. cuneata ssp. tahaaensis (Fig. 24) and the other taxa of Fitchia. In view of Cronquist’s (1955) dictum that anthers of Com- positae are primitively tailless, this would seem curious, because Fitchia possesses so many ana- tomical and morphological characteristics which appear primitive for the family. Cronquist’s dictum, for which no appreciable evidence is adduced, seems highly questionable. In transection the anthers of F. cuneata ssp. tahaaensis (Fig. 19) exhibit a feature of interest. One or two secretory canals may be observed in the connective of each anther. This condition occurs sparingly in F. speciosa, but was not observed in other species of Fitchia. The oc- currence of anther secretory canals has been noted in other genera which may be related to Fitchia, such as P etrohium and Oparanthus (Carlquist, 1957). STYLE: Like the anthers, the style of F. cuneata ssp. tahaaensis (Fig. 19) seems to exhibit more numerous secretory canals than those of other species of Fitchia. One canal (or occasionally a pair) is present exterior to each of the four style bundles. Liquid-preserved material is re- quired for accurate demonstration of this phe- nomenon. In the only other species for which such material was available, F. speciosa , canals were much less abundant in styles (Carlquist, 1957 ) . However, as indicated in the monograph, many other Heliantheae do show abundance of stylar secretory canals. pollen GRAINS: The drawings of pollen grains of Fitchia (Figs. 28-30) show some dif- ferences in representation compared with those of the monograph. Although the outer sculp- tured layer (ectosexine) is composed of fine rods and spaces, as illustrated earlier, a lacunose inner sculptured layer (endosexine) has now been observed. The endosexine consists of rods interspersed in large spaces. In addition, an inner and outer layer of nexine (below in each figure) may be distinguished. Size and spine shape are of especial interest, however, in com- parison of the species. The markedly blunt spines of F. cuneata ssp. tahaaensis (Fig. 29) 294 PACIFIC SCIENCE, Vol. XVII, July 1963 Figs. 25-27. Fitchia rapensis. Fig. 25: Tip of corolla lobe, showing exterior surface; sclerenchymatous cells in black. Fig. 26: Corolla, showing venation. Fig. 27: Stamen tip, showing inner surface. Figs. 25, 27, X 65. Fig. 26, X 8. Studies in Fitcbia—C ARLQUIST and GRANT match those of the typical F. cuneata ( Carlquist, 1957:32). In F. cordata (Fig. 28) the spines ate even more markedly blunt. Pollen grains of F. cordata (range, 39-51 u; average, 42 fx) are the smallest in the genus. The pollen grains of F. cuneata spp. tahaaensis (range, 45-51 fi; average, 48 /*) closely match those of ssp. cu- neata in size. The slightly smaller size of grains in the latter may be caused by the fact that flowers of F. cuneata ssp. cuneata from which grains were taken were not quite mature. Pollen grains of F. rapensis (Fig. 30) average 60 u in diameter, and have a range from 51 to 66 /x. They are thus slightly larger than those of F. 295 mangarevensis, although the difference is not significant. The spines of pollen grains of F. rapensis have an inverted-funnelform shape, like those of F. mangarevensis, but appear to be more sharply pointed. In addition, lacunae in spine tips were observed for the first time in the genus in F. rapensis. This feature, however, is not unexpected in Fitchia, since other Helian- theae have lacunae in spine tips (Carlquist, 1957:33). SPECIES CHARACTERISTICS: In the mono- graph three species groups were recognized: ( 1 ) F. speciosa; (2) F. nutans, F. tahitensis, and F. cuneata; and (3) F. mangarevensis and Figs. 28-30. Optical sections of equator of pollen grains, showing a third of each grain. Germ pores indi- cated at left and right in each. Fig. 28, F. cordata. Fig. 29, F. cuneata ssp. tahaaensis. Fig. 30, F. rapensis. 296 PACIFIC SCIENCE, Vol. XVII, July 1963 Figs. 31-32. Type specimens of Fitchia. Fig. 31, F.cordata. Fig. 32, F. cuneata ssp. tahaaensis. Studies in Fitchia — Carlquist and Grant F. rapensis. The new taxa clearly belong to the second group. The present study also emphasizes the distinctness of the third group and the close relationship between the two species it contains. Fitchia cuneata ssp. tahaaensis agrees with typical F. cuneata in most respects. Those by which it differs include: presence of sclerified tyloses in wood (not yet observed in wood of typical F. cuneata ) ; leaves with wider and longer laminae; secretory cavities present in leaf meso- phyll; involucral bracts thicker; sclerified tri- chomes on corolla-lobe tips longer and fewer; anther tips longer. These differences are rela- tively minor. Close identity of the two sub- species is revealed by such critical features as general leaf shape, pith structure, bract anatomy, corolla vascularization, awn shape and anatomy, and pollen size and ornamentation. The two taxa are therefore best regarded as subspecifically different. This situation is precisely what one would expect on account of the closeness of Tahaa to Raiatea, sister-islands (within a com- mon fringing reef). The nearest island to the sister-islands of Raiatea is Bora Bora. Qne might expect, there- fore, a similarity between F. cuneata and F. cordata. These two do appear similar in their small heads, relative lack of sclerenchyma in involucral bracts, venation and size of corollas, lack of secretory cavities in leaves (excepting ssp. tahaaensis ) and blunt shape of spines on pollen grains. They are dissimilar in that F. cordata possesses secretory canals and thin- walled nonlignified cells in the pith, has cordate leaves, and some sclerenchyma along veins of leaves. These characteristics are reminiscent of F. nu- tans. From F. nutans , however, F. cordata differs by such respects as those in which it resembles F. cuneata. Features of F. cordata which do not match those of any species in the F. nutans-F. cuneata-F. tahitensis group include the presence of nonlignified cells in pith, the small leaf size and cordate shape combined with relatively long petiole length, the presence of both long mul- tiseriate sclerified trichomes and sclerified epi- dermal cells on corolla-lobe tips, long stamen tips, small pollen size, and blunt spines on pollen grains. These characteristics, combined with the distinction furnished by its geographi- cal isolation on Bora Bora, mark F. cordata as worthy of recognition as a new species. 29 7 PHYLOGENETIC CONSIDERATIONS: The new taxa described here, although they contribute no anomalous features to the genus, do enlarge our picture of it and permit more detailed ob- servations in this regard. The monograph of the genus provided some notions on the relation- ships of Fitchia. These ideas do not seem to require alteration. The new taxa do enhance the diversity of the Society Islands Fitchias. Indeed, Papy ( 1955 ) notes that the Society Islands form a center of diversity for Fitchia, and that the variability of Fitchia on Tahiti is suggestive of this. The latter phenomenon remains a problem worthy of study when more material from Tahiti becomes available. Postulation of phylogenetic trends within each species group seems inadvisable, both because of the limited quantity of material known and the fact that the species seem to be variations on a basic theme rather than stages along a phylogenetic pathway. However, some interpretations of relationship among the species groups may be offered here. The larger size — in all parts of the plant — of F. speciosa, as compared with the remaining species, seems specialized. The larger pollen size suggests the possibility of a greater chromosome number, and the morphology of the awn base seems clearly a specialized feature. The fact that F. speciosa has such an isolated geographical position, the westernmost species of Fitchia, in contrast with the likelihood of an American ancestry of this helianthoid (e.g., Brown, 1935; Papy, 1955), reinforces this supposition. The extremely large seeds of F. speciosa seem less adapted to long-distance dispersal than the smaller seeds of the Society Islands species. The most likely interpretation, seemingly, is that seed size, and the peculiar adnation of achene summit to awn base in F. speciosa are charac- teristics acquired subsequent to arrival of its ancestors in Rarotonga. Fitchia speciosa may be regarded as a highly distinct derivative of the Society Islands stock. Although corollas of F. speciosa are the largest in the genus, their vena- tion is like that of the Society Islands species, not that of F. rapensis and F. man gar even sis, which have complex corolla venation. If pollen-grain size is an indication of chro- mosome number, then one would expect the Society Islands species to have the lowest num- 298 PACIFIC SCIENCE, Vol XVII, July 1963 ber, and progressively higher chromosome num- bers would be indicated for the F. rapensis-F. mangarevensis group and F. speciosa. If this were true, and it were a criterion of phylogeny within the genus, the complex floral venation of F. rap en sis and F. mangarevensis would be a specialization. Information on comparative cytology within the genus, in any case, is highly desirable. REFERENCES Brown, F. B. H. 1933. Flora of southeastern Polynesia, III. Dicotyledons. Bishop Mus. Bull. 130:1-386. Carlquist, S. 1957. The genus Fitchia (Com- positae). Univ. Calif. Publ. Bot. 29:1-144. 1958. Wood anatomy of Heliantheae (Compositae). Trop. Woods 108:1-30. Cronquist, A. 1955. Phylogeny and taxonomy of the Compositae. Amer. Midi. Nat. 53:478- 511. Erdtman, G. 1952. Pollen morphology and plant taxonomy. Chronica Botanica, Waltham, Mass. Hemsley, W. B. 1885. Report on present state of knowledge of various insular floras. Bot., Challenger 1 ( 1 ) : 1-75. Hoffmann, O. 1890. Compositae. In: Engler and Prantl, Die natiirlichen Pflanzenfamilien 4(5) : 87-391. Papy, H. R. 1955. Tahiti et les lies voisines. La vegetation des lies de la Societe et de Makatea. 2e partie. Trav. Lab. Forest. Toulouse 5(2/1: 3): 162-386. Riley, L. A. M. 1926. Notes on the flora of Rapa. Kew Bull. 1926:51-56. Lithoglyptes hirsutus (Cirripedia: Acrothoracica), A New Burrowing Barnacle from Hawaii Jack T. Tomlinson1 Two samples of coral from Kaneohe Bay, Oahu, Hawaii, have each yielded a number of specimens of a new species of acrothoracican burrowing barnacle of the family Lithoglypti- dae. Samples of Psammocora v err mm Vaughan collected by Stephen A. Wainwright,2 and of Porites compressa Dana collected by Charles Stasek,3 were referred to me by William A, Newman.3 These barnacles are the first repre- sentatives of the order Acrothoracica known from Hawaii. Family LITHOGLYPTIDAE Aurivillius 1892 Lithoglyptidae emend. Tomlinson and New- man I960. Mouth cirri well developed, on a 2-jointed pedicle; 4-5 pairs of terminal cirri, but if only 4 pairs, caudal appendage present; no gut teeth or gizzard in digestive tract; with adhesive disc on mantle; lateral bar absent; burrows in coral or mollusc hard parts. Lithoglyptes Aurivillius 1892 {emend.) Four pairs of terminal cirri on a 2-jointed pedicle with oblique sutures at first joints; cau- dal appendage present; mouth cirrus with 2 rami of 3 to 6 articles (5 to 6 in original description ) . Lithoglyptes hirsutus n. sp. DIAGNOSIS: Female (Figs. 1-7): Mantle aperture strongly arched on both sides of an acute projecting tip which is not a movable 1 Biology Department, San Francisco State College, San Francisco 27, California. Manuscript received December 19, 1961. 2 Zoology Department, University of California, Berkeley 4, California. 3 Scripps Institution of Oceanography, La Jolla, California. spine or hook; aperture length exceeds Vi of mantle width, aperture armed with numerous teeth and long flexible hairs, especially on the outer edge of the lip area; anterior and pos- terior rami of mouth cirri with 5 and 3 articles, respectively; caudal appendage with 2 seg- ments; head with acute projection opposite mouth area; burrow pointed oval in surface view. Holotype 1.2 X 0-67 mm; about 30 dried specimens in Psammocora verrilli from a depth of 3-6 ft on Sand Bar Reef and in Porites com- pressa from NE side Checker Reef, Kaneohe Bay, Oahu, Hawaii. The species is named for the presence of numerous hairs on the mantle aperture. TYPE MATERIAL: Holotype USNM 107544. Paratype material: San Francisco State College, San Francisco, California; California Academy of Sciences, San Francisco, California; Plymouth Laboratory, England; Seto Marine Biological Laboratory, Japan; Portobello Marine Station, New Zealand. DESCRIPTION: Female: The mantles of 16 adults averaged 1.95 mm in height x 1.18 mm in width (maximum 2.1 mm and 1.6 mm), and are flattened laterally. The mantle is studded with numerous small T-shaped teeth, short spinules, and three- or four-pointed star-shaped teeth, all scattered about on the surface, but particularly near the aperture. The mantle at- tachment disc adheres strongly to the burrow, but is readily removed if the coral is decalcified. The average length of the slitlike aperture of 12 specimens is 0.97 mm. Near the anterior end of the aperture two acute projections ex- tend to a point, but are not movable or hooked. The outer edges of the thickened apertural area are heavily set with teeth and long hairs. The inner edges have fewer teeth and hairs, but have a row of comblike projections which extend into the aperture and tend to occlude it. This structure has been called the "comb- collar” by previous workers. 299 FIGS. 1—7. 1, Lithoglyptes hirsutus, n. sp., female, side view of adult removed from burrow; abbreviations: AC, anal cirrus; AK, attachment knob; CC, comb collar; H, head; L1M, lip of mantle aperture, inner margin; LOM, lip, outer margin; MC, mouth cirrus; AIP, mouth parts; TC, terminal cirri; TO, tooth. 2-4, Mouth parts of Lithoglyptes hirsutus, n. sp., female; 2. mandible with palp; 3, first maxilla; 4, second maxilla. 5-7, Lithoglyptes hirsutus, n. sp., female; 5, middle portion of fourth (last) terminal cirrus, side view; 6, burrow in coral, surface view; 7, mouth cirrus, slightly foreshortened in comparison with some specimens. Lithoglytes hirsutus — Tomlinson 301 The mouth parts (Figs. 2-4) are typical for the genus, being composed of a pair of man- dibles with palps and two pairs of maxillae. Each mandible has four strong teeth, the inner one bearing short bristles. The mandibular palp exceeds the mandible in length; with long, soft bristles on the edges near the tapering point. Each first maxilla is armed with two strong teeth, numerous bristles, short teeth along the cutting edge, and has the usual apodeme. The second pair of maxillae are large and soft, and have numerous flexible bristles distributed along their edges. These close-set appendages serve as posterior limits to the mouth field. The mouth cirri ( Fig. 7 ) have a two-jointed pedicle upon which the two bristle-bearing rami articulate. They arise below the mouth parts but terminate above them, frequently curving outward at the tips. The anterior ramus has five segments and the posterior ramus has three segments. The segments of the rami are equipped with numerous bristles, which make the determination of the number of segments very difficult. There are four pairs of biramous, multiseg- mented terminal cirri, on two-segmented pedi- cles. The slanted articulation between these pedicular segments is characteristic of the genus. The rami are armed with single long setae on the outer edges of every second to sixth artic- ulation. The setae on the inner edge of the terminal cirri alternate between pairs of long setae at the articulation and pairs of shorter setae toward the middle of the segment (Fig. 5). A pair of two-jointed uniramous caudal ap- pendages occurs at the posterior end of the body. The total length of the caudal appendage is about Vi the length of the pedicle of the posterior terminal cirrus. The burrow of the female appears in the sur- face view as a pointed oval (Fig. 6). It is deeper than it is long. The walls are nearly perpendicular to the surface, and rounded at the bottom. The depth of the burrow appears to be slightly more than the greatest diameter of the animal in dried specimens. It is assumed that the living animal closes the burrow open- ing with the hardened, hairy apertural area of the mantle. MALE: Dwarf males resembling cyprid lar- vae were seen, but being dried they were much distorted, and whether or not they were truly mature stages could not be determined. Of the 11 males studied, none had the very reduced baglike shape of many acrothoracican males, nor the attenuated stalk of the male of Lit ho - glyptes spinatus. They were of a size and shape one would expect of a cyprid larva which had lost its bivalved carapace. The presence of a penis could not be ascertained. L. hirsutus is the first member of the genus to be found in the central Pacific. Of the other species in the genus L. indicus Aurivillius, L. ampulla Aur., and L. hicornis Aur. are from the Java Sea and Indian Ocean, and L. spinatus Tomlinson and Newman is from Jamaica. L. hirsutus can be distinguished from L. bi- cornis, L. ampulla , and L. spinatus by the ab- sence of the apertural hooks and/ or spines borne by the latter three species. This new species may be distinguished from L. indicus, which also lacks the hooks and spines, by the presence of the "hairs” on the mantle aperture and by its small size (2 X 1 mm rather than 6X^ mm). REFERENCES Aurivillius, C. W. S. 1892. Neue Cirripeden aus dem Atlantischen, Indischen und Stillen Ocean. Ofversigt af Kongl. Vet.-Akad. Forh. Stockholm 49 (3): 133-134. Tomlinson, Jack T., and William A. New- man. I960. Litho glyptes spinatus, a burrow- ing barnacle from Jamaica. USNM Proc. 112 (3443): 517-526. Toxicity of Dialyzed Extracts of Some California Anemones (Coelenterata)1 Edgar J. Martin2 ABSTRACT: Anemones of the California coast, collected from November I960 to March 1961, were compared with respect to their toxicity. Aqueous extracts of the anemones were dialyzed and assayed by intraperitoneal injection to mice, and the survival times of the injected mice were compared. The extracts of Anthopleura elegantissima and A. xantho gram-mica were by far more toxic than those of Metridium senile, Corynactis calif ornica, Tealia crassicornis, T. lofoiensis, and T. coriacea. The differences in potencies of the extracts may indicate species variations in ( 1 ) toxicity, (2) efficiency of the extraction method, or ( 3 ) both. The AQUEOUS extracts of many coelenterates are poisonous (Dodge, 1957). The poisons are mixtures of biologically active compounds ( Cre- scitelli and Geissman, 1962), some of them amines (Welsh and Prock, 1958), others heat- labile colloids (Martin, I960; Mathias et al., I960). The poison occurs in both the nema- tocysts and the nematocyst-free structures of the anemones (Lane, I960; Martin, to be pub- lished ) . Both the chemical composition and the biologically active substances of extracts vary from one species of anemone to another (Ma- thias et al., I960; Bergmann et al., 1956). The present study was designed to compare the lethality in mice of dialyzed extracts of some anemones of the California coast. MATERIAL AND METHODS SPECIES AND SITES OF COLLECTION: Ane- mones were collected at various sites on the California coast from November I960 to March 1961: Anthopleura elegantissima from Bodega Bay, Tomales Bay, Monterey, and Corona del Mar; A. xanthogrammica from Tomales Bay and 1 Supported in part by National Institutes of Health Grant RG 7626. Manuscript received February 19, 1962. 2 Laboratory of Comparative Biology, Kaiser Foun- dation Research Institute, Richmond, California; and Pacific Marine Station, University of the Pacific, Dil- lon Beach, California. Monterey; Metridium senile from Monterey; Corynactis calif ornica, Tealia lofotensis and T. crassicornis from Bodega Bay, all intertidal; and T. coriacea off Marineland from a 30-ft depth. In addition, the intertidal Nudibranch Diaulula sandiegensis and the Mollusca Crassostrea gigas and Mytilus calif ornianus were collected at To- males Bay. Anthopleura elegantissima is com- mon all along the California coast. Its extracts were used as the base of comparison with those of the other species available. The anemones were placed in the laboratory in aerated aquaria, with sea water which was re- newed twice a week. Water temperature varied between 13 and 20 C. The anemones were ex- posed to the light of the laboratory. They re- ceived no food supplement. Specimens which did not show normal vitality in the aquaria were discarded. After 2-4 weeks the anemones were removed from the aquaria and cleaned from ad- herent objects. During this manipulation they contracted firmly. Their body wall was then punctured to empty pockets of trapped water. Then the anemones were rinsed with distilled water for 10-15 sec, blotted with mild pres- sure, weighed in air, and their volume deter- mined. The specimens weighed from 0.3 to 34.2 g. EXTRACTION: The anemones were homoge- nized with three times their volume of distilled water at high speed in a Waring blender for 302 Toxicity of Anemone Extracts — Martin 303 two 1-min periods. After being centrifuged for 30 min at 1000 g the supernatant was decanted and centrifuged again. The resulting superna- tant extract was turbid but did not contain cells or debris. It was either decanted or, when it was found covered with a lipid layer, removed with syringe and needle; then its pH was deter- mined. The extract was dialyzed for 4 hr through commercial cellophane tubing against ten times its volume of 0.001 M sodium phosphate (pH 6.2) in 0.8% sodium chloride with 1% acti- vated charcoal (Weinberger, 1936). The dial- ysis was repeated once with fresh solution. All operations were performed at room temperature. Extracts of the other lower metazoa were pre- pared in the same manner. All extracts were as- sayed immediately after their preparation. The potency of extracts of each species was com- pared with that of an extract of A. elegantissima prepared on the same day. ASSAY: The extract was given intraperi- toneally to DAL-Swiss albino mice weighing 17 to 23 g. The mice were placed in cages in lots of four to eight (Russell et al., I960), and observed for 24 hr; their survival time was recorded. The abdominal cavities of the mice were inspected post mortem and mice with intraperitoneal hemorrhage were discarded. This condition was found in animals that showed severe pain reaction immediately after the in- jection and died within 4 min. The discarded animals were replaced by supplementary ones to complete the series. Each extract was as- sayed on a series of at least four mice. EXPERIMENTS AND RESULTS Preliminary experiments explored the effect of varying doses of extracts of A. elegantissima on the survival time of the injected mice. Only the area of LD<)<* was considered. A dose effect curve showed that with decreasing doses the survival time increased. At the dose of 20 cc of extract per kilo of mouse, both the range and the mean of the survival time were within practical limits; therefore, this dose was se- lected as a convenient reference. The pH of all the extracts varied from 6.0 to 6.6. At a dose of 20 cc of extract of A . elegantis- sima per kilo of mouse, the mean survival time of the injected mice varied from 8 to 36 min among the ten series tested. Between some of these series the difference in survival time was significant at the 2% level by the Chi-square test. The cause of these differences is unknown. The differences showed no correlation with vari- ations in the size, sites of origin, or dates of collection of the specimens, or with the pH of the extracts. They were not relevant for the present study. Table 1 shows that the potency of extracts may vary from one species of anemone to another. The highest potency, as estimated by the survival time of mice after intraperitoneal injection, was found for the extracts of A. ele- gantissima and A. xanthogrammica. Extracts of other anemones were far less potent even when their dose was tripled. With extracts of T. lofo- TABLE 1 Comparison of the Survival Time of Mice after Intraperitoneal Injection of Extracts of Anemones SPECIES NO. OF BATCHES NO. OF MICE DOSES CC OF extract/kg OF MOUSE SURVIVj M Mean \L TIME IN INUTES Range PROPORTION OF MICE SURVIVING 24 HR Anthopleura elegantissima 10 48 20 15.7 5-47 1/48 A. xanthogrammica 2 10 20 9.2 6-15 0/10 Metridium senile 1 6 20 174.1 80-240 0/6 Corynactis calif or nica 1 5 60 138.5 118-180 0/5 Tealia crassicornis 1 6 60 342.8 252-480 0/6 T. lofotensis 1 4 20 336 3/4 T. lofotensis 1 4 60 460-540 2/4 T. coriacea 1 4 20 264-640 2/4 T. coriacea 1 4 60 480-900 2/4 304 PACIFIC SCIENCE, Vol. XVII, July 1963 tensis and T. coriacea it was not possible to obtain an LD99. The extracts of the Nudibranch Diaulula sandiegensis and the Mollusca Crassoslrea gigas and Mytilus calif ornianus, each assayed on six mice at the dose of 60 cc extract per kilo of mouse, caused transitory depression of activity of the injected mice, but not death. The extracts of all anemone species lost their toxicity when they were heated at 90 C for 20 min. DISCUSSION The storing of the anemones alive for 2 weeks or longer before preparing the extracts was intended to minimize the amount of pos- sibly toxic products from food ingested by them. The dialysis of the extracts was aimed at minimizing their content of biologically active amines. The nature of the non-dialyzable tox- ins is not known. They may be proteins similar to those demonstrated by various authors in other coelenterates ( Mathias et ah, I960; Far- ber et al., 1961). We found information concerning the chem- ical components of the anemones here investi- gated only in the literature on A. elegantissima (Bergmann and Landowne, 1958). In general the lipid content of coelenterates Is high, and the proportions of various lipids and soaps and their chemical nature vary from one species to another (Bergmann et al, 1956; Bergmann and Landowne, 1958). The solubility of pro- teins is affected by lipids, and if it is assumed that the poisons here studied are proteins, it can be speculated that the solubility of these poi- sons may vary from one species to another. The data of Table 1 suggest that the toxicity of anemones may vary from species to species. This interpretation is supported by the great difference between species with respect to po- tency of extracts, which was observed even when the dose of the weaker extracts was tripled. But it is also possible that the efficiency of the extracting method, and hence the poison content of the extracts, may vary from species to species. This consideration applies equally to our study and to other authors’ reports of species variation in toxic compounds. These causes for variation may coexist. Lethal effects on mice were observed with extracts of the anemones but not with extracts of the Nudibranch and the two bivalves. This suggests that toxicity Is not a property of all littoral lower metazoa. REFERENCES Bergmann, W., S. M. Creighton, and W. M Stokes. 1956. Marine products. XL. Waxes and triglycerides of sea anemones, j. Org. Chem. 21:721-728. — — — and R. A. Landowne. 1958. Marine products, XLVI. Phospholipids of a sea anem- one. J. Org. Chem. 23:1241-1245. Crescitelli, F., and T. A. Geissman. 1962. Invertebrate pharmacology: Selected topics. Ann. Rev. Pharmacol. 2:143-192. Dodge, E. 1957. A study of the contents of the nematocysts of Metridium senile (L.), Thesis. University of Washington, Seattle, Wash. 63 pp. Farber, L., and P. A. Lerke. Preliminary ob- servations on the toxic property of the sea anemone, Rbodactis howesii. In: Abstracts of Symposium papers, 10th Pacif. Sci. Cong., Honolulu, p. 447. Lane, C. E, 1961. Pbysalia nematocysts and their toxin. In: H. M. Lenhoff and W. F. Loomis, eds., The Biology of Hydra and of Some Other Coelenterates. U. Miami Press, Coral Gables. Pp. 169-178. Martin, E. J. I960. Observations on the toxic sea anemone, Rbodactis howesii. Pacif. Sci. 14:403-407. Mathias, A. P., D. M. Ross, and M. Schach- TER. I960. The distribution of 5 -hydroxy ~ tryptamine, tetramethyl ammonium, homa- rine and other substances in sea anemones. J. Physiol. 151:296-311. Russell, F. E., J. A. Emery, and B. G. Bow- ers. I960. A comparison of mouse and chick in toxicity determination. Toxicol. AppL Pharmacol. 2:558-563. Weinberger, E. 1936. Improvement on dial- ysis. Brit. Patent No. 452,000. Welsh, John J., and P. G. Frock. 1958. Qua- ternary ammonium bases in the coelenter- ates. Biol. Bull. 115:551-561. Ecological Observations on the Sea Cucumbers Holothuria atra and H. leucospilota at Rongelap Atoll, Marshall Islands Kelshaw Bonham and Edward E. Held1 In a distributional study of the black sea cu- cumbers, Holothuria atra (Jager) and H. leu- cospilota ( Brandt ) , at Rongelap Atoll ( Fig. 1 ) the beaches of most of the larger islets were surveyed during September 1959, March 1961, and September 1961. Estimates of abundance and size were made, and observations on tem- perature tolerance, predation, asexual reproduc- tion by fission, and some other ecological aspects were considered. These two species of sea cucumbers are fre- quently, although not always, found together in the intertidal area of the seaward sides of the islets at the edge of the reef-flat nearest shore. They occupy pools of water from a few centi- meters to about one-half meter deep at low tide (Figs. 2, 3), or are occasionally entirely exposed. H. atra usually covers itself with a coating of light coral sand held in place by tube feet, al- though occasionally it is naked and black. The integument is thick and firm. This species lacks the Cuvierian organ and thus may be handled without the annoyance of adhesive threads. H. leucospilota is always naked and almost black, has a softer integument with fine protuberances giving it a prickly appearance. The intricate, microscopic calcareous deposits in the integu- ment differ markedly in the two species: in H. atra (Fig. 4 A) the "tables” are tall and there are no "buttons,” while in H. leucospilota (Fig. 4B ) the tables are short and squat, and there are many buttons. H. atra also occurs in deeper water in the lagoon, where it does not coat itself with sand, and to an undetermined extent on the vast areas of reef-flat between islets. At high tide the water at the seaward shores of the islets usually covers both species to a depth of I-IV2 m, and they 1 University of Washington, Laboratory of Radia- tion Biology, Fisheries Center, Seattle 5, Washington. Manuscript received December 20, 1961. are then frequently subjected to wave action from the surf whose initial force has been broken by the outer reef. A conspicuous feature of this habitat is the churning agitation and aeration of the water. H. atra appears to main- tain its position by inhabiting depressions and holding onto the relatively smooth reef-flat by means of its tube feet. H. leucospilota usually anchors the posterior portion of its body under- neath a stable rock or in a hole in the reef floor. In the absence of other cover it may find its way under a cluster of H. atra (Fig. 3). At appro- priate times it extends its highly mobile and extensible anterior end outward for feeding. Concentrations of these two species almost always occur where large slabs of beachrock are situated on the lower edge of the sandy beach slope bordering the inner edge of the reef-flat, and where, after the recession of the tide, the sea water may be observed to issue from the sand onto the reef-flat below the slabs of rock until the next tide. Below a sandy beach without slabs of rock no such prolonged runoff is evi- dent, and the sea cucumbers usually are few or lacking. It is in the pools fed by the relatively cool water from underneath the rock slabs that H. atra and H. leucospilota thrive. Concentra- tions of sea cucumbers were occasionally seen in the absence of slab-rock, and rarely slab-rock was found without sea cucumbers. They occupy an ecological niche almost devoid of other macroscopic organisms; occasionally small snails occur, but frequently the sea cucumbers are alone. There is no obvious food in the sand they ingest. CENSUSING OF SEA CUCUMBERS The surveying procedure consisted of observ- ing, photographing, and taking notes on the abundance of Holothuria while walking around 305 306 PACIFIC SCIENCE, Vol. XVII, July 1963 the periphery of 16 islets completely, and 5 others partially, thus covering 67% of the sea- ward shores of all islets in the atoll. The islets are numbered clockwise on the map of Rongelap Atoll shown in Figure 1 starting at the north- west extremity. Only about 2 % of the inter-islet reef area and shallower lagoon bottom regions supposedly suitable for H. atra was censused. Distances were measured by chain on the seaward shores of four islets, Aniiru, Kabelle, Namoen, and Rongelap, while at other islets distances were either determined by pacing, or, in most cases, estimated. On the east rim of the atoll high-altitude vertical, and elsewhere low- altitude oblique, aerial photos were employed to aid in orientation. Photographs taken on the reef-flat of undisturbed sea cucumbers were sometimes used in the estimations of abundance. The places of concealment under overhanging ledges were inspected, but rocks were seldom lifted to check for specimens. Table 1 shows the distribution of H. atra and H. leucospilota at the localities surveyed. Section numbers representing 100-m units of seaward shoreline appear under the appropriate fre- quency group. The central frequency groups in- crease by a factor of 10 while smaller ranges of numbers delimit the extremes. Along with the range in the column headings are the means in parentheses— geometric except for the first, where the zero limit precludes use of the geo- metric mean. Geometric and arithmetic means were tested for agreement with totals of the last two columns showing sums by localities taken directly from a previous tabulation, which also supplied the data for the body of Table 1. Sea Cucumbers at Rongelap Atoll— Bonham and Helb 307 TABLE 1 Distribution of H. atra and H. leucospilota on the Seaward Shores of Islets by 100-m Section Numbers (The body of the table shows section numbers in which the abundance of sea cucumbers fell within the range indicated in the column heading. Sections for each islet were numbered clockwise starting at the counterclock- wise extremity (left-hand end when facing seaward) of the seaward shore.) Islet Dat e 100-m section number Remarks 1 Total individu- als observed or estimated No Name it) Q« H. atra and H. leucospilota probably ingest and egest about 2.4 X 108 kg of sand yearly. It has been pointed out by Crozier (1918) for Stichopus moebii, by Yamanouti (1939) for H. atra and other species, and by Trefz ( 1958) for H. atra, that there is virtually no grinding action or reduction in size of sand particles upon passing through the gut. Trefz has demonstrated that even such minute and delicate calcareous structures as the integumental "anchors and plates” of the sea cucumber, Ophiodesoma spec- tabilis, do not undergo perceptible dissolution in the gut of H. atra, but that they appear the same after egestion as before ingestion, thereby negating the idea of chemical dissolution in the gut. In some of the scoured depressions on the seaward reef where very little sand is present, H. atra serves an additional function by retain- ing within and around itself much of the sand required in its feeding that might otherwise be washed away. SIZE OF SPECIMENS The length of undisturbed H. atra in the open ranged from about 2 to 60 cm, with estimated weights of 10 to 2000 g. We have not observed smaller ones. Frequent estimates of length dur- ing the censusing led to the opinion that the geometric mean length of H. atra on the islet shores is about 17 cm and of the more slender H. leucospilota about 25 cm. In contrast to the elongate shape of the larger specimens, the smaller specimens of H. atra were short and thick; in fact, the smallest (2 cm) were as thick as they were long. At the northeastern reef of Kabelle I. on September 14, 1961, 407 speci- mens of H. atra in a single pool were individ- ually measured or estimated for length and diameter in inches, while still undisturbed, and were then placed in a container of water for volume measurement in cm3 by displace- ment. The temperature of the pool ranged from about 35 to 39 C during the 5 hr re- 310 PACIFIC SCIENCE, VoL XVII, July 1963 TABLE 2 Length-Frequencies of 407 H. atra Measured in a Pool on the Northeastern Shore of Kabelle Island, Sept. 14, 1961 LENGTH INCHES NUMBER LENGTH INCHES NUMBER 2 5 6 98 3 8 7 29 4 78 8 14 5 159 9 8 10 2 quired for measuring. Both the high tempera- ture and the slight, unavoidable disturbance of adjacent individuals when picking up specimens for volume measurement resulted in a small amount of contraction. After handling, the speci- mens were deposited in another part of the pool. Table 2 gives the length-frequencies. These measurements show a single major peak at about 5 inches ( 12.7 cm), and fail to affirm the pres- ence of other expected frequency groups. Figure 5 is a histogram showing the volume- frequency. As with length-frequency there is a single peak ( at about 60 cm3 ) . From lines fitted by inspection to the scatter diagram of Figure 6 the relationship of volume (V) to length (L) in the size range 5-15 cm was estimated to be: V= 5.4 L0-94 The rather abrupt increase in slope beyond 15 cm is emphasized by the visually fitted line: V = 0.019L30 A few additional measurements of specimens longer than 25 cm suggest that this slope of 3.0 continues throughout the upper size-range. TEMPERATURE TOLERANCE Consideration of their habitat, exposed to the sun in shallow pools at low tide, would suggest a high temperature tolerance for H. atra. On partly cloudy days after at least 2 hr of con- tinuous sunshine, the temperatures of 69 pools containing H. atra ranged from 31.1 to 39.4 C, with mean and standard deviation of 36.19 and 2.14 C. The temperatures, determined with clin- ical centigrade thermometers, of eight black sea cucumbers were consistently higher than simul- taneous temperatures in the sand-coated sea cucumbers alongside them (by an average of 0.25 C, standard deviation 0.16 C). After more prolonged exposure to sunlight this differential would be expected to increase. Shaded thermom- eters on the bottoms of the pools gave readings lower than within the sand-coated sea cucum- bers, while thermometers in direct sunlight on the pool bottoms approximated those within the black sea cucumbers. Extreme differences between pool- and specimen-temperatures at any one place and time barely exceeded 1 C Some degree of reflective insulation from the radiation of the sun is imparted to H, atra by the coating of light-colored sand with which it habitually covers its body, enabling it to retain a slightly lower body temperature than it does when not sand-coated, or than does the naked, black H, leucospilota when out in the open. The warmest pool in which a H. leucospilota was found was 38.4 C, but the animal was not feeding. The usual habitat of H. leucospilota is cooler than that of H. atra because of its habit of seeking the shade, and protection of rocks. Water coming in over the reef at high tide has a temperature of about 29 C. When this cool water saturates the beach it remains cool if overlain by strata of slab rock, but is warmed by the sun if the sand is exposed. H. atra oc- curred in pools of higher temperature, 39.4 C. The upper limit was not determined. Crozier (1915: 281) gives the maximum temperature of the habitat of H. surinamensis as 31.8 C and states that muscle coagulates at 42 C. This is only about 3 C above the highest pool tempera- tures in which H. atra has been encountered. H. atra is unusually tolerant of heat, and is al- most the only macroscopic organism living in these warm pools at low tide on the warmest days. Fig. 4. Photomicrographs showing calcareous deposits in the integument. A, Tables of H. atra. B, Tables and buttons of H. leucospilota. Sea Cucumbers at Rongelap Atoll — Bonham and Held 311 312 PACIFIC SCIENCE, Vol. XVII, July 1963 PHOTOTROPISM H. atra, and to an even greater extent, H. leucospilota are negatively phototropic since both protect themselves from strong light, H . atra by its sand coating, and H. leucospilota by seeking shelter. They usually react to quick changes in light intensity, as by withdrawal of tentacles when covered by a shadow during feed- ing. H. leucospilota reacts to a shadow by sud- denly contracting under the protection of the rock where its posterior end is anchored, thus frequently requiring a quick grasp and a firm pull in order to collect this species. PREDATION Besides reflective insulation the sand coating of H. atra affords concealment. This function, however, would seem to be subordinate to that of insulation, because of the small amount of predation to which H. atra is apparently sub- jected, unless, of course, the converse obtains, that the small amount of predation may be due to the concealing effect of the sand coating. A damaged sea cucumber has been observed only once. This specimen, presumably pecked by a bird, occurred in an area where curlews had been active. In no case has an animal been ob- served actually preying upon H. atra. Grazier (1915:246) observed lightly colored parts of Fig. 5. Histogram showing volume-frequency of 407 H. atra from a pool at Kabelle I., Sept. 14, 1961. Fig. 6. Relationship of volume to length of 407 H. atra at Kabelle I., Sept. 14, 1961. Lines drawn by inspection. H. surinamensis which he suspected of being regenerated following either autotomy or injury by enemies such as fishes or crabs, but states that there is no evidence that Holothuria has enemies of this kind. Frizzell and Exline ( 1955: 29) ascribe predation on intertidal sea cucum- bers to sea gulls, and note the general concur- rence among authors that predation is slight, but propose that the apparent protective mecha- nisms of holothurians suggest predation to be more common than is supposed. TRANSVERSE FISSION Fission has been reported in holothurians by Crazier (1915:291), who found regeneration of either end of the body in about 10% of the H. surinamensis encountered. Frizzell and Exline (1955:15 and 27) state that a few forms in- cluding Holothuria parvula (Selenka), Cucum- aria lactea, and C. planci (Brandt) sometimes Sea Cucumbers at Rongelap Atoll — Bonham and Held 313 reproduce by transverse fission, and that Holo- thuria difficilis Semper of the tropical Pacific commonly does so. In September 1959 in a warm pool of about 37 C on the southwest shore of Burok I. two short specimens of H. atra were encountered lying end to end. The integument of the two adjacent terminal end portions was devoid of sand because of the absence of tube feet. The specimens were saved, and later examination revealed a calcareous oral ring in only one, in- dicating that the two individuals resulted from transverse fission. Another specimen in an ad- jacent pool was photographed in an advanced stage of constriction; but without disturbance by even so slight a stimulus as a shadow it vol- untarily relaxed the constriction and resumed normal shape within a period of about 5 min. During the September 1961 census, enough fissioning ( Fig. 7 ) and fissioned specimens were Fig. 7. H. atra, about 23 cm long, undergoing transverse fission in a pool 10-cm deep. Mellu I., sea- ward side near west end, Sept. 18, 1961, 1520 hr. encountered to verify that fission, or asexual re- production, does commonly occur in Holothuria atra . The importance of fissioning as a means of reproduction is emphasized in Figure 6 by the relative thickness of the small specimens and the consequent reduced slope (0.9) of their logarith- mic relationship of volume to length as com- pared to the usual cubic relationship (slope of 3 ) for the larger specimens. If it may be assumed that sexual reproduction would be seasonal in nature, while fissioning would occur at all seasons, then the lack of ap- parent age classes or frequency modes in the volume-frequency histogram of Figure 5 is also consistent with the concept of reproduction by fissioning. SUMMARY 1. The preferred habitat for H. atra and H. leucospilota on the seaward reef-flat is described and is linked with the presence of slabs or shelves of rock at the junction of the sandy beach with the reef-flat. On hot days runoff water from such areas is of relatively low tem- perature (because of the insulating effect of the slabs of rock). 2. Estimates of numbers of these two sea cucumbers are based upon notes and photo- graphs made while traversing the peripheries of the larger islets and some reef and lagoon areas, resulting in an estimated total of about 5 X 10° H. atra and 2 X 10 5 H. leucospilota. 3. The passage of sand through the gut of H. atra is at the rate of 1 g (dry weight) in from 5 to 10 min, resulting in an estimate for the atoll of about 2 X 108 kg of sand ingested and egested yearly. The primary ecological role of the sea cucumbers in this feeding action is to move the sand from one place to another. 4. Length of H. atra observed ranged from 2 to 60 cm. Measurements of 407 specimens in one pool showed that for specimens up to 15 cm in length, volume increased slowly with length, while for those from 15 to 25 cm, vol- ume increased almost as the cube of length. This is construed as evidence of reproduction by fissioning. 5. H. atra appeared healthy and feeding in small pools with temperatures up to 38.9 C. 314 PACIFIC SCIENCE, Vol. XVII, July 1963 The internal temperature is slightly lower in the sand-coated than in the naked black speci- mens because of the reflection of light and heat by the coating of sand characteristic of this species. 6. Evidence of predation upon H. atra or H. leucospilota is practically lacking. 7. Transverse fission in H. atra has been ob- served in various stages of progress, and is considered to be an important means of repro- duction. REFERENCES Crozier, W. J. 1915. The sensory reactions of Holothuria surinamensis . Zool. Jb. Phys. 35: 231-297. 1918. The amount of bottom material ingested by holothurians. J. Exp. Zool. 26: 379-389. Cuenot, L. 1948. Anatomie, Ethologie et Sys- tematique des Echinodermes. Classe de Holo- thuries. In: P. P. Grasse, Traite de Zoologie, Anatomie, Systematique Biologic. Tome 11, Echinodermes, Stomatocordes, Procordes, Paris. Pp. 82-120. Frizzell, D. L., and Harriet Exline. 1955. Monograph of fossil holothurian sclerites. Bull. Univ. Missouri School of Mines & Met- allurgy, Tech. Ser. 89: 1-204. Trefz, Shirley M. 1958. The physiology of digestion of Holothuria atra Jager with spe- cial reference to its role in the ecology of coral reefs. Doctoral thesis, University of Hawaii. Pp. 1-149. Yamanouti, T. 1939. Ecological and physio- logical studies on the holothurians in the coral reef of Palao Islands. Palao Trop. Biol. Sta. Stud. 4:603-635. Field Identification of Five Species of Californian Beach Hoppers (Crustacea: Amphipoda) Darl E. Bowers1 While studying the correlations between the distribution of five species of beach hoppers of the genus Orchestoidea and the physical and biotic factors of their sandy beach habitat, it became necessary to identify with certainty the animals collected from the many beaches sam- pled up and down the Californian coast. Since it was desirable tentatively to name the hoppers captured in the field, I made an effort to find characters which were easily observed on the beach and would allow such identification. As a result of my examination of many thousands of animals from over a hundred collecting locali- ties, I evolved a recognition of the five species mainly on the basis of pigmentation patterns, elements of which are relatively consistent in spite of the many variations to be found in the total pigmentation. The species under consideration are the two large hoppers, Orchestoidea calif orniana (Brandt) 1851, and O. corniculata Stout 1913; the two somewhat smaller species 0. columbiana Bousfield 1958, and O. pugettensis (Dana) 1853; and the small species, O. benedicti Shoe- maker 1930. Laboratory identification of these has been made possible by E. L. Bousfield ( 1957, and particularly 1959), to whom I am indebted for his help with some of the key characters. In these papers Bousfield has described a new spe- cies from southern California, 0. minor, similar to O. columbiana, but with a few subtle anatom- ical differences. These will require statistical analysis in order to show more clearly the affini- ties of the two species. This newly described form will not be considered here since I have been mainly concerned with the beaches of central and northern California. The pigmentation patterns may be relied on for identification of both sexes for almost all 1 Mills College, Oakland, California. Manuscript re- ceived January 30, 1962. sizes of animals, with the following cautions: ( 1 ) The largest individuals of calif orniana al- most invariably have lost the dorsal pigmenta- tion, but color and form of the second antennae, as described below, will usually make this no problem; (2) the largest individuals of cornicu- lata may have lost the dorsal pigmentation, but again, color and form of the antennae will usually suffice; (3) the smallest individuals of all species (under 4 mm in body length) may be so lightly pigmented as to make identification difficult or to require magnification and lighting available only in the laboratory; (4) animals living on white or very light-colored sands usually are only faintly marked, in which case combinations of other characters may have to be used. Some general beach characteristics and hop- per habits may also be of use in identification but are not dependable alone. O. calif orniana is typically found on long exposed beaches composed of fine sand, with few if any rocks present, beaches that are wide from the fore- shore to back shore with dunes into which the hoppers may retreat during high spring tides and for breeding, beaches which usually have a quite flat foreshore, which is correlated both with exposure to wave action and with fine sand particles. O. corniculata, on the other hand, is typically found on shorter, more protected beaches composed of coarser sand, possibly with rocks or boulders present, beaches which may be narrow from foreshore to back shore with cliffs or piles of boulders preventing retreat dur- ing high spring tides (breeding animals are commonly found mixed with the rest of the population), beaches which may have a fairly steep foreshore, which is correlated both with protection from wave action and with coarser sand particles. These two animals are not often found together, but occasionally on a calif or- niana-dommated beach, as at Carmel, California, 315 316 PACIFIC SCIENCE, VoL XVII, July 1963 corniculata hoppers may occur at the ends of the beach where conditions are not optimal for calif ornicma. I have never found calif orniana on corniculata-&omimx£<\. beaches. For columbiana and pugettensis I have not gathered enough data to make very meaningful statements here. They often occur together on coarse sand beaches, some of which are quite long, onto which relatively little seaweed is cast as food for the hoppers. 0. benedicti is com- monly found with calif orniana, and otherwise lives on beaches of the finest, silty sand. It is not always easy to find mature individuals in the daylight on a beach and much sample digging is often required to locate a ''bed” of hoppers. Small individuals are readily found but are not so readily captured. The distribution of hoppers on a beach is not of a random nature, but is dictated by the location of seaweed, the main food of these omnivores, on the shore, and by the tidal cycles, which they tend to follow up and down the beach. Since these animals feed mostly at night on the freshest drift present and many remain under this drift at dawn, the most productive searching is usually done under the seaweeds brought ashore by a high tide of the previous night. Large individuals, however, may move farther up the beach; and in calif orniana, at least, the mature males and some females in the breeding season (probably February to October) with eggs or young in their brood pouches will be found scattered in fairly high and dry zones that may not have been reached by the surf for some days. The location of the burrows of these large animals requires careful observation in the early daylight hours. In digging the burrow, or clean- ing out an old one, large calif orniana kick sand out onto the surface in two opposite elongated rays. If this sand is of a different color because of a higher moisture content or because it was obtained from different-colored material below, it may be quite noticeable. However, during the day, drying action of the wind and sun may reduce the color difference between the burrow sand and the surface sand, and the winds level the elongated heaps, making them increasingly difficult to see. The burrows may be open shafts up to 12 inches in depth, with a plug of sand in the mouth of the burrow. O. corniculata make burrow mounds much more like those of a pocket gopher. The sand is Fig. 1. Sketch of O. calif orniana to show features mentioned in the text. Californian Beach Hoppers — Bowers 317 pushed up less vigorously than by calif orniana and thus simply falls all around the burrow mouth in a rounded heap. Large individuals are usually found with smaller animals in mid- to high-tide zones rather than higher up on drier parts of the beach, as in calif orniana. The bur- rows are commonly open just around the hopper and are usually less than 6 inches deep, but I have found these hoppers more than 2 ft down. The three smaller hopper species dig less than the large hoppers and thus do not make very obvious surface markings. They are more often found associated with washed-up debris. I have found no sexual dimorphism with re- spect to pigmentation pattern, but the sex of these animals is easily distinguished by the structure of the 2nd gnathopods (the 2nd pair of large ventral appendages) except as noted below (see Fig. 1). In females the sixth seg- ment is a fleshy paddle-shaped structure. In young genetic males the same form is present, but it becomes modelled through a series of molts into a larger subchelate "hand” with a curved dactyl closing against an emminence of the "palm.” As far as I know, males can be dis- tinguished from females only when the change in the gnathopods begins to be apparent. In calif orniana and corniculata this occurs when the animals reach about 9.0 and 13-0 mm in body length, respectively (measured from the anterior of the head through the chord of the straight- ened but still somewhat curved body to the back edge of the 3rd abdominal segment ) . Sex- ually active females possess oostegites under the thoracic region. These are thin plates with a fringe of hairs, which overlap to form a mar- supium in which eggs and young are carried. However, some large females lack these hairs, their presence apparently being dependent on the breeding cycle rather than on size. The 2nd antennae of females of all species are shorter and less robust than those of males, and are less colorful. In midday, as the waves of a rising tide moisten a beach, it is common to see small ani- mals moving around in the just-wet area, feeding on fresh bits of seaweed, washed up sand crabs, or seemingly just wandering. Probably all species do this, but in my experience this movement has occurred mostly with columhiana, less fre- Fig. 2. Left 2nd antenna of O. corniculata male. Compare this with appendage in Figure 1. quently with calif orniana. Under such condi- tions, an observer lying face down on the sand can use pigmentation patterns to distinguish the animals without even touching them. The most obvious character that distinguishes large male calif orniana from male corniculata is the form and color of the second antennae. In calif orniana the flagellum of the antenna is rela- tively thin, longer than the combined length of the segments of the peduncle, and gradually tapers to the tip, which reaches back beyond the middle of the body when folded from the pe- duncle base. The color of the peduncle is usually rosy-red. The peduncle segments are longer but not as massive as those of corniculata. In cornic- ulata the flagellum of the antenna is thicker than in calif orniana, is shorter than the combined length of the peduncle segments, and tapers rapidly to the tip, which does not reach the middle of the body when folded ( see Fig. 2 ) . The color is usually salmon-pink, but sometimes, on a darker substrate, may be tinged with blue or brown. Newly molted individuals of both species may be white. The antennae of columhiana are calif orniana- like, but in large animals are bluish-white rather than red. The antennae of pugettensis are cor- 318 PACIFIC SCIENCE, Vol. XVII, July 1963 Fig. 3. Dorsal views of thoracic segments 6, 7, and 8, and abdominal segments 1, 2, and 3, showing two variants of each pigmentation pattern: calif orniana- like group— O calif orniana; b, O. Columbiana; c, O. benedicti — corniculata-\ike group: d, O. cornicu- lata; e, O. pugettensis. niculata-Yike but are more commonly bluish or brownish rather than salmon-pink. The anten- nae of benedicti are proportionally the shortest of the genus but are shaped like those of cornic- ulata. The longest males I have measured (see above for the reference points) are: O. calif orniana, 26.0 mm; O. corniculata, 25.0 mm; O. Colum- biana, 19.0 mm; O. pugettensis, 15.5 mm; and O. benedicti, 10.0 mm. The females are a few millimeters shorter. Because of exceptions to the above general statements about antennae, one finds that with subadult individuals, females, and indeed even mature males from some beaches, other obvious Californian Beach Hoppers — Bowers 319 traits are needed for field identification. Gen- erally, the overall pigmentation tone of these hoppers tends to match the color of the sandy substrate; but even on light sands where pig- ment is reduced, there are a few key spots that usually show up. I will be concerned mainly with the dorsum of the first three abdominal segments, which appear in a top view as the last three body segments ahead of the urosome, and the 7th and 8th thoracic segments. A study of the sketches in Figure 3 will show that there are two main groups of pigmentation patterns. The broad "butterfly” spots of the 1st and 2nd abdominal segments of the "calif or- niana- like” group are found with modifications in O. calif orniana, O. columbiana, and O. bene - dicti. The T-shaped spots of the 1st, 2nd, and 3rd abdominal segments of the " corniculata- like” group are seen in O. corniculata and 0. pugettensis. The calif orniana-/fi£e group. In calif orniana and benedicti there is a sagittal line showing in most variants of this pattern. Note the absence of pigment from the 3rd abdominal segment on calif orniana, but the presence of pigment in benedicti. In columbiana the midline marking is missing and the "butterfly” design is more nearly a flattened X, again with no pigment in abdominal segment 3. The characteristic mark- ings of abdominal segments 1 and 2 are suf- ficient to separate calif orniana from columbiana. Benedicti, besides being small, is more heavily pigmented with discontinuous spots, which thus give the impression of a checkerboard design along the whole body. The corniculata -like group. The two species in this grouping, corniculata and pugettensis, are the most difficult to separate, although they are easily distinguished from all the others. The most obvious dorsal pattern differences seem to be in the last two thoracic segments in which pugettensis has more delicate markings, but I have detected no easily recognizable differences here. A more consistent and reliable set of marks may be found at the lower margins of the lateral body wall in the last three thoracic segments of pugettensis (see Fig. 4). The third of these is missing in corniculata, which thus shows only two spots which are usually more diffuse and less intensely pigmented than in pugettensis. Fig. 4. Side views to show key characters of thoracic segments 6, 7, and 8. a, O. corniculata; b, O. pugettensis. (These spots do not serve to set off these two species from the other three, however, for these others may have such markings as well. ) These two hoppers are found on the same beach oc- casionally, and in such cases it may be necessary to rely on the key characters presented by Bous- field in the papers cited. However, these pig- mentation patterns, antennae colors, and other more subtle clues allow one acquainted with the animals to make fairly reliable preliminary de- terminations. Whatever "hedging” the reader may detect in this presentation is due to the wide range of variability to be found in these animals and to the fact that exceptions keep cropping up as 320 PACIFIC SCIENCE, Vol. XVII, July 1963 study continues. However, it is hoped that this discussion will make it easier to identify these animals in the field. I am grateful for financial support for this study which came in part from the American Academy of Arts and Sciences, and in part from the Mills College Faculty Re- search Fund. My thanks are here expressed to Dr. Joel Hedgpeth for helpful suggestions. FIELD KEY TO species OF Orchestoidea 1. Mature animals 2 1. Immature animals and others not distinguished by first part of key 5 2. Second antennae when folded reaching back to or beyond middle of body; flagellum longer than peduncle 3 2. Second antennae when folded not reaching back to middle of body; flagellum shorter than peduncle 4 3. Color of 2nd antennae rosy-red calif orniana 3. Color of 2nd antennae bluish-white columbiana 4. Color of 2nd antennae usually salmon-pink corniculata 4. Color of 2nd antennae not so 5 5. Dorsal pigmentation pattern containing "butterfly” designs 6 5. Dorsal pigmentation pattern containing T-shaped designs (Lower limb of T may be faint or missing) 8 6. Mid-dorsal line absent, "butterfly” spots flattened X’s columbiana 6. Mid-dorsal line present 7 7. No markings on 3rd abdominal segment; sides of body relatively free of pigmentation marks calif orniana 7. Markings on 3rd abdominal segment; sides of body blotched with checkerboard pattern ...benedicti I .corniculata pugettensis Ij 8. Two diffuse spots on sides of body.. 8. Three discrete spots on sides of body. REFERENCES Bousfield, E. L. 1957. Notes on the Amphipod genus Orchestoidea on the Pacific Coast of North America. Bull. So. Calif. Acad. Sci. 56(3): 119-129. 1959. New records of beach hoppers (Crustacea: Amphipoda) from the Coast of California. Nat’l Mus. Canada Contr. Zook, Bull. 172:1-12. A Simple Device for Making Successive Photomicrographic Records of Large Groups of Developing Organisms1 Sidney C. Hsiao, Walter K. Fujii, and Helen H. Fine2 In OUR analysis of the effect of ionizing radia- tion upon the cleavage of sea urchin zygotes we found it necessary to take successive photo- micrographs of a large number of eggs, in different samples, which had been exposed to graded doses of ionizing radiation (Hsiao and Daniel, I960). In order to estimate the rate of cleavage of the irradiated samples of ferti- lized eggs it is highly desirable to follow the cleavage of each egg in every sample and make photomicrographic records for later analysis. In other words, we need to take time-lapse pictures of the developing eggs subjected to different amounts of radiation so as to calcu- late the rate of cleavage and correlate it with dosage. After some preliminary trials we have put together, using commonly available ma- terials, a simple device capable of taking pho- tomicrographs repeatedly from the same field in a series of samples of irradiated eggs at specified time intervals. By lining up, accord- ing to time, prints made from each field, each egg can be identified and its cleavage followed from the first to the last frame in the series, and its rate of cleavage can be calculated. It occurs to us that investigators who have occa- sion to record developmental and other recur- rent phenomena may find this simple device useful. A brief description of its method of con- struction and manipulation is reported in this paper. PRINCIPLES OF CONSTRUCTION The simplest way to obtain a sequential rec- ord of a number of fertilized eggs after their exposure to irradiation or other treatment would 1 Work supported by U. S. AEC contract AT (04-3) 330. Hawaii Marine Laboratory contribution No. 189. 2 Department of Zoology, University of Hawaii. Manuscript received February 6, 1962. be to place the eggs on the stage of the photo- micrographic instrument and take cinemato- graphic or time-lapse pictures. This requires one set of instruments for each sample exposed to one specific dose. The cost of equipment and the limitation of laboratory space would rule out all sample series of reasonable size. How- ever, if the samples are immobilized and the photomicrographic camera is brought over an exactly predetermined point in each sample to take time-lapse photomicrographs during the course of cleavage, it would be possible to make photographic records of a number of samples with one instrument. This can be done by: ( 1 ) Immobilizing samples of eggs contained in uniform-sized culture vessels, such as a culture cell on a 1- X 3 -inch micro slide, in a straight line on a stage supported by a steady stand independent of the photomicrographic camera, so that when the microscope moves no motion is transferred to the samples. ( 2 ) Using a pho- tomicrographic camera carriage made to move back and forth along a straight track placed parallel to the line of the immobilized sample cultures and adjusted to bring the microscope objective across the middle of the cultures. (3) Employing a series of uniform spacers (such as 1-inch rectangular bars for use with the 1- X 3-inch micro slide containing samples of egg cultures), whose width equals the distance between the centers of the culture vessels of two neighboring samples, to stop the photo- micrographic camera carriage on the track each time the objective of the microscope reaches the center of a sample on the immovable stage. In this way a single photomicrographic camera can be used to photograph the eggs in the exact center of each sample culture vessel, and a series of the time-lapse photomicrographs is obtained for later analysis. 321 322 MATERIAL USED AND METHOD OF CONSTRUCTION PACIFIC SCIENCE, Vol. XVII, July 1963 The device built on the above principles con- sists of two independent parts: (A) a steady stand for supporting the cultures to be photo- graphed, and ( B ) a movable carriage on which a microscope and its camera and substage lamp are mounted so that the microscope and its accessories can travel from culture to culture without disturbing the stand or the organisms on it. The materials used for the construction of this device are limited to the few simple items listed below: 7-ft length of 2 X 2-inch wood piece of 28 X 8 X 1-inch plywood 12 -ft length of Vi -inch wide V^-inch thick angle aluminum 20 X 8 X Vi-inch sheet of aluminum 1 dozen No. 8, 1-inch wood screws 8 self-tapping sheet metal screws, size 3 /32-i/4-inch 3 aluminum sheaves, 5i-inch in diameter 11 machine screws with nuts, size 3/32- V4-inch 6-ft length of 3-sided 1 -inch-square ex- truded aluminum channel piece of 5 Vi X 28 X Vi -inch plate glass All these are easily obtainable from local hard- ware stores. A. The stand. This independent part of the instrument is made of a rectangular wooden base, two aluminum trapezoidal end supports, and a rectangular glass toppiece or stage. The construction of these three parts is summar- ized below. ( 1 ) From the 2 X 2 -inch lumber, two 11- and two 31-inch lengths are cut. These four pieces are joined at right angles by means of either end-lap joint or pinned tenon-and- tusk joint into a rectangular base measuring 31 X 11 inches on the outside. (2) From the Vs- -inch-thick sheet of aluminum two 8X8- inch pieces are cut and each is shaped into a near trapezoid (it will be referred to as trape- zoid in this paper for simplicity), as shown in Figure l A, by cutting off a triangular piece whose two sides adjacent to the right angle are 2 and 6 inches long (area ABC in Fig. 1 A). After rounding the corners, two 5/32-inch holes Fig. 1. Parts of the stand. A, Aluminum end sup- port used as left end of stand. 1, 2, 5/32-inch holes; 1 being 1 inch from vertical edge and 2 5 inches from 1; 3, 4, 3, are three beveled holes 3 inches apart, 3 being 1 inch from vertical edge, and all Yl inch from the bottom. B, Diagram of angle aluminum end frame for plate-glass toppiece or stage. Two upper holes 5/32 inches, two lower ones Ys inch. C, End portion of plate-glass toppiece or stage showing plate glass l , mounted between two 28Yl -inch strips of angle aluminum, 2, which are attached on end piece, 3. The two VXinch holes, H, are for mounting the toppiece or stage on the trapezoid end support. D, Sectional view of right end of stand, showing method of mounting trapezoidal end supports. 2, on wooden , base, 3; and toppiece or stage, 1, on the trapezoid, ; 2. S, No. 8 wood screw; H. sheet metal self-tapping i screw. Photomicrographic Records — Hsiao, FUJII, and Fine 323 5 inches apart are drilled Vi inch from the top and Vi inch from the vertical edge of one of the trapezoids. Three 3/ 16-inch holes spaced 3 inches apart are next drilled Vz inch from the bottom and 1 inch from the vertical edge. The holes on the second trapezoidal end support are similarly drilled but are done from the oppo- site surface so that this piece of aluminum forms a mirror image of the first one. ( 3 ) The rectangular glass toppiece or stage is made from a 5^2 X 28-inch strip of V4-inch plate glass and a rectangular frame made of VXinch wide angle aluminum. Two 554-inch sections are cut from the aluminum stock and four holes drilled in each piece. Along one side of the aluminum angle two 5/32-inch holes are drilled each Vs inch from the end, along the other side, two Vs-inch holes, each Vz inch from the end, are also drilled, as shown in Figure 16. Two 28]/2-inch lengths of aluminum are cut from the same stock and two Vs-inch holes are drilled on each piece, each hole being Ys inch from the end. The 28-inch long plate glass is placed lengthwise between the inner .surfaces of the long alumi- num angle strips whose ends are then placed across the two short (554 inches ) end pieces in such a way that the outer surfaces of the four angle strips are in contact and the pre- drilled pg-inch holes on the long strips match the 5/32-inch holes on the short ones, as shown in Figure 1C, leaving the Vs-inch holes of the short strips exposed at the ends of the frame. Four 3/32-1/4-inch self-tapping sheet metal screws are driven through the 5/32-inch holes in the short end aluminum section tapping into the Vs-inch holes in the long sections to hold the aluminum frame and glass together. It will be noticed that all the 5/3 2 -inch holes are for the passage of, and the Vs-inch holes for the tapping by, the self-tapping sheet metal screws. To assemble the stand the two trapezoidal end supports are used to join the glass toppiece or stage to the wooden rectangular base. After placing the wooden base on a flat bench or table top, with its long sides running from left to right in front of the operator, the right trapezoidal end support is clamped with a C- clamp on the inside of the 11 -inch-long end piece of the wooden base so that the vertical edge of the trapezoid faces the front of the stand, farthest from the operator and Va inch from the right distal corner of the rectangular base, the slanting edge nearest the operator, the bottom of the trapezoid flush with the bench top and the smaller openings of the three bev- eled holes against the inside surface of the wooden end piece. After drilling three holes into the wood, using the holes in the alumi- num trapezoid as guides, three No. 8, 1-inch, wood screws are driven into the wood until their heads are flush with the aluminum (see Fig. ID at S ). The C-clamp is released. Similarly the left trapezoidal end support is screwed on the in- side of the left end piece of the wooden base. The aluminum framed glass toppiece or stage is fastened to the trapezoidal end supports by apposing the two Vs-inch holes on each end angle aluminum piece (6 in Fig. 1C) to the two 5/32-inch holes near the top of each trape- zoid and a round-head self-tapping sheet metal screw driven from the outside through the trapezoidal end support, tapping into the angle aluminum as shown at H in Figure ID. When properly assembled this stand is very steady and can support a large number of culture slides or vessels on the plate glass. B. The carriage. This part of the instrument consists of a platform with track, a cart for carrying the microscope and its accessories, and a number of spacers of uniform size. ( 1 ) The platform is made from a piece of 1 X 28 X 8-inch plywood and two strips of angle aluminum. Each piece of aluminum is cut 28 inches long and two 5 /32-inch holes drilled through the apex of the angle, Vz inch from each end, and beveled. One piece of the drilled aluminum is placed along the edge of the ply- wood board with the apex pointing upward and a No. 8, 1-inch, wood screw driven through each hole into the wood. The second piece of angle aluminum is placed parallel to the first, but with its apex 5 Vs inches from that of the first, and fastened on the wood in a similar way. These two pieces of angle aluminum form the track on which the microscope-carrying cart moves. (2 ) The microscope-carrying cart is constructed from a piece of the V^-inch sheet aluminum, 5 X 6Va inches in size, with three pieces of angle aluminum and three Vs -inch aluminum sheaves attached to its under surface. Six 5/32-inch holes, as indicated by No. 1-6 in 324 PACIFIC SCIENCE, Vol. XVII, July 1963 Figure 2 A, are drilled through the sheet alumi- num which forms the floor of the cart. A 44 X 44-inch slit is cut at A and a round 44-inch hole at B, as shown in the same figure. Slit A is used to adjust and fix the substage illumi- nator in place, and B is for the microscope set screw, the same screw which came with the microscope for fastening it to the bottom of the microscope case during shipment. Two 5- inch sections of angle aluminum are cut, and two 5/32-inch holes drilled on each limb of the right angle. Each hole is 44 inch from the end of the aluminum section. One of these two sections is mounted on the front undersurface of the floor of the cart with two 3/32 machine screws and nuts, and the other section similarly mounted on the right undersurface at right angles to the first (Fig. 2 B at 4). A third piece of angle aluminum, 144-inch long is cut A. B. Fig. 2. Diagrams showing microscope carrying cart floor and top view of stand in use. A, Top surface of cart floor, showing openings 1 to 6 for attaching angle aluminum on undersurface; A, slit for adjusting and fixing illuminator; B, opening for microscope base set screw. Broken line indicates outline of the base of the microscope. B, Undersurface of cart floor. 1—3, sheaves; 4, 5 -inch angle aluminum for buffering against spacer. C, Right portion of plate-glass toppiece of stand show- ing culture slides, 1-6 in place. Stippled areas repre- sent masking tape. Other slides and glass (or metal) are positioning guides. and a 5/32-inch hole drilled at a distance of 44 inch from each end. It is mounted along the back border on the undersurface of the cart floor with metal screws and nuts through open- ings 3 and 4 (see Figs. 2 A, 2 B). Three alumi- num sheaves (1, 2, 3 in Fig. 2 B) are mounted with machine screws and nuts on the vertical limb of the angle aluminum pieces on the undersurface of the cart floor so that sheaves No. 1 and 2 are 444 inches apart and each 44 inch from the end of the 5 -inch section of angle aluminum. Sheave No. 3 is mounted on the center of the vertical limbs of the 144 -inch section of angle aluminum and at a distance 544 inches from a line joining sheaves No. 1 and 2. When the cart is placed on the platform the sheaves should ride smoothly but snugly on the track. (3) The spacers are six 5 3/ 16- inch sections cut from the 1-inch extruded rec- tangular aluminum channel. One spacer is mounted on the right-hand end of the plat- form with No. 8 wood screws, after the proper holes are drilled, to serve as the starting or ending point in the side to side movement of the microscope-carrying cart. OPERATION OF INSTRUMENT To employ this instrument for taking photo- micrographs a microscope and a substage il- luminator are mounted on the microscope-carry- ing cart. In our laboratory an old E. Leitz Wetzler microscope with its mirror removed is fastened on the car r floor by its set screw driven through opening B into the horse-shoe shaped base, and a Kohler illuminator through slit A. The light is centered and adjusted by standard procedures (Gray, 1958). The stand is placed on the laboratory table with its length parallel with the edge of the table and the slanting edges of the aluminum end supports near the operator. The platform is placed inside the stand so that a 44 -inch clearing is produced on all four sides between the platform and the base of the stand. The microscope with its cart and illuminator is carefully put in place by sliding the microscope stage immediately under the plate-glass toppiece of the stand from back forward across the stand until the sheaves are over the track. The cart is lowered onto the track and pushed toward the right until it is Photomicrographic Records — Hsiao, FUJII, and Fine 325 stopped by the fixed spacer at the end. A cul- ture slide containing the sample is placed on the plate glass with its length perpendicular to the long axis of the glass toppiece and directly under the low power objective of the micro- scope. In our experiments a glass or aluminum ring ^4-inch I. D. and 5/32-inch high is mounted in the center of a 1- X 3 -inch micro slide with epoxy cement to serve as a culture cell. When the culture slide has been centered, a piece of metal or glass is placed snugly against its right side and fixed on the plate glass with masking tape to serve as a slide end-stop. To facilitate lining up the other culture slides on the left side of this extreme right one the 1-inch wide space between the 28-inch-long aluminum frame on the side nearest the oper- ator and the culture slide is filled with another 1 X 3 -inch slide at right angles to the end- stop (see Fig. 2 C). Other culture slides are then easily placed in a straight line on the left of the original one as shown in Figure 2C by push- ing them toward the right lower corner against the slide end-stop. After all the culture slides are placed in a straight line from right to left, the microscope-carrying cart is moved toward the left by drawing it along the track and the 1-inch-wide aluminum spacers are placed be- tween the cart and the end piece on the plat- form, the number of spacers used being equal to the number of culture slides. As both the spacers and the culture slides are 1-inch wide the slide on the extreme left will be centered directly under the microscope. The culture slides and spacers are numbered correspondingly, both starting with No. 1 on the left and ending with the highest number on the right. A photograph of the assembled device with the photomicrographic instrument in place is shown in Figure 3. To take photomicrographs with this device, a camera is mounted on the microscope, and the specimens in the culture cell brought into view. Usually, before the first exposure is made, the specimens are centered and arranged on each slide, using a glass needle, so that a large number of individuals can be observed in each microscope field. Once the specimens are prop- erly arranged, no other manipulation is needed, and they will stay in place throughout the course of the experiment. After the specimens Fig. 3. Photograph of device, with camera, micro- scope, and illuminator in place. are brought into sharp focus, a photograph is taken with the time of exposure and light in- tensity adjusted by the usual method for photo- graphic work (Shillaber, 1944). To take a photograph of the second culture, the first spacer on the left is removed, the microscope pushed 1 inch toward the right tightly against the remaining spacers, the specimens brought into sharp focus, and a second frame of the film exposed. This procedure is repeated until the microscope-carrying cart is against the stopping spacer on the extreme right of the platform, and each culture has been photographed. To start the second series of photomicro- graphs, the microscope is moved to the left, all the aluminum spacers replaced on the plat- form, maintaining their original order, and the cart pushed tightly against them. This auto- matically places the microscope objective over the center of the first culture. After a predeter- mined time lapse, the process of photomicrog- raphy described above is repeated, taking a sec- ond photograph of each one of the cultures, covering, in each case, exactly the same field as in the first run. The exact time of day a photograph is taken is noted down, and later 326 PACIFIC SCIENCE, Vol. XVII, July 1963 1 H '•••••••Si : ' : • , 5JWs**1 v «»**•.*« • .*?*•*•* *A!‘*.»* • *««*.*. I*.* ••«•<*# ,*#♦*•■ *f*V *.♦•**«*;? «**;<*::* :&5 k *;*%:*:« *• » • i*i*y*«* ♦•**••, * * *****1 * & * , *.* • * * « IK M * n*v’Vt * **v ? *t* V.+t *.» * 5* #; ****»*>'*•* ■> ; a „ Fig. 4. Examples of time-lapse photomicrographs taken with the device. Each vertical column represents one sample culture in an X-irradiation experiment. Column 1 is the control, or Or; column 2 was subjected to 5r; and column 3 to 20r. The top horizontal row was taken at 1:25 P.M. or 85 min post-fertilization. The second horizontal row was taken at 1:30 P.M.; the third row at 2:15 P.M.; and the fourth row at 3:00 P.M. or 180 min post-fertilization. Photomicrographic Records-— Hsiao, Fujii, and Fine 327 transcribed onto the print. In one of our ex- periments, for instance, six cultures of eggs, each culture having been treated with a differ- ent dose of X-irradiation, were photographed successively within 3 min. In this way, a series of time-lapse photomicrographs, each bearing the time of day when it was taken, was made for each sample culture from the zygote to a desired stage, such as morula, and the resulting photographic prints analyzed and the rate of cleavage calculated. In Figure 4 selected exam- ples of sequential photographs are reproduced to show the stability of the relative position of the eggs and their structural change during cleavage. To identify each egg and estimate its rate of cleavage, a tracing of the first photograph of each sample is made on transparent paper. The drawing of each egg to be followed is given a number. By superimposing the tracing on the second or any subsequent time-lapse photograph of a particular sample culture, the eggs can be individually identified. The time it takes for an egg in a given sample culture subjected to a specific treatment to attain a certain developmental stage, (for instance, the attainment of the four-cell stage by a zygote after acute exposure to 20 roentgens), is deter- mined by subtracting the time of fertilization from the time the picture was first taken which shows the desired stage of the egg. Thus, in one experiment the eggs were fertilized at 12:00 noon, the two-cell stage of egg No. 38 first appeared in the photograph taken at 1:30 P.M.; the four-cell stage at 2:15 P.M.; and the eight-cell stage at 3: 15 P.M. Thus we can deduce that it took 90 min for 1st cleavage, 135 min for 2nd cleavage, and 195 min for 3rd cleavage. DISCUSSION 1. Blum and Price (1950) described an ap- paratus for following photographically the cleavages of sea urchin eggs. They pointed out two advantages of the photographic method: ( 1 ) more accurate timing of the cleavages than fixing followed by counting; (2) the cleavages of each individual egg can be followed and recorded. Blum and Price used two microscopes in the inverted position, one for U-V-ray irra- diated eggs, and the other for controls. The present device requires only one microscope. 2. In our instrument, observation is not lim- ited to one treatment, i. e., one sample of treated animals compared with its control. In the pres- ent device, a large number of samples, each re- ceiving a special treatment, can be compared with one or more untreated control or controls, using only one photomicrographic apparatus. A stand 28-inch long as used in our apparatus, can accommodate 20 or more samples at one time. The only limit to the number of samples is the time required for making one run of pho- tography through the whole set of samples, and that required for making photographic print-analyses later. With this apparatus (as with Blum and Price’s) the change in cleavage rate due to treatment such as X- irradiatic® 'Can be recorded. For example an increase or de- crease in the rate of cleavage can be observed; or initial retardation followed by recovery is made apparent. Similarly, the course of abnor- mal development can be followed; some of the abnormalities may disappear, others may per- sist. Fixing of eggs and later examination can not provide a sequential series of recorded events, especially recoveries. Furthermore, the eggs from a single female can be used in all of these samples, thus eliminating individual variations resulting from the use of batches of eggs from different females. 3. The instrument described here is one ver- sion of our device. Other modifications both in the material used and in details of construction are possible. For instance, we have also applied two coiled springs symmetrically to the micro- scope-carrying cart in order to pull the entire carriage tightly against the aluminum spacers. Another modification is the use of a light cool- ing and filtering device. A shallow transparent tray containing cold water can be placed on the plate glass between the culture samples and the light source. ACKNOWLEDGMENT We are indebted to Dr. Walter R. Steiger and Mr. George Yokotaki of the University of Hawaii physics department for their advice and machine-shop help. 328 PACIFIC SCIENCE, Vol. XVII, Inly 1963 SUMMARY A simple device for taking time-lapse pho- tomicrographs of a group of cultures of eggs to show the development of each individual has been described. This device is made of readily available material. It consists of two parts: (1) A piece of plate glass supported by a frame which is steady and independent of the motion of the photomicrographic appara- tus. It serves to replace the stage of the micro- scope and to carry the eggs in cultures in a motion-free state. (2) A carriage for the pho- tomicrographic apparatus which can be made to travel back and forth in a straight line from left to right, repeatedly bringing the micro- scope and its camera to a specific field in each culture for sequential photomicrography. The device has the advantage of being sim- ple, easy to construct and operate, and it can be used to follow photographically an individ- ual egg during its early development from zy- gote to morula as well as to record the appear- ance, disappearance, or persistence of develop- mental abnormalities. Other recurring phe- nomena can also be studied photomicrograph- ically by this instrument. REFERENCES Blum, H. F., and J. P. Price. 1950. Delay of cleavage of the Arbacia egg by ultraviolet radiation. J. Gen. Phys. 33:285-304. Hsiao, S. C., and P. C. Daniel. I960. Studies on the effect of ionizing radiation upon de- veloping sea urchin. Anat. Rec. 137: 365. Gray, P. 1958. Handbook of Basic Microtech- nique. McGraw-Hill Book Co., New York. Pp. 24-39. Shillaber, C. P. 1944. Photomicrograph in Theory and Practice. John Wiley & Sons, New York. Revision of the Genus Pandanus Stickman, Part 15 Malayan Species Described by H. N. Ridley Harold St. John1 As IN MANY OTHER GENERA Henry N. Ridley, director of the Botanic Garden, Singapore, made noteworthy contributions to the knowledge of the Malayan species of Pandanus. His species were validly published in local journals or in his books. He described them briefly, usually giving some details of the stem, leaves, appear- ance of the syncarp, length of style, and width of the exposed tip of the drupes. He cited specimens but did not adopt the type method. None of his species were illustrated. The writer has investigated Ridley’s species, sought the holotypes or has chosen lectotypes in the Singa- pore herbarium. These specimens were kindly made available by the present director, Dr. H. M. Burkill. Ridley’s species of Pandanus are nearly all good, and these are to be maintained. He placed most of his species in either the section Rykia or in Aero stigma. As will be noted in the treat- ment that follows, additional sections are to be found among his species. The large majority of Ridley’s species are treated in this article; and the few remaining ones will be illustrated and described in subsequent parts when their type specimens have been located and studied. SECTION Aero stigma Pandanus aurantiacus RidL, Roy. Asiat. Soc., Straits Branch, Jour. 41: 49 (1904) = 1903; FI. Malay Penin. 5: 81, 1925; Martelli, Webbia 4(2): t. 30, figs. 10- 13, 1914; (sect. Aero stigma) Fig. 172 DIAGNOSIS OF LECTOTYPE: "Large branching shrub with stems 3.6 m. tall, 5-7.5 cm. through”; leaves "glaucous green”; the lower bracteal leaf 1.26 m long, 4.2 cm wide, coriaceous, 1 -ribbed 1 B. P. Bishop Museum, Honolulu 17, Hawaii, U. S. A. Manuscript received July 19, 1961. but 2 -pleated and in section M-shaped, the blade sword-shaped, from the base gradually tapering to the approximately 20 cm long subu- late apex which about 10 cm down is 1.5 mm wide, at midsection each side with 42 parallel secondary nerves, the base amplexicaul and un- armed but beginning 7-8 cm up the margins with serrations 0.6-0.9 mm long, 1-2.5 mm apart, pale; the midrib below from 21 cm up with serrations 0. 3-0.5 mm long, 2-10 mm apart, subulate-tipped; at midsection the mar- gins and midrib below with serrae 1-1.5 mm long, 1-3 mm apart; near the apex the margins and midrib below with rigid serrulations 0.4-0.6 mm long, 2-4 mm apart; pistillate inflorescence 43 cm long, 12-13 mm in diameter, trigonous, bracted, spicate, bearing in the terminal 20 cm 5 syncarps which are orange and subequal; syn- carps 6-7 cm long, 5-5.5 cm in diameter, ovoid to broadly ellipsoid, bearing about 160 drupes, these 21-22 mm long, 3.5-5 mm wide, 3-4 mm thick, lower 34 oblanceoloid, upper lA subulate, 4—6-angled, the body sides smooth, later with exposed fibers; pileus 10-11 mm long, of a lower half narrowly conic-pyramidal, the surface smooth, sharply angled, the angles continuing along the stout subulate style form- ing the upper half, arcuate proximally; stigma 3 mm long, linear, in a distal cleft running al- most to the apex; endocarp centering in lower lA, and 7-8 mm long, the walls pale, 0.3 mm thick; apical mesocarp a broad cavern 2 mm long; basal mesocarp fibrous up the sides, fleshy within. EXPANDED DESCRIPTION: Shrub to 5 m tall; stem with the leaf scars inconspicuous; prop roots 60 cm long; leaves to 2 m long, 4-5 cm wide, above clear green, below paler and glau- cous; foliage leaves little broadened at base, the margins beginning at 4 cm up with prickles 1.6-2 mm long, 2-5 mm apart, stout subulate, ascending; the midrib below beginning at 9 329 a cm. o zcvei FIG. 172. Pandanus aurantiacus Ridl., from lectotype. a, Infructescence, X Vl\ b, drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; d, drupe, lateral view, X 4; e , drupe, longitudinal median section, X 4; /, drupe, style, and stigma, apical view, X 4; g, leaf base, lower side, X 1; b, leaf middle, lower side, X 1; h leaf apex, lower side, X 1- Page 274: Revision of Pandanus, 15. Malaya — St. John 331 cm with prickles 3-3.5 mm long, 8-20 mm apart, very stout subulate, strongly reflexed; pis- tillate scape 15-40 cm long, bearing 3-9 syn- carps, these 5-7 cm long; drupes very hard until ripe when they become bright reddish orange and soft. LECTOTYPE: Singapore, Bukit Timah [Bukit Mandai Road], 1894, H. N. R[idley ], (SING). Specimen examined and herein designated as lectotype. SPECIMENS EXAMINED: Singapore, Pasir Pan- jang, 17 Oct. 1912, I. H. Bur kill & M. R. Hen- derson 6,802 (bo, SING). Malaya: Johore, Pulau Kukub, 1904, H. N. Ridley (paratype), (SING). Ridley, when publishing this species, cited three collections and mentioned Sarawak, but did not designate a holotype. The most com- plete of his original specimens is here chosen as the lectotype. Pandanus collinus Ridl., Mater. FI. Malay Penin. 2: 228, 1907; FI. Malay Penin. 5: 79, 1925; Martelli, Webbia 4(1): 10, 1913; (sect. Acrostigma ) Fig. 173 DIAGNOSIS OF LECTOTYPE: Shrub 1. 2-2.1 m tall; stem at summit 6 mm in diameter, brown- ish, smooth; leaves 25-45 cm long, 8-11 mm wide, firm chartaceous, apparently pale green on both sides and flat except near the base where 2 -pleated, 1 -ribbed, above unarmed, at the middle with 10-11 secondary parallel veins in each half, below the tertiary cross veins barely visible, gradually narrowed in the last 10 cm to a subulate tip 15-20 mm long and 0.6-0.8 mm wide, the base amplexicaul, un- armed, blackish purple, beginning at 2 cm from the base the margins with prickles 0.7-1 mm long, 3-7 mm apart, subulate, ascending, yellowish but red-tipped; the midrib below un- armed to well beyond the middle; at mid- section the margins with subulate tipped ser- rulations 0. 2-0.3 mm long, 1.5-3 mm apart, appressed ascending, red-tipped; near the apex the margins with serrulations 0. 1-0.2 mm long, 0.3-0. 5 mm apart; the midrib below with ser- rulations 0. 2-0.3 mm long, 0.5-1. 5 mm apart; pistillate inflorescence terminal, erect, bearing 1 head; peduncle 8 cm long, 3 mm in diameter, bracted; syncarp 3 cm long, 2.5 cm in diameter, subglobose, bearing about 88 drupes, these 14-15 mm long, 4-5 mm wide, 3-4 mm thick, the body ellipsoid to obovoid, 5-6-angled, 10 mm long; pileus with the base 3.5-4 mm high, semiorbicular or slightly narrower, rather smooth, but with longitudinal ridges and lines; style 4-5 mm long, abruptly subulate, angled, sharply proximally curved, bony; stigma 3-4 mm long, elliptic-linear, brown, papillose, ex- tending to the tip; endocarp centering in lower Vb but extending to the base, the walls 0.1 mm thick, cartilaginous, stramineous; seed 7.5-8 mm long, lance-obovoid; upper mesocarp an oblate hemisphere, filled with white medullary mem- branes; basal mesocarp fibrous up the sides, fleshy within, sparse. LECTOTYPE: Malaya, Kedah, Kedah Peak, woods, 1893, H. N. R[idley }, (SING). Specimen examined and here designated as lectotype! DISCUSSION: P. collinus belongs in the section Acrostigma. It was described by Ridley from three collections from Malaya. The best of these and the one labeled P. collinus! by Ridley is here chosen as the lectotype. Contrasting dif- ferences are given for this in our treatment of P. alticola. Pandanus glaucophyllus Ridl., Roy. Asiat. Soc., Straits Branch, Jour. 41: 50, (1904) = 1903; ( sect. A cro stigma ) Fig. 174 DIAGNOSIS OF holotype: "Stems 3-4 m. tall, 12-25 mm. in diameter,” the internodes brown to yellowish, smooth, mostly higher than the nodes; leaves 52-60 "or more” cm long, near the base about 4 cm wide, near the apex 5. 7-6.5 cm wide, appearing narrowly oblong, rather flaccid and when dry chartaceous, glaucous, 1- ribbed but 2 -pleated and in section low M- shaped, the secondary parallel veins conspicuous and 34 of them in each half, the apex abruptly contracted to a 6-7 cm trigonus subulate caudate prolongation which near its base is 1 mm wide, the base of the blade amplexicaul and unarmed, but beginning at 4 cm up the margins with prickles 0. 5-0.8 mm long, 1-3 mm apart, slen- der subulate, pale, ascending or diverging and some of them doubles or triples; the midrib 5 mm Fig. 173. Pandanus collinus RidL, from lectotype. a, Habit and syncarp, X 1; drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1;^ drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe, style, and stigma, apical view, X 4; g, style and stigma, distal view, X 10; k, leaf base, lower side, X 1; i, leaf middle, lower side, X 1 \ j, leaf apex, lower side, X i. acm. Fig. 174. Pandanus glaucophyllus Ridl., from holotype. a, Habit and syncarp, X Vl\ b, drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; d, drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe, style, and stigma, apical view, X 4; g, leaf base, lower side, X 1; h, midrib, lower side, near base, X 4; i, leaf middle, lower side, X 1; i> leaf margin, near middle, X 4; k, leaf apex, lower j side, X 1* 334 PACIFIC SCIENCE, Vol. XVII, July 1963 below beginning at 4-9 cm with prickles 0.8-1 mm long, 0.5-6 mm apart, slender subulate, brownish, ascending, oblique, closely appressed; at midsection the margins with prickles 0.5-0.7 mm long, 1-3 mm apart, slender subulate, as- cending flat appressed and occupying recesses in the margin; the midrib below glabrous or with a few prickles like those lower down; at the bend of the apical contraction the margins with prickles 0.6-0.8 mm long, 0.6-1. 5 mm apart, slender subulate, somewhat ascending or divergent; the midrib with prickles 0.6-1 mm long, 1-2 mm apart, arcuate subulate, ascend- ing; on the caudate tip the margins and mid- rib below with prickles 0. 3-0.7 mm long, 0.3-1. 5 mm apart, stout arcuate subulate, as- cending; pistillate peduncle 8 cm long, 7 mm in diameter, erect, bearing a single syncarp 5.7 cm long, 4.5 cm in diameter, broadly ellipsoid, bearing about 320 drupes, these 12-13 mm long (or 16-18 mm following the curve of the stigma), 5-6 mm wide, 3. 5-4.5 mm thick, the body narrowly oblong ellipsoid, 5-6-angled, the sides smooth; pileus smooth, the base 3-3.5 mm long, conical or broadly so, bearing the terminal style 5-6.5 mm long, stout subulate, strongly curved proximally, having a distal cleft nearly its whole length in which is the stigma 3-4 mm long, linear, brown, papillose, running to the tip; endocarp pale brownish, centering in the lower (4, its walls 0.1 mm thick; seeds 7-8 mm long, oblance-oblong; apical mesocarp a cavern 2-2.5 mm long, invaded by trichomes and hairs; basal mesocarp with fibers up the sides and the center fleshy. HOLOTYPE: Malaya, "Perak, Thaiping Hills, in dense forest about 2500 feet altitude," above tea border, Dec. 1898, H. N. Ridley (sing). Holotype examined! Pandanus parvus Ridl., Roy. Asiat. Soc., Straits Branch, Jour. 33: 71, Jan. 1900; FI. Malay Penin. 5: 78, 1925; P. flagellifer Warb., in Engler’s Pflanzenreich IV, 9: 80, Dec. 21, 1900; Martelli, Webbia 4(1): 14, 1913; 4(2): t. 33, f- 7, 1914; (sect. Aero stigma) Figs. 175, 176 DIAGNOSIS OF HOLOTYPE: "Stem 2.92 cm. tall, 12 mm. in diameter, slender, branched above,” rooting throughout; leaves 27-37 cm long, near the tip 32-36 mm broad, near the base 16-20 mm broad, appearing ligulate, but actually tapering downwards, thin subcoriaceous, dark green above, pale and glaucous beneath, 2 -pleated and in section low AA-shaped but only 1 -ribbed, the midrib below narrow and raised, the secondary parallel veins prominent through- out, and at midsection 18-21 on a side, the terti- ary cross veins faintly visible below near the tip, making small squarish meshes, near the apex ab- ruptly rounded contracted to a 3-6 cm caudate deltoid subulate tip 0.8 mm wide, the base am- plexicaul, unarmed for 4-5 cm, then the margins with prickles 0.7-1. 3 mm long, 2-7 mm apart, slender subulate, straight or slightly curved, brownish, ascending; the midrib below unarmed to near the apex; at midsection the margins with similar prickles 0.5-1 mm long, 4-12 mm apart; on the rounded apical contraction the margins with prickles 0.5-0.8 mm long, 0.5-2 mm apart, stouter subulate, slightly ascending; the midrib below with prickles 0.8- 1.4 mm long, 1-7 mm apart, arcuate subulate, slightly ascending; on the caudate tip the margins and midrib below with stout prickles 0. 3-0.8 mm long, 0.5-1. 5 mm apart, subulate, arcuate, ascending; in the series of bracteal leaves the first several 2-4 cm long, lance-navicular, entire; pistillate in- florescence 1 -headed; peduncle 5 cm long, 4-5 mm in diameter, 3 -angled, few bracted; syncarp erect, 3.5 cm long, 3.3 cm in diameter, ovoid- globose, bearing about 150 drupes, these 11-12 mm long or following the curve 14-15 mm long, 3. 5-5. 5 mm wide, 2. 5-3. 5 mm thick, fusi- form, the body 7-8 mm long, oblanceoloid, the surface smooth, later fibrous; pileus smooth, apparently glaucous, 9-10 mm long, the base oblong-pyramidal and with the apex oblique and proximally curved, tapering into a hornlike style which is 5-6 mm long, all (except the apical ones) sharply proximally curved, stigma 4-5 mm long, lance-linear, dark brown, papil- lose, extending to the tip; endocarp in lower V4, stramineous, bony, the walls 0.3 mm thick, and the lateral walls extending upwards to bound the apical cavity; seed 5 mm long, ob- lanceoloid, truncate at tip; apical mesocarp a cavern filled with white medullary ascending hairs; basal mesocarp sparse, fibrous and fleshy. o Fig. 175. Pandanus parvus Ridl., from holotype. a, Habit and syncarp, X 1| b, drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1 \ d, drupe, lateral view, X 4; e, drupe, longitudinal median sec- tion, X 4; f, drupe and stigma, apical view, X 4; g, leaf base, lower side, X 1; h, leaf middle, lower side, X 1; i, leaf apex, lower side, X 1- 336 PACIFIC SCIENCE, Vol. XVII, July 1963 1 1 7 cm O-J Fig. 176. Pandanus parvus Ridl., from Singapore, cult., 1895, H. N. Ridley, a , Staminate inflorescence, X 1; b, stamens, lateral view, X 10; Q floral bract, lateral view, X 1- expanded DESCRIPTION: Pistillate stems 0.8-2.92 m tall, 6-12 mm in diameter; prop roots up to 50 cm long, 4-5 mm in diameter, smooth, green to brown, nearly vertical; leaves 27-50 cm long, 21-48 mm wide, the caudate apex 3-9 cm long; peduncle 5-12 cm long; syncarp 2.5-4 cm long, 2.5-33 cm in diameter, globose or ovoid-globose. DESCRIPTION OF staminate PLANT: Stami- nate inflorescence 15-18 cm long, bracts white and somewhat fleshy; peduncle 3-4 mm in di- ameter, fleshy, soon withering and caducous; lowest bracts 8-13 cm long, 26-30 mm wide, sterile, the body 6-9 cm long, navicular, with prominent parallel veins, the lower half with margins unarmed; near the apex the margins Page 280: Revision of Pandanus, 15. Malaya — St. John 337 with prickles 0.5-1 mm long, 0.3-1 mm apart, subulate, diverging or subascending; the mid- rib with stouter prickles 0.7-1 mm long, 0.5-1 mm apart, slightly ascending; on the caudate tip the margins and midrib below with prickles 0.5-0.9 mm long, 0.5-1 mm apart, arcuate subulate, ascending; median bracts 3-4.5 cm long, 1.6-2 A cm wide, lanceolate, the margins and the midrib below towards apex minutely prickly; spikes 5-8 and 1.5-2. 5 cm long, 0.8- 1.3 cm in diameter, ellipsoid, dense; stamens very numerous, attached singly and directly to the axis; filament 0.2 mm long; anther 4.5-5 mm long, 0.2 mm wide, bearing an apical subulate prolongation of the connective 0.4-0.7 mm long. HOLOTYPE: Singapore, Kranji, 1892, H. N. R[idley J, (SING). Holotype examined! SPECIMENS EXAMINED: Singapore, Hinterin- dien, Dec. 16, 1905, A. Engler 3,862 (b); Singa- pore, Bidadari, 1898, H. N. R{idley ] 9, 171 ( FI ) ; Singapore, cult., first male flowers seen, 1895, H. N. R[idley ], (fi, sing); without col- lector, ex Hb. (SING), Forest Iurong, low grow- ing plant, Jan. 10, 1889, Flora of Singapore 92 ( B ) , the holotype of P. flagellifer. Malaya: Johore, Sungei Pelepah Kiri, 26 June 1928, E. J. H. Corner 33,336 (sing). Pandanus perakensis Ridl., Mat. FI. Malay Penin. 2: 231, 1907; FI. Malay Penin. 5: 81, 1925; (sect. Aero stigma) Fig. 177 diagnosis of LECTOTYPE: (Probably acau- lescent); leaves 1.5-1. 8 m long, 2-3.6 cm wide, subcoriaceous, light green, linear, rather ab- ruptly narrowed to the 9 cm subulate, trigonous apex, this 10 cm down 6 mm wide, in section the blade AA-shaped, at midsection with 39—41 secondary veins in each half; near the base the margins with spines 2-3 mm long, 5-13 mm apart, stout subulate, arcuate, pale, broad based, ascending; the nearby midrib below with spines 2.5-3 mm long, 6-13 mm apart, similar but even thicker subulate and reflexed; at mid- section the margins with subulate-tipped serrae 0.5-1. 5 mm long, 4-8 mm apart, the midrib below with prickles 0.5-1 mm long, 8-15 mm apart, subulate; near the apex the margins with serrulations 0. 3-0.5 mm long, 2-3 mm apart; the midrib below with similar ones but 2-4 mm apart; peduncle 15 cm long, 8-9 mm in diameter, straight, bracted; syncarps 3-4, spicate, crowded, 4. 5-6. 5 cm long, 3. 5-4. 5 cm in di- ameter, elliptic-cylindric (rarely subglobose), appearing spiny from the numerous curved as- cending styles ( but a few at the very base reflexed), bearing about 430 drupes; drupes 20-22 mm long, 5-6 mm wide, 4-5 mm thick, oblance-fusiform, compressed, apparently green, upper l/z free, the body 13-14 mm long, ob- lanceoloid, the apex truncate, pileus at last shedding, the base semiorbicular, bearing a style 6-7 mm long, stout subulate, ridge angled, gently arcuate towards apex of syncarp; stigma 4-5 mm long, linear, distal, running to the very apex; endocarp 6-7 mm long, centering in lower V3, ellipsoid, truncate, stramineous, car- tilaginous, the sides 0.15 mm thick; apical meso- carp 3-3.5 mm long, a fat discoid cavern with a few white membranes; basal mesocarp fibrous along the margins, fleshy within. EXPANDED DESCRIPTION: Pistillate plants with peduncle 13-15 cm long; syncarps 2-3 and 4.5-7 cm long, 3. 5-5. 5 cm in diameter. LECTOTYPE: Malaya, "Perak, Maxwell’s Hill”; June 1893, H. N. R{idley ], (sing). Lectotype examined! SPECIMENS EXAMINED: Malaya, Pahang, Ku- ala Lipis, 1892, A. S. Machado 11,633 (SING); Kedah, Baling Hill, 21 Nov. 1941, E. J. H. Corner & J . C. Nauen (SING). Pandanus spinulosus (Ridl.) comb. nov. (sect. Acrostigma) \ P. collinus Ridl. var. spinu- losus Ridl., FI. Malay Penin. 5: 79, 1925. Fig. 178 DIAGNOSIS OF HOLOTYPE: Stem slender, just below the leaves 4 mm in diameter; leaves 67-68 cm long, 16 mm wide, subcoriaceous, green above, below paler and apparently some- what glaucous, with a central furrow over the midrib, 2 -pleated, and towards the tip the 2 pleats spinulose above, at midsection with 16-18 parallel secondary veins in each half, these prominent throughout, below the tertiary cross veins visible, remote and forming elongate meshes, the blade ligulate, in outer (4 tapering Fig. 177. Pandanus perakensis Ridl., from lectotype. a, Infructescence, X 1; b> c, drupe, lateral view, X 1; d, drupe, longitudinal median section, X 1; e, drupe, lateral view, X 4; f, drupe, longitudinal median sec- tion, X 4; g, drupe, style, and stigma, apical view, X 4; h, leaf base, lower side, X h leaf middle, lower side, XI \ i> apex, lower side, X 1- Fig. 178. Pandanus spinulosus (Ridl.) St. John, from holotype. a, Infructescence, X 1; ^ drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; d, drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe, style, and stigma, apical view, X 4; g, style and stigma, lateral view, X 10; h, leaf base, lower side, X 1; i, leaf middle, lower side, X 1; j, leaf apex, upper side, X 1; ^ leaf apex, lower side, X 1- o loom. J i _i _J „ 1 , -i.,.. ... L___ _l Fig. 179- Pandanus unguiculatus Ridl., from holotype. a. Syncarp, X 1; b, drupe, lateral view, c, drupe, longitudinal median section, X 1; d, drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe, style, and stigma, apical view, X 4; g, leaf base, lower side, X 1; h, leaf middle, lower side, X 1 ; i, leaf apex, lower side, X 1 • Page 284: Revision of Pandanus, 15. Malaya-— St. John 341 gradually to the subulate, deltoid apex about 3 cm long and 1.5 mm wide, the base amplexi- caul, unarmed, partly dark purple, beginning at 4 cm up the margins with prickles 0.7- 1.5 mm long, 3-5 mm apart, subulate, ascending at 45°, the base thickened, pale; the midrib below un- armed to beyond the middle; at midsection the margins with prickles 0.7-0.9 mm long, 2-4 mm apart, subulate, ascending; near the tip the margins and midrib below with subulate ser- rations 0.4-0. 6 mm long, 0.7-1 mm apart; pis- tillate inflorescence erect, bearing a single syn- carp; peduncle 10 cm long, 4-5 mm in di- ameter, 3-sided, bracteate; syncarp 4.5 cm long, 3.1 cm in diameter, oblong-ellipsoid, bearing about 120 drupes, these 15-17 mm long, 3-4.5 mm wide and thick, pale green, fusiform but the upper end contracted and more slender, 5-6- angled, the body 11-12 mm long; pileus 8-9 mm long, the base 2.5-3 mm high, narrowly semiorbicular, with 5-6 sharp angles, these mostly single; style 5-5-5' mm long, sharply curved proximally, flattened and ridged below, pale, bony; stigma 3-4 mm long, distal, linear, running to the apex, brown, papillose; endocarp centering in lower but extending to the base, the walls 0.1 mm thick, cartilaginous, pale; seed 8-9 mm long, extending to the base; apical mesocarp spherical, filled with white, medullary membranes; basal mesocarp sparse, fibrous. HOLOTYPE: Malaya, Kelantan, along the ridge of Gunong Sitong, 2,500 ft. alt., 6 March 1924, Mo harried Nur & Foxworthy 12,252 (sing). DISCUSSION: P. spinulosus is a member of the section Aero stigma, and in that section is found its closest relative, P. collinus Ridl., which species has the leaves 25-45 cm long, 8-11 mm wide, unarmed above, each half at midsection with 10-11 secondary parallel veins, at mid- section the margins with subulate-serrations 0.2-0. 3 mm long; syncarp 3 cm long; drupes 14-15 mm long, the pileus base hemispheric, abruptly contracted to the subulate style which is 4-5 mm long. P. spinulosus has the leaves 67-68 cm long, 16 mm wide, above towards the tip the two pleats spinulose, at midsection with 16-18 secondary parallel veins in each half, at midsection the margins with subulate prickles 0.7-0.9 mm long; syncarp 4.5 cm long; drupes 15-17 mm long, the pileus base narrowly hemi- spheric, gradually tapering into the thick subu- late style which is 5-5.5 mm long. Pandanus unguiculatus Ridl., Mat. FI. Malay Penin. 2: 229, 1907; FI. Malay Penin. 5: 78, 1925; (sect. Aero stigma) Fig. 179 DIAGNOSIS OF HOLOTYPE: "Plant small”; leaves 26-48.5 cm long, near the base 2-2.3 cm wide, near the middle 2. 4-2. 7 cm wide, ligulate, but perceptibly tapering both ways from the middle, dark green above, pale green below, thick chartaceous, only the midrib thick- ened, but the blade pleated and M-shaped in section, the secondary parallel veins conspicu- ous, at midsection 17-19 in each side, the ter- tiary cross veins visible in upper third below, oblique, forming short, rather oblong meshes; near the apex abruptly contracted to a subulate tip more than 20-25 mm long, 0.9 mm wide, the base amplexicaul and entire but beginning at 4-5 cm the margins with prickles 0.7-1 mm long, 2-6 mm apart, subulate, ascending, pale; the midrib unarmed for lower 34 s; at the mid- section the margins with prickles 0.5-0.7 mm long, 2-5 mm apart, subulate, appressed as- cending, "red-tipped” but when dried brownish; near the apex the margins and midrib below with subulate-tipped serrulations 0.2-0. 3 mm long, 0.5-2 mm apart; peduncle 5 cm or more long, 4 mm in diameter, bracted; syncarp sin- gle, 5.5 cm long, 4 cm in diameter, broad ellips- oid, bearing about 220 drupes, these 16-18 mm long (or 18-20 mm, following the curve), 4-5 mm wide, 3.5-4 mm thick, oblance-fusiform, the body 14-15 mm long, narrowly ellipsoid; the pileus smooth, pyramidal-ovoid, 5-7 mm high, tapering into the hornlike style which is 2.5-3 mm long, mostly upwards curved to about right angles; stigma 2. 5-3. 5 mm long, distal, linear, papillose, brown; endocarp slightly sub- median, 9—1 1 mm long, oblanceoloid, the apex truncate, but with ascending shoulders partly surrounding the apical cavity, stramineous, carti- laginous, the lateral walls 0.15 mm thick; seed 7-8 mm long; apical mesocarp forming a broadly ovoid cavern with a few white medul- lary membranes; basal mesocarp scant, fibrous and fleshy. Fig. 180. Pandanus stelliger Ridl., from lectotype. ar-d, Drupe, lateral view, X 1; e> drupe, longitudi- nal median section, X 1; f, g, drupe and stigma, apical view, XI \ h, drupe, lateral view, X 4; i, drupe, longitudinal median section, X 4; j, drupe and stigma, apical view, X 4; /, leaf base, lower side, XI \ m, leaf middle, lower side, X 1; n, leaf apex, lower side, X 1- Page 286: Revision of Pandanus , 15. Malaya- — St. John 343 EXPANDED DESCRIPTION FROM ALL SPECI- MENS EXAMINED: Stems up to 1.8 m tall, 9-13 mm in diameter, smooth, yellowish, erect or ascending, with aerial roots; leaves arching out- ward and the tips pendent, glaucous below, the subulate tip 20-46 mm long; inflorescence erect; peduncle 5-6 cm long. HOLOTYPE: Malaya, Selangor, Bukit Kutu, May 1896, H. N. R{idley'] 7,659 (sing). Holo- type examined! SPECIMENS EXAMINED: Malaya, Pahang, Fra- sers Hill, 19 Aug. 1937, */. H. Corner 33,233 (SING). SECTION Aster o stigma Pandanus stelliger Ridl., Roy. Asiat. Soc, Straits Branch, Jour. 41: 49-50, (1904) = 1903; FI. Malay Penin. 5: 76, 1925; Mar- telli, Webbia 4(2) : t 26, f. l-2a, 1914; (sect. Aster o stigma) Figs. 180, 181 DESCRIPTION OF LECTOTYPEl Shrub 1.5-33 m tall, the slender stems pale brown, 2.5 cm in diameter, branched, bearing a few mucros; leaves dark green, 55-60 cm long, near the base Fig. 181. Pandanus stelliger Ridl., from lectotype, habit and infructescence, X Vi. 344 PACIFIC SCIENCE, Vol. XVII, July 1963 1.8 cm wide, near the middle 2. 8-3. 8 cm wide, in section with a broad, central channel, and hence M-shaped, the secondary parallel veins at midsection 19-20 in each half, prominent, the tertiary cross veins visible below, transverse or oblique, forming short, somewhat oblong meshes, the actual base amplexicaul and un- armed, but beginning at 2-2.5 cm the margins with prickles 1.5-2. 5 mm long, 4-9 mm apart, broad subulate, ascending, pale; the midrib unarmed except near the tip; at midsection the margins with prickles 2. 1-2. 5 mm long, 7-23 mm apart, similar; near the 3-8 cm apex the margins and midrib below with serrulations 0.2-0. 3 mm long, 2-6 mm apart, brown; pe- duncle terminal, 5 cm long, leafy bracted, bear- ing 4 syncarps in a dense spike about 10 cm long; syncarps subequal, 5-6 cm long, 3-5 — 4.5 cm in diameter, ovoid to broadly ellipsoid, bear- ing 64-88 drupes, these 19-21 mm long (in- cluding the stigma), 7-10.5 mm wide, 6-8 mm thick, oblanceoloid with an obtusely convex apex, the sides 5-6-angled, smooth, when dried brown, the body 15-17 mm long, cuneiform; pileus 6-7 mm long, oblate semiorbicular; style subterminal, 0.5-1 mm long, vertical except on the basal drupes where sharply reflexed distally; stigmas on the expanded apical surface, 4—7 mm Jm diameter, 4-7 -pointed, horizontal or mushroom-shaped, papillose, brown; drupe body 14-17 mm long; endocarp centering in lower 34, bony, light brown, the inner surface shining, the lateral walls 0.5 mm thick, the apical plug 1—1.3 mm thick; seeds 6-7 mm long, obdeltoid; apical mesocarp with a cavity 6-7 mm long, with a few white membranes near the margins; basal mesocarp sparse, fibrous and fleshy. EXPANDED DESCRIPTION OF ALL PISTILLATE PLANTS EXAMINED: Plants 1.5-4 m tall; leaves 55-113 cm long, at midsection with 1 9—2 9 sec- ondary parallel veins in each half; peduncle 5—7 cm long, bearing 2-5 syncarps, these 4-6 cm long; drupes 7-13 mm wide, 6-10 mm thick, bright yellow. DESCRIPTION OF STAMINATE PLANTS: Leaves 63-70 cm long, near the base 20-21 mm wide, near the middle 26-29 mm wide, at midsection the parallel secondary veins 22-24 in each half, and at midsection the margins with appressed ascending serrulations 0. 3-0.4 mm long, 7-12 mm apart; staminate inflorescence 47 cm long, leafy bracted; lowest bract 47 cm long, 2.2 cm wide, for the lower 10 cm slightly distended, sheathing and paler, the rest foliaceous; floral bracts thin, apparently white, the median one 1 5 cm long, 4.6 cm wide, lanceolate, on the outer 34 the margins and midrib below with serrula- tions 0.1 mm long, 0.2-1 mm apart; the 7 spikes 3-6 cm long, 12-15 mm in diameter, cylindric, dense; fascicle 5-8 mm long, with 19-22 stamens, the naked common column 3-4 mm long, rhachis 2-2.5 mm long, clavate; free filament tips 1-2 mm long, divergent; anthers 1.5-2 mm long, linear-oblong, bearing a pale subulate projection of the connective 0.2-0.3 mm long. LECTOTYPE: Malaya, "Selangor, on the Tras Route at the 20th mile (8775),” 2,000 ft. elev., July 1897, H. N. Ridley (SING). Lectotype here designated; specimen examined! Isotype ( FI ) ! SPECIMENS EXAMINED: Malaya: Perak, Larut, top of mountains, open junfgle], 2,000 to 3,000 ft. alt., April 1883, H. Kunstler 4,166 (cal); Haram Parak, June 1884, Scortechini ( CAL ) . Pahang: Gunong Tahan, 7 Sept. 1937, E. J. H. Corner (SING) ; Sungei Chelia, Chegar Perah, 14 Oct. 1927, M. R. Henderson 19,370 (sing). Penang: Government Hill, Dec. 1895, C. Curtis (SING); Waterfall Hill, 500 ft. alt., June ’88, L. Wray Jr. 2,227 (cal) . SECTION Hombronia Pandanus irregularis Ridl., FI. Malay Penin. 5: 76-77, 1925; Henderson, M. R., Roy. Asiat. Soc., Malay Branch, Jour. 17: 82, 1939; (sect. Hombronia) Figs. 182, 183 DIAGNOSIS OF HOLOTYPE: "Leaves long, very broad,” 9-3 cm wide, thick and leathery, 1- ribbed, 2 -pleated, in section depressed, AA- shaped, the secondary parallel veins visible be- low, scarcely so above and at midsection 83-92 in each half, the base amplexicaul and unarmed, but beginning at 12 cm up the margins with thorns 4.5 mm long, 1-2 cm apart, heavy deltoid subulate, black-tipped, ascending; the midrib below with similar reflexed thorns 2-3 cm apart; to cm. Fig. 182. Pandanus irregularis Ridl., from holotype. a, Infructescence, X lA ; b, drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; d, drupe and stigmas,' apical view, X 1; e> drupe, transverse median section, X 1; f, leaf base, lower side, XI \ g> leaf middle, lower side, X 1- 346 PACIFIC SCIENCE, VoL XVII, July 1963 Fig. 183- Pandanus irregularis Ridl., from holotype, leaf apex, lower side, X I- on a leaf section cut higher up (exact place unknown) the margins with spines 3-3.5 mm long, 25-36 mm apart, heavy subulate, ascend- ing; the midrib below with scars of fallen spines 3-6 cm apart; pistillate inflorescence 30 cm long, with about 7 syncarps; peduncle 15 mm in diameter, obtusely trigonous, bracted; syn- carps 6. 5-7.5 cm long, subglobose, bearing about 36-44 drupes, these 1-3- but mostly 2- celled, 3-3.5 cm long, 1.8-2.4 cm wide, 1.2-1. 5 cm thick, "dark green,” narrowly obovoid to oblanceoloid, compressed, the apex entire, ob- tuse, the sides smooth, convex, lateral sutures none; stigmas almost all eroded but 1 basal drupe bears a single stigma 5 mm long, lanceo- late, sharply bent to the horizontal, the recep- tive surface distal; endocarp 22 mm long, cen- tering in upper 36, at midsection filling the drupe, bony, massive, dark brown but including pale fibers, the lateral walls 4-5 mm thick; seeds 8-9 mm long, 3 mm in diameter, ellipsoid; api- cal mesocarp cavernous, with strong, longitudi- nal fibers and soft pith; basal mesocarp fibrous and fleshy. EXPANDED DESCRIPTION OF ALL SPECIMENS EXAMINED: Leaf more than 1.4 m long, near the base 9 cm wide, near the middle 12 cm wide, near the apex 10.5 cm wide, broad ligu- late; margins near the base with thorns 3. 5-4.5 mm long, 1-2.5 cm apart; midrib below with thorns 3-3.5 mm long, 2-5 cm apart, subulate, reflexed; at midsection the margins with spines 4-4.5 mm long, 20-25 mm apart, sharply as- cending, subulate but the base much thickened; the midrib below concealed; the apex apparently damaged in growth and without a caudate apex, rather contracted to 2 broadly obtuse lobes and deeply emarginate, and even prickly on the margins of the cleft, all this, however, appar- ently an abnormality; peduncle 30 cm long; Page 290: Revision of Pandcmus, 15. Malaya — St. John 347 spike 14 cm long, bearing 4 syncarps; stigmas 4 mm long, sharply proximally bent, the recep- tive surface usually proximal. HOLOTYPE: Malaya, Kelantan, Sungei Ketah, Batu Bau (Bow), 15 Feb. 1924, Mo h anted Nur & F oxworthy 12,104 (SING). Holotype exam- ined! SPECIMENS EXAMINED: Malaya, Pahang, Gua Tipus, the common pandan on the limestone, 17 Oct. 1927, M. R. Henderson 19,468 (sing). DISCUSSION: P. irregularis was placed by Dr. Ridley in the section Rykia. It has, however, the drupes 1-3 -celled, and usually 2 -celled, with the cells in linear arrangement, and the large stigmas terminal, but distal and laterally di- rected. There is no question but that it must now be placed in the section Hombronia. It is one of the numerous plant species known to occur only on limestone knobs. These pre- cipitous pinnacles or narrow ridges rise spec- tacularly from the rolling uplands of Thailand and Malaya. Their sides are often precipices of shining, white limestone, sparsely vegetated, but their crests have a dark evergreen, virgin forest cover. To a botanist they are very attractive. Four species of Pandanus in Malaya are re- stricted to these calcareous habitats. Besides this species published by Ridley, two more are now being described, and one more has yet to be found in fruit. A good account of this localized flora is given by Dr. M. R. Henderson (Roy. Asiat. Soc., Malay Branch, Jour. 17: 13-87, pi. Ill — XI, 1939). Malaya, sect. nov. Phalangibus pluricarpellatis, stylis terminali- bus adscendentibus vel arcuatis corniformatis in- tegris vel bifidis, stigmatibus centripetalibus el- lipsoideis vel lanceolatis; plantis masculis in- cognitis. Phalanges of several united carpels; styles ter- minal, ascending or arching, stout, hornlike with a subulate apex or compressed and more or less oblong, with the apex sharply bifurcate; stig- mas centripetal, ellipsoid or lanceolate; stami- na te plants unknown. HOLOTYPUS: Pandanus Klossii Ridl., Fed. Malay States Mus., Jour. 6: 190-191, 1915. DISCUSSION : The single known species occurs on the mountains of Pahang in Malaya. It com- bines the characteristics of the section Rykia in its hornlike, simple or bifurcate styles, and of the section Pandanus, in its many-carpellate, large phalanges. Since these two sections are very distinct and are only remotely related, it is best to make the very bizarre P. Klossii the basis of a new section.2 Pandanus Klossii Ridl., Fed. Malay States Mus., Jour. 6: 190-191, 1915; FI. Malay Penin. 5: 74, 1925; (sect. Malaya) Figs. 184, 185 DIAGNOSIS OF HOLOTYPE: Shrub 2-7 m tall; stems usually solitary, 7.5 cm in diameter, gray, erect or arcuate; leaves 1.3-1. 8 m long, 6.8-72 cm wide, thick coriaceous, above dark green, below pale green, 1 -ribbed, but 2-pleated and in section low M-shaped, the longitudinal paral- lel secondary veins 0.5-0.8 mm apart, 49-52 in each half, evident on both sides as are the tertiary cross veins that form a network of meshes, oblong or rhombic, longer than broad, the blade ligulate, and only near the tip rather abruptly narrowing to a 5 cm stout subulate tip, trigonous and 2-3 mm wide, the very base not collected, near the base the margins with stout thorns 5-8 mm long, 8-30 mm apart, arcuate ascending, black; the midrib below with similar but reflexed thorns; at midsection the margins with prickles 3 mm long, 15-50 mm apart, subulate, ascending, flat appressed; the midrib below with prickles 2 mm long, 35-50 mm apart, heavy, arcuate, ascending; near the tip 2 The new section can be inserted in the author’s key in this Revision of the Genus Pandanus (p. 2) by replacing the 11th line from the bottom by: L. Stigmas elongate, acerose, or stout hornlike, simple or bifurcate, a. Stigmas stout hornlike, simple or bifurcate Malaya a. Stigmas elongate, acerose, FIG. 184. Pandanus Klossii Ridl., from holotype. a, Phalange, lateral view, X 1; phalange, longitudi- nal median section, X 1', c, phalange, and stigmas, apical view, X 1; d, carpel apices and stigmas, apical view, X 4; e, style and stigma, oblique view, X 4; f, leaf base, lower side, X 1; ^ leaf middle, lower side, X 1; I), venation in outer third, lower side, X 4; i, leaf apex, lower side, X 1- Page 292: Revision of Pandanus, 15. Malaya — St. John 349 the margins and midrib below with serrations 0. 3-0.5 mm long, those of the margin 1-1.5 mm apart, those of the midrib below 2-5 mm apart; pistillate peduncle 15-30 cm long, 12-18 mm in diameter, 3-sided, leafy bracted; syncarp sin- gle, globose or ellipsoid, about 15 cm in di- ameter; phalanges numerous and 6-6.5 cm long, 3. 6- 3. 9 cm wide, 2-2.7 cm thick, cuneiform, 4-6-angled, "showing signs of turning red or orange,’’ the sides smooth, gently convex, free in upper J4, the apex low convex, lateral sutures visible only in upper Vy central apical sinuses 1- 4 mm deep, straight or arcuate, narrow; car- pels 9-10 (in a double 16-17 and the phalanges 5.7- 6 cm wide), the apices semiorbicular , faintly angled; styles 4-6 mm long, dark, bony, either simple and hornlike or flattened and sharply bifid, mostly centripetal; stigmas as long as the styles, ellipsoid or lanceolate, brown, pap- illose, covering the proximal face; proximal sinus none; endocarp in lower 14 and 2.7 cm long, bony, massive, dark brown, the lateral walls 2 mm thick, the inner surfaces shining and rugulose; seeds 14-16 mm long, 2-3 mm in diameter, oblong-ellipsoid; upper mesocarp 2- 2.5 cm long, of dense pith and a few fibers; basal mesocarp fibrous and fleshy. ADDED DESCRIPTION FROM ALL SPECIMENS EXAMINED: Phalanges about 96; carpels 5-10; syncarp 9-15 cm in diameter. HOLOTYPE: Malaya, Pahang, Gunong Tahan, July 1911, H. N. Ridley 16,037 (SING). Iso- type (bo). Ridley stated, "Common all over the Padang. In the more open exposed spots the stem is short and erect, about 6 to 8 feet tall; in the woods the stems are long and weaker, often falling about at all angles, 20 feet or more long. No trace of male flowers could be seen anywhere. The plant is very abundant, al- most filling up the woods in some places." The collection, Ridley 16,037, was made by Ridley previous to his publication of the species in 1915, was labeled P. Klossii by him, and seems to have been the only collection in the Singa- pore herbarium up to that time. In his Flora of the Malay Peninsula (1925) he accepted the species and cited only his collection from Gu- nong Tahan, so this one seems acceptable as his holotype. SPECIMENS EXAMINED: Malaya, Pahang, Fig. 185. Pandanus Klossii Ridl., from holotype, syncarp, X Vl- Gunong Tahan, 14 Sept. 1937, E. J. H. Corner (SING) ; Pahang, Gunong Tapis, Kuantan, 4,600 ft. alt., 14 June 1934, C. F. Symington & Kiah 28,843 (sing). DISCUSSION: P. Klossii Ridl, a very unique species, is here made the holotype for the new section Malaya. This combines the characters of the two sections Pandanus and Rykia. When Ridley described the species in 1915 he did not place it in a section, but later in 1925 in his Flora of the Malay Peninsula he put it in the section Keura, which is now called Pandanus. As usual he had a good eye for species, but his description seems to have been hastily drawn. He stated ( 1915: 191 ) "the drupes are separate 350 PACIFIC SCIENCE, Vol. XVII, July 1963 till the fruit is nearly ripe, when from 6 to 7 become adnate and remain so as the whole fruit breaks up.” The holotypic collection consists of two sheets, one with overripe fruit, the other with a small, immature syncarp. In both the phalanges are of several fused carpels. The car- pels are fused from the base to within 4 mm of the apex, and the sides lack sutures.The writer sees no evidence on these specimens to confirm Ridley’s statement that the drupes are separate and only tardily connate; on the contrary, all the structural evidence indicates that the pha- langes are of several carpels united from the beginning. Though Ridley did not explain the source of his specific epithet, it was obviously given in honor of Cecil Boden Kloss, an English zoolo- gist and to a lesser degree a botanist. In 1903 and 1907 he was employed by the Singapore Botanic Garden. He was later director of the Raffles Museum. Multidens, sect. nov. Frutices, foliis angustis, syncarpio solitario el- lipsoideo, stylo singulo terminali osseoso prox- ime curvato basi cylindrico margine (2-) 3-6- lobato lobis deltoideis, drupis 1-loculatis. Plantis masculis cum inflorescentia bracteata 1-spicata, staminibus distinctis in axili affixis, filamentis 4-plo tarn longis quam antheris. Bushes; leaves narrow; syncarp solitary, el- lipsoid; style single, terminal, proximally curved, bony, shining, the base cylindric, the margin (2-) 3-6-lobed, the lobes deltoid; drupes 1- celled. Staminate plants with inflorescence brac- teate, bearing a solitary terminal spike; stamens separate, attached directly to the axis; filament 4 times as long as the anthers. HOLOTYPUS: P. pentodon Ridl., FI. Malay Penin. 5: 75, 1925. Also in the section is P. tetrodon Ridl. of Singapore. Pandanus pentodon Ridl., FI. Malay Penin. 5: 75, 1925; (sect. Multidens) Fig. 186 diagnosis OF lectotype: Stem short, branched, up to 2 cm in diameter; leaves said to attain more than 100 cm in length, 11-12 mm wide, coriaceous, dark glossy green, pale green beneath, channeled above the midrib and with 2 lateral pleats, in section low M-shaped, narrowly ligulate, outer halves of leaves missing, at midsection with 17-22 secondary parallel veins in each half, the tertiary cross veins barely visible below near the base, the base amplexi- caul and unarmed, apparently copper-colored, beginning at 3-5 cm up the margins with pale serrulations 0. 5-0.8 mm long, 1-6 mm apart; the midrib unarmed to at least beyond the mid- dle; at midsection the margins with slender ser- rulations 0.3 -0.6 mm long, 4-8 mm apart, the upper surface unarmed; the apex tapering to a 10 cm trigonous subulate tip which 10 cm down is 4 mm wide, the margins and midrib below with serrations 0. 2-0.3 mm long, 0.7-3 mm apart; pistillate peduncle 13 cm long, 8 mm in diameter, bracted, bearing a single, erect syncarp; the highest peduncular bract 30 or more cm long, at base 2.2 cm wide, narrowly lanceolate and tapering noticeably right from the base, the margins with serrulations similar to those of the leaves, but on the upper surface beginning 10 cm up the 2 lateral pleats with similar serrulations, the lower ones 5-13 mm apart, the upper ones 2-5 mm apart; partly hiding the fruit are 3 series of floral bracts, 3 deep, and subequal, 5 cm long, 2 cm wide, lance- elliptic, fleshy, cream-colored; syncarp 5 cm long, 3 cm in diameter, bearing about 400 drupes (when young these are cream-colored), 12-13 mm long, 2-3.5 mm wide and thick, linear-oblanceoloid, 5-6-angled, the body 10-11 mm long, the summit rounded; pileus 2.5-3 Insertion in Key On p. 225 (p. 2 of reprint) for the second J, read J. Styles 1 (-2), often cartilaginous or bony, z. Styles toothlike to broad subulate, entire or bifid; staminate spikes bearing stamens fascicled at apex of column Rykia z. Style with cylindric base, then curving proximally and the margin (2-) 3-6-lobed with deltoid lobes; staminate spike solitary, bearing separate stamens attached directly to the axis Multidens Fig. 186. Pandanus pentodon Ridl., from lectotype. a, Infructescence, X 1; ^ drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1 \ d, drupe, lateral view, X 4; e, drupe, longitudinal median sec- tion, X 4; /, drupe and style, apical view, X 1; g, drupe apex and style, apical view, X 10; h, stigma, oblique view, X 10; i, drupe apex, style, and stigma, lateral view, X 10; j, leaf base, lower side, XI; ^ leaf middle, lower side, X 1; l, leaf apex, lower side, Xl;», staminate inflorescence, X 1; «, stamens and axis, X 10. io cm. | —I — T“ ' ' I o 5 cm. FIG. 187. Pandanus tetrodon Ridl., from holotype. a, Syncarp, X 1; b, drupe, lateral view, X 1; drupe, longitudinal median section, X 1; bl-g, drupe, style, and stigma, apical view, X 1; b, drupe, lateral view, X 4; i, drupe, longitudinal median section, X 4; j, k, drupe, style, and stigma, apical view, X 4; /. stigma, proximal view, X 4; m. leaf base, lower side, X 1; n> middle, lower side, X 1- i cm. Page 296: Revision of Pandanus, 15. Malaya — St. John 353 mm long, the base pyramidal-semiorbicular, smooth, when dried somewhat yellowish; style 1-1.5 mm long, apical, bony, yellowish, shining, its base short cylindric, then the rest sharply curved proximally and flabellate with the mar- gins (3-) 4-5 (-6) dentate; stigma 0.5-1 mm long, obdeltoid to flabellate, papillose, brown, on the distal side of the style; endocarp central, bony, pale brown, the upper lateral walls 1 mm thick; seed 5-5.5 mm long; apical mesocarp an empty cavern; basal mesocarp fibrous up the sides, fleshy within. STAMINATE PLANT: Herbage not seen; stam- ina te inflorescence dense, leafy bracted, the main part 14 cm long; lowest floral bract about 30 cm long, the upper part green, foliaceous, 10 mm wide, with serrulations 0.2 mm long, 2-5 mm apart, the 10 cm basal part 19-20 mm wide, folded, firm, apparently pale, with similar or smaller serrulations; median floral bract 7 cm long, 3 cm wide, navicular, elliptic, acute, ser- rulate near the apex on margins and midrib below, firm, apparently pale; spike single, ter- minal, 7.5 cm long, 13 mm in diameter, cylin- dric, dense; stamens distinct, attached directly to the axis; filaments 4.2-4.4 mm long, anthers 0.8 mm long, narrowly oblong, bearing at tips a 0.15-0.2 mm subulate prolongation of the connective. LECTOTYPE: Malaya, Johore, Ulu Kahang, abundant along edges of Kahang river, 250 ft. alt., 1 June 1923, R. E. Holttum 10,861 (sing) — the pistillate plant here designated as lecto- type. Mingled on the same sheet is a staminate plant. Pandanus tetrodon Ridl., Roy. Asiat. Soc., Jour. 68: 13-14, 1915; P. singaporensis Ka- nehira, Jour. Jap. Bot. 14: 173-177, figs. 6-8, 10, 1938, a superfluous name; non Barrotia tetrodon Gaud., Bot. Voy La Bonite, t. 13, figs. 1-8, 1841, an invalid name; non P. tetrodon Balf. f., Linn. Soc. Bot. Jour. 17: 63, 1878, a name not accepted by its author; non P. tetro- don Balf. f. ex Kanehira, Bot. Mag. To- kyo 52: 236-239, fig. 70, 1938, a later synonym of P. compressus Mar tel li ( 1905 ) ; ( sect. Multidens ) Fig. 187 DIAGNOSIS OF HOLOTYPE: "Large bushy plant; branches at apex 2.5 cm in diameter”; leaves well over 97 cm long, 2.6 cm wide at the middle, 2 cm wide near the base, coriaceous, 1 -ribbed, 2 -pleated, in section low M-shaped, about at the midsection with 25-26 secondary parallel veins in each half, these visible at least below throughout, tertiary cross veins visible in outer half, transverse, forming oblong meshes, mostly short, the blade ligulate, the apex not preserved, the base amplexicaul, unarmed, be- ginning at 5 cm up the margins with prickles 1.5-2. 5 mm long, 5-10 mm apart, subulate- tipped serrae, pale, with reddish tips; the mid- rib below unarmed for at least lower %’s; near the midsection the margins with similar serrae, 1.3-1. 5 mm long, 6-13 mm apart; leaf apex "acuminate,” but not preserved; pistillate in- florescence terminal, more or less concealed; pe- duncle 19 cm long, 7 mm in diameter, 3-sided, sinuous, leafy bracted, bearing 1 syncarp, this 8.8 cm long, 5 cm in diameter, cylindric with obtuse ends, bearing about 176 drupes, these 22-24 mm long, 6—10 mm wide, 5-10 mm thick, increasing upwards in size and the apical ones definitely the largest, cuneate-oblanceoloid, obtuse, 5-6-angled, upper % free, the sides nearly smooth; style 3-4 mm long, bony, dark shining brown, at base cylindric, then quadrate and curved proximally, and all the lateral ones with an oblique plane to the 2-3-4-dentate margins; stigma 1—2 mm wide or long, rugose, of various shapes, filling the cavity proximal of the style apex, with 2-4 (-5) deltoid lobes; endocarp centering in lower Vs, bony, dark brown; seed 6 mm long, obdeltoid; apical meso- carp 6-7 mm long, a rounded cavity with a few pale membranes; basal mesocarp fibrous and fleshy. EXPANDED DESCRIPTION OF PISTILLATE PLANTS: Stems at apex 1.5-2. 5 cm in diameter, brown, striate, covered with ascending adventi- tious rootlets 1-2 cm long; leaves 78 to more than 97 cm long, the tip gradually diminishing to a trigonous subulate apex about 10 cm long, and at base 4 mm wide, on the subulate apex the margins and midrib below with serrations 0.3-0. 5 mm long, 2-5 mm apart; syncarp sin- gle or with a secondary one below and 4.5 cm long, ovoid. Fig. 188. Pandanus immersus Ridl., from holotype. a, Syncarp, X 1; h, drupe, lateral view, XI \ c, drupe, longitudinal median section, X lj d—i, drupe, style, and stigma, apical view, X 1 ’> j> drupe, lateral view, X 4; k, drupe, longitudinal median section, X 4; l, drupe apex and style, apical view, X 4; m, drupe apex, style, and stigma, proximal view, X 4; n, leaf middle, lower side, X 1; venation of leaf middle, lower side. X 4; p, leaf apex, lower side, X 1- Page 298: Revision of Pandanus, 15. Malaya — St. John 355 HOLOTYPE: Singapore, on the west coast road near Pasir Panjang, Dec. 11, 1911, H. N. Ridley 15,465 (SING). Holotype examined! SPECIMENS EXAMINED: Singapore, Botanic Gardens, Aroid Rockery, cult., 20 April 1936, E. J. H. Corner 30,993 (kep, sing). SECTION Rykia Pandanus immersus Ridl., Asiat. Soc., Straits Branch, Jour. 41: 48-49, 1903; Mater. FI. Malay Penin. 2: 224, 1907; FI. Malay Penin. 5: 76, 1925; Martelli, Webbia 4(1): 18, 1913; (sect. Rykia ) Fig. 188 NOM. VERN.: "mengkuang ayer.” DIAGNOSIS OF HOLOTYPE: Aquatic, "the stem rising but little above” the surface of the river; leaves "many feet long,” 10.2 cm wide, coria- ceous, green above, pale and glaucous beneath, the longitudinal veins strong, visible above and conspicuous below, about 1 mm apart, at mid- section with 70 secondary parallel veins in each side, the venation also consisting of tertiary cross veins 1-3 mm apart, these easily visible above and conspicuous below, the base not pre- served but the blade 1 -ribbed, 2 -pleated, in sec- tion M-shaped, ligulate, narrowing upwards and in the upper quarter tapering to a subulate del- toid apex about 30 cm long, and at a point about 10 cm down from the tip only 1.3 mm wide; at about the midsection the margins with thorns 2.5-4 mm long, 9-18 mm apart, subulate but the base thickened, nearly straight, ascend- ing, pale except for the very tip which may be reddish; near the tip the margins with serrae 0.8-0. 9 mm long, 1.5-3 mm apart, stout; the midrib below nearly unarmed; pistillate inflores- cence apparently emersed; peduncle more than 8 cm long, 15 mm in diameter, 3 -sided, brac- teate; syncarp perhaps a third mature, 9.5 cm long, 6 cm in diameter, ellipsoid, 3 -sided, bear- ing about 1,440 drupes, these 17-19 mm long, 5-6 mm wide, 3-4 thick, the body 12-13 mm long, arcuate spatulate, upper free, cuneate to the base; the pileus 7-8 mm long, the base subsemiorbicular, minutely papillose; the style 4-6 mm long, hornlike, mostly gently curved proximally, the apical ones unbranched, of the median and lower ones about half are entire with a single horn but the other half are bifur- cate and the 0.5-2 mm lobes often are wide- spreading; stigma 3.5-5 mm long; endocarp (immature) oblanceoloid, nearly as long as the body; mesocarp fibrous and fleshy. HOLOTYPE: Malaya, "Selangor, in the Labu River, forming dense thickets,” May 1891, H. N. R[idley] , (SING). Holotype examined! Pandanus johorensis Martelli, Soc. Bot. Ital., Bui. 302, 1904; P. muarensis Ridl., Mater. FI. Malay Penin. 2: 226-227, 1907, and FI. Malay Penin. 5: 77, 1925; (sect. Rykia ) Figs. 189, 190 NOM. VERN.: "rassau.” DIAGNOSIS OF ISOTYPE: Leaves 1.3-1.45 m long, 4.2 cm wide, coriaceous, paler beneath, 1 -ribbed, 2 -pleated, at midsection with 43 sec- ondary parallel veins in each half, no visible cross veins, ligulate, gradually tapering to a 27 cm subulate deltoid apex, this 10 cm down O. 7 mm wide, the base amplexicaul, unarmed, but beginning at 10-11 cm on one side, and at 18 cm on the other the margins with prickles 1.5-2 mm long, 4-11 mm apart, subulate, as- cending, dark reddish; the midrib below begin- ning at 8 cm with prickles 3-3.5 mm long, 12-22 mm apart, stout subulate, arcuate, re- flexed; at midsection the margins with prickles 1-1.5 mm long, 4-14 mm apart, stout subulate, arcuate, closely ascending; the midrib below with prickles 1.5-1. 8 mm long, 5-45 mm apart, stout subulate, arcuate, reflexed or ascending; on the subulate apex the margins with serrae 0.5 mm long, 1-3 mm apart; those of the midrib below similar but 3-7 mm apart; peduncle 29 mm long, 8 mm in diameter, 3 -sided, slightly sinuous, leafy bracted; syncarp single, 6 cm long, 3 cm in diameter, narrowly ellipsoid (very young), bearing about 4,096 drupes, these (im- mature) 11-12 mm long, 1.5-2. 5 mm wide and thick, 5-6-angled, upper 2A free, the body 4.5-5 mm long, oblong-ellipsoid, papillose, with thick longitudinal fibers; style 7 mm long, setiform, hard, shining and yellow; stigma linear, proxi- mal, brown, as long as the style; endocarp and seed not yet formed, but inside the ovary wall Fig. 189. Pandanus johorensis Martelli, from isotype, a, Syncarp, XI \ b, c, drupe, lateral view, X 1; d, drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; /, drupe apex and style, apical view, X 4; g, style and stigma, proximal view, XI 0; h, leaf base, lower side, X 1; i> leaf middle, lower side, X 1; j, leaf apex, lower side, X 1. i UtOOT Fig. 190. Pandanus johorensis Martelli, from Corner 28,161. a, Staminate inflorescence, X l/3\ b, spike, X 1; c, staminate column, scale, and stamens, X 10; d, leaf base, lower side, X 1; ^ leaf middle, lower side, X 1; f, leaf apex, lower side, X 1 • UUUS io cm. 358 PACIFIC SCIENCE, Vol. XVII, July 1963 Fig. 191. Pandanus pevangensis Ridl., from holotype. a, Infructescence, X 1/5; b, drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; d, drupe apex and style, X 1; drupe apex and stigma, proximal view, X 4; /. leaf base, lower side, X 1; g, leaf middle, lower side, X 1; ^ leal apex, lower side, X I- Page 302: Revision of Pandanus, 15. Malaya — St. John 359 of tissue and heavy longitudinal fibers there is a single large cavity, with indefinite traces of inner tissue. EXPANDED DESCRIPTION FROM ALL PISTIL- LATE PLANTS EXAMINED: Shrub, to 6 m tall, upper branches 3-4 cm in diameter, dark brown, forming dense thickets, with a few thornlike axillary rootlets 4-6 mm long, divergent; leaves 1-1.8 m long, 2.9-4. 1 cm wide; syncarp solitary, pendulous, 12-16 cm long, 6-10 cm in diame- ter, ellipsoid. DESCRIPTION OF STAMINATE PLANTS: BraC- teal leaves up to 80 cm long, 3.5 cm wide, coria- ceous, apparently pale, ligulate, the upper lA long tapering to a trigonous subulate apex, this 10 cm down 2 mm wide, the lower wholly unarmed, at midsection with 5 1 secondary veins in each half, no visible tertiary veins; near the apex the margins and midrib below with subu- late serrulations 0.5-0.7 mm long, 2-6 mm apart; staminate inflorescence 45 cm long, with 8 spikes, these 15-25 cm long, 12-15 mm in diameter, densely flowered but somewhat in- terrupted; flowers sweet and sickly scented; fila- ment column 4 mm long, stout, clavate, bearing an apical parasol-like scale 3-5 mm in diameter, the apex hispidulous with hairs 0. 3-0.4 mm long; free filament tips 0.5-2 mm long, pendent from the lower surface, glabrous or sparsely hispidulous; anthers 10-16, and 0.8-1. 3 mm long, mostly cylindric, bearing an apical mucro 0.2 mm long, formed of a projection of the connective (description from Corner 28,161). ISOTYPE: Malaya, Johor, Muar, Bukit Wiligu, Sungei Pauh, 1902, Fox 11,326 (SING). Isotype examined! SPECIMENS EXAMINED: Malaya, Johore, Ma- wai, generally flowering of staminate and pistil- late on riverside, flowers visited by bees and beetles, May 13, 1934, E. J. H. Corner 28,161 (SING). DISTRIBUTION: Malaya and Sumatra, form- ing dense thickets in estuaries, forming a zone beginning where Nip a stops, in tidal fresh wa- ter. Plants with similar leaves have been seen from Borneo. DISCUSSION : P. johorensis Martelli is a mem- ber of the section Rykia. It was briefly pub- lished in 1904 by Martelli as only a few phrases in a key to certain species. His holotype was: "Stato di Johor. Penis. Malese, no. 11326 Ridl. (H. Becc.).” The specimen above cited is an isotype in the herbarium in Singapore. It bears the data: "Malay Peninsula, Johor, Bukit Wilgu, Sungei Pau, 1902, Fox 11,326.” In his Materials for a Flora of the Malayan Peninsula 2: 226, 1907, Ridley accepted Martelli ’s species, and cited the single collection, "Muar: Bukit Wilgu, Sungei Pauh (W. Fox) 11326.” Later, Ridley in his Flora of the Malay Peninsula 5: 77, 1925, treated this plant as P. johorensis sensu Ridl., not of Martelli, and redescribed it as the new species P. muarensis Ridl. For this he had the holotype: "Muar, Sungei Pauh (Fox).” What- ever misconception Ridley thought there was, he had no right to rename the species using the same holotype. Certainly the collection Fox 11326 from Bukit Wilgu, Sungei Pauh, Johor, is the same as [Ridley] 11,326 with the same locality data. The number is a Singapore Her- barium number, as the staff members were re- quired to use this single series of numbers, not their own personal collection numbers. The hol- otypic specimen in the herbarium at Firenze was not a duplicate distributed from Singapore. Rather, it is a fragment consisting of about two score of the drupes placed in a paper pocket on which in Martelli’s hand is written the data. These fragments were taken from the syncarp of the isotypic specimen, then in Calcutta, now in Firenze. It lacks stem, leaves, inflorescence, and syncarp. Nevertheless, the actual holotype is this fragmentary specimen, Fox (or Ridley ), once in Calcutta, and now in Firenze. However, there seems no doubt but that P. muarensis Ridl. is a typonym and must be placed in the syn- onymy of P. johorensis Martelli. The isotype in Singapore has a syncarp with immature drupes that have not matured their endocarp and seed. Pandanus penangensis Ridl., Asiat. Soc, Straits Branch, Jour. 41: 50, (1904) = 1903; FI. Malay Penin. 5: 81, 1925; (sect. Rykia ) Fig. 191 diagnosis OF HOLOTYPE: "Tree 7 m. tall, 10 cm. in diameter”; leaves 2 m or more long, near the base 12 cm wide, when young marbled light and dark green, beginning at 15-16 cm 360 PACIFIC SCIENCE, Vol. XVII, July 1963 the margins with prickles 1.3- 1.5 mm long, 4-7 mm apart, pale, subulate, flat appressed ascend- ing; the midrib below sharp salient and with similar ascending prickles, the secondary parallel veins conspicuous and in lower part 69 in each half, and except at base the tertiary cross veins prominent on both sides, transverse or oblique and forming short, rather oblong meshes, mid- section not seen; at 45 cm down from the apex the blade 11.5 cm wide, the apex abruptly nar- rowed to a subulate tip about 5 ( ? ) cm long, the margins with subulate serrae 1.5-2. 5 cm long, 3-10 mm apart; the midrib below with serrae 0.8-1 mm long, 7-20 mm apart; drupes 42-47 mm long, 7-9 mm wide, 6-7 mm thick, 5-6- angled, the sides smooth, gently curved, prob- ably only the pileus free, the body 37-40 mm long, drupes and pilei shedding separately; pileus 9-11 mm long (or if measured along the curve of the style 13-16 mm long), 10-13 mm wide, 8-10 mm thick, 6-angled, broad ovoid-pyramidal, the tip proximally curved; style 5-7 mm long, heavy subulate, curved proximally almost to a right angle; stigma 4-5 mm long, lanceoloid, papillose, brown; endocarp 32-33 mm long, narrow ellipsoid, bony, ebony- colored, the inner surface shining, the lateral wall 1-1.5 mm thick; seed 15-17 mm long, 4-6 mm in diameter, ellipsoid; apical and basal mesocarp fibrous and fleshy. EXPANDED DESCRIPTION FROM ALL SPECI- MENS examined: Leaves to 4 m long, the mid- section 10 cm wide, the margin with prickles 2.8-3 mm long, 5-14 mm apart, stout, heavy based, arcuate subulate, appressed ascending, brown; pistillate inflorescence of 4-5 syncarps, each with numerous drupes; syncarps 15-20 cm long, 9.5-10 cm in diameter, ellipsoid. HOLOTYPE: "Malaya, Penang Hill from base to the top, in woods.” The specimen is labeled: Malaya, Pulau Penang, Fort Hill, 1900, H. N. Ridley (SING). Holotype examined! SPECIMENS EXAMINED: Malaya, Pulau Pe- nang, Government Hill, common on the hill top, but not often in fruit, 29 July 1907, L H. Burkill 2,666 (sing). Lord Howe Island, A Riddle of the Pacific, Part III S. J. Paramonov1 In this final part (for parts I and II see Pacif. Sci. 12 ( 1 ) : 82-91, 14 (1): 75-85) the author is dealing mainly with a review of the insects and with general conclusions. INSECTA Our knowledge of the insects of Lord Howe Island is only preliminary and incomplete. Some groups, for example butterflies and beetles, are more or less sufficiently studied, other groups very poorly. Descriptions of new endemic species and records of the insects of the island are dis- persed in many articles, and a summary of our knowledge in this regard is lacking. However, a high endemism of the fauna is evident. Al- though the degree of endemism is only at the specific, or at most the generic level, the con- nection with other faunas is very significant. Olliff (1889) wrote an interesting review of Coleoptera of the island. In his time about 80 species of Coleoptera were recorded, but among them were 46 new species and 5 doubt- fully new. That is, more than 56% were en- demic species. Similar reviews for other groups are lacking. An interesting large, wingless phasmid with a body length of 106 mm, Dryococelus (for- merly Carabidion) australis Montr., is now prob- ably extinct. Gurney (1947: 391) wrote: "The present status of Dryococelus is uncertain. A note in Ward’s Natural Science Bulletin (vol. 8, No. 2, p. 11, 1935) indicated that the species is extinct, but I have been informed by John W. H. Rehn, of the Academy of Sciences of Philadelphia, that he has seen a recently pub- lished note suggesting that the species is main- taining itself.” 1 Commonwealth Scientific and Industrial Research Organisation, Canberra, Australia. Manuscript received February 27, 1962. During two visits to the island, in 1954 and 1955, the author failed to find the insect. An official enquiry was made recently to the Ad- ministration staff of the island, and the author received a letter from the Superintendent of the Island, Mr. H. Ward, on Nov. 3, 1961, in which he states: "A number of the old inhabi- tants have been questioned and all have advised that it is at least 30 years and possibly 40 years since this insect has been seen on the Island. A member of the staff, aged 33 years, has never seen or heard of the insect, nor has any pupil of the local School.” The only possibility is that the insect may still exist in one of the biggest banyan trees on the slope of Mt. Gower, on the lagoon side. The area is well isolated from the settlement where the rat concentration was probably the greatest, and may have survived in crevices of the tree. The presence of the insect in the past is evidence of an ancient connection with Aus- tralia or some other continent by a land bridge; any other type of transportation is highly improbable. A very interesting example of distribution is found in an archaic Homopterous family, the Pelorididae, with very limited possibilities of movement (nearly wingless), and adapted to very restricted habitat. There are 15 species recorded now for the whole world. Thirteen of them are associated in Chile, New Zealand, Tasmania, and eastern Australia with the south- ern beech Nothofagus. The plant also grows in New Caledonia, and Dr. J. W. Evans (Di- rector of the Australian Museum) visited this island especially to search for Pelorididae. None were found, nor was an environment discov- ered which would be favourable for their existence. In 1959 J. W. Evans, having been informed that a single nymph was recorded from Lord Howe Island, visited the summit of Mt. Gower 361 362 PACIFIC SCIENCE, Vol. XVII, July 1963 Fig. 1. A typical cloud "carpet” covering the summit of Mt. Gower while the rest of the island is in sun- shine. This "carpet” preserves the humidity and coolness over long periods, in strong contrast with other parts of the island. (Photo by Miss Z. Liepa.) and during 3 hr (sic) found two new species of this family. It is interesting to note that Nothofagus is absent from Lord Howe Island, but 24 adults and 7 nymphs of Pelorididae were found in moss. Probably the Pelorididae are not espe- cially associated with Nothofagus , but require, above all, high humidity with cool temperatures. It is evident that this family has lived on the island in a very specialised habitat for millions of years. The possibility of transportation must be excluded. It means that the general condi- tions for life are unchanged: the summit of Mt. Gower is a refuge for species ecologically close to recent New Zealand fauna and flora ( see Fig. 1 ) , whereas at sea level are preserved elements from the north, from purely tropical areas. It is noteworthy that, of the endemic palms, two dwarf species grow not at sea level, as do two others (see Fig. 2), but much higher up, closer to the summits of the two large moun- tains. The submerged land "Howeania” was iso- lated from Australia, or connected only in the far north, and the present "Australian” elements we must regard as casual elements, not as hav- ing come by a direct land bridge. The Austra- lian elements play an important role in the composition of the flora and fauna of the island, but it is remarkable that the most typical Australian forms, the Australian "sui generis,” are absent. The Diptera of the island have not been reviewed, the main obstacle being the lack of knowledge of the fauna of surrounding areas. The dipterofauna in general is very im- poverished, because the island lacks fresh water basins, and the running streams are of tem- porary character. Therefore the families and Lord Howe Island, Part 3 — PARAMONOV 363 genera connected in their life history with water are absent. Sandy ecological niches are also poorly represented, and in general the eco- logical uniformity is a factor which limits the richness in flies. However, many new species were discovered from the families Asilidae, Muscidae, Ortali- dae, Leptidae, Calliphoridae, Tachinidae, etc. (see references). The author has the impression, however, that most of the endemic species are repre- sented in the surrounding areas by closely re- lated species. Nothing extremely old or "extra” was found; however, only the lower part of the island has been comparatively studied. The middle zone (about 1,000 ft) and the high zone (2,000 ft and above) are practically un- touched. The small fresh-water streams and pools deserve the special attention of collectors. The author observed the hatching of enor- mous numbers of hmttia and Sarcophaga from the sand on the beach. They emerged early in the morning, and with wings still undevel- oped walked onto the small rocks on the shore, the surface of some of the rocks being abso- lutely covered with them. Here they waited until their wings were developed and then flew away. The sandy area is covered by sea during high tide, but the puparia apparently are not harmed. The author observed no hatch- ing from the sandy areas not reached by the high tide. To illustrate the spate of our knowledge of the Lord Howe Island fauna we may quote the results of J. D. Bradleys collecting in 1953: in Wi days of collecting, among 22 specimens of Microlepidoptera there were 12 species, 8 of which were new, and 4 others which were Aus- tralian species not previously recorded for the island. The island is also very suitable for experi- ments in insect biological control. In October Fig. 2. Two endemic species of palms which dominate the lower part of the island. Before the last war the export of their seeds was an important industry of the island. (Photo by Miss Z. Liepa. ) 364 PACIFIC SCIENCE, Vol. XVII, July 1963 Fig. 3. An "Arcade” on Admiralty Island. The greyish dots in the air are sea birds which nest there. The basalts form the main mass of the island. (Photo by Miss Z. Liepa.) 1959, specimens of Optus oophilus were liber- ated there to control the Queensland fruit fly, Strumeta tryoni, and the parasite was recov- ered in March 1961. The small size of the island greatly facilitated the experiments. It is impossible to go into details about the insect fauna of the island as the available data are too fragmentary. However, some points are very significant, especially those concerning wingless insects (and birds); they show a close relationship to the New Zealand fauna and to the eastward land masses in general. The Australian element, considered numerically, may be larger than that of New Zealand or Polynesia, but the proximity of the Australian continent, its great extension northwards and southwards, and the direction of the prevailing winds in the southwest Pacific no doubt ac- count for this preponderance. If we take as an example the plant genera which are confined to Australia and Lord Howe Island (Notelaea, Melaleuca , Lyonsia, Lagunaria, and W estringia) , we must not overestimate the importance of this fact, because the most characteristic Aus- tralian genera are either- entirely absent from Lord Howe Island or represented by only one or two species. This means that these species have had more facilities for transportation, natural or by man. Summarizing all the data in our hands, we can say from its fauna that Lord Howe Island is not a part of the Australian zoogeographical region, but is closer to the New Zealand sub- region. GEOLOGY In order to understand the history of the fauna and flora the geology also must be con- sidered. The general geology of the island is very simple; the island consists mainly of two principal formations: the volcanic rock which Lord Howe Island, Part 3”— PARAMONOV 363 forms the general mass (see Fig. 3), and the stratified beds resting on it (see Fig. 4). Two thirds of the island is composed of volcanic rocks, comprising three isolated masses. Edgeworth (1889: 132) summarized his study of rocks as follows: 1. All the igneous rocks of Lord Howe Island (so far as represented by the collection examined) belong to the Basalt group. 2. A vast period of time must have elapsed between the eruption of the diabasic basalt and that of comparatively recent olivine basalt. 3. All the basalts, with the exception of the diabasic types, are probably not earlier than Tertiary, and some may be Post -Tertiary. 4. The diabasic basalt is probably Pre-Tertiary, or may be Paleozoic. Immediately overlaying the volcanic rocks, and between them and the succeeding coral- sand rock series, occurs a bed of stiff unctuous red or yellow clay. It does not appear to be fossiliferous. More interesting is the coral-sand rock se- ries; this deposit is often from 30 to 40 ft high. The coral-sand rock is the chief fossilifer- ous deposit of the island, and has yielded the remains of the interesting reptile Meiolania, eggs of turtles, bird-bones, and recent species of both land and marine shells. A systematic boring of this deposit, as well as of coral beds in the sea, may give us a de- tailed picture of the island’s physical history. Unfortunately, geological study is only at a preliminary stage, and when it was stated that no minerals of economic value existed on the island further investigations were stopped. The island deserves, however, our special attention as a very convenient object where the history of Australia, of the surrounding areas to the east, and of the Pacific may be studied much more easily than at other points. The impoverished fauna and flora and the simple geological structure may give us all Fig. 4. A typical stratified deposit at the eastern seashore east of Old Gulch; the coral-sand strata are prominent. (Photo by Miss Z. Liepa.) 366 PACIFIC SCIENCE, Vol. XVII, July 1963 Fig. 5. Map showing the area of "Howeania.” the necessary data, because of the small volume of material for study. Lord Howe Island may be the focal point for the study of this area of the Pacific. An enormous number of facts about the Australian continent remain to be studied, and this will require many years of work, but the general outline of the history of organisms and environment may be elucidated much more rapidly if we know the history of Lord Howe Island. For developing synthetic conclusions the island represents an extremely useful area. COMPOSITION OF THE FAUNA AND FLORA The fauna and flora of the island probably consists of three different layers of animals and plants: 1. The remnant from the very large land mass, now submerged, which we have called "Howeania.” This remnant is the oldest part of the fauna and flora. 2. Immigrants from surrounding areas, ar- riving after the land was submerged, i. e., very long ago, in the geological sense. These ele- ments have had time to develop into new species. 3. Recent immigrants, helped directly or in- directly by man. We have some data which indicate that in recent time several species arrived on the island, but finding conditions unsuitable for permanent settlement they dis- appeared. The history of the fauna can be illustrated with the help of Figure 5. If we suppose that the shallow area of the sea, shown by the white area, was the outline of the submerged land Lord Howe Island, Part 3-— Paramonov 367 ( Howeania ) , then the peculiarities of the fauna and flora of the island can be explained easily. The absence of typical Australian elements was caused by the very wide and deep part of the sea in between, which acted as an impas- sable barrier. The presence of Australian elements which are not typical for the Australian fauna can be explained by the very narrow strait in the north (see dotted line of Tropic). It permit- ted the penetration of tropical elements, whilst the elements more adapted to moderate cli- mate could reach the island from the south, where there was a broad land connection with New Zealand. The very essential difference between the faunas of Lord Howe and Norfolk Island may be explained by the very large and deep sea area extending eastwards from Howeania. Since it is about halfway between New Caledonia and New Zealand and was connected to them by land, Norfolk Island contains elements of both countries. The New Zealand elements are not repre- sented so well on Lord Howe Island because the ecological conditions are very limited spa- tially, and they are confined mostly to the moun- tain summits of the island. We have reached all these conclusions after attentive study of the animals and plants, de- tails of which can not be demonstrated in a work of this size. The author reached these conclusions before consultation of the above map showed the distribution of the deep and the shallow portions of the sea. The illustrated structure of the sea bottom fully supports the author’s considerations, which are not purely theoretical speculations based on the study of biology, but reflect also the history of the land masses in the discussed area. Here we have a working theory for reconstruction of the his- tory of the island and its life. The theory also explains satisfactorily the history of the basic part of the island’s fauna. With regard to the immigrants which reached the island after the submergence of Howeania, the position is less satisfactory: first, there is a lack of data about the fauna of the island; second, data are lacking about the dis- tribution of organisms in the surrounding areas. Without these data, reconstruction of the his- tory of the fauna and flora is impossible. Regarding comparatively recent immigrants, we must consider the role of man in the dis- tribution of plants and animals. Early whalers must have played a considerable part. If, for example, we find on the island a plant which has a strange, widely interrupted distribution, such as South Africa and the island, and grows near the sea shores, we have grounds to sus- pect that the plant has been transported by ships. As an example, we may cite the case of the so-called "African water lily” in Aus- tralia ( Aponogeton distachyum, or Cape Pond Lily). The author found the lily in 1947 in the lagoon at Lome, Victoria. This plant is very abundant there, and in July the entire lagoon is covered with the white flowers. The lily is rapidly diminishing in numbers, however, and we must predict that sooner or later it will disappear from the lagoon. During the last century whalers introduced many animals and plants along the sea shores which they visited. One of the island’s chief industries began in this fashion, i. e., the cul- ture of the high quality onion, some bulbs of which had been washed ashore, found by an island woman, and planted. COMPOSITION OF THE FLORA Oliver (1917) gave an analysis of the com- position of the island’s flora. Although the data are outdated in nomenclature, the conclusions may be regarded as valid. Of the 169 genera of vascular plants repre- sented on Lord Howe Island 4 are endemic. Of these 4 genera, Colmeiroa and Hedyscepe are allied to New Zealand forms, Negria to both New Zealand and New Caledonian genera, and Howea to Malayan and tropical Australian genera. If the five species belonging to these 4 genera be taken as modified descendants of species which arrived via a land bridge, then they would indicate a New Caledonia-New Zealand migration, with the land connection severed first at the southern end, thus accounting for a greater degree of peculiarity for the species related to New Zealand forms. 368 PACIFIC SCIENCE, Vol. XVII, July 1963 Of the nonendemic genera 93 are widely distributed, occurring in New Zealand, Aus- tralia, and Polynesia; 47 others range widely through tropical countries, but do not reach New Zealand; 1 occurs in New Zealand only; 5 in Australia only; 3 in Polynesia only; 11 in New Zealand and Australia only; 2 in New Zealand and Polynesia only; 1 in Africa only. It is interesting to state, and Oliver omitted to do so, that most of the genera are repre- sented only by a few species: 169 genera are represented by 209 species, which means that speciation is not strong on the island. Progres- sive evolution was almost absent on the island, and we have only transformation, probably due to genetic impoverishment in the popu- lations. It is possible also that some endemics of the island are only relics of species more widely distributed in the past, and now preserved only on the island. If we compare the number of the very widely distributed genera (see above) and the genera with very restricted areas, we see a great dif- ference: of 165 nonendemic genera 95 are distributed in New Zealand, Australia, and Polynesia; 47 only in Australia and Polynesia; and only 23 belong to the different countries, but are isolated. The presence of a large portion of widely distributed genera and species might have been expected in the flora of an isolated island, be- cause species possessing facilities for wide dis- persal would naturally form the bulk of immi- grants after the land connection had been severed. Fig. 6. View of the lagoon from the south; in the background are Rabbit Island and Mt. Eliza; on the seashore are three Araucarias introduced from Norfolk Island, growing well among the native plants. Taken at low tide. (Photo by Miss Z. Liepa.) Lord Howe Island, Part 3 — Paramonov 369 Fig. 7. The lagoon viewed from the south during low tide. In the foreground are coral reefs, of interest to the marine zoologist; behind these is low Rabbit Island, and in the background is the northern part of the island with tooth-shaped Mt. Eliza. (Photo by Miss Z. Liepa. ) This means that we must be very careful in making conclusions based only on numerical data. For the history of the flora and fauna the easily distributed elements are practically value- less. Let us turn to the analysis of the species. Of 209 species of the flora 70 (or 33%) are en- demic. The percentage of specific endemism is TABLE 1 TOTAL NO. OF SPECIES ENDEMIC SPECIES INCLUDED IN TOTAL PERCENTAGE OF ENDEMIC FORMS Whole flora 209 70 33 Australia 168 41 24 New Zealand 109 38 35 Polynesia 116 39 34 very high. Oliver gives the following table which is very instructive: From Table 1 (from Oliver, 1917) it is evi- dent that endemic forms of Lord Howe Island are distinctly more closely related to New Zea- land and Polynesia than to Australia. As a re- sult, the line of Gressitt (1956) must be trans- ferred westwards of Lord Howe Island, separat- ing it from the Australian continent. The zoo- logical data also have shown that the fauna of the island is more closely related to the lands to the east than to Australia. A direct connection of Lord Howe Island with Norfolk Island, however, is doubtful, be- cause the Araucarias of Norfolk Island have not reached Lord Howe Island naturally, although those transplanted by man are growing very well (see Fig. 6). 370 PACIFIC SCIENCE, Vol. XVII, July 1963 NORTH ROCK <& ADMIRALTY ISLETS LORD HOWE ISLAND .SUGARLOAF PASSAGE SUGARLOAf * -Hi MALABAR WIRELESS STN WELL CEMETERY WIND SOCK WILD GOATS WILD PIGS BANYAN TREE JETTY MUTTON BIRD ISLAND SAIL ROCK •/*£> i6S LION POINT © -> ROCKY POINT WOLF ROCK \\ BOAT EDMANOCH POINT O lHAVEN / * > CUT GRASS POINT SUGARLOAF POINT 1 Mt GOWER iSHfe) t» 2840' 3;^ LARGEST t BANYAN TREE GEORGE ROCK Q (awash) TRACK GOWER ISLAND O Lord Howe Island, Part 3— PARAMONOV 371 CONCLUSIONS 1. Lord Howe Island is not an "oceanic” island owing its existence to the activity of corals, although some part was played by them; the island’s fauna and flora are not a casual "mixtum compositum” of elements arrived by various methods from different directions, dur- ing different epochs. 2. Lord Howe Island is a small part of a sunken continent or large island, preserving a very specific fauna and flora, with very high endemism (in some groups as high as 70%). 3. Some endemic species developed here probably as a result of very long isolation from the closely related population in the surround- ing areas; other endemic species (some birds) probably are unchanged species preserved from the sunken land, being mostly destroyed by man in the surrounding areas. 4. The sunken land ("Howeania”) was never connected with the Australian continent after creation of Eucalyptus and Acacia species, because on the island they are absent, although growing very well when introduced. It appears that the island’s isolation is very old. 5. Lord Howe Island was probably never directly connected with Norfolk Island. The Araucarias of Norfolk Island, so typical of that island, are absent from Lord Howe Island. The species common to both islands are probably species with very strong capacity for dissemi- nation. 6. Having no land connection with Australia and Norfolk Island, Lord Howe Island received its fauna and flora mostly from the north and south. Warmth-loving elements of the flora and fauna probably arrived from the northeast ( palms, for example ) . All the tropical elements show a connection with this direction. 7. The elements of subtropical or moderate areas, liking cool climate and high humidity, probably arrived from southeastern and south- ern directions (the so-called New Zealand ele- ments have their origin in these sectors). 8. Since it is a highly elevated island (up to 2,800 ft) Lord Howe Island has preserved its character, and its flora and fauna, for a long period of geological time. The flora and fauna of its summits (for example the plant Dixonia Fig. 8. 1. Mt. Eliza 2. Caves in this area 3. North Hills 4. Dawson Peak 5. Poole’s Lookout 6. North Ridge 7. Grassy Hill 8. North Scab 9. Old Settlement Creek 10. Thomson’s Lookout 1 1 . Settlement Creek 12. Hine’s Curio Shop 13. "Ocean View’’ Guest House 14. The Peg 15. "Somerset" Guest House 16. Thomson’s General Store 17. Electric Power Station 18. "Waverley’s” Tea Rooms 19- "Dignam’s” Tea Rooms 20. The Hall 21. "Leanda Lei’’ Hotel 22. Doctor 23. Post Office & Bank Agency 24. Hospital 25. Government House 26. School 27. "Pine Trees’’ Guest House 28. Nichol’s Clear Place 29. Blinky Beach Picnic Hut 30. Lagoon 31. Soldier’s or Big Creek 32. "Smoking Tree" 33. "Goat House” Asterisks indicate localities known by more than one name. Those currently used on the island are shown on the map and alternative names are listed below: Admiralty Islets: Roach Islets Blinkenthorpe Bay: Ross Bay Lion Point: Mutton Bird Point Mt. Malabar: North Peak North Bay: Callam’s Bay North Rock: North Islet Old Gulch: Collin’s Cove Rabbit Island : Blackburn or Goat Island Rabbit Island Passage: Boat Passage South Entrance: Erscott’s Passage Sugarloaf: Soldier’s Cap Transit Hill: Lookout Mountain 372 antarctica, and the Pelorididae insects) are very old elements, preserved in a nearly un- changed condition for many millions of years. It is evident that after the ancient land sank, there was no re-elevation of it which deserves serious attention (there were only slight oscil- lations of ocean). 9. At the present stage of our knowledge of the geology, paleontology, and biology of the island, it is impossible to draw any conclusions concerning the theory of drifting continents: there are no positive and no negative data. 10. Finding of interesting data can be ex- pected after a study of Ball’s Pyramid. This colossal rock has never been touched by ex- ploration, and no specimens either of plants or of animals have been collected there. The pos- sibility is not excluded that some forms which have been destroyed on the island may have survived on this rock. 11. There are numerous caves on the island, some very long, which have not been explored; we have reason to think that interesting re- mains of recent and past fauna may be pre- served there. 12. The presence of a coral reef gives an opportunity to gather additional data about oscillations of the sea level, and their magni- tude (see Fig. 7). 13. The preliminary data at our disposal give evidence that in the past there existed a large land mass, probably southeastward from Lord Howe Island, where the flora and fauna of the island developed. On the island itself we can find only remnants of this life of the past. 14. Lord Howe Island is a particularly in- teresting subject for the study of the Pacific and surrounding countries. The scientific value of the island has been underestimated, almost neglected. Only one expedition, organised by the Australian Museum in 1887, has worked on the island. Since then any study that has been done has been private, and by amateurs. This article shows that the island deserves more serious attention: its interest is not local but international. Only longer expeditions of geologists and biologists, working together, can supply us with the necessary data. PACIFIC SCIENCE, Vol. XVII, July 1963 REFERENCES Ashton, J. H. 1921. A revision of the Aus- tralian Cicadidae, I. Proc. Roy. Soc. Vic. 33:87-107. Bradley, J. D. 1956. Microlepidoptera from Lord Howe Island. Bull. Brit. Mus. Ent. 4 (4) : 145-164. Carter, H. F. 1920. Notes on some Australian Tenebrionidae, etc. Proc. Linn. Soc. N.S.W. 45 (2): 222-249, figs. 1920. Revisional notes on the Family Cistelidae. Trans. Roy. Soc. S. Aust. 44:198- 217. 1921. Description of . . . New Mosquito from Lord Howe Island. Proc. Zool. Soc. Lond. 1920 (4) : 62 3-628. 1926. Revision of Athemistus and Mi- crotragus (Fam. Cerambycidae) with notes, and description of other Australian Coleop- tera. Proc. Linn. Soc. N. S. W. 51 (4) : 439, 506. Chopard, L. 1951. A revision of the Austra- lian Grylloidea. Rec. S. Aust. Mus. 9:397- 533. Cunningham, G. H. 1926. The Gasteromy- cetes of Australasia, IV. Species of the genus Geaster. Proc. Linn. Soc. N. S. W. 51 (2) : 87. Dodd, A. P. 1924. Chalcidoidea and Procto- trupoidea from Lord Howe, etc. Trans. Roy. S. Aust. 48:162-186. ( Evans, J. W. 1941. Concerning the Pelorididae. Aust. Jour. Sci. 4(3) : 95-97. 1956. Insect distribution and conti- nental drift. A symposium . . . Geolog. Dept. Univ. Tasmania, pp. 134-161. 1959. The Pelorididae of Lord Howe Island. Rec. Aust. Mus., Sydney 25(3): 57- 61, figs. Gressitt, J. L. 1956. Some distribution patterns of Pacific Island Faunae. System. Zool. 5(1): 11-33, 47. Lord Howe Island, Part 3 — Paramonov 373 Gurney, A. B. 1947. Notes on some remark- able Australian Walkingsticks, etc. Ann. Ent. Soc. Amer. 40(3) : 373-396. Hale, H. M. 1926. Studies in Australian Aqua- tic Hemiptera, No. 7. Rec. S. Aust. Mus. 3 (2): 195-217, figs. Hincks, W. D. 1947. Dermaptera from New Hebrides and Lord Howe Island. Ent. Mo. Magazine, London 83:65-67. Johnston, H. T., and Hardy, G. H. 1923. Some Sarcophagid flies from Lord Howe Island. Rec. Aust. Mus. 14(1): 62—7 1, figs. Laing, F. 1925. Descriptions of some new genera and species of Coccidae. Bull. Ent. Res. 16(1): 51-66, figs. Lea, A. M. 1909. Australian and Tasmanian Malacodermidae. Trans. Ent. Soc. Lond., 1909 ( 1 ) : 45—2 5 1, pis. — 1909. Revision of the Australian Cur- culionidae belonging to the subfamily Cryp- torhynchidae ( Coleoptera ) . Proc. Linn. Soc. N.S.W. 33 (4), 1908:723. 1916. Notes on Lord Howe Island Phasma and on associated Longicorn Beetle. Trans. Roy. Soc. S. Aust. 40:145-147, pis. 1917. Notes on some miscellaneous Coleoptera, with descriptions of new species, Part 3. Trans. Roy. Soc. S. Aust. 41:121-322 (with list of distribution, pp. 212-214). 19 18. On Australian Coleoptera, Part 1. Rec. S. Aust. Mus. 1(1): 96. Miscellanea. 1917. Notes on Diatomaceous Earth from Lord Howe Island. Trans. Roy. Soc. S. Aust. 41: 659. Ollxff, A. S. 1889. The insect fauna of Lord Howe Island. In: Lord Howe Island, its Zool- ogy, Geology and Physical Characters. Mem. Aust. Mus. 2:77-98, pi. 6. (The first, most important record about the island.) Paramonov, S. J. 1957. On some new Ortalids ( Acalyptrata ) . Ann. Mag. Nat. Hist. (12) 10:779-781. 1958. A review of Australian species of Laphria (Asilidae), with descriptions of three new species from Lord Howe Island. Pacif. Sci. 12 (1):92— 105, figs. I960. A review of Australian Pygo- phora-species (Muscidae). Ann. Mag. Nat. Hist. (13)3:505-512. Ramsay, E. P. 1882-3. Description of a new species of Coris from Lord Howe Island. Proc. Linn. Soc. N.S.W. 7(2): 301. Tonnoir, A. L. 1923. Notes on Australian Rhypliidae. Ann. Mag. Nat. Hist. (9) 12: 502-505. Turner, A. J. 1917. On some moths from Lord Howe and Norfolk Islands in the South Australian Museum. Trans. Roy. Soc. S. Aust. 41:53-120. 1918. Further notes on some moths from Lord Howe Island, etc. Trans. Roy. Soc. S. Aust. 42:276-289. 1922. Some Australian moths from Lord Howe Island. Proc. Linn. Soc. N.S.W. 47(4): 439-440. Tilly ard, R. J. 1917. Odonata, Plannipennia, and Trichoptera from Lord Howe and Nor- folk Islands. Proc. Linn. Soc. N.S.W. 42(3) : 529-544, figs. Waterhouse, G. A. 1897. The Rhopalocera of Lord Howe Island. Proc. Linn. Soc. N.S.W. 22(2) :285-287. 1920. Description of new forms of butterflies from the South Pacific. Proc. Linn. Soc. N.S.W. 45(3): 468, 470. Wheeler, W. M. 1919. The ant genus Lodo- myma. Psyche, Boston 24:97-106. 1919. The ants of the genus Metapone. Ann. Ent. Soc. Amer. 12(3) : 173—191, figs. 1927. Ants of the genus Amhlyopone. Proc. Amer. Acad. Arts and Sci., Boston 62: 1-4, 15-17, figs. NOTES Additional Plants from the Midway Islands A collection of plants from the Midway Is- lands, made by Dr. Hubert W. Frings, Depart- ment of Zoology, University of Hawaii, on April 16, 1962, includes four species not previ- ously recorded from there. These are: GRAMINEAE Polypogon monspeliensis (L.) Desf., Eastern Island. CRUCIFERAE Coronopus didymus (L.) J. E. Smith, Eastern Island. EUPHORBIACEAE Euphorbia peplus L., Sand Island. PRIMULACEAE Anagallis arvensis L., Sand and Eastern Islands. All four of these species, although not native to Hawaii, are now found on the major islands of the Hawaiian chain and were probably in- troduced accidentally to Midway from Oahu. At various times ornamental plants, and the soil in which to grow them, have been taken to Midway from Honolulu. It is likely that seeds of the plants reported here reached Midway in such soil. Dr. Frings’ collection included 27 species of vascular plants, which he described as the more common species on the atoll. It is of interest to note that only 5 of these species ( Boerhavia diffusa L.; Tribulus cistoides L.; Ipomoea indica (Burm. f.) Merr.; I. pes-caprae (L.) Sw.; Scaevola sericea Vahl) are native to Midway. This pro- vides some indication of the changes in the flora, and concomitant changes in the ecosystem, which are related to man’s activities on Midway. This collection has been deposited in the herbarium of the B. P. Bishop Museum in Ho- nolulu— Charles H. Larnoureux, Department of Botany, University of Hawaii, Honolulu, Hawaii. On Malayan Shores: A Review S. H. Chuang. 1961. On Malayan Shores. Pp. xvi + 225; 112 numbered plates and frontispiece (13 plates and frontispiece in color), 28 text figures. 5.5 X 7.5 inches (14 X 19 cm). Pub- lished by Muwu Shosa, P. O. Box 1813, Singa- pore. It is for this incredibly rich center of the great Indo-West Pacific marine flora and fauna that Dr. Chuang’s small volume is intended as an introduction. He did not write the book for the specialist in any one group, or even for the trained marine biologist, but for the "amateur naturalist.” With this aim he has avoided as much of the specialized language of zoology as possible; his descriptions of plants and animals are short and point out only one to several salient characteristics; he uses no keys and gives no synonyms. On the other hand his illustrations are numerous and excellent: some 50 species are illustrated by line drawings in the text fig- ures and over 500 are represented by clear reproductions of photographs in the 112 plates. He has limited himself to the more common and conspicuous species, as would be expected; he does not touch upon the fishes which have been dealt with in previous books. The book is divided into two parts, the first is a general review of aspects of marine biology 374 On Malayan Shores: A Review 375 applicable within the scope of the work, the second is a systematic review. The first section includes chapters on the chemistry of sea water, zonation of the shore, coral reefs, plankton, etc. There is also a chapter valuable for a trained biologist visiting the peninsula which discusses Malayan collecting localities ( Chapter VI ) . The second portion is divided among seven phyla, although eight or nine additional phyla are dis- cussed, with species named, in the chapter on animal life of Part I. The appendix gives faunal lists for four typical shore habitats. A glossary gives Malayan names for over 100 species. The bibliography is short and covers only those books cited in the text, not the systematic works from which the nomenclature was derived. The index is very complete, listing all systematic names applied to all taxa, English common names, topics discussed in the first part of the book, and authors cited; it even includes a glossary to some of the more technical terms used, as "siphonoglyph” and "periostracum.” While this book is addressed only to the amateur, it obviously will be used by two other groups: students of biology in Malaya, and ma- rine biologists visting these waters for the first time. Having spent a year working on the ad- jacent Gulf of Thailand, and having briefly vis- ited Malayan shores, I can attest how extremely valuable this volume will be to visitors. More- over, because the fauna from India to Indonesia, from the Coral Sea to the South China Sea is basically the same, it will be of use to all visitors to the Western Pacific and adjacent Indian Ocean. Admittedly the book has limitations: it is not an exhaustive scientific monograph on the flora and fauna. Yet if it were, it would not be one small volume, but many large volumes. Rather than disparage the book for omissions, we should commend the author for producing a volume of considerable value to biologists, and regret that books such as this are not available for other areas in the tropical Pacific. — A. H. Banner, Department of Zoology, University of Hawaii, Honolulu, Hawaii. Manuscript Form. Manuscripts should be typed on one side of standard-size, white bond paper and double-spaced throughout. Pages should be consecu- tively numbered in upper right-hand corner. Sheets should not be fastened together in any way, and should be mailed flat. Inserts should be either typed on separate sheets or pasted on proper page, and point of insertion should be clearly indicated. Original copy and one carbon copy of manuscript should be submitted. The author should retain a car- bon copy. Although due care will be taken, the editors cannot be responsible for loss of manuscripts. Introduction and Summary. It is desirable to state the purpose and scope of the paper in an introductory paragraph and to give a summary of results at the end of the paper. Dictionary Style. It is recommended that authors fol- low capitalization, spelling, compounding, abbrevia- tions, etc., given in Webster’s New International Dic- tionary (unabridged), second edition; or, if desired, the Oxford Dictionary. Abbreviations of titles of pub- lications should, if possible, follow those given in World List of Scientific Periodicals. Footnotes. Footnotes should be used sparingly and never for citing references (see later). When used, footnotes should be consecutively numbered by supe- rior figures throughout the body of the paper. Foot- notes should be typed in the body of the manuscript on a line immediately below the citation, and sepa- rated from the text by lines running across the page. Citations of Printed Sources. All references cited should be listed alphabetically by author at the end of the paper, typed double-spaced. References to books and to papers in periodicals should conform to the following models : Batzo, Roderick L., and J. K. Ripkin. 1849. A Treatise on Pacific Gastropods. Rice and Shipley, Boston, vii + 326 pp., 8 figs., 1 map. Crawford, David L. 1920^. New or interesting Psyllidae of the Pacific Coast (Homop.). Proc. Hawaii. Ent. Soc. 4(1) : 12-14. 1920&. The sandalwoods of Hawaii. Proc. Hawaii. Ent. Soc. 4(2): 374-375, 13 pis. In the text, sources should be referred to by author, date, and page, as follows: "It was noted (Rock, 1916: 18) that . . .” or "Rock (1916: 21-24) says . . .” Quotations. Quoted matter of fewer than five printed lines (about 200 characters) should be given in the text in the usual form, using double quote marks. Longer quotations should be set flush with left mar- gin. The author is responsible for the accuracy of quoted material. Numbers. Decimals, measurements, money, percent- ages, time; enumerations in which any figure is 10 or over; and isolated enumerations of 10 and over should be given in Arabic figures, rather than spelled out, except when the number begins a sentence. ILLUSTRATIVE MATTER Only the minimum number of illustrations required to supplement the text will be accepted by the editors. Reproduction costs of illustrations in excess of the number allowed by the editors will be paid by the author. Artwork for illustrations should be 8Y2 x 11 inches or smaller, and it should accompany manuscript, on separate sheets. Often more valuable than a photo- graph is a good line drawing. Figures and Graphs. Copy for figures and graphs should always be drawn large enough to allow for at least one-third reduction by the engraver. Copy should consist of carefully prepared line drawings in one color only, drawn in India ink on plain white draw- ing paper or tracing cloth. Co-ordinate paper with lines in light blue (a color which will not show in a photograph) may be used; but co-ordinates which should appear in the finished graph must be drawn in India ink. If original figures may not be conven- iently submitted with manuscript, duplicate rough sketches or photographic prints may be furnished to aid the editors in their decisions. It is strongly urged that an indication of scale be incorporated as a part of all drawings in which mag- nification and size are critical considerations. Photographs. Photographs should be chosen for clarity in portraying essential information. They should be printed for contrast, on glossy paper, and should be sent unmounted. They should be identified with serial number written in soft pencil on the back to corre- spond with list of captions. Illustrations will be returned to the author. Tables. Tabular matter should be kept to a minimum. Each table, prepared to conform with Pacific Science style, should be typed on a separate page, and its posi- tion indicated on the manuscript. Mathematical Formulas. Complicated formulas cannot be set by the printers. Authors should submit them as illustrations. Captions. Readily identifiable captions for figures, graphs, photographs, and other illustrative matter should be supplied on a separate page. PROOF Proof should be corrected immediately and returned at once to Robert Sparks, assistant to the editors. Authors are reminded that the editors will allow only a minimum number of corrections on galley proof. Additions to the printed text and changes in style and content are not permitted. All queries on proof should be answered. If cor- rected proof is not received within four weeks after being sent to the author, author’s changes cannot be accepted. REPRINTS Reprints or separates should be ordered on the form provided and returned with author’s proof. All correspondence concerning separates must be directed to the printer, Star-Bulletin Printing Company, 420 Ward Avenue, Honolulu 14, Hawaii. NO. 4 VOL. XVII OCTOBER 1963 PACIFIC SCIENCE A QUARTERLY DEVOTED TO THE BIOLOGICAL AND PHYSICAL SCIENCES OF THE PACIFIC REGION JEN-HU CHANG Climatology in Hawaiian Sugar-Cane Industry D. L PLUCKNETT, J. C. MOOMAW, and C. H. LAMOUREUX Root Development in Aluminous Hawaiian Soils BENJAMIN C. STONE New and Critical Species of Pelea ROBERT L. WILBUR A Prior Name for Gouldia terminalis WILLIS E. PEQUEGNAT Population Dynamics in a Sublittoral Epifauna THOMAS S. HIDA and ROBERT A. MORRIS The Marquesan Sardine in Hawaii DONALD C. MATTHEWS Folliculinids from Submerged Wood CHARLES P. HOYT Rhinoceros Beetles in West Africa V. S. SOBOLEV Features of Vo Icanism of the Siberian Platform MAXWELL S. DOTY Gibsmithia hawaiiensis gen. n. et sp. n. HAROLD ST. JOHN Revision of the Genus Pandanus Part 16. Thailand and Vietnam INDEX UNIVERSITY OF HAWAII PRESS BOARD OF EDITORS O. A. BUSHNELL, Editor-in-Chief Department of Microbiology, University of Hawaii Robert Sparks, Assistant to the Editors Office of Publications and Information, University of Hawaii L. H. Briggs Department of Chemistry University of Auckland Auckland, New Zealand Maxwell S. Doty Department of Botany University of Hawaii Ai Kim Kiang Department of Chemistry University of Malaya, Singapore Gordon A. Macdonald Department of Geology University of Hawaii John J. Naughton Department of Chemistry University of Hawaii Martin Sherman Department of Entomology University of Hawaii Walter R. Steiger Department of Physics University of Hawaii Donald W. Strasburg Bureau of Commercial Fisheries, Hawaii Area (U. S. Fish and Wildlife Service) Honolulu, Hawaii Donald G Matthews Department of Zoology University of Hawaii Sidney J. Townsley Department of Zoology University of Hawaii Thomas Nickerson, Managing Editor Assistant to the University Provost INFORMATION FOR AUTHORS Contributions to Pacific biological and physical science will be welcomed from authors in all parts of the world. (The fields of anthropology, agriculture, engineering, and medicine are not included.) Manu- scripts may be addressed to the Editor-in-Chief, PACIFIC SCIENCE, University of Hawaii, Honolulu 14, Hawaii, or to individual members of the Board of Editors. Use of air mail is recommended for all communications. Manuscripts will be acknowledged when received and will be read promptly by members of the Board of Editors or other competent critics. Authors will be notified as soon as possible of the decision reached. Manuscripts of any length may be submitted, but it is suggested that authors inquire concerning possi- bilities of publication of papers of over 30 printed pages before sending their manuscripts. Authors should not overlook the need for good brief papers, presenting results of studies, notes and queries, com- munications to the editor, or other commentary. PREPARATION OF MANUSCRIPT It is requested that authors follow the style of Pacific Science described herein and exemplified in the journal. Authors should attempt to conform with the Style Manual for Biological Journals, Am. Inst. Biol. Sci. Washington. ( Continued on inside back cover) PACIFIC SCIENCE A QUARTERLY DEVOTED TO THE BIOLOGICAL AND PHYSICAL SCIENCES OF THE PACIFIC REGION VOL. XVII OCTOBER 1963 NO. 4 Previous issue published September 17, 1963 CONTENTS PAGE The Role of Climatology in the Hawaiian Sugar-Cane Industry: An Example of Applied Agricultural Climatology in the Tropics. Jen-hu Chang 379 Root Development in Aluminous Hatvaiian Soils. D. L. Plucknett, J. C. Moomaw, and C. H. Lamoureux 398 Studies in the Hawaiian Rutaceae , IV. New and Critical Species of Pelea A. Gray. Benjamin C. Stone A Prior Name for the Hawaiian Gouldia terminalis (Ruhiaceae). Robert L. Wilbur Population Dynamics in a Sublittoral Epifauna. Willis E. Pequegnat Preliminary Report on the Marquesan Sardine, Harengula vittata, in Hawaii. Thomas S. Hida and Robert A. Morris Hawaiian Records of F olliculinids ( Protozoa ) from Submerged Wood. Donald C. Matthews Investigations of Rhinoceros Beetles in West Africa. Charles P. Hoyt Characteristic Features of the Volcanism of the Siberian Platform. V. S. Sobolev Gibsmithia hawaiiensis gen. n. et sp. n. Maxwell S. Doty Revision of the Genus Pandanus Stickman, Part 1 6. Species Discovered in Thailand and Vietnam. Harold St. John Index 407 421 424 431 438 444 452 458 466 493 Pacific Science is published quarterly by the University of Hawaii Press, in January, April, July, and October. Subscription price is $4.00 a year; single copy, $1.25. Check or money order payable to University of Hawaii should be sent to University of Hawaii Press, Honolulu 14, Hawaii, U. S. A. Printed by Star-Bulletin Printing Company, 420 Ward Avenue, Honolulu 14, Hawaii. fliftHSORUff issTirunon FEB 2 6 1964 The Role of Climatology in the Hawaiian Sugar-Cane Industry: An Example of Applied Agricultural Climatology in the Tropics1 Jen-hu Chang Climatological study in the Hawaiian sugar- cane industry has a long, noteworthy history. Meteorological observations on the plantations were initiated in 1883, preceding the establish- ment of the first official weather bureau station in Hawaii by fully 20 years. The climatological network in the cane-growing areas expanded by leaps and bounds to 50 stations at the turn of the century, and to 500 stations in I960 in an area of only 350 sq miles (Fig. 1). In the late 1920’s and early 1930’s Das ( 1928, 1931 a, 1931 b9 1932) used the rainfall and tem- perature records to define the effect of climate on crop yield and juice quality in a simple, direct manner. He also advocated the use of day-degree as a guide to irrigation control (Das, 1936). Investigations along the same line of climate- and-plant complex were carried further by Wil- liams (1933), Borden ( 1940, 1949), Clements (1940), Swezey (1942), and others into the early 1940 s. At the end of World War II meteorologists, freshly relieved from their wartime duties, were able to turn their attention to a multitude of peacetime problems. The Hawaiian sugar in- dustry promptly seized this opportunity. The Experiment Station and the Pineapple Research Institute founded their joint Meteorological De- partment in 1946 and contracted with a group of meteorologists at the University of Chicago to investigate dynamic and regional climatology of Hawaii. The culminating results of this in- tensified research were a series of papers in the Meteorological Monographs, discussing the gen- eral circulation, weather types, local flow pat- terns, rainfall statistics, and the like. Such studies inevitably led to the improvement of forecasting and furnished valuable information for opera- tional planning. At the half-century mark, when weather 1 Contribution No. 121, Experiment Station, Ha- waiian Sugar Planters’ Association, Honolulu, Hawaii. Manuscript received June 14, 1962. modification was a new subject, the industry embarked upon an intensive study of cloud seed- ing. Leading scientists from seven research in- stitutes in the U.S.A. and abroad participated in the so-called "Project Shower.” It was hoped that the Hilo coast, with its humid trade winds, would provide an ideal environment where rain- fall could be induced by the addition of chemi- cals to the warm clouds; it was soon realized, however, that scientific know-how of artificial rainfall was probably still decades away. The project nevertheless produced valuable informa- tion in the field of cloud physics. Realizing the limitations upon man’s ability to modify weather on a grand scale, the industry sought instead to improve the efficiency of water use through the study of micrometeorology. In 1957 an evapotranspiration project was initiated at the Experiment Station. Apart from its ob- vious application in irrigation planning, the de- termination of potential evapotranspiration pro- vides a vital link for solving the energy budget and water balance equation in the soil-plant system. The solution of these equations renders the climate-and-yield relationship amenable to quantitative treatment. Climatological study in the Hawaiian sugar industry thus encompasses a wide variety of topics. More than 50 papers have appeared in a dozen scientific journals. The industry ranks among the leaders in the study of agricultural meteorology. The experience gained during the past three decades may very well benefit research workers in other parts of the world, especially those in the tropics. This paper attempts to summarize these studies. It is hoped that such a stock-taking will not only serve as a reminder of our past accomplishments but also as a guide- post for planning future research. TRADE-WIND WEATHER The subtropical high pressure cell in the Pa- cific Ocean with its attendant trade wind is the 379 380 PACIFIC SCIENCE, Vol. XVII, October 1963 Fig. 1. Location map. basic circulation in Hawaii; all other weather types are perturbations in this basic current. The eastern portion of a subtropical anticyclone is characterized by divergence and subsidence and consequent adiabatic heating and pronounced dryness aloft. There is, however, a mixing layer near the surface where the air in contact with the cold ocean currents becomes cold and humid. These two layers of air are separated by a zone of temperature inversion which, along the Cali- fornia coast, occurs at an altitude of about 1,500 ft. Downstream, toward the west, as water tem- perature rises and subsidence weakens, the trade- wind inversion reaches a height of about 6,000 ft in Hawaii. Figure 2 shows the vertical structure of trade wind in Hawaii. The air below the inversion is moist, with a mixing ratio of 12-15 gm/kg at the ground surface. In the absence of orographic uplift this moist air is a poor rain-producer be- cause the trade inversion acts as a lid to oppose the development of convective clouds; therefore, in lowland areas, the trade wind is atmospheri- cally moist in terms of humidity but ecologically dry in terms of rainfall. The air below the trade inversion has a lapse rate close to dry adiabatic. It requires only a little uplifting to produce rainfall. Baer ( 1956) has demonstrated theoretically that if the rela- tive humidity is approximately constant below the inversion layer, rainfall would increase with height exponentially up to the base of inversion. The logarithmic distribution of rainfall with height is a unique climatic feature in Hawaii. The prevalence of trade-wind weather in Ha- waii is dependent upon the location and strength of the Pacific subtropical anticyclone. In July and August, when the subtropical ridgeline in the high troposphere is located to the north of the islands, the trades prevail during 97% of Climatology and Sugar Cane — CHANG 381 the time (Yeh et al., 1951^) . In winter, when the subtropical anticyclone moves southward, the frequency of trade-wind weather decreases to 40-50% (Fig. 3). OTHER WEATHER TYPES During one-third of the year, when the trades are not prevalent, weather in Hawaii falls into two general types: mixed and cyclonic. The mixed type is characterized by the presence of an essentially east-west polar front north of the islands. This type is a very poor rain-producer as the front is located at a distance. The mixed type often ushers in cyclonic weather. Cyclonic weather occurs in Hawaii during 22% of the year. The cyclonic weather may be further divided into four subtypes: (1) extra- tropical cyclones, ( 2 ) kona storms, ( 3 ) easterly waves, and (4) tropical cyclones or hurricanes. The first two subtypes occur most frequently in winter while the latter two subtypes are pri- marily summer disturbances. Uncomplicated extratropical cyclones occur only when the zonal westerlies move southward into the tropics. They are, however, weakened on their journey southward and produce only TYPICAL LAPSE RATE CURVE SHOWING THE STRUCTURE OF TRADE WIND IN HAWAII I4°F 23 32 41 50 59 68“F FIG. 2. Structure of the trades in Hawaii. Fig. 3. Monthly variation of frequencies of three weather types. small rains. Some of these cyclones reach only the northern island of Kauai. The kona storm is essentially a cold-core low developed in the tropics. Simpson (1952) has described two main processes in its formation: ( 1 ) the transformation of mid-latitude cyclones which have been trapped at low latitudes by blocking action of a warm high; (2) the cy- clogenesis in the easterlies triggered by the building down of a pre-existing cold upper low usually at the southern extremity of a polar trough. Once formed, all kona storms seem to assume the same characteristics. The term kona indicates that the wind during such a synoptic- period is southerly. Kona storms occur about three or four times a year, bringing heavy rain- fall throughout the islands. In the dry lowland areas they may account for more than half of the annual rainfall. The occurrence of winter cyclones and kona storms in Hawaii is closely related to the general circulation. Yeh et al. (1951 b) have shown that winter rainfall in Hawaii increases as the lati- tude of the jet stream increases, unless the latter penetrates to the latitude of the islands them- selves. They explained that when the jet stream is weak and far to the north, Hawaii will ex- perience a period of low zonal index. Low index 382 flow pattern is characterized by large meridional flow and frequent interaction between tropical perturbations and troughs in the westerlies. On a hemispherical scale the mid-tropospheric westerlies are composed of roughly sinusoidal waves. In winter an upper air trough is climato- logically anchored off the east coast of Asia. The location of the next downstream trough depends upon the zonal velocity according to the well-known Rossby formula. The zonal ve- locity in the western Pacific is such that the trough is usually located to the west of the is- lands in January and to the east in February (Namias and Mordy, 1952). Since the trough is characterized by convergence ahead and di- vergence behind, cyclones in Hawaii are more numerous in January than in February. The low rainfall in February is especially evident in the PACIFIC SCIENCE, Vol. XVII, October 1963 northern island of Kauai (Leopold and Stidd, 1949). The easterly waves are trade-wind perturba- tions that may form during any part of the year but with a maximum during summer. In August the easterly waves occur along the east coast of Hawaii every 3 or 4 days, bringing with them decks of high clouds and rainfall; however, they dissipate readily over the rugged land and sel- dom reach Oahu and Kauai. Since 1904 there have been only four tropical storms with wind of 74 rnph or more passing through the islands, all in the 1950’s. The for- mation of a hurricane in the central Pacific is somewhat of a climatic anomaly and requires a combination of favorable conditions. In 1957, when two hurricanes passed through Hawaii, Frazier (1957) noted that a well-developed Fig. 4. Median annual rainfall (inches) in Hawaii. Climatology and Sugar Cane — CHANG 383 trough extending to low latitudes persisted along the west coast of North America throughout the summer and fall season. This trough, through an energy dispersion mechanism, sharpens the next downstream subtropical trough, thus pro- moting cyclogenesis in the latter area ( Ramage, 1959). The increase in radiation and tempera- ture in Hawaii during the last decade, which will be discussed in a later section, was probably also an important cause for the increased hurri- cane activity. RAINFALL If the islands of Hawaii did not exist, the annual rainfall over the ocean would be in the neighborhood of 30 inches; the actual average rainfall over the islands is about 70 inches. This increase of 40 inches of rainfall is the result of orographic uplifting of trade wind and its per- turbation-easterly waves. The effect of easterly waves is strongest near Hilo and diminishes northward toward the Hamakua coast. The me- dian annual rainfall at Hilo is 139 inches, con- siderably higher than any other coastal area in the state (Fig. 4). The trade-wind rainfall increases with alti- tude exponentially up to a certain point and then decreases again. The belt of maximum rainfall varies from 3,000 to 4,000 ft, depending upon the effect of local topography on the flow patterns. This effect has been discussed in detail by Leopold (1949). The steep isohyetal gradient in a trade-wind climate is best illustrated in Kauai, where Mt. Waialeale, with a mean annual rainfall of 465 inches, is only 15 miles away from the semiarid west coast. Most of the sugar plantations are located in lowlands with a me- dian rainfall of less than 50 inches. A few cane- growing areas on Hawaii, however, receive as much as 200 inches of rainfall in a year. Except along the Kona coast of Hawaii, where the upslope sea breeze produces frequent sum- mer afternoon showers, winter is the wet season throughout the state. The contrast between sum- mer and winter rainfall is most accentuated in the dry lowlands, where the monthly rainfall in summer is often less than 0.5 inch. Summer rainfall minima are found only in a very few places in the tropics and are designated as "As” climate in the Koppen classification. In spite of the large areal and seasonal varia- tions, the rainfall distribution for any particular month bears a close resemblance to that of the annual. Thus Stidd and Leopold (1951) were able to express the monthly rainfall as a function of annual rainfall in the following manner: y = a (x — 30) -f b where y is monthly rainfall; x, annual rainfall; a, the gradient factor of observed orographic in- crease of rainfall through increment of average annual rainfall; b, a geographic constant quan- tity derived from a rainfall blanket of uniform thickness over the islands and adjacent ocean. The constant 30 is inserted because the mini- mum annual rainfall in the dry lowland is about 30 inches. This same concept was later expanded to de- scribe daily rainfall distribution during trade- wind weather. For the island of Oahu, for ex- ample, a dry index of zero is assigned to a theoretical station having a zero mean annual rainfall, and a wet index of 100 to a station having 250 inches mean annual rainfall. A daily forecast chart is constructed by plotting the in- dices as the abscissa and the daily rainfall amounts as the ordinate. The forecaster fore- casts the daily rainfall for the dry and wet index stations. From the straight line connecting these two reference points, daily rainfall for any actual station may be read off as the ordinate corre- sponding to the abscissa of the station’s mean annual rainfall. The trade-wind rainfall is more likely to occur during the night or in early morning than during the day. Loveridge (1924) attributed the noc- turnal rainfall to the radiative cooling at the top of the clouds. Leopold (1948), however, added that cooling at night would lower the condensation level. Nocturnal rainfall, in distinct contrast to afternoon showers in many tropical countries, is in many ways beneficial to agricul- ture. Trade-wind showers are very light, with drop size less than 2 mm in diameter (Blanchard, 1953), and with intensities usually much less than 0.2-0.3 inch per day. Only kona storms and hurricanes are capable of causing severe crop damage. On January 24, 1956, a kona storm deposited 38 inches of rain at Kilauea, Kauai. Such severe storms are rare, however, averaging CUMULATIVE PERCENTAGE OF RAINFALL Fig. 5. Relationship between per cent of days with rain and per cent of total rainfall. one in every 5 or 6 years. Furthermore, storm damages are usually restricted to Kauai and the east coast of Hawaii. The frequency distribution of daily rainfall in Hawaii is extremely skewed. Figure 5 shows the relationship between the percentage of rainy days and the percentage of rainfall amounts cumulated from the least to the heaviest. At Ewa, for example, 10% of the days with the heaviest rain accounted for nearly 60% of the total, while 50% of days with least rainfall amounted to only 6% of the total rain. The skewness of the daily rainfall frequency curve decreases slightly with the increase of annual rainfall. The skewed rainfall distribution has at least two important implications. First, a large amount of the annual rainfall will probably be lost as runoff. Second, the mean monthly rain- fall will be considerably higher than the median, and the former is a poor indicator of the "normal” condition. For this reason median rainfall is used extensively in Hawaii. As rainfall variability is very high in the tropics (Biel, 1929), even the use of the median is inadequate for many agricultural pur- Climatology and Sugar Cane — CHANG 385 poses. A knowledge of the extremes and fre- quency probabilities is indispensable. Because the basic cause of rainfall variation in Hawaii is rather uniform, the rainfall probabilities for different stations can be related in a simple, empirical manner (Landsberg, 1951). Figure 6 shows the probabilities (slanting lines) of hav- ing an annual amount of less than a given quan- tity (ordinate) as a function of the median annual amounts (abscissae), based on at least 60 years’ records of 20 plantation stations. It is evident that Figure 6 could be used as a risk chart. Monthly rainfall probabilities could also be presented in the same manner. TEMPERATURE The mean annual temperatures in the low- land plantation areas vary from 72.5 F along the east coast of Hawaii to 75 F for the drier stations. The temperature decreases with eleva- MEDIAN ANNUAL RAINFALL (INCHES) g co Fig. 6. Probabilities of having individual rainfall amounts in Hawaii as a function of median annual rainfall. tion at an average rate of 4 F per 1,000 ft. The coldest plantation stations, at an elevation of about 3,000 ft, have a mean annual temperature of 62-63 F. All the stations in Hawaii below an altitude of 5,000 ft have an annual temperature range of less than 9 F (Jones, 1942) ; thus a station at an elevation of 2,000 ft with an annual temper- ature of 65 F would have a minimum monthly temperature as high as 62 F. The isothermal climate is favorable for the growth of a peren- nial crop like sugar cane. In Hawaii, as in many other tropical coun- tries, the daily temperature range exceeds the annual mean daily temperature range and ex- hibits greater areal differences than does the mean annual temperature. Figure 7 shows the distribution of the annual mean daily tempera- ture range in the cane-growing areas. In general the dry leeward stations have the greater tem- perature range. The daily temperature range is subject to a small seasonal variation, being slightly higher in the winter. According to the results of phytotron experiments at the Cali- fornia Institute of Technology, daily tempera- ture range exerts a profound influence on fruit quality (Went, 1957). The distribution of soil temperature in the tropics can be deduced fairly accurately from observations of air temperature (Chang, 1958). The mean annual soil temperature at any depth differs only slightly from the mean annual air temperatures. The annual temperature range at the soil surface under the cover of a sugar-cane crop is reduced to half of the air temperature range. Thus, in most of the lowlands in Hawaii soil temperature exceeds 72 F throughout the year. Studies at the Experiment Station have established that temperatures of 62 F are ex- tremely limiting for cane growth and nutrient- and water-uptake. Such low temperatures are observed only in the mountains above 2,000 ft. The lowest soil temperature recorded on sugar cane land was 61 F at Hamakua plantation at an elevation of 3,000 ft. RADIATION Radiation measurements have been taken at 35 plantation stations by photochemical tubes which use oxalic acid as agent and uranyl sul- 386 PACIFIC SCIENCE, Vol. XVII, October 1963 Fig. 7. Annual mean daily temperature range (°F) in Hawaii. fate as catalyst. The photochemical method is accurate enough for general agricultural pur- poses (Chang, 1961). Apart from the direct effect on local climate and plant growth, the radiation data could be used for assessing yield potential, estimating irrigation water needs, directing fertilizer practices, etc. (Borden, 1940). Figure 8 shows the distribution of mean daily radiation in Hawaiian sugar plantations. Maui has the highest radiation. Pukalani, in central Maui, receives 607 langleys/day or over 220,000 langleys/year. This is exceeded only by a very few desert areas in the world (Budyko, 1958). The average radiation for the cane-growing areas in Hawaii is in the neighborhood of 510 langleys/day, well above Houghton’s (1954) estimate of 415 langleys/day for the land areas in the latitudinal zone 0°-20° N. In Hawaii the minimum radiation is recorded in December, while the maximum is usually in June or July. The maximum monthly radiation is about 15% higher, and the minimum some 2 5 % lower, than the annual mean. Radiation in December is about two-thirds that of June or July- SECULAR CLIMATIC CHANGES Since the cause for climatic variation in Ha- waii is rather uniform, and since the leeward stations in particular are sensitive to circulation regimes (Landsberg, 1951), the climatic change at Makiki, in Honolulu, can be considered to 1 illustrate that of the islands in general. The average daily radiation at Makiki in- creased some 10% from 491 langleys/day in the 1930’s to 544 langleys/day in the 1950 s (Fig. 9). The rising trend probably started in 1936, but records before 1932 are not available. Climatology and Sugar Cane — CHANG 387 Increased radiation results in a rise of tempera- ture, though with a large time lag in a maritime climate. The temperature at Makiki (Fig. 10) did not start to rise until 1948. Increased radiation could conceivably bring about a change in general circulation. Went- worth (1949) noted that the prevailing wind in Honolulu shifted from northeast to east from 1907 to the late 1930’s and veered back there- after. Increased frequency of east wind suggests a southerly location of the subtropical anti- cyclone, less frequent trade-wind weather, but not necessarily an increase in cyclonic activities; therefore, the trade-wind rainfall should de- crease when the east wind is prevalent. At Makiki there was a trend to decreased rainfall from the 1920’s to 1941 and to increased rain- fall thereafter (Fig. 11). This explanation of the rainfall trend is tentative, however, and re- quires further study. Whatever the causes for climatic change may be, its impact on agriculture is varied and pro- found. In general, the climate in Hawaii has become more favorable for sugar-cane culture during the last 20 years. This must account for part of the yield increase during that period. MODIFICATION OF CLIMATE The climate of Hawaii is, in general, well suited to the growth of sugar cane. Water econ- omy is the one area where the climatic environ- ment could be altered significantly to bring about a higher yield. Irrigation in the Hawaiian sugar industry started in 1852. At the present time 54% of the plantation area is irrigated. In Fig. 8. Solar radiation (langleys/day) in Hawaii. 388 PACIFIC SCIENCE, Vol XVII, October 1963 Fig. 9- Solar radiation at Makiki, 1932-60. general the irrigated areas have less than 60 inches of rainfall in a year. Intelligent management of irrigation requires a knowledge of the water need by a particular crop. Das (1936) advocated the use of day- degree as a measure of water need and as a guide to irrigation interval control. Numerous field experiments were conducted in Hawaii to determine the relationship between sugar yields and the number of day-degrees between irriga- tion. It was found that temperature is a poor indicator of the solar energy, which determines to a large extent the water needs of a crop. In fact, the monthly temperature and radiation at Makiki for the period 1932-60 have a correla- tion coefficient of only 0.57. The correlation is even poorer when the maximum temperature is used instead of the mean. In other parts of the world also, the use of day-degree as a phenological index has been refuted (Schneider, 1952; Wang, I960). As interest in the day-degree approach waned, Baver (1954) properly called attention to the rapid development of the evapotranspiration and other micrometeorological concepts and their application in irrigation. He contended that the meteorological approach has the ad- vantage of simplicity of operation when com- pared with methods based upon measurement of soil moisture change. More important, how- ever, are the many applications of evapotrans- piration in determining the regional water balance and in disentangling the climate-yield relationship. With these ideas in mind the Ex- periment Station started an intensive study of micrometeorology in 1957. EVAPOTRANSPIRATION AND ENERGY BUDGET Sugar cane is a tall, ungainly plant with an aerodynamically rough canopy. Wind profile measurements indicate that the roughness pa- rameter of a mature sugar cane of 4 cm height is 9 cm, as against 2.3 cm for thick grass of 10 cm height and 0.1 cm for short lawn grass (Sutton, 1953). The high roughness of a sugar- cane crop renders the aerodynamic approach to evapotranspiration a difficult task (Deacon et al., 1958). This, together with the lack of a suitable instrument for measuring vapor flux, accounts for the fact that the aerodynamic method was not assayed in our experimental work. The potential evapotranspiration of sugar cane was measured by drainage lysimeters and the data were analyzed by the IBM computer according to the Penman (1948) and Thorn- thwaite (1948) formulae. Detailed discussion of the instrumentation and the results has been reported elsewhere (Chang, 1961). It needs only to be emphasized that in a tropical mari- FlG. 10. Five-year moving mean of temperature at Makiki, recorded at middle year, for the period 1918-61. Climatology and Sugar Cane — CHANG 389 Fig. 11. Five-year moving mean of rainfall at Makiki, recorded at middle year, for the period 1920-58. time climate the Penman estimates are far better than the Thornthwaite formula. The various radiation components were meas- ured at Makiki to evaluate the energy budget approach to evapotranspiration. Over a mature cane field the partition of incoming radiation is approximately as follows: 16% reflected radia- tion, 17% back radiation, and 67% net radia- tion (Fig. 12). The relationship between the net and incoming radiation is almost a constant throughout the year and over different vegeta- tive surfaces. It was observed at Wahiawa that over short grass and pineapple fields, which have an albedo of about 5%, the net radiation remains two-thirds of incoming radiation. Theoretically it is difficult to reconcile the fact that the net radiation is not affected by the de- crease of albedo, but from a practical standpoint this is convenient. In the tropics, especially under the cover of tall vegetation, heat flux to and from the soil is negligible; therefore, the net radiation is consumed either in heating the air or in evapo- transpiration. At Makiki 82 % of the annual net radiation, or 55% of the incoming radiation, is used in evapotranspiration. There is a small sea- sonal variation of this percentage value, which is some 10% higher in summer than in winter. There is also a regional variation of the energy partition. For instance, the percentage of insolation used in evapotranspiration increases with total radiation. In Hawaii the regional variations of energy budget are rather small, and the equation at Makiki could be used to esti- mate with reasonable accuracy the monthly potential evapotranspiration for other cane- growing areas in Hawaii. For estimating short term, say weekly, potential evapotranspiration it is advisable to use an evaporimeter which integrates not only radiation but other meteoro- logical elements as well. The simple energy budget presented here is applicable only in the humid tropics. In middle and high latitudes the energy budget is subject to enormous seasonal variations. In an arid cli- mate the advective heat, which is extremely dif- ficult to evaluate, may introduce a significant error. PAN EVAPORATION AS A MEASURE OF EVAPOTRANSPIRATION Until a cheap and readily installable piece of equipment capable of recording vertical vapor transfer is available, agricultural meteorologists will continue to use evaporation pans. In an arid climate pan evaporation is accentuated by the oasis effect to such an extent that its usefulness as a climatic parameter is greatly impaired. In humid climates, pan evaporation has been found to be more accurate than the Penman and Thornthwaite estimates (Suzuki and Fukuda, 1958), and a satisfactory guide to irrigation control (Krogman and Lutwick, 1961). In this INCOMING RADIATION (100%) FIG. 12. Energy budget over a sugar-cane field at Makiki. 390 PACIFIC SCIENCE, Vol. XVII, October 1963 respect Hawaii is ideally suited to the use of evaporation pans, thanks to the existence of a moist mixing layer below the trade inversion. Lysimeter and pan-evaporation records in Ha- waii indicate that the potential evapotranspira- tion of mature sugar cane is approximately the same as evaporation from a ground-level U.S. Weather Bureau Class A pan. The ground-level pan, however, evaporates some 10% less than an elevated pan at cane top level, and some 15% more than a buried pan in an irrigated plot. The one-to-one ratio obtained by using the ground-level pan is very much in line with ratios found for many other crops. From the standpoint of water use, crops may be divided into two groups: conventional and nonconven- tional. The potential evapotranspiration of a nonconventional crop is influenced to a large extent by its physiology. Rice, with a ratio of 1.2 (Suzuki and Fukuda, 1958), and pineapple, with a ratio of 0.35, are good examples. The water need of a conventional crop is deter- mined primarily by the ^weather conditions, al- though it increases with the roughness of the crop, especially in areas of large advective heat (Tanner and Pelton, I960). Thus the ratio in- creases from 0.75 for short grass (Penman, 1948) to 0.87 for corn (Fritschen, 1960^), and to 1.0 for sugar cane. The fact that potential evapotranspiration of sugar cane approximates pan evaporation does not necessarily mean that irrigation based on a ratio of one-to-one is most economical. A new experiment has been set up at Waipio to de- termine sugar-cane growth and yield by using different pan ratios, i.e., 1.30, 1.15, 1.00, 0.85, 0.70, and 0.55. The yield data over a number of years should permit determination of the most economical level of irrigation. Fig. 13- Monthly water balance at Opaeula. MONTHLY AND ANNUAL WATER BALANCE The monthly water balance of a place can be portrayed by comparing the pan evaporation and the median rainfall. At Opaeula, for ex- ample, rainfall exceeds pan evaporation during the winter months from November to February (Fig. 13). From March to October, however, supplemental irrigation is needed to fill the water deficits, which total 29.6 inches in a year. By analyzing the water balance of some 30 stations we have been able to construct an an- nual water-deficit map for the cane-growing areas in Hawaii (Fig. 14). The map is tentative, as the stations are not well distributed. Recently we have added 20 pan stations in order to strengthen the network. In constructing the map the radiation records were also used to fill the gap. By planimetering the deficit areas in the map, TABLE 1 Estimated Annual Water Deficit for Growth of Sugar Cane ISLAND AREA ( acre ) AVERAGE DEFICIT ( inches ) TOTAL DEFICIT ( acre-inches ) Hawaii 98,601 16.5 1,628,390 Maui 42,424 72.8 3,087,612 Oahu 33,223 41.0 1,361,141 Kauai 47,088 33.4 1,573,502 Total 221,336 34.6 7,650,645 Climatology and Sugar Cane — CHANG 391 the annual water deficits of the four islands are estimated as in Table 1. It is noted that there is a large regional variation. The average annual deficit for plantation areas on Hawaii, which are almost exclusively unirrigated, is only 16.5 inches. By contrast, the average deficit for Maui is as high as 72.8 inches. The total annual water deficit for all the plantations is estimated at 7.6 million acre- inches, or 208 billion gallons. The industry uses about 400 billion gallons of water a year for irrigation. At first glance, the camp crop in Ha- waii seems to be irrigated more than adequately. This is, however, not true. Much of the irriga- tion water is wasted due to maldistribution. For instance, a plantation irrigation at a high level of adequacy often loses as much as 60% of its water through deep penetration or runoff. Fur- thermore, the water supply is often such that a plantation over-irrigates in one season and suf- fers from drought in another. The monthly water-balance charts and the water-deficit maps are a valuable guide to agri- cultural planning. Such questions as whether or not to construct a reservoir, what irrigation sys- tem to adopt, and what crop to grow can be answered, at least in part, through the use of climatic data. The monthly water balance presented above is by no means precise because the median rain- fall is only an approximation of the effective rainfall. The amount of effective rainfall varies with the rainfall intensity as well as with the moisture storage capacity of the soil. To credit rainfall correctly, it is necessary to compute the daily water balance. Fig. 14. Annual water deficit (inches) in Hawaii. 392 PACIFIC SCIENCE, Vol. XVII, October 1963 I960 JULY AUG SEPT OCT NOV DEC FIG. 15. Running daily soil-moisture balance at Waipio. Dashed lines indicate irrigation. DAILY WATER BALANCE If the soil storage capacity is known, the daily water balance can be computed by comparing the daily rainfall and potential evapotranspira- tion. Figure 15 shows the running daily soil moisture balance at Waipio, where the storage capacity is 2.5 inches. The rainfall in excess of the amount that a soil can hold is regarded as surplus. By subtracting the surplus from the rainfall, the effective rainfall can be determined. When the daily water balance reaches zero, drought occurs and irrigation is called for. The water deficits during the drought days can be added to determine the total water deficit dur- ing the growing season of a crop. This daily water-balance scheme assumes that the depletion rate of soil moisture is equal to potential as long as the soil moisture is above the wilting point. This assumption is adopted primarily for simplicity of computation. It should not be taken to mean that we endorse the well-known Veihmeyer and Hendrickson argument (1955). Although the relationship between the depletion rate and the moisture tension is still debatable, there are indica- tions that the relationship probably varies with weather conditions (Denmead, 1961). The con- stant depletion rate reported by Veihmeyer and Hendrickson is probably less in error in a humid, cloudy climate, especially under the cover of tall vegetation, than in a dry continental climate. The computed daily water balance at Waipio has been compared with gypsum-block readings. They agree in general. When the water balance reaches zero, the gypsum-block resistance is usually between 3,000 and 5,000 ohms. The latter value calls for irrigation. In general, the irrigation date determined by the pan evapora- tion is one or two days earlier than that indi- cated by the block reading. KOHALA RESULTS Kohala is a plantation on northern Hawaii with an annual water deficit varying from 20 to more than 50 inches. In I960 only about 27% of the plantation was irrigated. The ques- tion then arises as to whether an expanded irri- gation program will be profitable in the long run. Apparently this question cannot be an- swered without an investigation of the climate- and-yield relationship. The daily water balances during more than 10 years for one irrigated and five unirrigated stations in Kohala were computed by using a storage capacity of 3 inches. The total water deficits for different crops were summarized and correlated with the yields (Fig. 16). The correlations are highly significant for Hawi, Niulii 9, Halawa 3, and Upolu 6, but are poor for Puakea 6 and Union 8. In general they are very good for field crops, especially in view of Climatology and Sugar Cane — CHANG 393 the fact that there were considerable differences in crop age, solar radiation, cultural practice, variety, etc. The average slope of the regression lines for the five unirrigated fields is 4.6 tons of cane per acre for every 10 inches of water. This is an important figure in determining the economy of irrigation. It is thought that by extending the regression lines to the point of zero deficit the approximate yield potential of the area could be estimated. The average yield for the five unirri- gated fields was 65.2 tons of cane per acre. The estimated potential would be 91.4 tons cane per acre, or an increase of 40% over the present yield. This estimate is probably reasonable when compared with the yield of 109 tons of cane per acre at Waialua, Oahu, an irrigated plantation with slightly better climatic conditions. Admittedly the use of linear extrapolation as a means of estimating yield potential is a crude procedure. Yet this seems to be the most rea- sonable method we can adopt at present. The- oretically the relationship between water and yield should be curvilinear, as shown in Figure 17. The exact shape of the curve cannot be determined until results of the Waipio experi- ments become available. The difference in yield potential between the two curves in Figure 17 is caused primarily by radiation, if soil and other factors are constant. Experiments are also underway to assess the effect of solar radiation on yield. It is hoped that in future this con- certed research program will enable us to draw with certainty the climate-and-yield relationship now presented hypothetically in Figure 17. By analyzing the daily water balance at Ko- hala, we also derived an empirical relationship between the actual rainfall and the effective rainfall for plant use. The computation was car- ried out by assuming varying soil-moisture storage capacities of 2, 3, and 4 inches. Figure 18 shows the curvilinear relationships between the annual rainfall and effective rainfall for all the five stations combined. Multiple regression analyses were carried out to the fourth power and a few selected values are given in Table 2. It is noted that the effective rainfall increases with the storage capacity only slightly, espe- 90 80 UJ cr 70 (T LU CL U 60 < O CO -I — J 1 1 I 1 I _L l 10 20 30 40 50 60 70 80 90 WATER DEFICIT (INCHES) IN 24 MONTHS FIG. 16. Relationships between cane yield and water deficit for six fields at Kohala. 394 ANNUAL RAINFALL (INCHES) Fig. 17. Hypothetical curves showing the relation- ship between cane yield and climate in Hawaii. dally when the annual rainfall is low. The Ko- hala results could be applied to other areas for estimating effective rainfall, as the rainfall characteristics are more or less similar through- out the islands. JUICE QUALITY The dry matter production of a sugar-cane crop can be related in a quantitative manner to the solar radiation and the amount of effective water applied, if cultural practices remain the same. The problem of juice quality is more complicated, however, and at present can be dealt with only in a rough empirical manner. The weather factors that favor high sucrose content are: PACIFIC SCIENCE, Vol. XVII, October 1963 1 ) High radiation favors photosynthesis and formation of sucrose. 2) Water supply should be adequate during the growing season, but should be followed by a dry period of about two months before har- vest. High rainfall during the ripening period will dilute the juice and cause a part of the sucrose to break down into other sugars. 3) It is generally known that cool weather preceding the harvest season improves the su- crose content of the cane. Less well known, however, is the beneficial effect of large diurnal temperature range. Harrington (1923) and Morinaga (1926) have shown that alternating high and low temperatures hasten the germina- tion of seeds and bulbs. If the buds of sugar cane react as favorably towards a great range of temperature, then earlier suckering could be ex- pected, which would mean a greater number of mature stalks at harvest and hence a higher sucrose content. In summary, the ideal weather for high juice quality would be abundant sunshine, high tem- perature range, adequate water supply during the growing season, and dry and cool weather during the ripening period. A study of Figures 7 and 8 would indicate that the Ewa, HC&S, Pioneer Mill, and Kekaha plantations have the best climate for good juice quality. It is true that we cannot modify the climate to such an extent as to appreciably improve the juice quality. Nevertheless, an understanding of 0 10 2030 40 50 60 70 80 90 100 WATER DEFICIT (INCHES) IN 2 YEARS Fig. 18. Relationships between annual rainfall and effective rainfall in Kohala, assuming varying soil- moisture storage capacities of 2, 3, and 4 inches. Climatology and Sugar Cane- -CHANG 395 TABLE 2 Effective Rainfall as a Function of Annual Rainfall at Kohala (Assuming varying soil moisture storage capacities of 2, 3, and 4 inches) Annual rainfall (inches) Effective rainfall for: 40 50 60 70 80 90 100 2 -inch storage 34 - 40 46 52 56 60 61 3 -inch storage 37 43 49 55 58 62 65 4~inch storage 37 44 50 56 60 65 69 the climatic effect on juice quality has at least two practical applications. First, we can schedule planting and harvest dates accordingly, in order to best fit the seasonality of a climate; in this connection a study of climatic singularity would be valuable. Second, an understanding of the role played by climate could help us to interpret the results from many agronomic experiments. CONCLUDING REMARKS ON CLIMATOLOGICAL RESEARCH This review has demonstrated the broad scope of climatological research in the Hawaiian sugar industry. The application of this research, as Curry (1952) pointed out 10 years ago, could ultimately give new life to many an economic activity. Toward that end we have made only a beginning. Much work remains to be done. Future research should be planned in three general areas: L GENETIC CLIMATOLOGY: Only 20 years ago climatology was treated primarily as a study of statistical meteorology, replete with records but almost devoid of explanations. The post-war studies of the climate of Hawaii are among the pioneers in the field of genetic climatology. These studies bring a fair measure of systematic order into the otherwise incoherent climatic facts and explain many seemingly local phenom- ena in the light of general circulation. A better understanding of the dynamism and genesis of climate could conceivably lessen or avert agri- cultural hazards in many other parts of the world. It is hoped that Trewartha’s new book (1961) will stimulate interest and hasten de- velopment in this field. 2. ENERGY BUDGET AND WATER BALANCE: Genetic climatology focuses its attention on the free atmosphere. Local variations, caused by the interaction of the atmosphere and the ter- rain, can best be understood by a study of the energy budget and water balance. Conven- tional observations, tied to the needs of synoptic forecasting, are often inadequate for the study of topoclimate. Urgently needed is an inter - nationally-standard field instrument for the measurement of solar radiation. The recent de- velopment of an instrument which measures net radiation (Soumi et al, 1954; Fritschen, 1960£), or even net radiation minus soil heat flow (Portman, 1954), is most encouraging. Measurements of evapotranspiration by in- expensive field instruments are also needed on a wider scale. In humid climates evaporimeters, such as evaporation pans or atmometers (Halkais et al, 1955), have been of value. In arid climates, however, the problem is vexing. There the very concept of potential evapotranspiration is elu- sive and unrealistic. For potential evapotranspi- ration requires, by definition, a homogeneous soil moisture regime infinite in horizontal extent, and once the area upwind is adequately watered the climate is no longer arid. In actuality the effect of advective heat exists even in a very large irrigated area. Gal’tsov (1953), for in- stance, has observed decreasing water require- ments from the border towards the center of an irrigated region in Kazakhstan. It is difficult to evaluate the effect of advective heat. Perhaps an approach similar to but more refined than Be- lasco’s study ( 1952 ) on the modification of air mass is in order. To solve the water-balance equation the soil moisture storage capacity needs to be known. This requires the cooperation of soil scientists. On a global scale, representative figures of moisture storage capacity of the major geo- graphical regions are useful for many purposes. 396 PACIFIC SCIENCE, Vol. XVII, October 1963 In view of the importance of energy budget and water balance in characterizing the regional climate, it is a pity that almost all the textbooks in English on climatology make no reference to the subjects. The Russian work by Budyko (1958) is the only general reference. In this respect, Thornthwaite’s sustained interest in these areas is commendable, and his recent plea ( 1961 ) should be well heeded. 3. EFFECT OF CLIMATE ON GROWTH AND YIELD: We have demonstrated that determina- tion of water balance is a step toward solving the climate-and-plant relationship. More field experiments are needed, however, to define the hypothetical curve as presented in Figure 17. Eventually we hope that a method will be de- veloped to estimate dry-matter yield and juice quality from climatic data. It has been argued that climatologists are not responsible for the study of the effects of climate on vegetation (Trewartha, 1957). To be sure, plant scientists are better qualified to run the field experiments. But the development of tech- niques for evaluating the climatic elements that affect plant growth falls squarely within the responsibility of a climatologist. Without these techniques the meaning of climatic data is often obscured. Whoever solves the problem of cli- mate-and-plant relationship will contribute work of enormous practical value. REFERENCES Baer, L. 1956. Orographic rainfall distribution with application to Hawaii. Trans. Amer. Geophys. Un. 37:546-547. Baver, L. D. 1954. The meteorological ap- proach to irrigation control. Hawaiian Plant. Rec. 54:291-298. Belasco, J. E. 1952. Characteristics of air masses over the British Isles. Geophys. Mem. 11: 1-34. Biel, E. R. 1929. Die veranderlichkeit der jah- reeummon des niederschlage auf der Erde. Geogr. Jber. Ost. 14:46. Blanchard, D. C. 1953. Raindrop size-distri- bution in Hawaii rains. J. Met. 10:457-473. Borden, R. J. 1940. Nitrogen-potash-sunlight relationships. Hawaiian Plant. Rec. 44:237- 241. 1949. Sunlight and sugar yield. Ha- waiian Plant. Rec. 53:43-45. Budyko, M. I. 1958. The Heat Balance of the Earth’s Surface. U.S. Weather Bureau. 259 pp. Chang, J. H. 1958. Ground Temperature. Har- vard University. 300 pp. 1961. Microclimate of sugar cane. Ha- waiian Plant. Rec. 56:195-225. Clements, H. F. 1940. Integration of climatic and physiologic factors with reference to the production of sugar cane. Hawaiian Plant. Rec. 44:201-233. Curry, L. 1952. Climate and economic life: A new approach. Geogr. Rev. 42:367-383. Das, U. K. 1928. The influence of weather on the production of sugar in a typical unirri- gated plantation of Hawaii. Hawaiian Plant. Rec. 32:79-107. 1931^. The problem of juice quality. Hawaiian Plant. Rec. 35:163-201. 1931 b. Weather and the quality of juice at Ewa. Hawaiian Plant. Rec. 35:135-162. 1932. Weather and crop relationships at the Honokaa Sugar Company. Hawaiian Plant. Rec. 36:255-271. 1936. A suggested scheme of irrigation control using the day-degree system. Hawaiian Plant. Rec. 40:109-111. Deacon, E. L., C, B. Priestley, and W. C. Swin- BANK. 1958. Evaporation and water balance. Climatology, UNESCO, pp. 9-34. Denmead, O. T. 1961. Availability of soil water to plants. Iowa University Ph.D. thesis. Frazier, H. M. 1957. The weather and circula- tion of October, 1957. Mon. Weath. Rev. 85:341-349. FRITSCHEN, L. J. 1960^. Transpiration and evap- otranspiration as related to meteorological factors. Iowa State University Ph.D. thesis. I960 A Construction and calibration of the thermo-transducer type net radiometer. Bull. Amer. Met. Soc. 41:180-183. Gal’tsov, A. P. 1953. O Klimatisheskom vzaimodeiztvii oroshaemykh i neroshaemykh ploshchadei. Akad. Nauk SSSR, Ser. Geogr. 3:11-20. Halkais, N. A., F. J. Veihmeyer, and A. H. Hendrickson. 1955. Determining water needs for crops from climatic data. Hilgardia 24:207-233. Climatology and Sugar Cane — CHANG 397 Harrington, G. T. 1923. Use of alternating temperatures upon the germination of seeds. J. Agric. Res. 23:295-332. Houghton, H. G. 1954. On the annual heat balance of the northern hemisphere. J. Met. 11:1-9. Jones, S. B. 1942. Lags and ranges of tempera- ture in Hawaii. Ann. Assoc. Amer. Geogr. 31: 68-97. Krogman, K. K., and L. E. Lutwick. 1961. Consumptive use of water by forage crops in the upper Kootenay River Valley. Canad. J. Sci. 41:1-4. Landsberg, H. 1951. Statistical investigations into the climatology of rainfall on Oahu. Met. Monogr. 1:7-23. Leopold, L. B. 1948. Diurnal weather patterns on Oahu and Lanai, Hawaii. Pacif. Sci. 2: 81-95. — 1949. The interaction of trade wind and sea breeze, Hawaii. J. Met. 6:312-320. and C. K. Stidd. 1949. A review of con- cepts in Hawaiian climatology. Pacif. Sci. 3: 215-225. Loveridge, E. H. 1924. Diurnal variation of pre- cipitation at Honolulu, Hawaii. Mon. Weath. Rev. 52:584-585. Morinaga, T. 1926. Effect of alternating tem- perature- upon the germination of seeds. Amer. J. Bot. 13:141-166. Namias, J., and W. A. Mordy. 1952. The Feb- ruary minimum in Hawaiian rainfall as a manifestation of the primary index-cycle of the general circulation. J. Met. 9: 180-186. Penman, H. L. 1948. Natural evaporation from open water, bare soil and grass. Proc. Roy. Soc Ser. A. 193:120-145. PORTMAN, D. J. 1954. Direct recording of the difference between the radiation and conduc- tion components of the energy balance of the earth’s surface. Bull. Amer. Met. Soc 35:4. Ramage, C. S. 1959. Hurricane development. J. Met. 16:227-237. Schneider, M. 1952. Sumen, mittel und mitt- lere extreme der temperatur von phanologis- chen zeitspannen. Ber. dtsch. Wetter. 7:276- 281. Simpson, R. H. 1952. Evaluation of the kona storm, a subtropical cyclone. J. Met. 9:24-35. Soumi, V. E., M. Franssila, and N. F. Is- litizer. 1954. An improved net radiation instrument. J. Met. 12:276-282. Stidd, C. K., and L. B. Leopold, 1951. The geographical distribution of average monthly rainfall, Hawaii. Met. Monogr. 1:24-33. SUTTON, O. G. 1953. Micrometeorology. Mc- Graw-Hill Book Co. 240 pp. Suzuki, S., and H. Fukuda. 1958. A method of calculating potential evapotranspiration from pan evaporation data. J. Agric. Met. 13: 81-85. Swezey, J. A. 1942. Rainfall evaluation as an aid to irrigation interval control. Hawaiian Plant. Rec. 46:75-100. Tanner, C. B., and W. L. Pelton. I960. Po- tential evapotranspiration estimates by the ap- proximate energy budget method of Penman. J. Geophys. Res. 65:3407. Thornth WAITE, C. W. 1948. An approach toward a rational classification of climate. Geogr. Rev. 38:55-94. 1961. The task ahead. Ann. Assoc. Amer. Geogr. 51:345-356. Trewartha, G. T. 1957. A reply to what one geographer wants from climatology. Prof. Geogr. 9:8-9. I96I. The Earth’s Problem Climates. University of Wisconsin Press. 334 pp. Veihmeyer, F. J., and A. H. Hendrickson. 1955. Does transpiration decrease as the soil moisture decreases? Trans. Amer. Geophys. Un. 36:425-448. Wang, J. Y. I960. A critique of heat unit ap- proach to plant response. Ecology 41:785- 790. Went, F. W. 1957. Climate and agriculture. Sci. Amer. 196:83-94. Wentworth, C. K. 1949. Directional shift of trade winds at Honolulu. Pacif. Sci. 3:86-88. Williams, W. L. S. 1933. Weather charts for plantation use. Hawaiian Plant. Rec. 37:97- 106. Yeh, T. C, C. C. Wallen, and J. E. Carson. 1951a. A study of rainfall over Oahu. Met. Monogr. 1:34-46. J. E. Carson, and J. J. Marciano. 1951 b. On the relation between the circum- polar westerly current and rainfall over the Hawaiian Islands. Met. Monogr. 1:47-55. Root Development in Aluminous Hawaiian Soils1 D. L. Plucknett,2 J. C Moomaw,3 and C. H. Lamoureux4 ABSTRACT: Roots of Rhodomyrtus tomentosa and Melastoma malabathricum were excavated in three soil series from the bauxitic area of Kauai. Root systems of R. tomentosa and M. malabathricum in Kapaa and Halii soils were very shal- low, with tap roots turning laterally at shallow depth and with long lateral roots very close to the soil surface. Deeper tap-root penetration of R. tomentosa and M. malabathricum was observed in the Koolau soil. Lime and phosphorus treatments were added to bauxitic subsoils of the Kapaa and Halii series in pots and Leucaena glauca ( L. ) was planted in the pots. Tap roots of L. glauca were stimulated by phosphorus treatment, but were restricted in untreated subsoils. Increased root development with phosphorus treatment seemed to be more related to phosphorus supply than to decreased aluminum effects. No evidence of root damage due to aluminum was found. L. glauca roots were sectioned with a freezing microtome and stained, using hematoxylin without a mordant. Although all staining obtained could not be at- tributed to aluminum, since other metals can act as mordants for hematoxylin, intensity of staining was assumed to be related to aluminum concentration in the tissues. Cell walls, nuclei, and cytoplasm stained in all tissues, and outer walls of epidermal cells stained very heavily. Staining was more intense in roots from check and P-treated plants than in roots from lime-treated plants. Studies of the root development of natural plant communities (Weaver 1920, Weaver and Albertson 1943, Weaver and Darland, 1949) and of cultivated plants (Weaver, 1926; Troughton, 1957; Crider, 1955) under a vari- ety of conditions and treatments, are well known to ecologists. Normally, a well-devel- oped root system is essential to the vigor- 1 Published with the approval of the Director of the Hawaii Agricultural Experiment Station, Univer- sity of Hawaii, as Technical Paper 583. Manuscript received June 8, 1962. 3 Department of Agronomy and Soil Science, Uni- versity of Hawaii. 3 Department of Agronomy and Soil Science, Uni- versity of Hawaii. Present address: The Interna- tional Rice Research Institute, Manila Hotel, Manila, Philippines. 4 Department of Botany, University of Hawaii. ous growth and successful competition of the dominants in most plant associations. Shallow- rooted plants, however, are known from many different habitats, especially those with poor physical properties but also from those with a low nutritional status with respect to nitrogen, phosphorus, calcium (Fox, Weaver, and Lipps, 1953) and potassium (Haynes, 1943). Root damage and restricted root development in such crop plants as barley and tobacco have been associated with high levels of soil aluminum, but little is known of root development in un- cultivated plants growing in soils with high concentrations of active aluminum. In a study of vegetation on gibbsitic Ha- waiian soils Moomaw and Takahashi (I960) re- ported shallow root systems, but detailed studies of root development and distribution in these 398 Root Development in Aluminous Soils— PLUCKNETT, Moomaw, and Lamoureux 399 soils were not made at that time. Two particu- larly aggressive plant species, Rhodomyrtus tomentosa (Ait.) Hassk. and Melastoma mala- bathricum L., are known to grow well on Ha- waiian bauxitic soils, where Rhodomyrtus tends to form dense thickets excluding all other vegetation. These two species were introduced on Kauai about 50 years ago and Rhodomyrtus has now been declared a noxious weed. It is the object of this study to characterize root growth and distribution of these and other tap-rooted species on the bauxitic soils of Kauai. Aluminum has been shown to be toxic to plants. One of the frequently-reported toxic ef- fects of aluminum is root injury (Lignon and Pierre, 1932; McLean et al., 1926; Bortner, 1935) in which roots may be brown in color with few rootlets and discolored root tips (Gil- bert and Pember, 1931) or with root tips blackened and thickened to twice normal size (Bortner, 1935). Restricted lateral root de- velopment in rye has been reported using water cultures (Magistad, 1925). Trenel and Alten (1934) concluded that aluminum may be a root poison. Using a divided-root technique corn plants were exposed to nutrient solutions with and without aluminum. Injury was restricted to roots in the high-aluminum solutions. Nagata (1954) found that over 5 ppm aluminum in culture solutions hindered barley growth and that aluminum seemed to accumulate in the roots. Growth hindrances were decreased by adding phosphorus or calcium. He concluded that translocation of phosphorus from the root to the top in the barley plant was hindered by aluminum in culture solutions. The precipitation of phosphate and alumi- num in the plant as an aluminum phosphate has been suggested. Burgess and Pember (1923) proposed that aluminum was fixed as relatively insoluble aluminum phosphate in plants, espe- cially in roots. McGeorge ( 1925 ) suggested that internal precipitation of aluminum by phos- phorus may be important in plants but listed no specific location. Wright ( 1937 ) divided root systems of bar- ley plants, placing each half in different culture solutions with and without aluminum. Plant analysis indicated plant damage resulting from poorly developed root systems in solutions con- taining aluminum, and internal precipitation of phosphorus and aluminum where large amounts of aluminum and phosphorus were present in roots. Wright (1943) found a higher percent- age of phosphorus in aluminum-treated barley plants than in nontreated; this was particularly marked in the roots. The water-soluble phos- phorus in the aluminum-treated plants was low, while a H2SO4 solution (pH 3.0) extracted practically all P from untreated plants but much smaller amounts from plants grown in contact with Al. The precipitation was listed as occur- ring primarily in roots, and sharp reductions in yield were attributed to P deficiency in meri- stematic regions due to root precipitates. Wright ( 1945 ) , using microchemical tests to determine inorganically and organically bound P, found abundant inorganic P in roots grown in contact with Al and little or none in roots from solu- tions without Al. Problems of plant growth on acid soils have long been ascribed to the "active” Al in the soil and to problems of phosphate nutrition due to fixation of phosphates by Al. Longnecker and Merkle (1952) studied root development of crimson clover in relation to lime placement and found most root growth in layers which had been limed. The beneficial effect of liming was attributed to decrease in solubility of Al and Mn and an increase in solubility of P. Rag- land and Coleman (1959) applied lime at sev- eral rates to subsoils of the Norfolk catena in pots and found grain sorghum root growth into unlimed subsoils was related inversely to amounts of exchangeable Al. Root growth into subsoils increased substantially with lime treat- ment. Root development of sorghum grown in suspensions of acid clay was restricted severely unless 80% of the acidity was neutralized. DESCRIPTION OF EAST KAUAI The area studied is referred to by McDonald et al. (I960) as the Lihue Depression. It is a nearly circular basin with the rim being formed by the Haupu ridge on the south, the main mountain mass of central Kauai on the west, the Makaleha mountains on the north, and Nonou and Kalepa ridges on the east. The basin is floored with lavas of the posterosional Koloa 400 volcanic series. Two vents from the Koloa vol- canic series, Hanahanapuni Crater and Kilohana Crater, lie within the basin. The general topography of the basin is of gently sloping to moderately steep ridges and plains dissected by perennial streams, notably the Wailua River and its tributaries. The average annual rainfall in the Lihue De- pression ranges from 40 or 50 inches near the ocean to over 170 inches near the mountains. Rainfall is usually highest in winter months but there are no months during which no rain falls. Mean monthly temperatures from nine sta- tions below 300 ft elevation on Kauai range from 69 F in February and March to about 77 F during August through October (McDonald et al., I960). Although no temperature records are available for higher elevations, there is a decrease in temperature with increase in eleva- tion, of about 3 F for each 1,000 ft. Prevailing winds are the northeast trade winds but cyclonic storms occasionally upset this pattern, especially in winter months. DESCRIPTION OF SOILS STUDIED Three soils occurring in the Lihue Depression were selected for study. These soils are located in or near the main area of infestation of R. tomentosa and Al. malabathricum and either comprise or are associated with the major baux- itic soils of East Kauai. The principal mineral form of Al present in these bauxitic soils is gibbsite, the trihydrate of aluminum oxide (Sherman, 1958). Detailed soil descriptions have been made by the Soil Conservation Service (Womack, I960). Kapaa Series The Kapaa series is a deep, well-drained, Aluminous Ferruginous Latosol developed on gently sloping to steep uplands on Kauai. These soils occur mainly in association with the Halii soils which lie above, and Puhi soils which lie below. The Halii series is developed from parent material similar to that of the Kapaa series, namely the melilite and nepheline basalts of the Koloa volcanic series. The Kapaa soils occur be- tween 200 and 1,000 ft elevation with mean annual rainfall from 60 to 100 inches. They are PACIFIC SCIENCE, Vol. XVII, October 1963 clayey in texture but feel like silty clay in the A horizon (0-6 inches) owing to the strong, very fine granular structure. They are extremely hard when dry but sticky and plastic when moist, containing many roots and few pebbles in the surface layers. With depth, some mottling occurs on the blocky structures and pebbles are more numerous. These soils are used mainly for pasture, nonirrigated sugar cane, and pineapple. Halii Series The Halii series is a deep, well-drained Alu- minous Ferruginous Latosol on gently sloping to moderately steep uplands on Kauai. This series is associated with the Koolau series at higher elevations and with Kapaa series below. It occurs in belts at about 300 to 1,000 ft eleva- tion with mean annual rainfall from 80 to 120 inches. The A horizon (0-9 inches) of this series is a gravelly clay that feels like silty clay and is grayish-brown in contrast to the yellow- ish-brown of the Kapaa series. It is somewhat less hard when dry and more plastic when wet than the Kapaa. It is used principally for non- irrigated sugar cane and small acreages of pine- apple. Koolau Series The Koolau series is a deep, poorly-drained Hydrol Humic Latosol developed on gently slop- ing to moderately steep uplands. It is associated with the Halii soils, and occurs between 400 and 4,000 ft elevation with mean annual rain- fall of 120 and 200 inches. Most of the Koolau series is covered by rain forest, but some is used for sugar cane and pasture. The Koolau series is a grayish-brown clay that feels like a silty clay loam, with a structure that appears to be massive to very weak medium granular. It is only slightly sticky and plastic when moist, with matted roots and many worm holes and casts making it quite porous. Lower horizons have distinct strong brown mottles. EXPERIMENTAL PROCEDURE Roots of Melastoma malabathricum and Rho- domyrtus tomentosa were excavated in the Halii, Koolau, and Kapaa soil series and de- scriptions and measurements were made of Root Development in Aluminous Soils — Plucknett, Moomaw, and Lamoureux 401 depth of penetration of tap roots, location and length of lateral roots, relation of roots to soil profile, and possible evidence of causes of thicket formation. A few plants of Norfolk Island pine ( Araucaria excelsa) (Lamb.) R. Bn. were also excavated in the Wailua Game Refuge. Root description and measurement were also made of R. tomentosa seedlings used in a pot experiment designed to measure plant and soil Al. Notes taken included color, thickening, black tips, number of lateral roots, length of tap roots, and number of active buds. Because tap roots were observed turning laterally in Halii and Kapaa soils in the field, a pot experiment was established using the soil layer in which these roots turned as "subsoils.’’ These soils were collected from horizons where tap roots were observed to turn laterally and were sacked carefully to prevent dehydration. They were screened through wire mesh con- taining approximately 4 meshes to the inch. Weighed samples of the screened soil were used to form the "bottom” 5 -inch soil layer in 11- inch plastic pots, and were treated with six lime and phosphate treatments. After the treatments were mixed thoroughly in the subsoils, a 5 -inch layer of untreated Kapaa surface soil was added. Leucaena glauca 5 seeds were planted in the sur- face soil and after germination plants were thinned to two per pot. At harvest the soils were carefully removed from the pots and washed from the roots. Meas- urements of tap-root penetration and lateral root development were made. Roots were ex- amined for blunted and blackened tips, and root tips from each treatment were preserved for staining studies. Yields of tops and roots were recorded and plant Al concentrations were de- termined. The pH, extractable Al, exchangeable calcium, and cation exchange capacity were de- termined for each soil. Root tips of L, glauca plants from treated pots were sectioned on a freezing microtome and stained, using hematoxylin without a mordant (Johansen, 1940). Slides of the root sections were made and photomicrographs were taken. 6 In a personal communication Dr. F. R. Fosberg has indicated the correct name for this plant should be Leucaena leucocepbala. Fig. 1. Rhodomyrtus tomentosa excavated in the Kapaa soil series, Wailua Game Refuge. The tap root turned laterally at 4-inch depth and lateral roots pene- trated diagonally before ascending toward the surface. RESULTS Six R. tomentosa plants were excavated in the Kapaa soil series. All plants were extremely shallow-rooted, with tap roots turning laterally at depths of from 3 to 10 inches below the sur- face (Fig. 1 ) . Lateral root development irnrhese plants was especially pronounced and one large 10-ft shrub had a lateral root 24 ft long. Lateral roots displayed a tendency to grow downward and outward for 2 to 4 inches and then to as- cend toward the surface. Lateral roots were frequently found just at, or slightly under, the soil surface. Two thickets of R. tomentosa were excavated in the Kapaa soil series. These thickets con- tained shrubs up to 10 ft in height with trunks 1.5 to 2.25 inches in diameter. Roots were ob- served with diameters up to 2 inches. Tap roots turned at a depth of 10 inches and no roots of the thickets were found below this depth. There was no evidence of root fusion. M. malabathricum plants excavated in the Kapaa soil were found to have root development similar to that in R. tomentosa, with tap roots turning laterally at shallow depth and long lat- eral roots. One plant (Fig. 2) had a small, twisted and deformed tap root with 2 main lateral roots 5 ft long. Fibrous roots were almost lacking in both R. tomentosa and M. mala- bathricum. Norfolk Island pine trees (A. excelsa) planted about 10 years ago in the Kapaa soils were 402 excavated to determine root development in species planted in these soils. Tap roots of these trees penetrated to what appeared to be the bottom of the planting hole before turning up- ward toward the surface. Two individual plants and a thicket growth of R. foment os a were excavated in the Koolau soil near Hanahanapuni Crater. Tap roots of the two individual plants penetrated 9 inches downward before turning diagonally for 1 or 2 inches. Small plants of M. malabathricum nearby, like the R. tomentosa, had tap roots which penetrated from 8 to 10 inches. The main tap root of the R. tomentosa thicket penetrated 24 inches downward in the Koolau soil without turning, even though water from the soil filled the hole at a depth of 1 ft, placing it well below the water table at the time of sampling. One lateral root of this thicket was observed from which numerous stems had arisen. This was the only observation of this type in the plants excavated. It is possible that this "root” could have been a stem buried by road construction since the thicket was located dose to a forest preserve road. The Koolau soil in this area has a grey surface 8 to 10 inches in thickness with a reddish-brown, yellow-mottled layer below. A 5 -ft M. malabathricum plant about 20 ft from the R. tomentosa thicket was also ex- cavated and the tap root was traced to a depth of 18 inches where water rapidly filled the trench. A series of R. tomentosa plants in the Halii Fig. 2. Melastoma malabathricum excavated in the Kapaa soil series, Wailua Game Refuge. Note the shallow lateral roots and the small tap root. PACIFIC SCIENCE, Vol. XVII, October 1963 Fig. 3. Plants of Rhodomyrtus tomentosa from the Halii soil series on Kilohana Crater, Kauai. The lateral root of the larger plant was 11 ft long. soil series were excavated on the northern slopes of Kilohana Crater. These plants had extremely shallow root systems with tap roots turning laterally at about a 4-inch depth and with lateral roots almost at the soil surface (Fig. 3). R. to- mentosa shrubs in this area were easily pulled up without digging, and tracing of lateral roots was accomplished by pulling. Excavation of such roots was difficult because of long overlapping lateral roots of surrounding plants. One 40-inch plant had a lateral root which arose from the tap root about 2 inches below the soil surface and which was traced at depths of 1 inch or less for 11 ft. No evidence of root injury was found in R. tomentosa plants grown in pots except for an unusual blunting of root tips in the Halii soil. This blunting is best described as a curving and thickening of the root tip which then re- sembled a miniature chicken’s head. Some black- ened tips were also found in the Halii soil, but since the concretionary iron-rich surface of the Halii soil was used in these pots it is doubtful that these root abnormalities were due to A1 injury. The only evidence for thicket formation of R. tomentosa in this pot study was the presence of a number of buds on the tap root just below the soil surface. Young shoots were observed arising from these buds and often 6 to 10 young shoots were growing simultaneously on one plant. The number of shoots arising from buds increased with time. Bud numbers of single plants ranged from 5 to 30. Root Development in Aluminous Soils— Plucknett, Moomaw, and LAMOUREUX 403 Root: Development of Leucaena glauca •Root development of Leucaena glauca in treated subsoils of the Halii and Kapaa soil series was investigated separately in pots. In both the Kapaa and Halii subsoils root develop- ment was usually restricted to the untreated top- soil if subsoils were untreated. The greatest root development in subsoils was produced by P treatments. Treatments in which lime was added in addition to P stimulated root growth more than lime alone. The addition of 1,000 lb of ele- mental P without lime produced the most root penetration into the subsoils and also the highest plant yields. Comparative root systems produced by treat- ments in the Kapaa subsoil are illustrated in Figure 4. Tap roots of L. glauca in the P treat- ments were straight and penetrated to the bot- tom of the pot. Tap roots of check plants did not develop in the untreated subsoils. The effect of treatment on tap-root penetration and root de- velopment in treated Kapaa subsoil ranked as follows: P > lime plus P > lime > check. Figure 5 shows comparative root systems in Halii subsoils. Roots of check plants in the Halii subsoils penetrated slightly into the sub- soil, but total root development in the check was much less than total development in subsoils treated with P and with lime plus P. Two types of tap-root development with treatment were observed: A long straight tap root was characteristic of P treatment, while a Fig. 4. Plants of Leucaena glauca grown in treated subsoils of the Kapaa series. Treatments from left to right are: 500 lb P, 1,000 lb P. 5 tons lime, 5 tons lime plus 1,000 lb P, and check. Fig. 5. Plants of Leucaena glauca grown in treated subsoils of the Halii series. Treatments from left to right are: check, 5 tons lime plus 1,000 lb P, 5 tons lime, and 1,000 lb P. branching tap root was characteristic of the lime plus P treatment. An interesting result of treatments added to Halii subsoils was the number of nodules pro- duced on roots of L. glauca. Most nodulation occurred with lime plus P, but lime alone also stimulated nodule formation. Only one nodule was found with P treatment, and no nodules were found in the check. Staining of Leucaena glauca Root Tips Hematoxylin staining of cell walls, nuclei, and cytoplasm was evident (Fig. 6d) . Outer walls of epidermal cells were especially heavily stained ( Fig. 6a, b) , which may indicate a precipitation of aluminum in this region. In P- treated plants two darkly stained areas at the periphery of the stele were observed, which may also represent areas of aluminum precipitation. One of these areas is shown in Figure 6c. The cells of the stele appeared to stain more intensely than those of the cortex (Fig. 6a, b), but the stelar cells are less highly vacuolated and thus contain more stainable material per unit volume than the cortical cells. Staining was more intense in roots from the check (Fig. 6a) than in roots from lime-treated soils (Fig. 6b). Phosphate-treated roots stained more heavily than Jime-treated roots. Roots treated with a combination of 5 tons lime plus 1,000 lb P stained more deeply than those treated with lime alone, but in most cases 404 PACIFIC SCIENCE, Vol. XVII, October 1963 stained less deeply than those treated with P alone. Staining of nuclei was especially pro- nounced in the 1,000 lb P treatment. The results obtained in this study differ from those reported by McLean and Gilbert (1927) for corn and cabbage, and by Wright and Donahue (1953) for barley, in that cells from stelar regions stained readily in all our prepara- tions. Although McLean and Gilbert noted stain- ing of nuclei and cytoplasm, this staining was restricted to the epidermis and outer cortex, and none of the stelar cells were stained. Wright and Donahue found that staining occurred from the epidermis to the outer wall of the endodermis, but that there was very little staining in the stelar region. DISCUSSION The importance of the effect of high soil A1 on root growth cannot be minimized, but shal- low root development in the Halii and Kapaa soil series was not interpreted as resulting from "A1 toxicity.” High soil A1 can cause conditions in the soil which may limit root development, however. From previous work, it is known that A1 can interfere with phosphate nutrition of the plant both by precipitation of phosphorus in the soil and possibly by precipitation within the plant. In addition, A1 is thought to con- tribute to the acidity of the soil. If high alu- mina content of these subsoils is important in limiting root development of plants in bauxitic soils, any decrease in extractable A1 and plant A1 Fig. 6. Photomicrographs of Leucaena glauca roots sectioned and stained with hematoxylin. A, Root sec- tion from the untreated (check) subsoil of the Kapaa soil series. B, Section from Kapaa subsoil treated with 2.5 tons lime. C, Root from Kapaa subsoil treated with 500 lb P. Note heavily stained area in xylem region. D, Cross-section of root from Kapaa subsoil treated with 5 tons lime plus 1,000 lb P. Note staining of nuclei, cell walls, and cytoplasm. Root Development in Aluminous Soils — Plucknett, Moomaw, and Lamoureux 405 should result in stimulated root growth. An ex- amination of results of liming treatments shows, however, that although pH was increased and extractable A1 and plant A1 concentrations were decreased by liming, no marked stimulation of root development or plant growth of L. glauca occurred with liming. The increased root de- velopment of L. glauca in P- treated Halii and Kapaa subsoils in pots is interpreted more as a response to P than a decrease in A1 effects. In addition, of course, a mass-action effect may be operating in which active aluminum in the soil is being supplied with enough phosphorus to permit complete precipitation as aluminum phosphate with enough remaining to supply the plant with adequate P. Deep tap-root development in the Koolau soils was unexpected because of the extremely poorly-drained condition of this soil. Root growth in the wet Koolau soil in pots also ap- peared normal even though figures for extract- able A1 were high. The lack of root damage in plants used in this study was probably related to the evolutionary background of the plants, which seem to thrive in areas of low fertility and high rainfall. Roots of plants sensitive to Al, like rye or barley, would probably be severely injured in these soils. But Leucaena glauca, like many other tropical plants, does not respond markedly to lime. Thicket formation in R. tomentosa is prob- ably caused by the large number of adventi- tious shoots which arise from buds on the tap root just below the soil surface. Heavy staining of roots with hematoxylin cannot be definitely attributed to Al alone, since iron and other metals may also act as a mordant for hematoxylin. However, since in the present study chemical analysis showed Al to be present in large amounts, much of the staining obtained is attributed to Al. The particularly intense staining of the outer walls of the epidermis, and of two areas in the outer part of the stele, are interpreted as an indication that Al precipita- tion may have occurred in these areas. Our results appear to differ significantly from those of McLean and Gilbert (1927) and of Wright and Donahue (1953). The reasons for these differences are not immediately apparent. In all three studies root sections were stained in hematoxylin without the addition of a mor- dant, and the staining which occurred was inter- preted as an indication that aluminum was already present in the tissues. REFERENCES Bortner, C. E. 1935. Toxicity of manganese to Turkish tobacco in acid Kentucky soils. Soil Sci. 39:15-33. Burgess, P. S., and F. R. Pember. 1923. Active Al as a factor detrimental to crop production in many acid soils. R. I. Agri. Expt. Sta. Bull. 194. 40 pp. Crider, F. J. 1955. Root growth stoppage re- sulting from defoliation of grass. USDA Tech. Bull. 1102. 23 pp. Fox, R. L., E. J. Weaver, and R. C. Lipps. 1953. Influence of certain soil profile charac- teristics upon the distribution of roots of grasses. Agron. J. 45:583-589. Gilbert, B. E., and F. R. Pember. 1931. Fur- ther evidence concerning the toxic action of aluminum in connection with plant growth. Soil Sci. 31:4, 267-273. Haynes, J. L. 1943. Effects of pasture practices on root distribution. J. Amer. Soc. Agron. 35:10-18. Johansen, D. A. 1940. Plant Microtechnique. McGraw-Hill, New York. Lignon, W. S., and W. H. Pierre. 1932. Sol- uble Al studies, II. Soil Sci. 34:307-322. Longnecker, D., and F. G. Merkle. 1952. Influence of placement of lime compounds on root development and soil characteristics. Soil Sci. 73:71-75. Magistad, O. C. 1925. The Al content of the soil solution and its relation to soil reaction and plant growth. Soil Sci. 20:181-225. McDonald, G. A., D. A. Davis, and D. C. Cox. I960. Geology and Groundwater re- sources of the Island of Kauai, Hawaii. Ha- waii Div. Hydrogr. Bull. 13. 406 PACIFIC SCIENCE, Vol. XVII, October 1963 McGeorge, W. T. 1925. The influence of Al, Mn, and Fe salts upon the growth of sugar cane, and their relation to the infertility of acid island soils. Hawaiian Sugar Planters’ Assoc. Expt. Sta. Bull. Agri. Chem. Sec. 49. McLean, F. T., and B. E. Gilbert. 1927. The relative aluminum tolerance of crop plants. Soil Sci. 24(3): 163-174. Moo MAW, J. Q, and M. Takahashi. I960. Vegetation on gibbsitic soils in Hawaii. J. Arnold Arb. 61:4, 391-411. Nagata, T. 1954. Influence of Al-ion concen- tration on the absorption of nutrients by plants. I, II. J. Sci. Soil Manure Japan 24: 255-262. (C A. 48, 7710). Plucknett, Donald L. 1961. Root growth in bauxitic soils. Hawaii Farm Sci. 10(1) :8-9. Ragland, J. L., and N. T. Coleman. 1959- The effect of soil solution aluminum and cal- cium on root growth. Soil Sci. Proc. 23:35 5— 357. Sherman, G. D. 1958. Gibbsite-rich soils of the Hawaiian Islands. Hawaii Agri. Expt. Sta. Bull. 166. TRENEL, M., and F. ALTEN. 1934. Die physi- ologische Bedeutung der mineralischen Bo- denaziditat. Angew. Chem. 46:313-320. Troughton, A. 1957. The underground or- gans of herbage grasses. Commonwealth Bur. Pastures and Field Crops Bull. 44. 163 pp. Weaver, J. E. 1920. Root development in grassland formations. A correlation of root systems of native vegetation and crop plants. Carnegie Inst., Wash. Publ. 292. 151 pp. 1926. Root Development of Field Crops. McGraw-Hill Book Co. Ltd., N. Y. 291 pp. and F. W. Albertson. 1943. Resur- vey of grasses, forbs, and underground plant parts at the end of the great drought. Ecol. Monogr. 13:63-117. and R. W. Darland. 1949. Soil-root relationships of certain native grasses in vari- ous soil types. Ecol. Monogr. 19:303-338. Womack, Durwajrd W. I960. Tentative De- scription Legend of the Soil Survey of the Island of Kauai. May, I960. Soil Conservation Serv., State of Hawaii. (Mimeo.) Wright, K. E. 1937. Effects of phosphate and lime in reducing aluminum toxicity of acid soils. Plant Physiol. 12 ( 1 ): 173-181. — - — — 1943. Internal precipitation of P in relation to Al toxicity. Plant Physiol. 18:708- 712. — 1945. Aluminum toxicity: Microchem- ical tests for inorganically and organically bound phosphorus. Plant Physiol. 20:310- 312. — and B. A. Donahue. 1953. Aluminum toxicity studies with radioactive phosphorus. Plant Physiol. 28:674-680. Studies in the Hawaiian Rutaceae, IV New and Critical Species of Pelea A. Gray1 Benjamin C. Stone2 The following notes, including descriptions of four proposed new species, are the partial outcome of monographic studies of Hawaiian genera. Pelea is a genus confined to the Hawai- ian and Marquesan archipelagoes, with all but two of its species endemic to the Hawaiian Islands. There*are four relatively natural sections of the genus, characterized primarily by features of the mature fruits and to a lesser extent by leaf arrangement. Systematic treatment of sec- tion Pelea , which includes the type species, Pelea clusiaefolia, is completed. The following notes apply to sections Apocarpa, Megacarpa, and Cubicarpa, as defined by the writer (Stone in Degener, 1962 ) . Related studies have recently appeared or are now in press (Stone 1962^, 1962b, Stone in Degener, 1962). Casual collect- ing of species of Pelea was begun by the author in 1955; in 1958 and 1959 and summer, 1961, intensive field and herbarium studies were car- ried out. Through the courtesy of the U. S. Na- tional Herbarium, Smithsonian Institution, loans of historically important collections were ob- tained, and I am grateful to Dr. H. R. Fletcher, of the Royal Botanic Gardens, Edinburgh, Dr. George Taylor of the Royal Botanic Gardens, Kew, Dr. J. E. Dandy of the British Museum (Natural History), Dr. Alicia Lourteig of the Museum d’Histoire Naturelle, Paris, Dr. Richard A. Howard of the Arnold Arboretum of Harvard University, and Dr. A. J. Eames of Cornell Uni- versity for their aid. I am particularly indebted to the former Director of the Bishop Museum, Dr. Alexander Spoehr, Botanist Marie C. Neal, and Curator of Collections E. H. Bryan, Jr., for their generous assistance. The following comments are arranged by ge- neric section. 1 The greater part of this work was accomplished at the University of Hawaii and at the U. S. National Herbarium. Manuscript received February 27, 1962. 2 College of Guam, Agana, Guam. SECTION Apocarpa STONE 1. Pelea ovata St. John & Hume in Lloydia 7: 272, 1944. P. Forbesii St. John & Hume, l.c. Syn. nov. Examination of the holotype specimens shows that the pubescence which is found on Forbes 369.K, the type of P. Forbesii, is not constant, but is found only on one leaf, perhaps from an abnormal cause. It has been noted that leaf-galls of other species, even ordinarily glabrous ones, will often be densely puberulent. Even the new leaves and buds of P. Forbesii are found to be glabrous. Since no other character or combina- tion of characters appears to distinguish this from Pelea ovata, it is concluded that but one species is represented. Since the description of P. ovata calls for glabrous leaves, and since there is no question of priority involved, both descrip- tions first appearing on the same page of the publication, the name P. Forbesii is relegated to synonymy. Pelea ovata is a vinelike shrub endemic to the island of Kauai. 2. Pelea hawaiensis Wawra in Flora 56:110. 1873. P. cinerea var. y Hillebrand, FI. Haw. Ids. 69, 1888; Rock, Indig. Trees Haw. Ids. 239, I913 (in part, excluding Oahu plants). P. cinerea var. hawaiensis (Wawra) Rock in Bot. Gaz. 65:265, 1918. A tree with smooth light-brown bark, op- posite leaves, and generally pubescent innova- tions; the pubescence of fulvous or reddish hairs, rarely pale; petioles, twigs, and leaves sometimes soon glabrate, commonly with a more or less per- sistent indument; inflorescences usually densely puberulent, or tomentose, cymose, with generally 3 to 27 flowers, axillary; flowers with deltoid acute thick tomentellous sepals; petals deltoid- 407 408 PACIFIC SCIENCE, Vol. XVII, October 1963 lanceolate, tomentellous, often reddish within; stamens glabrous; ovary densely tawny or golden tomentellous; style hirsutulous; stigma dark maroon, glabrous; capsule commonly 16-34 mm broad, often subtended by the persistent sepals, the carpels with a dense persistent fulvous or reddish pilosity; endocarp firm, thickly pilose with pale hairs. TYPE: Hawaii: Kawaihae-iuka, in 1862, Wm. Hillebrand (in the Wawra Herbarium, Vienna). A photo shows capsules 27 mm broad. (a) var. hawaiensis The typical variety, which is certainly distinct from Pelea cinerea, as was stated (in herb.) by the late C. N. Forbes, is nonetheless a southeast counterpart of that species. The distribution of the typical variety includes Hawaii and Maui. Other varieties, some of them eminently distinct, others only poorly differentiated, are found on Hawaii, Maui, Lanai, and Molokai. ( b ) var. rubra ( Rock ) B. C. Stone, comb. nov. P. cinerea var. rubra Rock in Bot. Gaz. 65: 264, 1918. P. oblon^ifolia var. /3 Hillebrand, FI. Haw. Ids. 65, 1888. P. cinerea var. 8 Rock, Indig. Trees Haw. Ids. 239, 1913. P. cinerea var. 8 Hillebrand, op. cit. 69. (as to Kau specimen). Hillebrand tells us that this is a glabrous shrub with long rambling branches. The poorly KEY TO VARIETIES OF Pelea hawaiensis Capsules mostly 27-34 mm diameter (rarely smaller); inflorescences often few-flowered (3-9). Leaves puberulent beneath, at least on the costa, at first ( and often persistently ) uniformly tomentellous to pilose; inflorescence compact, with stout axes, peduncles ca. 2 mm thick, densely tomentellous; capsular pubescence fulvous to brownish-orange. Hawaii var. hawaiensis Leaves glabrous, glossy; inflorescences elongate, the axes slender, peduncle to 1 mm thick, glabrous or nearly so; capsular pubescence reddish. Hawaii var. rubra Capsules smaller, mostly 16-22 mm diameter; inflorescences mostly compact and multiflorous with 9-27 flowers, or rarely more open and with fewer flowers. Blades densely shaggy-pilose beneath, short-lanceolate; branchlets puberulent; inflorescences mostly 3-9-flowered. Blades cordate at base; tomentum fulvous, the hairs extremely minute. Molokai var. molokaiana Blades subcuneate to rounded at base; tomentum brownish or olivaceous, the hairs up to 0.7 mm long. Maui, Lanai var. pilosa Blades tomentellous to glabrate beneath, sometimes at last glabrous; inflorescence several- to many-flowered (with up to 27 flowers), or less commonly only 3-7 -flowered. Inflorescences mostly only 3-7-flowered. Blades cordate or emarginate at base, slightly puberulent on the costa dorsally, elon- gate and often acuminate. Lanai... var. sulfurea Blades cuneate at base, glabrate to glabrous beneath, rather broad, ovate. Maui var. Brighamii Inflorescences mostly 9-27-flowered. Blades cordate or subcordate at base, or sometimes merely emarginate; inflorescences about 4-6 cm long. Blades glabrate beneath; petioles mostly 20-45 mm long. Maui var. Remyana Blades puberulent beneath; petioles mostly 7-18 mm long. Maui ..var. racemiflora Blades cuneate or rarely slightly emarginate at base; inflorescences compact, 2-3 cm long; costa puberulent or glabrous dorsally. Hawaii var. Gaudichaudii Critical Species of Pelea—~$ TONE 409 developed capsule of the specimen which he described (with a question mark) as a variety of Pelea ohlongifolia (a species of sect. Mega - carp a) misled him; on closer examination, it can be seen that only one follicle is developed, and it is thus difficult to ascertain whether the fruit is apocarpous or syncarpous. However, the tomentum of both the follicle and the endocarp, as well as the leaf venation, suffice to establish the relationship of this plant with P. hawaiensis , rather than with the similar P. cinerea of Oahu (which has pubescent endocarp but glabrous follicles), or the greatly different P. oblongifolia of Hawaii ( with glabrous endocarp and sparsely puberulent syncarpous capsules). TYPE: Hawaii: North Kona; Huehue, lava- beds, 6 June 1909, Rock 3565 (Bishop). Distribution: Kona and Kau, Hawaii; Olo walu, Maui. SPECIMENS EXAMINED: Maui: Central ridge of Olowalu valley, 12 May 1920, Forbes 2345. M. (Bishop). Hawaii: North Kona, Puuwaawaa, 6 June 1909, Rock 3561 (Bishop). Kau, Hille- brand 251 (Kew), 252 (Kew), without number (Kew). Kau Desert, 2 August 1911, Forbes 385, H. (Bishop). (c) var. pilosa St. John in Lloydia 7:272, 1944. P. cinerea (3 var. Hillebrand, FI. Haw. Ids. 69, 1888 (as to Lanai specimens with pilose blades). This variety is limited to Lanai and Maui. Hillebrand’s description is based mostly on the Lanai specimen, although he cites Mann & Brig- ham 371 from Makawao, Maui. This is a mis- print for 377; Hillebrand wrote 377 on the label of his own collection as a note citing the Mann and Brigham specimen for comparison; but their specimen is a different variety (var. Brig- hamii) . Rock erred in calling his Maui speci- mens the same, and, after coining the name sulfurea, applied it to both the Hillebrand and the Mann and Brigham specimens; var. sulfurea is restricted to Lanai. Unfortunately, the type of var. pilosa was destroyed at Berlin during World War II. SPECIMENS EXAMINED: Maui: South slope of Haleakala, March 1920, Forbes s.n. (Bishop); Auahi (Auwahi), March 1920, Forbes 2099 .M 2098M., 2097 M., 2115 M., and s.n. (Bishop). (d) var. molokaiana B. C. Stone, var. nov. Folia base cordata, infra fulvo-pilosa. HOLOTYPE: Molokai: Ridge below Puu Ko- lekole, July 1912, Forbes 126 Mo. (Bishop; iso- type at US ) . DISTRIBUTION: Restricted to Molokai. The innovations are densely tawny-hirsutu- lous; blades lance-ovate, rounded or acute at apex, subcordate to cordate at base, up to 8 X 3.5 cm, with petioles densely hirsutulous, as are the lower surfaces of the blades. (e) var. sulfurea (Rock) B. C. Stone, comb, nov. P. cinerea var. sulfurea Rock in Bot. Gaz. 65 : 265, 1918. P. sulfurea (Rock) St. John & Hume in Lloydia 7:274, 1944. TYPE: Lanai: Without locality, July 1870, Hillebrand (Kew; isotype, Bishop). DISTRIBUTION ; Restricted to Lanai. Abundant collections are in the Bishop Museum Herba- rium. (f) var. Brighamii (St. John) B. C. Stone, comb. nov. P. Brighami St. John in Lloydia 7:271, 1944 (sub sect. Cubicarpae) . TYPE: Maui: Makawao, Mann & Brigham 377 (Bishop, GH). DISTRIBUTION: Maui. Described as a species but, no doubt through a typographic error, misplaced under section Cubicarpae . This is the variety that Rock had in mind when he discussed var. sulfurea (in Bot. Gaz. 65:265, 1918), although he did not typify that variety, but left Hillebrand s original intent as applying to Lanai plants. A number of specimens have been seen, mostly from Makawao, Olinda, and Olowalu. (g) var. Remyana B. C. Stone, var. nov. Arbor, innovationibus dense fulvo hirtellis; foiiis maturis supra glabris infra glabris costa media sparse puberulo vel glabrato exceptis, laminis 5-16 cm longis, 2-6.5 cm lads, suban- guste ellipticis, apice rotundatis vel emarginatis, base subcordatis vel cordatis ( vel emarginatis ) ; inflorescentiis 15-21 (-27-) floriferis; capsulis 410 PACIFIC SCIENCE, Vol. XVII, October 1963 dense fulvo-puberulentibus, c. 20 mm diametro; endocarpio hirtello. HOLOTYPE: Hawaii: Without definite locality, Jules Remy 626 (Paris). DISTRIBUTION: Known only from Hawaii. (h) var. racemiflora (Rock) St. John in Lloydia 7:272, 1944. P. cinerea var. racemiflora Rock, Indig. Trees Haw. Ids. 241, 1913. type: Maui: Auahi, on aa lava, at about 1300 feet alt., November 1910, Rock 8676 (Bishop). DISTRIBUTION: Maui. SPECIMEN EXAMINED: Maui: Pakiloi forest, March 1920, Forbes 2082.M. (Bishop). (i) var. Gaudichaudii (St. John) B. C. Stone in Pac. Sci. 16:369-371, 1962. P. Gaudichaudii St. John in Lloydia 7:272, 1944. brunelia sandivicensis Gaud. nom. nud. in Bot. Freycinet Voy. in obs. 1826. TYPE: 'Sandwich Islands,’ Gaudichaud (Paris; fragment at Geneva ) . DISTRIBUTION: Hawaii, especially around Ki- lauea and vicinity. 3. Pglea rnakahae B. C. Stone, sp. nov. Arbor; innovationibus minute puberulenti- bus; foliis maturis infra dense cinereis, laminis olivaceis, trichomiis ad 0.1 mm longis, cum squamulis interspersis; petiolis 10-22 mm longis, 1-2 mm crassis, glabrescentibus; laminis anguste ovatis vel ellipticis, rare late lanceolatis, coriaceis, ca. 3.5-12 cm longis, 1.5-5 cm latis, base cuneatis, apice subacutis; inflorescentiis axillaribus, cymosis unifloris (vel trifloris? ) , axibus cinereis, ad 12 mm longis, binodosis; floris ignotis; sepalis caducis; capsulis 20-24 mm diametro lobis discretis follicularibus glabris punctatis, apice minute mucronulatis; endocar- pio glabro. HOLOTYPE: Oahu: Waianae Mountains, Ma- kaha Valley, near top of ridge, at about 4000 ft. alt., 21 October 1954, Gordon A . Pearsall s.n. ( Bishop ) . A characteristic species closely allied to and simulating P. cinerea, but with glabrous endo- carp; similar also to P. cinereops but with glab- rous ovary. The disk may however be slightly puberulent. The cinereous indument is extremely fine, composed of waxy scales intermingled with hairs less than 0.1 mm long, obscuring the densely glandular-punctate surface of the lamina. The indument tends to persist on the younger peti- oles and stems, also. The leaf margins are usually very slightly inrolled; the lateral nerves are ascending, parallel, and united by a sinuate marginal nerve 1-8 mm or more in from the edge. SECTION Mega carp a stone 4. Pelea Balloui Rock, Indig. Trees Haw. Ids. 228, 1913. P. ukuleleensis St. John (as to flowering branch of type sheet), in Lloydia 7:267, 1944. Syn. nov. The type specimen of P. ukuleleensis is a mixture of two species: the flowering material is from a plant of P. Ballouii; fruits of P. clusiae- folia are mounted on the same sheet. Hence the species is based on mixed material, of which both elements already bear valid names, and the name ukuleleensis is relegated to synonymy. The type of P. Balloui is: Maui: Haleakala, trail from Ukulele to Waikamoi Gulch at ca. 5000 ft. alt., 25 October 1910, Rock & von Temp sky 8609 (Bishop). The type sheet of P. ukuleleensis is: Maui: Ukulele, July 1919, Forbes 749.M. (Bishop). 4. Pelea Hiiakae B. C. Stone, sp. nov. Arbuscula; innovationibus minute puberu- lentibus glabratis; foliis, petiolis, ramulisque glabris; petiolis filiformibus 9-21 mm longis, 1 mm crassis; laminis ellipticis base cuneatis vel subrotundatis apice obtusis vel rotundatis, 3.5- 6.5 cm longis et 2-3.5 cm latis, glabris, costa media anguste dorsaliter salientis brunneis, nervis lateralibus paralleliter ascendentibus, cum nervio sublineare a margine distans 1 mm coali- tis; inflorescentiis axillaribus 2-3-nodosis, ad 3.5 cm longis in fructu, axibus glabris, pedun- culis 3-8 mm longis, in crassitudine petiolo aequis, pedicellis clavoideis, sub calyce expansis; cymis unifloriferis; floris ignotis; capsulis ca. 25 mm diametro profunde (2A-3A) lobatis lobis Critical Species of Pelea — Stone 411 Fig. 1. Pelea kauaiensis H. Mann. Holotype specimen, Mann & Brigham, Cornell Univ. Herb. A leaf, showing pubescence and venation of lower surface; a flower (the entire 1 -flowered inflorescence), showing the glandular-punctate appearance; floral parts; immature capsule in side view; and reconstructed conjectural top view of a mature capsule. rotatis; exocarpiis ad basim sparse puberulenti- bus glabrescentibus; endocarpio sparse hirtello trichomiis ad suturas dorsales aggregatis; lobis diseminatis; seminis ovoideis, ca. 5-6 mm longis; testa Crustacea nigra nitida. HOLOTYPE: Oahu: Koolau Range, Kipapa Gulch, Waipio, on southerly ridge in woods at 1800 ft. alt., 16 October 1932, Edward Y. Ho- saka 809 ( Bishop ) . This species seems to be related to Pelea descendens St. John, which however is charac- terized by smaller capsules 13-17 mm in diame- ter, with glabrous exocarp and quite glabrous endocarp. Pelea Hiiakae is also related to P. Wawraeana Rock, which differs in that the pubescence of the endocarp is restricted to the innermost part of the sutures (and may some- times be lacking), and in the more compact, shorter, multi-flowered pubescent inflorescences. The species is named in honor of Hiiaka, youngest sister of the goddess Pele, heroine of the tales of Pele and Hiiaka. It must be noted that this specific name was first applied to certain specimens from the island of Kauai which were thought to represent an undescribed species ( Stone 1626, Faurie 225, and Lydgate s.n. ) . Since doubt remains as to the fruiting state of these plants and their specific distinctness from Pelea anisata, they must re- main undescribed. The name Pelea Hiiakae must now be borne by the Oahu plant described above, as this is the first publication of the name. Should the Kauai plants ultimately prove dis- tinct, another name will be given to the species. 5. Pelea kauaiensis H. Mann Jr. in Proc. Boston Soc. Nat. Hist. 10:313, 1866. Fig. 1 A small tree to 5 m high, with opposite leaves; innovations minutely but densely tawny - puberulent, but the branches quickly glabrate, at the 3rd or 4th node quite glabrous. Leaves distinctly petiolate, petioles 12-30 mm long, flattened ventrally and narrowly and shallowly channelled and nearly glabrous or with a few scattered minute trichomes at the margins; the rounded dorsal surface especially near the distal 412 PACIFIC SCIENCE, Vol. XVII, October 1963 end hirsute with trichomes 0.8- 1.0 mm long; but the whole petiole glabrate and ultimately glabrous. Blades 5-10 cm long, 3-6 cm broad, el- liptic, usually rounded at base, broadly subacute at apex or there minutely retuse, ventrally gla- brous even in bud, densely villous beneath all over the dorsal surface and densely hirsute along the costa, with pale stramineous trichomes ca. 1 mm long, the indument more or less perma- nent; costa raised and rounded dorsally; lateral nerves 8-14 per side, nearly straight and parallel, the connecting marginal nerve deeply arched, with several minor veinlets intervening between it and the margin; marginal nerve ca. 2-8 mm from the edge. Inflorescence a highly reduced cyme with usually 1 or rarely 2 to 3 flowers, axillary among the leaves; peduncle and pedicel sparsely and minutely puberulent in flower, later subglabrate; peduncle reduced to a slight pro- jection no higher than the 2 or 4 bractlets it bears, and hidden by them; pedicel ca. 5 mm long, 0. 3-0.4 mm thick, slightly flaring at calyx to a breadth of 07-0.9 mm, with bractlets tightly appressed at base of pedicel on the ob- solete peduncle; bractlets deltoid, ca. 0. 3-0.4 mm long, opposite, decussate; pedicels and perianth segments copiously opaque-punctate. Pistillate flowers ca. 3 mm high, sepals broadly ovate with rounded summit and sharp apex, 1.3 mm long, 1.5 mm broad, glabrous except for the minutely ciliolate margins. Petals glabrous, broadly ovate, acute, 2-4 mm long, ca. 1.7 mm broad. Stamens sterile, all shorter than the style, in 2 subequal quartets, the shorter quartet just under 1 mm long, the longer quartet bare more than 1 mm long. Ovary 4-lobed, glabrous, ca. 1 mm high, 1.3 mm broad, on a glabrous shallow somewhat reddish-speckled disk, the style 0.5 mm high, stigma 4-lobed, ca. 0.7 mm broad, the lobes dark purplish and minutely papillate. Staminate flow- ers unknown. Immature capsule nearly 10 mm diameter, the lobes (some abortive) still more or less ascending, 5 mm long, deeply parted (16-%), sepals caducous; exocarp glabrous, co- piously glandular-punctate; endocarp glabrous; lobes 2 -seeded. TYPE: Kauai: Waimea, 2000-3000 ft. alt., Mann & Brigham s.n. (Cornell). DISTRIBUTION: Kauai. Only one further collection of this species is known; it is from the general area of the type locality (Waimea, Kauai) and was discovered by Otto Degener. Unfortunately, it is a sterile specimen. This species has been consistently misinter- preted since its first description. Hillebrand (1888) described under the name Pelea kauai- ensis a species superficially similar to that which Mann had actually described, but clearly distinct; Hillebrand’s material was correctly segregated by Rock (1918) under the name Pelea recurvata Rock. In 1897 Heller described Pelea cruciata, which was however considered by later authors, such as Rock, to be a synonym of Pelea kauai- ensis Mann. Such was not the case; Heller’s species was justly distinguished, and is quite distinct from both Pelea kauaiensis or Pelea recurvata. Skottsberg (1944) noted the discrep- ancies between descriptions and specimens of these three entities, and published in tabular form some of their differences. Upon comparing type specimens of the three species it becomes quite apparent that they are distinct. The above extended description of Pelea kauaiensis is based on the original material, and should serve finally to clarify this least understood member of a trio of superficially similar species. Pelea kauaiensis may be quickly distinguished from the two other species formerly confused with it as follows: 1. Endocarp glabrous. 2. Blades permanently pilose beneath; capsule lobes ascending or rotate; capsules 10-? (perhaps 20) mm di- meter; inflorescence reduced, 1-3- flowered kauaiensis 2. Blades ultimately slightly glabrescent; capsule lobes recurved; capsules 21- 23 mm diameter; inflorescence stout, 3-9-flowered. recurvata 1. Endocarp pubescent; blades densely pu- berulent cruciata Mann’s original description indicated the name as kavaiensis; since, however, the Latin v is more commonly spelled out either as a true v or as u in modern usage, and since the u is es- sential to the meaning as well as the pronun- ciation of the name, it is thought best to retain with Hillebrand the name of the island in its correct form, as kauaiensis. See art. 73, note 6, Critical Species of Pelea — Stone 413 of the 1961 International Code of Botanical Nomenclature. Little need be added to the descriptions of P. recurvata and P. cruciata, but it should be men- tioned that Skottsherg 2863, commented upon by Skottsberg (1944) as P. cruciata, is actually a specimen of P. recurvata, as the endocarp is quite glabrous. 6. Pelea Lakae B. C. Stone, sp. nov. Arbor ad 5 m alta, ramis divaricatis, trunco ad 5-10 cm diametro, innovationibus dense hir- sutulis, trichomiis pallide albobrunneis adpressis ad 1 mm longis; petiolis ramulisque glabratis; petiolis 10-65 mm longis, 1.5-3 mm crassis; laminis ellipticis vel subovatis vel subobovatis, coriaceis, 4-25 cm longis, 2.5-11 cm lads, base cuneatis vel obtusis, apice subacutis, obtusis vel subrotundatis, supra glabris, infra puberulentibus in pagine juvente glabrescentibus, costa media salientis stramineis vel brunneis puberulentis ( trichomiis cum squamulis glaucis interspersis ) , nerviis lateralibus curvate ascendentibus in ex- tremis cum nervo marginale arcuato 2-10 mm intra margine coalitis, reticulis venulosis minori- bus intervenantibus; inflorescentiis axillaribus cymosis, cymis ad 2 cm longis axibus puberulen- tibus glabratis, 2-4-nodosis, 3-5-floriferis, pe- dunculis 2-7 mm longis pedicellis ca. 3 mm longis; floris ignotis; bracteis deltoideis extus puberulentibus; capsulis maturis 22-32 mm di- ametro profunde (34) lobatis lobis recurvatis pre- et postdehiscentionem et paullo reflexis, uni- vel diseminatis, base 8 mm altis, lateraliter compressis, apicem versus angustatis subfalcatis; exocarpio minute denseque puberulento non- ruguloso; endocarpio glabro in suturam basem centralem excepto, trichomiis paucis aggregatis; seminis ovoideis, 5-6 mm longis, testa Crustacea nigra nitida. HOLOTYPE: Oahu: Koolau Range, Pupukea, at about 2000 ft. alt., at edge of small boggy region, 17 September 1961, B. C. Stone & A. K. Chock 3633 (Bishop; duplicates to be distributed). A species reminiscent of Pelea recurvata of Kauai, but with a puberulent rather than gla- brous ovary, and a few hairs clustered near the innermost angles of each suture on the endocarp. The specific name honors Laka, goddess of the hula. 7. Pelea Lohiauana B. C. Stone, sp. nov. Fig. 2 Arbuscula? ramis virgatis elongatis, innova- tionibus ramulis petiolisque dense tomentellis, trichomiis stramineis val pallide brunneis ca. 0.8-0.9 mm longis; foliis maturis glabratis, ju- venilis tomentellis; petiolis 20-30 mm longis, glabrescentibus, supra canaliculatis glabris; lam- inis ellipticis coriaceis 7-15 cm longis 3.5-7 cm latis base apiceque rotundatis vel paullo emarginatis, marginibus paullo revolutis, costa media prominente, nervis lateralis 5-9 (-12) per latere unico, cum nervo marginale continue vel paullo arcuato intra margine 1-2 (-5) mm coalitis; inflorescentiis cymosis (3-) 5-7-flori- feris longe-pedunculatis tomentellis, pedunculis 12-38 mm longis ad nodo primo, 1-2-nodosis, bracteis oppositis lanceolatis 1.5-2. 5 mm longis concavis; axibus majoribus 5-14 mm longis; pedicellis 2-5 mm longis; floris masculis pro genere magnis sepalis ad 5.5-6 mm longis et ca. 4.5 mm latis, extus hirsutulo-ciliolulatis intus Fig. 2. Pelea Lohiauana Stone. Staminate flowers and floral parts of Degener 23984. 414 PACIFIC SCIENCE, Vol. XVII, October 1963 glabris, deltoideis, 7-9-nervatis; petalis toto gla- bris sparse obscure punctatis ovato-lanceolatis ca. 11 mm longis et 5 mm latis apice intus un- guiculatis; staminibus fertilibus 8, (4 ad 10 mm longis, 4 ad 7.5 mm longis, antheris 1.7— 1.8 mm longis, filamentis ligulatis ca. 1 mm latis); ovario glabro ca. 2 mm alto, disco glabro ca. 1 mm alto; stylo ca. 2 mm longo, stigma ca. 1.4 mm lato lobis rufidis minute papillatis cruciatis; floris foemeneis minoris sepalis ca. 3-5 mm longis et 3 mm latis, petalis ca. 7-8 mm longis et 3.5 mm latis, staminibus sterilibus ca. 2 mm longis subaequitantibus; ovario discoque glabro, sytlo glabro ca. 4 mm longo; capsulis quadrilo- batis, glabris, ad 20-25 mm diametro, endo- carpio glabro. HOLOTYPE: Kauai: Kokee, August 1924, Degener 8586 (Bishop; duplicates at Cath, NY). DISTRIBUTION: Kauai, in the Waimea high- lands. SPECIMENS EXAMINED: Kauai: Near Lehua- makanoe Bog, 18 September 1955, Degeners, Hansen & Hanner 23984 (Bishop; staminate fls. ). Kaholuamano, October 1895, Heller 2869 (formerly determined as P. resiniflora ) (GH, edinb, US). Halemano, 14 February 1909, Rock 2294 (A). Waialeale, October 1911, Rock (a). Kaholuamano, September 1909, Rock 5551 (A, Bishop), 5552 (gh, Bishop). This is by no means a complete citation of specimens; more are extant, particularly in the Bishop Museum, but many lack data, particularly localities and collectors’ numbers. Most of these specimens had been determined as Pelea macro- pus Hillebr., which however is quite distinct in having puberulent ovaries ( and hence capsules ) and clavoid pedicels. Because of this interpreta- tion, this species has consistently been mistaken for P. macropus, and this error has tended to obscure knowledge of both of them. The gla- brous endocarp and rotate capsular lobes place P. Lohiauana in a group with P. oblanceolata St. John (of Hawaii), P. manukaensis St. John (of Hawaii), and P. pseudoanisata Rock (of Hawaii ) . The specific name honors Prince Lohiau of Kauai, hero of the tales of Pele and Hiiaka, and one of the memorable characters of Hawaiian mythology. 8. Pelea macropus Hillebrand, FI. Haw. Ids. 65, 1888. P. acutivalvata Leveille in Fedde Repert. Sp. Nov. 10:153, 1911. Innovations minutely and closely pale cinere- ous-puberulent, hairs appressed, ca. 0.2 mm long; branchlets at youngest nodes cinereous but soon glabrate; leaves glabrous ventrally, at first minutely cinereous dorsally (mostly on the costa ) , but soon glabrate, the older leaves mostly quite glabrous; petioles soon glabrate. Leaves opposite, the petioles mostly 9-15 mm long; blades thinly coriaceous, rounded or emarginate at apex, contracted and subcuneate at base (the margins sometimes slightly revolute); costa ventrally pale, shallow-sulcate; dorsally pale and prominent with a rather sparse and minute ephemeral puberulence; lateral nerves at right angles to costa or slightly ascending, about 7-9 per side, conspicuous, prominulent on both sur- faces, connected at a distance of 2-10 mm from the margin by a moderately to deeply arcuate marginal nerve, with several sets of intervening meshes. Inflorescences small, axillary or ramu- line, mostly 1-3 -flowered, the peduncle very short, about 1-5 mm long, bibracteolate at apex, puberulent, the bractlets deltoid, rounded, 0.5 mm long, puberulent; axes minutely cinereous, short ( to 5 mm ) in flower, somewhat elongated in fruit; pedicels 2-5 mm long, flared at base of calyx and clavoid, bibracteolate, cinereous; pistillate flowers with broadly deltoid sepals about 1.5 mm long and 1.7 mm broad, minutely puberulent externally; petals about 2.5 mm long and 1.7 mm broad, with a faint close puberu- lence of few minute hairs along the midregion, glabrous within; rudimentary stamens glabrous, less than 1 mm long; disk reddish, glabrous, shallow, ca. 1.7 mm broad; ovary puberulent, ca. 1 mm high and 1 mm broad; style glabrous, ca. 1.6 — 1.7 mm long (including stigma), stigma lobes red, each 0.25 mm long, minutely papillate, obovoid. Capsules 25-35 mm diameter, about 9 mm high, sparsely appressed-puberulent (es- pecially near the base), the lobes deeply separated ( Y<$ ) , somewhat acuminate and laterally com- pressed, often unequally developed; endocarp glabrous, cartilaginous; seeds 5-6 mm long, ovoid, the testa crustaceous, blackish. Critical Species of Pelea— -Stone 415 HOLOTYPE: Kauai: Waimea, Knuds en 189 (Berlin; now destroyed). DISTRIBUTION: Kauai. SPECIMENS EXAMINED: Kauai: "Robinson’s summer house,” February 1910, Faurie 112 (type of P. acutivalvata; type at Edinburgh; isotype at A). Waimea, Kaholuamano, October 1895, Heller 2870 (bm, k. Paris, us). The description given above sufficiently dis- tinguishes Pelea macropus from the newly de- scribed P. Lohiauana. SECTION Cubicarpa stone 9- Pelea peduncularis Leveille in Fedde Repert. Sp. Nov. 10:443, 1912. Fig. 3 P. sandwicensis var. /3 sensu Hillebrand, FI. Haw. Ids. 66, 1888 (not P. sandwicensis [Hook. f. & Arn.] Gray). P. sandwicensis var. macrocarpa Hillebr. ex Rock in Bot. Gaz. 65:265, 1918. P. nodosa Leveille in Fedde Repert. Sp. Nov. 10:443, 1912. P. grandipetala Leveille l.c. (in part; specimen typicum mixtum; in part P. honoluluensis) . Syn. nov. P. singuliflora Leveille l.c. (in part; see P. Wawraeana ) . Syn. nov. P. Rockii St. John in Lloydia 7:271, 1944. Syn. nov. A small or sometimes shrubby tree, the in- novations finely scurfy and somewhat sparsely cinereous-puberulent; trichomes soon caducous, the mature branchlets and leaves glabrous, or the petioles sometimes puberulent then glabrate. Leaves petiolate, opposite; petioles 13-40 mm long, lenticellate in age especially at the ex- tremities; blades coriaceous, elliptic, rounded to slightly emarginate at apex and at base, mostly 3-18 cm long and 2-11 cm broad (commonly 5-9 X 3-6 cm ) , darker above, the margin near the base usually tightly revolute; midrib above sulcate, raised beneath, often reddish or purplish; major lateral nerves mostly about 7-10 per side, united distally by a slightly sinuate marginal nerve close to (about 1-5 mm from) the mar- gin. Inflorescence a pedunculate usually 5—2 1- flowered minutely cinereous glabrate cyme often FIG. 3. Pelea peduncularis Levi. Inflorescences; at left, a long inflorescence with a dehisced capsule, from Faurie 189 (isotype); at right, inflorescences in flower and in bud from Remy 621, with enlarged view of one staminate flower. longer than the adjacent petiole, the peduncle up to 4 cm long and with as many as 8 nodes, but usually about 3-4-nodose, stout ( 1-2 mm thick), the pedicels short and stout, 3-4 mm long, the bractlets lanceolate, the upper ones deltoid, minutely ciliolulate and sparsely pu- berulent, 1-3 mm long; pistillate flowers with sparsely scurfy-puberulent deltoid-ovate sepals about 2 mm long and 1.8 mm broad, lanceolate- ovate glabrous punctate petals about 5 mm long, reduced sterile stamens barely higher than the glabrous 4-lobed ovary, the disk glabrous, the style filiform reddish about 2 mm long, the stigmas 0.5 mm long, slightly clavate and mi- nutely papillate, rotate. Staminate flowers similar but larger, the petals about 7 mm long, the gynoecium much reduced, the fertile stamens longer, in 2 subequal quartets, the longer sta- mens about 8 mm long. Capsules mostly about 416 PACIFIC SCIENCE, Vol. XVII, October 1963 20—25 mm broad and 10 mm high, quadrate or slightly lobed ( less than or barely halfway ) , the exocarp glabrous, punctate, mostly dark green or tinged with dull reddish-purple; endocarp pale and glabrous; seeds mostly 2 per cell; carpels after dehiscence reflexed in age, the axis per- sistent only at the extreme base. TYPE: Oahu: Koolau Range, Kalihi, October 1909, Faurie 189 (Edinburgh; isotype at Paris). DISTRIBUTION: Restricted to Oahu; found on both the Waianae and Koolau mountain ranges. This common Oahu species has been misin- terpreted since it was first described. This is not surprising, however, since the confusion involved with the identity of the true Pelea sandwicensis (H. & A.) Gray had obscured the concept greatly, and since Leveille himself did nothing to clarify the species, and presumably did not realize that the other "species” so briefly and inadequately described by him were either the same or mixtures (the type specimens often being composed of branchlets clearly from two or three different species). In addition, some rather clearly defined varieties of the species occur in particular areas of Oahu, and some of these have been described as distinct species. Dozens of collections from various localities both in the eastern and western mountain ranges are known. A few representative specimens are cited here; full citation of specimens will be presented in the forthcoming monographic treatment. SPECIMENS EXAMINED: Oahu: Koolau Range: Pupukea, January 192 7, MacDaniels 547 (Bishop); February 1928, Degen er & Shear 8580 (NY, Catholic, US). Kalihi, October 1909, Faurie 1 90 ( bm ) . Konahuanui, Famie 194 ( bm, isolectotype of Pelea singuliflora ) . Manoa-Pauoa hills, April 1861, Hillebrand 251 (Kew). Nuu- anu, May 1861, Hillebrand 238 (Kew). Nuuanu- Kalihi ridge, August 1922, Skottsberg 173 (Goth, Kew); Niu, Hillebrand 1797 (US, isotype of Pelea molokaiensis (3 var. Hbd. ) . Waianae Mountains: Puu Kaala, Mann & Brigham 600 (Cornell, Kew). Makaha Valley, Feb. 1909, Forbes (Bishop). Puu Kanehoa, May I960, Stone 3268 (Bishop, Goth, Kew, Leiden, us). Without locality: Mann & Brigham 208, in part (Cornell). Remy 621 (Paris, 3 sheets). (a) var. peduncularis Fig. 3 Found mostly along the southern stretches of the Koolau Range, from Punaluu down through Waialae and Niu valleys. The typical variety, easily confused with Pelea oahuensis Levi, and sometimes with Pelea Wawraeana Rock, but differing in many respects from both; from the former in the larger and flatter capsules, the more ample and stouter cymes, and generally larger vegetative parts, and the larger reddish (not yellowish) flowers; and from the latter in the larger capsules with glabrous endocarp, and mostly glabrous petioles and branchlets as well as cymes. (b) var. niuensis (St. John) B. C. Stone, comb, nov. Pelea niuensis St. John in Lloydia 7:272, 1944. Branchlets glabrate; petioles glabrate; blades narrowly elliptic or oblong-elliptic, glabrous, 6-12 cm long and 2-5 cm broad; inflorescences commonly 7-9-flower ed, glabrous; capsules nearly quadrate when fresh, slightly lobate when dry, 21-25 mm broad and about half as high, glabrous; endocarp glabrous. KEY TO VARIETIES OF Pelea peduncularis 1. Capsules slightly lobed (up to halfway), green, usually dull, sometimes reddish; blades elliptic to elliptico-oblong. 2. Leaves opposite. 3. Blades very narrowly oblong-elliptic var. niuensis 3. Blades elliptic to broadly oblong-elliptic. 4. Cymes mostly 7-21 -flowered var. peduncularis 4. Cymes mostly 3-7-flowered. . var. pauciflora 2. Leaves whorled in fours - var. paloloensis 1. Capsules unlobed, quadrate, dark glossy green. var. quadrata Critical Species of Pelea — Stone 417 HOLOTYPE: Oahu: Koolau Range; Niu, Oc- tober 1940, St. John 20111 (Bishop). DISTRIBUTION: Endemic to the southeastern- most part of the Koolau Range. This variety differs from the typical P. pedun- cularis only in the much narrower leaves and perhaps slightly fewer-flowered cymes. SPECIMEN EXAMINED: Oahu: Northeast slope of Puu Kumakalii, April 1936, Degener, Tam, Takamoto & Martinez 10579 (Bishop, Catholic, NY). (c) var. pauciflora (St. John) B. C. Stone, comb. nov. Pelea Rockii var. pauciflora St. John in Lloydia 7:271, 1944. Cymes mostly 3-7-flowered. HOLOTYPE: Oahu: Koolau Range; Kaukona- hua Gulch, Wahiawa, May 1909, Rock 3046 (Bishop). DISTRIBUTION: Known only from the type locality. A poorly defined variety which may be only an anomalous form of the species. (d) var. paloloensis (St. John) B. C. Stone, comb. nov. Pelea paloloensis St. John in Lloydia 7:271, 1944. Leaves whorled in fours at the nodes. HOLOTYPE: Oahu: Koolau Range; Palolo, Waialae-iki, Jan. 1911 , Forbes 2404.0. (Bishop). DISTRIBUTION: Known only from the type locality. The variety, which probably is nothing more than an anomalous form, is an example of the occasional exception to the common taxonomic criterion for distinguishing species which are members of the section Pelea, with ordinarily whorled leaves, and the other sections, with generally opposite leaves. (e) var. quadrata B. C. Stone, var. nov. Fig. 4 Capsulis elobatis quadratis fere ca. 23 mm latis. HOLOTYPE: Oahu: Waianae Mountains; Puu Hapapa, August 1932, Degener 8521 (Bishop). DISTRIBUTION: Known from the northern end of both ranges of Oahu. A characteristic variety, found both in the northern Waianae mountains and in the north- ern Koolau Range (as for instance around Pu- pukea), sometimes occurring in company with the typical variety. 10. Pelea olowaluensis St. John in Lloydia 7:2 66. 1944. A tree up to 3 m high, the branchlets at first puberulent; leaves opposite; petioles glabrous; blades 2.8-8 cm long, oval, subcoriaceous, gla- brous; cymes mostly 3 -flowered, glabrous, the peduncle ca. 9 mm long; capsules 16-20 mm broad, flattened, lobed halfway or less, glabrous; endocarp glabrous. HOLOTYPE: Maui: Olowalu Valley, on the cen- tral ridge, May 1920, Forbes 2326.M. (Bishop). DISTRIBUTION: West and East Maui. SPECIMEN EXAMINED: West Maui: Mauna Huuma, May 1910, Forbes & Cooke 22.M. (Bishop) . This species is closely allied to Pelea pedun - cularis of Oahu and falls properly into section Cubicarpa . The characteristically revolute basal leaf-margins are especially noticeable. When fresh the capsules might perhaps be unlobed or barely notched. The disjunction of the two col- lections would imply that the species is, or was, fairly widespread on Maui. Fig. 4. Pelea peduncularis var. quadrata Stone. Three views of a capsule in the fresh state, from Stone 2832 ( Pupukea, Oahu ) . 418 PACIFIC SCIENCE, VoL XVII, October 1963 Fig. 5. Pelea oahuensis Levi, emend. Stone. Neotype, Stone 3282. Above, capsules and pistillate flowers; below, staminate flowers. 11. Pelea oahuensis Leveille in Fedde, Repert. Sp. Nov. 10:442, 1912, emend. B. C. Stone. Figs. 5, 6 A small to middle-size tree, pungent of anise, with glabrous reddish branchlets and opposite leaves; innovations scurfy, the waxy white ephemeral scales mostly concealing an extremely minute sparse puberulence of white hairs usually less than 0.2 mm long, ephemeral also, the older leaves and branchlets glabrous except for a few rare hairs persisting in the axils or on the dorsal midrib near the base; leaves petiolate, the peti- oles mostly 6-20 mm long, reddish in age, gla- brous, flattened ventrally, lenticellate; blades thin coriaceous, mostly (3) -4-12 cm long and 2.5-7 cm broad, glabrous and moderately veiny, darker above, moderately shining, mostly elliptico- obovate, rounded to slightly emarginate at both ends, the costa pale and shallowly canaliculate toward the base ventrally, beyond prominulous, beneath raised and reddish; lateral main nerves slightly curvate-ascendent, joined marginally by a lightly arched or nearly straight nerve about 1-5 (-9) mm from the edge. Inflorescence cymose, mostly 3-5 -flowered (rarely with 7 flowers), rather short (less than 2 cm long), on short glabrous greenish sometimes scurfy peduncles 3-4 mm long, bibracteolate at apex (the bracteoles minute, deltoid, minutely pu- berulent or glabrous ) ; axes about 2-3 mm long, similar to peduncles; pedicels about 3 mm long, bibracteolate near the middle or above; flowers functionally staminate or functionally pistillate, or perfect, frequently with both staminate and perfect flowers on the same plant; pistillate flowers greenish, petals yellowish-green; sepals deltoid-ovate, about 1.8 X 1.8 mm, the margins minutely ciliolulate, otherwise glabrous; petals 4.5-5. 5 mm long and 2 mm broad, glabrous; stamens rudimentary, about 1 mm long (some- times in two slightly unequal quartets); ovary and disk glabrous, punctate, the disk about 2.8 mm broad and nearly 1 mm high, the ovary about 2 mm broad and 1 mm high; style slender, glabrous, about 2 mm long including the 4-lobed stigma; staminate and perfect flowers similar but the stamens greatly elongated, the longer quartet as long as the petals; petals slightly longer than those in pistillate flowers, and the ovary and disk greatly reduced, the style short, only 1 mm long. Capsules subcuboid, 10-14 mm broad, about 9 mm high, unlobed or nearly so. Critical Species of Pelea— Stone 419 Fig. 6. Pelea oahuensis Levi, emend. Stone. Habit of fruiting specimen, from Stone 2805 ( Pupukea, Oahu). At left, perfect flower from Lane 243 (Pupukea). At right, above, capsules and seed from Stone 3200 (Kaala, Oahu). At right, below, capsule and pistillate flower from Stone 2826 (Pupukea). 420 PACIFIC SCIENCE, Vol. XVII, October 1963 quite glabrous, often pale, greenish, or after dehiscence brown, dehiscent along the upper and lateral sutures; endocarp glabrous, pale, thin, late detaching; seeds often only one per carpel; testa crustaceous, shining black. TYPE: Leveille cites three specimens, all from Kalihi, Oahu, collected by Faurie: numbers 11, 217 , and 217 bis. No type is designated. A search was made for these specimens in the herbaria of the Bishop Museum, the British Museum, the Royal Botanic Gardens at Kew and at Edin- burgh, the Museum d’Histoire Naturelle, Paris, Gothenburg Botanical Garden, U. S. National Museum, Gray Herbarium, Arnold Arboretum, Cornell University, etc., but without avail. Many of Leveille s "types” are at Edinburgh or at the British Museum, but none of the cited specimens can be located. The original description is so short and noncommittal that alone it is meaning- less: "Affinis praecedenti a quo tantum differt foliis longius 10-25 mm petiolatis, oppositis; petiolo nec rugoso, nec vulnerato, corymbis 2-5 floris, pedicellis bis bibracteolatis, apice incras- satis.” Rock (1914) reduced Pelea oahuensis to P. molokaiensis Hbd. (2 var. Hbd., on the basis of Paurie 217, which perhaps was at the Berlin Museum. The preceding species to which Le- veille refers in the above description is his own Pelea waianaiensis, which is more fully de- scribed. It is based on a single number ( Faurie 215, isotype at Bishop), and is apparently a form of Pelea peduncularis Levi. The reference, however, is helpful in interpreting the present species, as the two are similar in appearance. In the absence of any type material, it seems necessary to designate a neotype, which would of course be replaced by original material should any turn up. The species, as interpreted here, is a very distinct and common one on Oahu, and is the Oahuan counterpart of the "moki- hana” or Pelea anisata Mann of Kauai; the same strong anisate odor is present in all parts, the flowers and fruits are very similar, and the specific differences are mainly in characters of the leaves and habit. There are a great many collections of this species, most of which have been determined as Pelea Wawreana Rock, which is, however, a very different species of Sect. Megacarpa. The following collection may be selected as neotype: NEOTYPE: Oahu: Waianae Mountains; Kunia trail, 26 March I960, B. C. Stone & G. Pearsall 3282 (Bishop; duplicates to be distributed). DISTRIBUTION: Endemic to Oahu. This is a very common species on both the Waianae and Koolau ranges. It may be identified in the field by its strong anise fragrance, very short essen- tially glabrous 3-5 -flowered cymes with small greenish flowers, and small cuboid green or whitish-green capsules. In the Waianae Moun- tains it occurs in company with Pelea peduncu- laris, P. kaalaensis , P. clusiaefolia var. crassiloba, and P. elliptica. REFERENCES Heller, A. A. 1897. Observations on the ferns and flowering plants of the Hawaiian Islands. Minnesota Bot. Studies 9:760-922, pis. 42-69. Hillebrand, W. 1888. Flora of the Hawaiian Islands. Heidelberg, London, & New York, i-xciv, 1-673. Rock, Joseph F. 1914. Revisio plantarum Ha- waiiensium a Leveille descriptarum. Repert. Sp. Nov. ed. Fedde 13:352-361. — — — 1918. Pelea and Platydesma. Bot. Gaz. 65:261-267. Skottsberg, C. 1944. Vascular plants from the Hawaiian Islands, IV. Phanerogams collected during the Hawaiian Bog Survey 1938. Acta Horti Gotob. 15:275-531. Stone, B. C. 1962a. Studies in the Hawaiian Rutaceae, I. Taxonomic and nomenclatural notes on Platydesma (Hawaii) and Melicope (Solomon Islands). Madrono 16:161-166. 1962 A Studies in the Hawaiian Ruta- ceae, II. On the identity of Pelea sandwicensis. Pacif. Sci. 16:366-373. 1962c Studies in the Hawaiian Ruta- ceae, III. On the New Caledonian species of Pelea, and a misunderstood species of Platy- desma. Adansonia (Paris) : I, 2:94-99. l%2d. Rutaceae: Genus Pelea. In: O. Degener, Flora Hawaiiensis. Honolulu (pri- vately printed). A Prior Name for the Hawaiian Gouldia terminalis (Rubiaceae) Robert L. Wilbur1 Among the most frequently encountered woody plants in the wetter, forested portions of the Hawaiian Islands are members of the extremely variable genus Gouldia. Fosberg (1937) pre- sented the results of his detailed study of this baffling genus and concluded that the variability could be properly categorized in not less than three species composed of more than 90 varieties and forms. However, even this number of for- mally named taxa failed adequately to represent the variability, for hybridization was so rampant that at that time more than 50 hybrids were also recognized and characterized. It is therefore not surprising that Gouldia has acquired a repu- tation, among botanists working on Hawaiian plants, not unlike that of Crataegus and Rubus in the eastern United States. Like those genera, it is naturally felt that its taxonomy can now be handled only by a specialist. The present note, written far from Hawaii, is therefore merely concerned with the nomenclature of the most widespread and variable species of this endemic genus.2 Although members of the genus undoubtedly must have been collected by botanists on several expeditions prior to that of the "Rurik” led by Kotzebue, the first description of a species is apparently the detailed analysis provided by Chamisso and Schlechtendal (1829) of their "Kaduae affinis.” Chamisso was the botanist on Kotzebue’s voyage, and the original collection apparently was made on the slopes of the Koolau range of Oahu. Their account, as pointed out by Heller ( 1897), Fosberg (1937:4,26) and Bul- 1 Department of Botany, Duke University, Durham, North Carolina. Grateful acknowledgment is made to the National Science Foundation for a grant of re- search funds to Duke University (NSF-Grant 18799) which made the present study possible. Manuscript received March 29, 1962. 2 Dr. F. R. Fosberg’s thoughtful advice is here acknowledged with appreciation, but this does not imply that he is necessarily convinced of the change proposed in this paper. lock (1958), did not result in a published bi- nomial at that time as was inferred by A. Gray (I860) and the Index Kewensis (1895). In spite of the unusually detailed analysis of the sixth species appearing in their newly described genus Kadua , Chamisso and Schlechtendal failed to provide a binomial for this plant; they merely indicated its close affinities to Kadua , from whose species it differed in its indehiscent fruit and toothed stipules. The following year DeCandolle (1830) listed each of Chamisso and Schlechtendal’s species and condensed the original detailed accounts of each into but a few lines. Their ”6. Kaduae affinis ” appeared in DeCandolle’s Prodromus in the form quoted below: 6. K? AFFINIS (Cham, et Schlecht. 1. c. p. 164.) ramis tetragonis transversim rugosis, foliis elliptico-lanceolatis acutis basi obtusis breve petiolatis, stipulis membranaceis utrinque sub- dentatis deciduis, cyma thyrsoidea terminali, drupa subglobosa, limbo calycis obliterato infra apicem coronata, indehiscente. [Woody tree or shrub} in insula O-Wahu. Flor. ignoti. Fosberg (1936:4) dismissed DeCandolle’s publication as a nomenclatural source in the statement quoted below: DeCandolle, in 1830, a year after the publication of Chamisso and Schlechtendal’s work, credited the latter with a f Kadua - affinis Cham, and Schlecht.,’ appending a description which is an obvious condensation of the description pub- lished by Chamisso and Schlechtendal. DeCan- dolle’s entire treatment of Kadua is based directly on the original treatment of the genus by Chamisso and Schlechtendal, with the same species arranged in the same order and with descriptions which are identical but somewhat condensed ... It is obvious that f Kadua affinis’ is the result of a misinterpretation of the intent of the original authors of the genus Kadua, as DeCandolle added nothing to the descriptions and no discussion. Therefore, it is evident that 421 422 PACIFIC SCIENCE, Vol. XVII, October 1963 DeCandolle did not have any intention of mak- ing a new species 'Kadua affinis / making it pos- sible to judge it on the basis of Chamisso and Schlechtendal’s original intent. The Cambridge Rules of 1930, Article 68, state that terms which are merely words, not intended as names, should be rejected; thus, fortunately, a means is provided for disposing of this meaningless name. It would seem, however, that Fosberg’s stated reasons are not sufficient justification for reject- ing the name. Chamisso and Schlechtendal con- cluded in their final sentence concerning this species that "the whole internal structure of the fruit agrees therefore with Kadua.”3 Although they did not publish a binomial, it certainly would appear that DeCandolle did. Chamisso and Schlechtendal expressed their doubts as to generic position in one fashion and did not provide the taxon with a name; DeCandolle was no more certain as to the generic position of the plants than the original authors, but he ex- pressed his doubts as to generic position in a different manner and did provide a binomial. He was perfectly free to utilize the epithet affinis, which he did. The question mark follow- ing the abbreviation of the generic name merely indicated that he too was uncertain that the clearly described species actually was congeneric. Chamisso and Schlechtendal more than ade- quately described the species but failed to pro- vide a binomial; DeCandolle provided a binomial even though admitting that the plant might eventually prove not to be a congener of the other five species. Although DeCandolle "added nothing to the description and no discussion,” he did provide the binomial that the previous detailed description of the species lacked. Fos- berg’s conclusion that "it is evident that De- Candolle did not have any intention of making a new species Kadua affinis ” seems unwarranted. DeCandolle provided a binomial, and there is certainly no evidence that his intentions were at variance with his accomplishment. In the Pro dr omus DeCandolle did not cus- tomarily place any authority at all after names that he was proposing. Species and new combi- nations being there published by him were not followed by any authority or reference. Since 3 "Convenit ergo omnis interna fructus fabrica cum Kadua.” abbreviations for Chamisso and Schlechtendal. together with a reference to the place of publi- cation, were all included parenthetically by De- Candolle after K? affinis , it might be argued that he was merely accepting Chamisso and Schlechtendal’s treatment and had no intention of publishing a new name. This is impossible to prove one way or the other, but certainly Article 34, Note 2, of the International Code of Botanical Nomenclature (1961) was never meant to be applied in such a case, as is shown by the examples provided. Furthermore, DeCandolle employed the parenthetical citation to convey a variety of information in addition to his custom- ary indication of author and place of publication. Other information characteristically conveyed parenthetically by DeCandolle concerned authors who had merely indicated a species as new on a herbarium specimen (e.g., Eryngium Haenkei Presl ex DC, Prodr. 4:94. 1830); or mention that a description of a new species has been provided in a letter by another (e.g., Eryngium prostratum Nutt, ex DC., Prodr. 4:92. 1830); or that a drawing had been seen in the herbar- ium (e.g., Cornus disciflora Moc. & Sesse ex DC., Prodr. 4:273. 1830). Certainly, since De- Candolle had not seen specimens of this Ha- waiian species, it was in accord with his rather liberal usage of parenthetical citation to indicate the source of his information. It is certainly not necessary to conclude that DeCandolle was not aware that Chamisso and Schlechtendal actually had not published a binomial. The conclusion therefore seems to me inescapable that DeCan- dolle did originate a binomial that was the first published for this species. A combination based upon DeCandolle’s bi- nomial, therefore, appears necessary for this extremely common Hawaiian species. Combina- tions for the multitudinous varieties and forms within this species, of which more than 85 were originally proposed by Fosberg, are not provided here; it would seem most undesirable for anyone not thoroughly familiar with these variants to make the numerous transfers apparently re- quired. It seems probable that a restudy of the problem with the benefit of the numerous col- j lections made during the past quarter of a number of taxa worthy of recognition. However, Gouldia terminalis — WlLBUR 423 Skottsberg (1944 a, b) has expressed the opinion that there are at least several times as many species as accepted by Fosberg, and he has pro- vided binomials for a number of them by ele- vating certain of Fosberg’s varieties and forms in addition to accepting some binomials of earlier authors. And, recently, the Degeners (1961) have also expressed a different view and, in so doing, provided 22 new combinations largely as a result of elevating Fosberg’s taxa to the next highest rank. Therefore, it would not seem desirable to make wholesale transfers at the present time. Gouldia affinis (DC.) comb. nov. "Kaduae affinis” Cham. & Schlecht., Linnaea 4: 164. 1829. Kadua affinis [K? affinis] DC., Prodr. 4:431. 1830. Petesia terminalis Hook. & Arn., Bot. Beechy’s Voy. 85. 1832. Petesia coriacea Hook. & Arn., Bot. Beechy’s Voy. 85. 1832. Gouldia sandwicensis A. Gray, Proc. Am. Acad. 4:310. I860, nom. illegit. Art. 63 and 67. Gouldia terminalis (Hook. & Arn.) Hbd., FI. Haw. Is. 168. 1888. REFERENCES Bullock, A. A. 1958. Nomenclatural notes: VI. Type species of some generic names. Kew Bull. 13:98 {Gouldia']. Chamisso, L. C. A., von, and D. von Schlech- TENDAL. 1829. De plantis in expeditione speculatoria Romanzoffiana observatis. Lin- naea 4:157-165 {Kadua] . DeCandolle, A. P. 1830. Prodromus systematis naturalis regni vegetabilis 4:430 + 431. [Kadua]. Degener, Otto, and Isa Degener. 1961. Gouldia in Hawaii. Phytologia 7:465-467. Fosberg, F. Raymond. 1937. The genus Goul- dia (Rubiaceae). Bishop Mus. Bull. 147:1-82. Gray, A. 1860. Notes upon some Rubiaceae, collected in the South Sea Exploring Expedi- tion under Captain Wilkes. Proc. Am Acad. 4:310 + 311. [Gouldia]. Heller, A. A. 1897. Observations on the ferns and flowering plants of the Hawaiian Islands. Minnesota Bot. Stud. 1:896-899. [Gouldia]. Index Kewensis. 1895. 2:1. International Code of Botanical No- menclature. 1961. Utrecht. 372 pp. Skottsberg, C. 1944^. Vascular plants from the Hawaiian Islands. Acta Horti Gothoburg. 15: 466, 517. [Gouldia]. — 1944 A On the flower dimorphism in Hawaiian Rubiaceae. Arkiv for Botanik 31A (4): 11-14. [Gouldia]. Population Dynamics in a Sublittoral Epifauna Willis E. Pequegnat1 At PRESENT we have little evidence that suc- cessional changes following predictable patterns occur among epifaunal communities living on natural rock-reefs in the shallow sublittoral of the open ocean. If this phenomenon does occur here, it can be detected by sustained observa- tions at a single study site. In the period from 1957 to I960 I conducted an intensive study of the epifaunas of two submarine hogbacks lo- cated at different depths off the coast of Corona del Mar, California. Some observations made during this period indicate that both gradual and disruptive changes do occur in these epi- faunal communities. Attention is called in this paper to an abrupt change in population density of a predominant species that occurred on part of the shallower reef, and to the widespread biotal adjustments that ensued. The epifauna of the deeper reef (9.5-18.5 m) remained quite stable throughout the study pe- riod (Fig. 1, Reef 500). Nevertheless, this is a dynamic stability with an appreciable flow of individuals through the community structure. For example, in the rock oyster, Chama pellucida, which is dominant on the upper part of this reef, natality and growth counterbalance mor- tality from all causes and such erosive factors as Lithophaga burrowings, resulting in a stable community as determined by consecutive sam- plings over a prolonged period. Some seasonal population fluctuations were observed, especially among sponges and ectoprocts, but I have not detected there any changes that can be construed as serial stages moving toward a climax differing from any extant community. But this stability appears not to be charac- teristic of a significant part of the epifauna that 1 Formerly at National Science Foundation, Wash- ington 25, D. C. Present address: Department of Oceanography and Meteorology, Texas Agricultural and Mechanical University, College Station, Texas. Manuscript received May 2, 1962. was present on the shallower reef (2-11.2 m) when the first observations were made (Fig. 1, Reef 200 ) . Although the temporal changes in communities observed here cannot be presented as a typical ecological succession, they have had profound influences upon the composition of the epifauna of this rock-reef. My purpose in calling attention to these changes is to stimulate other investigators to follow the course of simi- lar population shifts that may occur in regions under their purview. Additional data may make it possible to delineate the role of successional heterochrony in creating the complex epifauna present on submarine reefs at any given point in time. The central species in the present study is the common mussel, Mytilus edulis . Beginning in late January, 1959, the Mytilus population on the inshore reef began a decline in density, which was not recognized as of singular im- portance at the time. However, before this de- cline leveled off it resulted in an almost complete disappearance of this species from the reef. The basic cause is unknown, although the early sharp- ness of decline suggests that disease may have been a contributing factor. Still, the decline did not occur on the deeper reef, some 300 m distant, although it must be pointed out that the maxi- mum population densities here were only a hundredth of those on the shallower reef. In addition to the possibility of disease, it is known that an unusually large population of Pisaster giganteus was preying on the dense mussel popu- lation during preceding weeks. Although the decline extended through a year, the bulk of the Mytilus population disappeared in half this time. Unfortunately observations were not made during part of this period, nor has it been pos- sible to follow developments after Mytilus was nearly wiped out. Nevertheless, sufficient data are available to provide a picture of the biotal shifts that accompanied the decline. The dis- 424 Population Dynamics— Pequegn AT 425 Fig. 1. Location of the two sublittoral rock-reefs under study, in relation to the city of Corona Del Mar and the entrance to Newport Harbor, Orange County, California. appearance of certain predominant species ap- parently triggered increases and decreases in associated species. And these changes proved to be very extensive, involving some species that would not have been expected to react to changes in Mytilus populations. STUDY SITE The rock mass under study is composed of siltstone that has been subjected to complex folding. Gorsline (1962) advises that these promontories may properly be called submarine hogbacks. The dip of the strata ranges from a few degrees to 90°, but the larger part exhibits the latter dip. Because it is located approximately 200 m offshore, I shall refer to it hereinafter as Reef 200. Situated about 2.5 km southeast of the en- trance to Newport Harbor, Reef 200 ( Fig. 1 ) rises from a sand bottom at a greatest depth of 11.2 m to a shallow point of 2 m at standard mean sea level. It has a minimum surface area of 1,200 m2 distributed over a length of 58 m and an average width of 27 m (ranges from 8 to 44 m). Because its long axis parallels the shore, the hogback lies athwart the direction of surface wave propagation. This results in in- terference with wave transmission and a high degree of turbulence at shallow points. The magnitude of turbulence is inversely related to depth; hence it falls off rapidly down the walls and is barely perceptible at the reef base on days of average swell. Most of the Mytilus population was confined to the reef’s flat top, between depths of 2 and 4 m. The same was true of such brown algae as Eisenia arborea , Egregia australis , and Laminaria farlowi. STUDY METHODS Samples of the Mytilus populations of Reef 200 were obtained under water by using SCUBA. Quadrats were established by placing metal rings encompassing 0.1 m2 against the biota and re- moving all components by hand. Where re- quired, hammers, chisels, and. forceps were also utilized. All specimens were placed in canvas bags, which were sealed prior to ascent and re- turned to the laboratory. Subsequently species were identified, and individuals of numerable species (noncolonial) were counted and meas- ured. Prior to the Mytilus decline, study sites were selected by dropping metal rings onto the reef from a moving boat. Samples were removed from the sites where the rings came to rest. During and after the decline, conscious attempts were made to remove samples within a few meters of predecline quadrats. This was possible since chisel marks were still evident on the rocks. The first samples were taken on October 29, 1958, and the last on February 22, I960. The longest period without samples extended from February 15 to August 19, 1959, when the author was in Europe. For most purposes I have divided the dates of sampling into three periods of time: Predecline (October, 1958, to January 15, 1959), Decline (January 16 to October 15, 1959), Postdecline (October 16, 1959, to February, I960). Although Mytilus had not disappeared completely from the reef at the end of the study, its average population per 0.1 m2 quadrat was then only 0.5% of its original value. Twelve 0.1 m2 quadrats were studied during the above periods. Large species, such as Pisaster giganteus, Strongylocentrotns purpuratus, and Parasticho- pus parvimensis, were sampled with metal rings encompassing 1 m2. Specimens of these species were not removed from the reef. Samples were 426 PACIFIC SCIENCE, VoL XVII, October 1963 taken at random distances along transverse and longitudinal transects. Some 100 sq m quadrats were established during this study. RESULTS During the decline, the Mytilus edulis popu- lation at depths of 2 m dropped from an average of about 1,100 to less than 10 individuals per 0.1 m2. Even more drastic reduction eventually occurred at greater depths (Table 1). And as Mytilus declined at specific sites, populations of the red alga Corallina chilensis, which are stated in terms of percentage of rock surface covered in Table 1, increased on the same sites, particu- larly above depths of 4 m. There was, however, a considerable lag between Mytilus depletion and the spread of Corallina. This lag permitted rapid population increases of a few invertebrates. But, as will be discussed later, some of these increases were cut back after the encroachment of Corallina got underway. In turn these popula- tion shifts triggered other reciprocal changes among less noticeable species, especially among platyhelminths, nemerteans, mollusks, annelids, and crustaceans (Table 2). Fortunately sufficient samples had been taken prior to, or early in, the Mytilus decline to establish a basis of com- parison with subsequent unforeseen events. The samples of November, 1958 (Table 1), reveal ( 1 ) that most of the Mytilus population lived on horizontal surfaces at depths between 2 and 4 m, ( 2 ) that its density decreased sharply with increasing depth to the 7-m level, and (3) that it was absent from this point downward. These relationships persisted unchanged as late as Jan- uary 15, 1959. Soon thereafter, however, it was noted that Mytilus shells were beginning to ac- cumulate in unusual numbers on the shoreward bottom of the reef. Unfortunately an impending trip to Europe precluded additional sampling for several months. But by August, 1959, the Mytilus population at the 2-m level (samples were taken only a few meters horizontally from the predecline quadrats) had dropped about 80%. And, as can be seen in Table 1, by Febru- ary, I960, considered here to be early post- decline, Mytilus had almost disappeared from the entire reef. Apparently at depths below 2 m the initial rate of decline was not as rapid, but complete disappearance eventually occurred. The later occurrence of small individuals at depths of 8 and 9 m is probably explained by dislodgment of small clumps from the reef-top. As seen in Table 1, Corallina increased markedly during this time, and came to form a uniform mat, some 6 cm thick, over much of the reef’s upper surface. This trend continued until at the culmination of the study it covered 65% of the shallowest part of the reef with a mat some 9 cm thick. Meanwhile significant changes occurred in molluscan populations. Whereas an average of 25 molluscan species per plot (among a total of some 35 species on all plots) occurred with Mytilus (at 2-m depths) prior to the decline, this average dropped to 11 by September, 1959 TABLE 1 Changes in Population Densities, with Depth and Time, of Mytilus and Corallina (Dashes indicate absence of data) DEPTH (m) Mytilus edulis: IND /0.1 M2* BY DATE Corallina chilensis: % COVERf BY DATE 11/58 9/59 10/59 2/60 11/58 9/59 10/59 2/60 2 1,086 176 95 5 5 25 50 65 4 180 90 0 0 5 10 15 35 6 95 50 0 0 - 5 5 10 7 7 20 0 0 5 - - 0 8 0 0 7 0 - 5 5 - 9 0 7 0 0 0 0 0 - * Individuals per quadrat, f Per cent of quadrat surface covered by plant. Population Dynamics — PEQUEGNAT 427 TABLE 2 Data Gathered from 0.1 m2 Quadrats at 2-m Level Before, During, and After Major Part of Mytilus Decline NO. SPECIES IN PHYLA (Average) PHYLA Predecline Decline Postdecline Porifera 2.5 1.5 2 Coelenterata 3 3 3 Platyhelminthes 2 3 4 Endoprocta 1 0.5 0 Ectoprocta 5 3.5 6 Sipunculoidea 1 1 1 Nemertea 1 1 3 Mollusca 25 11 25 Annelida 12 7 21 Arthropoda (Crustacea) 8 6.5 11 Echinodermata 4 2.5 5 Chordata (Ascidiacea) 1 1 1 TOTAL ANIMAL SPECIES 66 48.5 82 Numerable Species 56 42 72 Motile Species 36 28 47 TOTAL ANIMAL INDIVIDUALS/0.1 M2 1759 1754 841 (Table 2). Then, as the result of the encroach- ment of different molluscan species, it increased again to an average of 25 by February, I960. Four types of population shifts were exhibited by molluscan species during this time: (1) the numbers of one species originally present in- creased, (2) several species declined but per- sisted, (3) others declined in numbers and eventually disappeared, and (4) previously ab- sent species invaded the Corallina mat. Chama pellucida is the only species that increased and retained its population gain, although, as would be expected, the new individuals were very small. Twelve of the original 35 species suffered severe population drops, but were still present in February, I960. Illustrative of these, in addi- tion to Mytilus edulis, are Hiatella arctica , which dropped from an average of 135 to 10 indi- viduals / 0.1 m2, and Pterorytis nuttalli, which dropped from an average of 16 to 2 / 0.1 m2. Twenty species of mollusks that were originally present disappeared completely. Among the more prominent of these are Mytilus calif ornia- nus, Modiolus capox, Tegula ligulata, Amphissa hicolor, and Gians carpenteri . Among the 15 invading species are Anomia peruviana, Chlamys hastatus, Flabellina iodinea, Doriopsilla fulva, and Erato vitellina. The following three species were unaffected by the other changes: Crepipa- tella lingulata, Pholadidea penita, and P. ovoidea. Similar changes took place among the poly- chaetes. Whereas an average of 12 species oc- curred on Mytilus quadrats (2-m depth) prior to the decline, this dropped to 7 in October, 1959, and then rose sharply to 21 by February, I960 (Table 2). The greatest increase was ex- hibited by Nereis grubei, which shot from an early average of 8 to 96 individuals / 0.1 m2 by October, 1959. It persisted at or above 80 / 0.1 m2 into February, I960. The populations of Anaitides mucosa and Platynereis sp. more than doubled in this time. Arabella iricolor declined at first, but recovered its earlier population size by February, I960. On the other hand, Eunice rubra, Typosyllis pulchra, Polycirrus sp., and Polydora sp. declined sharply and eventually dis- appeared. Eleven species of polychaetes, not previously present, invaded the Corallina mats. Among the more interesting of these are Lum- brinereis zonata, Anaitides sp., and Nainereis dendritica. None of the invading species came to be represented by many individuals. Among the crustaceans, Balanus tintinnabu- lum displayed a marked population surge from 428 PACIFIC SCIENCE, Vol. XVII, October 1963 a predecline average of 30 individuals to 800 / O. 1 m2 during the decline. This dropped back to an average of 350 living individuals per 0.1 m2 by February, I960. Curiously, Balanus tri- gonus populations remained unchanged through- out the period, as did those of Pachycheles rudis, P. holosericus, and Crangon dentipes. Four spe- cies of crustaceans disappeared during the My- tilus decline: Paraxanthias taylori, Cirolana harfordi , Lophopanopeus leucomanus, and L. diegensis, the latter of which had been repre- sented only by young. These losses were more than offset by the appearance of six species, among which are Pelia clausa and Pugettia dalli. Among the smaller echinoderms, Ophiactis simplex reacted quickly by nearly doubling its population at the 2-m level, going from 180 individuals to 340/0.1 m2 by October, 1959. This dropped back to an average of 154/0.1 m2 by February, I960. The plankton-feeding holo- thuroid Cucumaria lubrica appeared on the 2-m level for the first time. While working with the above data I was struck by the fact that throughout the entire period from November, 1958, to February, I960, over 72% of the numerable individuals at the 2-m level were found in only four species (Table 3). We see, however, that the relative percent- ages that each contributed to the total shifted considerably during the decline. Now let us turn to an analysis of shifts among the larger species existing on Reef 200. Not only did populations of some of these decline significantly, but there were also shifts of popu- lation centers on the vertical axis. One might easily have predicted some, but not all, of these changes. Prior to the decline, 70% of the Pisas- ter giganteus population was found on the upper half of the reef in close association with Mytilus (Table 4). Subsequently this percentage dropped to 46. At the same time the Pisaster population on the entire reef dropped about 45%, presum- ably as a result of emigration to new sites. Most of the Pisaster emigrants must have come from the upper reef. Quite unexpectedly the sea urchin Strongylocentrotus purpuratus appears to have been markedly affected by the Mytilus loss. Its population dropped 34% and data in Table 4 point also to a downward movement on the reef. Strongylocentrotus franciscanus, on the other hand, suffered only a small loss of total popula- tion, and the center of population remained on the lower reef as before. The quadrat densities of the sea cucumber Parastichopus parvimensis dropped 30% during the Mytilus decline, but the center of population remained on the lower half of the reef. DISCUSSION Four kinds of population changes occurred during the Mytilus decline: (1) some species previously present with the mussel increased in numbers and held their gains; (2) some species TABLE 3 Percentage of Total Individuals of Numerable Species on Each Quadrat Provided by Only Four Species SPECIES* PERCENTAGES OF TOTAL POPULATION BY DATEf 11/21/58 1/16/59 9/4/59 10/4/59 2/14/60 Mytilus edulis 61 62 10 5 0.2 Balanus tintinnahulum 2 2 40 46 42 Nereis grubei 0.2 0.7 4 5 10 Ophiactis simplex 11 10 20 22 20 PERCENTAGE OF POPULATION 74.2 74.7 74 78 72.2 NO. OF OTHER NUMERABLE SPECIES 53 50 50 26 68 * Note that these four species accounted for an average of 75% of the total animal population, but their relative contribu- tions to the total changed markedly during the decline. t Data obtained prior to (11/21/58) and during the Mytilus decline on the dates heading columns. All quadrats situated at depths of 2 m and within 3 m of the first. Population Dynamics— -PEQUEGN AT 429 TABLE 4 Relative Population Densities of Large, Motile Species Before and During the Decline of Mytilus Populations SPECIES INDIVIDUALS /M2* Predecline During Decline depth (m) f depth (m)f 2-7 7.1-11.2 2-7 7.1-11.2 Pisaster giganteus 33 1.7 1.2 1.5 Strongylocentrotus purpuratus 3.7 0.5 1.6 1.0 Strongylocentrotus franciscanus 0.5 5.8 1.1 4.8 Parastichopus parvimensis 0.1 0.6 0 0.5 * Average number of individuals per square meter. f Reef is divided roughly into upper (2-7 m depth) and lower parts to show some redistribution of individuals of some species with depth. appeared only after Corallina became predomi- nant; (3) others maintained their populations at previous levels, though most underwent tem- porary fluctuations; and (4) still others de- creased and, in some instances, disappeared. Most of the major population shifts occurred where Mytilus was previously dominant, but a few significant shifts were noted below the 4-m line. The degree of these effects may provide some measure of the closeness of relationship of in- dividual species to the mussel (and to Corallina ), both in positive and negative (inhibitory) ways. Our knowledge of these relationships will be increased significantly when careful studies are made of the feeding habits of the many species involved. Some species increased in numbers simply because space became available and they had the reproductive capacity to spread out. Balanus tintinnahulum exemplifies this; its larvae were available to settle prior to the encroachment of Corallina. It is apparent, also, that Corallina is unable to invade an established Mytilus bed, but when the latter is gone the alga moves in after a few weeks’ lag. Other species increased after this alga had become established, apparently because it provided a more suitable milieu. Quite possibly some species increased as a result of the withdrawal of predators not adapted to living in algal mats. The marked decrease in disaster giganteus resulted from an abrupt loss of food. The shift of remaining individuals to the lower part of the reef is probably related to the pres- ence there of Chama pellucida, an alternate food source. The anemone Anthopleura sp. dropped from an average of 40 individuals to 17/0.1 m2 because of a reduction- of substratum rather than of food. Prior to the decline most anemone in- dividuals lived on mussel valves, with a few living in the rock niches. The number increased on the rock as the mussel declined, but the total available space was not sufficient to sustain the previous populations. The reduction of Para- stichopus parvimensis may be accounted for by the greater production of detritus by Mytilus than Corallina, but this is purely conjectural. The different reactions of the sea urchins Stron- gylocentrotus purpuratus and S. franciscanus to the Mytilus decline indicate that broad generali- zations are not appropriate. Whereas 5'. purpura- tus suffered a severe decline, the other species was unaffected. It is perhaps significant that S. purpuratus existed primarily on the upper half of the reef, whereas S. franciscanus held forth on the lower half. The withdrawal of S. pur- puratus may have resulted from the reduction of niches provided by the Mytilus beds and the fact that Corallina produced a very compact cover over the rocky surface. Some interesting relationships between the numbers of species and individuals occurred during this disruptive change in community structure. These can be ascertained from the last four line items of Table 2. Note, for example, 430 PACIFIC SCIENCE, VoL XVII, October 1963 that while the number of numerable (non- colonial) species per quadrat dropped about 25% during the decline, the total of individuals persisted unchanged for many months. But, as can be seen, by the beginning of the postdecline the total number of individuals in the numerable species fell markedly, in spite of the marked increase in numbers of species per quadrat. We find part of the explanation of this in the fact that when Mytilus was at its peak there was an average of 31 individuals per numerable species (aside from Mytilus) on each quadrat; that during the decline this jumped to 41, and then dropped to 12 in the early postdecline. These phenomena are based upon the following oc- currences: (1) the rapid population increases of a few such species as Nereis grubei, Balanus tintinnabulum, and Ophiactis simplex; while (2) other species were leaving the sites; (3) formerly absent species, principally motile ones (Table 2), invaded the sites, resulting in an increase of 30% over the original number of species; (4) those species that underwent rapid population increases overshot the mark and dropped back; and (5) all of the invading species, at least during the period of this study, were represented by a small number of indi- viduals. It is of interest that no single animal species equalled the former population density of Mytilus . Balanus came closest, but it is only a fraction the size of the mussel. This indicates that the animal contribution to the total biomass in the Hyperbenthal Zone (Pequegnat, 1961) was markedly reduced; the emigrations of the echinoderms mentioned above also account for a considerable loss. Reciprocally, the plant com- ponent must have risen through the Corallina increase. For several reasons the appearance of the Corallina mat and associated species appears to comprise an unstable complex. Judging from the population changes discussed above and from observations made elsewhere in the subtidal and intertidal, one gains the impression that ex- tremely dense growths of Corallina are asso- ciated with disturbed conditions. Though one might expect this predominance of the corallines to be temporary, Dawson (1959) points to the distinct possibility that the corallines encroach upon and later dominate areas subject to pol- lution from human wastes. Hence interest is heightened in obtaining additional samples in the future. REFERENCES Dawson, E, Yale. 1959. A Primary Report on the Benthic Marine Flora of Southern California. Oceanogr. Survey of the Continen- tal Shelf Area of S. Calif. Publ. No. 20. State Water Pollution Control Board. Sacramento, California. Gorsline, Donn S. 1962. Personal communi- cation. HARTMAN, 0. 1944. Polychaetous annelids. Rep. Allan Hancock Pacif. Exped. 3:1-33. 1950. Goniadidae, Glyceridae, Nephyti- dae. Rep. Allan Hancock Pacif. Exped. 15: 1-181. — 1951. The littoral marine annelids of the Gulf of Mexico. Publ. Inst. Mar. Sci., Univ. Texas 2:7-124. Pequegnat, Willis E. 1961. New world for marine biologists. Nat. Hist. 70(4). Preliminary Report on the Marquesan Sardine, Harengula vittata, in Hawaii Thomas S. Hida and Robert A. Morris1 The Marquesan sardine, Harengula vittata, was introduced to Hawaiian waters in eight plantings from 1955 through 1959 by the Bu- reau of Commercial Fisheries in an attempt to establish this species as a supplementary bait fish for skipjack fishing. The details of the first seven introductions have been reported by Murphy (I960) and the eighth by Brock (I960). The eighth introduction comprised an estimated 4,000 sardines ranging from 7.0 to 9.4 cm in standard length2 and averaging 8.2 cm. This brought the estimated total number of sardines introduced to Hawaii to 144,000. All of the releases have been made around the island of Oahu (Fig. 1). SOURCES OF INFORMATION The introduction of the sardines was publi- cized by means of posters and letters. Commer- cial and sport fishermen and game wardens were asked to cooperate in supplying information on sightings and captures. Jars of formalin and labels for recording data concerning captures were supplied to the game wardens and the skip- jack fishermen, whose bait nets were considered a likely source of specimens. From 1956 through I960, 54 samples comprising 336 fish were turned in, the majority of them by skipjack fish- ermen. DISTRIBUTION Recaptures of Marquesan sardines have been reported from six of the eight major islands in 1 U. S. Bureau of Commercial Fisheries Biological Laboratory, Honolulu, Hawaii. Manuscript received May 21, 1962. 2 All of the length measurements appearing in this report are expressed in standard length, which is the distance from the tip of the snout to the end of the hypural. Hawaii ( Fig. 2 ) . The lack of reports from Lanai and Niihau may be due to the fact that skipjack fishermen rarely fish for bait in the waters around those islands. Tabic 1 lists all the sardine recoveries from the Hawaiian Islands. The first recapture was made in Keehi Lagoon, Honolulu, in 1956. Subsequent recoveries were from Barber’s Point and Kaneohe Bay, Oahu, in 1957. In 1958 sar- dines were taken at the islands of Kauai and Maui, where no releases had been made, and also in Honolulu Harbor and Pearl Harbor, Oahu, for the first time. In 1959 the sardine was re- ported from the islands of Hawaii and Kahoo- lawe, and for the first time from Haleiwa, Oahu. The first specimens from Molokai and from Na- wiliwili and Hanamaulu bays, Kauai, came in I960. Sardines have been taken from both the leeward and windward shores of Oahu and Kauai but only from the leeward shores of Hawaii, Maui, Kahoolawe, and Molokai ( Fig. 2 ) . Neither our knowledge of the life history of this species nor our observations of it in Ha- waiian waters are adequate to tell us whether the extension of its distribution from Oahu to the other islands has come about through migra- tion of the adults across the channels or through the drifting of eggs or larvae with the currents. The habitat occupied by the sardine seems to coincide with that of the nehu, Stolephorus purpureas, the most commonly used tuna bait fish in Hawaii. This distribution pattern may be only a sampling artifact, since most of the fishing with gear likely to take sardines is done by skipjack fishermen fishing for nehu. However, because of the chronic shortage of tuna bait in Hawaii, the fishermen are alert for reports of bait supplies even in areas outside the usual nehu fishing grounds. If sardines were present in conspicuous abundance in any accessible area, it appears highly probable that the fishermen 431 432 PACIFIC SCIENCE, Vol. XVII, October 1963 TABLE 1 Records of Marquesan Sardines Collected in Hawaii LOCALITY DATE OF CAPTURE METHOD OF CAPTURE NUMBER CAUGHT NUMBER PRESERVED STANDARD LENGTH (cm) Range Mean Oahu I. Keehi Lagoon 9/27/56 night net1 1 9.97- 9.97 3/27/57 night net1 1 8.50- 8.50 10/4/59 night net1 est. 17 3.59-5.60 4.40 11/3/59 night net1 500 7 5.25-5.68 5.46 11/5/59 night net1 27 4.08-6 07 4.88 11/25/59 night net1 16 4.03-6.46 5.51 12/1/59 (?) 2 3.05-3.51 3.28 7/20/60 day seine est. 3 1 2.92- 2.92 9/14/60 day seine 3 3.39-4.72 4.27 Kaneohe Bay 3/26/57 day seine est. 7 1 8.75- 8.75 4/8/57 night net / 4 8.07-8.75 8.46 6/4/57 night net 1 9.50- 9.50 6/30/58 (?) 1 9.75- 9.75 9/23/58 day seine 1 5.60- 5.60 11/3/58 day seine 1 3.46- 3.46 11/27/58 day seine est. 6 3.53-4.30 3.96 7/20/59 day seine 12 16 8 4.30-13.21 7.66 9/4/59 day seine 2 4.90-8.56 6.73 9/6/59 day seine 2 3.04-3.22 3.13 9/11/59 day seine 1 8.30- 8.30 9/14/59 day seine 2 8.21-8.67 8.44 10/2/59 day seine 1 5.40- 5.40 2/26/60 day seine 1 12.01- 12.01 6/9/60 day seine 2 3.34-3.68 3.51 Barbers Point 9/9/57 gill net 6 11.20-13.50 12.55 Honolulu Harbor 2/28/58 night net 4 7.71-9.40 8.72 6/27/58 night net 1 9.54- 9.55 Pearl Harbor 4/29/58 day seine 3 10.05-10.71 10.45 Haleiwa 9/2/59 day seine 11 4.46-6.10 5.27 6/22/60 day seine est. 5 3.98-5.14 4.54 7/22/60 day seine 20 2 4.71-8.64 6.67 Maui I. Kihei 7/5/58 (?) 1 10 not exam. not exam. 10/7/58 day seine 29 3.49-5.32 4.67 11/3/58 day seine 4 6.32-7.00 6.59 5/19/59 day seine est. 1 9.12- 9.12 5/28/59 day seine o 7 9.10-10.91 9.91 5/29/59 day seine 2 9.29-10.83 10.06 6/2/59 day seine 1 8.56- 8.56 Maalaea Bay 8/13/59 day seine 3 5.62-6.00 5.87 1 The net used in Hawaii for taking tuna bait at night is a rectangular lift net pulled under the fish which are attracted to an underwater light suspended alongside the fishing boat. 2 A "bucket” of bait in the Hawaiian skipjack fishery is estimated to represent about 7 lb of fish. Marquesan Sardine — HlDA and MORRIS 433 TABLE 1 — Continued LOCALITY DATE OF CAPTURE METHOD OF CAPTURE NUMBER CAUGHT NUMBER PRESERVED STANDARD LENGTH (cm) Range Mean Kauai I. Port Allen 9/10/58 night net 6 6.17-6.69 6.41 10/2/58 night net 6 3.90-7.90 6.60 5/24/60 day seine est. 100 3 3.80-7.34 5.05 7/5/60 day seine est. 500 10 4.00-6.87 5.45 7/9/60 day seine est. 100 58 5.85-8.75 7.24 7/23/60 day seine est. 20 7 4.67-5.99 5.54 Hanalei Bay 9/25/58 (?) . 6 6.90-7.90 7.50 Waimea Bay l/P/59 pole-and-line 2 9.80-11.48 10.64 10/14/59 pole-and-line 1 14.62- 14.62 7/23/60 day seine est. 40 2 11.34-13.30 12.32 Nawiliwili Harbor 6/2/60 night net 8 10.57-12.47 11.75 Hanamaulu Bay 9/6/60 gill net 1 12.21- 12.21 Hawaii I. Kawaihae Bay 3/14/59 day seine est. 3 bkts.2 12 7.56-10.05 8.73 Kahoolawe I. SW Point 6/17/59 day seine est. Vl bkt. 11 9.76-12.08 10.83 Molokai I. Laau Point 9/3/60 day seine 5 5.61-6.70 5.80 would find them. The main nehu baiting areas are in the relatively protected waters of bays, harbors, and canals where the water is brackish and turbid owing to the influx of streams and ground water. The Kihei area on the island of Maui and Waimea on Kauai are exceptions, the baiting grounds being exposed to the open ocean and moderate surf action. Baiting, in general, is done close to shore in shallow water, although night-lighting for bait may be done in deeper waters of channels and harbors. Since it appears that the sardine has established itself in the nehu grounds, it is important, from the stand- point of the tuna fishery, to keep a close watch on the population trend of the species in Hawaii and the effect that it may have on the bait supply in the future. ABUNDANCE Our only source of information on the abun- dance of sardines in Hawaiian waters has been the reports of observations by commercial and sport fishermen. The numbers of specimens turned in by fishermen are not a good index of abundance, because often they are only a small and arbitrarily selected fraction of the catch. There have also been times when sardines were seen but not caught. The first observation of a large school of sardines, 20 to 30 buckets,3 was made in Wai- mea, Kauai, late in the summer of 1958 by skip- 3 A "bucket,” the unit commonly used by skipjack fishermen for measuring bait, contains an average of 7 lb of fish. 434 PACIFIC SCIENCE, VoL XVII, October 1963 Fig. 1. Map of Oahu showing areas where sar- dines have been released. jack fishermen. The school remained in Waimea Bay until January 1959; its disappearance from the Bay coincided with a storm which occurred in that month. In March, 1959, a skipjack boat caught 3 buckets of sardines at Kawaihae, Ha- waii, and reported catching 900 lb of skipjack with them, a bait-catch ratio comparable to that ordinarily obtained using nehu. The fishermen commented that the sardines were an excellent bait. In October, 1959, an estimated 500 sardines were caught in Keehi Lagoon, Oahu, by skipjack fishermen. These were used for bait, but no report was submitted on the results. The latest report of sizeable sardine catches came from Port Allen, Kauai, where a catch of 100 sardines was made in May, I960, and a catch of 500 sardines and another of 100 was made in July, I960. Aside from these few instances of fairly large catches, most of the reports from the fisher- men have indicated that the sardine was taken or sighted only in small numbers. Thus, although the sardine appears to be well established in Hawaii, it is apparently not abundant, at least not in areas frequented by fishermen, and is not as yet making any significant contribution as a bait fish. FOOD We examined the stomach contents of 132 sardines caught in Hawaii to study their food habits in their new environment. This examina- tion covered samples representative of all areas from which specimens had been turned in. Only fish with identifiable food organisms were in- cluded in the analysis. We did not consider as food items nematodes, wood, and other foreign matter, or material which we could not readily identify. Thirty-nine (30%) of the stomachs examined were considered to be empty. This high percentage of empty stomachs was prob- ably due to the fact that many of the fish had been held captive for several days in baitwells. In terms of the percentage frequency of oc- currence of various organisms in the stomachs of all samples (Table 2), copepods were first, followed by gastropod larvae, adult and larval shrimp, crab larvae, fish larvae, amphipods, and polychaetes. Copepods were also observed in the greatest numbers. The following list of the numbers and kinds of organisms found in the well-distended stom- ach of a 9.4-cm sardine caught at Kihei, Maui, represents the quantity and variety of food that may be taken by this fish: 689 copepods 24 lucifers 34 amphipods 7 ostracods 30 shrimp larvae 3 crab megalops 27 crab zoeae 3 stomatopod larvae 1 unidentified fish larva In contrast, the stomach of another 9.4-cm sardine appeared filled to capacity with three nehu larvae about 3 cm long. Of the fish examined, 50% were between 3-5 and 6.0 cm long, the rest ranging up to 14.5 cm. Fig. 2. Map of the Hawaiian Islands showing the areas from which sardines have been recovered. Percentage Frequency of Occurrence of Various Organisms in the Stomachs of 93 Sardines Marquesan Sardine — HlDA and MORRIS 435 vaaooaviD ^ i i • ! i I - vaodonaxd : : : : : : : : o : : : : : : : : : : /N 3VA3VT <30dA33T3d 17 6 10 33 \o 33313111 17 100 00 (• — i : : r- : o o : : : : m : : : cm *-i ; : ON VXaVHOATOd cot o */n : r — : cn : : : : m v~\ cn : r-H : i— < : : : : l/N VdOdlHdDtV : irs o : o o m: : cn i-i (N t-=i : oo r-t rr> : r- 3VAHV1 HSI3 12 8 27 100 100 r-~- 3VA3V1 9V33 mi'— • o ro m : o o : m:mooro »-4 : xj< ex vo : vo CN avAHVT * snnav dTtlHHS 8 38 60 17 20 20 80 VO CN 3VA3V1 aOdORXSVO : o 5— i o : m : : o m : : m cn : m : : xp oo : 00 CN vaodadoo 67 100 94 60 100 73 12 100 100 83 CN 00 LOCALITY (No. of fish examined ) Oahu I. Honolulu Harbor (3) Haleiwa (12) Kaneohe Bay (16) Barbers Point (5) Keehi Lagoon (12) Maui I. Kihei (15) Hawaii I. Kawaihae Bay (3) Molokai I. Laau Point (5) Kahoolawe I. SW Point (10) Kauai I. Port Allen (6) Waimea Bay (1) Ail areas combined 436 PACIFIC SCIENCE, VoL XVII, October 1963 The food habits of the larger and smaller sar- dines appeared to be essentially similar, but fish larvae were found only in sardines 8.0 cm long or larger, and there seemed to be a size- associated difference in the composition of the copepod component of the diet. The smaller fish fed more commonly on small cyclopoid cope- pods, such as Corycaeus sp. and Oncaea sp., while the larger sardines had more often been 10 9 8 7 6- 5- 4 3 2 ! : 7 6. 5 4 3 2 0 13 119581 1959 2- 1 19601 n W V .1. JAN. FEB. MAR. APR. MAY JUNE JULY AUG. SEPT. OCT. NOV. DEC. Fig. 3. Size frequencies of sardines collected in Hawaii in 1958, 1959, and I960. feeding on larger calanoids, such as Candacia sp., Labidocera sp., and Pleuromamma sp. The samples were inadequate to carry on a detailed study of food habits among the different areas. As indicated in Table 2, 17% of the stomachs with food had fish larvae in them. Of 34 fish larvae found, 24 (71%) were identified as nehu. The largest number of larval nehu found in a stomach was 6, in a 10.8-cm specimen from Kihei, Maui. The largest nehu found measured 3-3 cm, in a 9.4-cm sardine, also from Kihei, Maui. The food habits of H. vittata in its native environment were extensively studied by Naka- mura and Wilson (ms). They also found cope- pods to be the most frequently occurring or- ganism (79.1%), followed by pelecypods, gas- tropods, barnacle cypris, pteropods, amphipods, and megalops. Fish larvae occurred in only 1.6% of the stomachs. The only distinct difference between the feeding habits of the species in Hawaii and in the Marquesas seems to be the low percentage of occurrence of fish larvae in the Marquesan fish. The food habits of the sardine in Hawaii are similar to those of the nehu, both species feed- ing largely on the crustacean elements in the plankton. Hiatt (1951) found that nehu feed primarily on copepods* barnacle larvae, mysis larvae of shrimps, ghost shrimp ( Lucifer sp.), crab larvae, and palaemonid shrimps. SPAWNING The size of the first sardines taken at Maui and Kauai in 1958 was such that, as pointed out by Murphy (I960), it seemed more likely that they were the products of spawning in Hawaiian waters rather than members of the original transplanted stock. A collection of 29 sardines taken at Kihei, Maui, in October, 1958, con- tained 7 measuring less than 4.3 cm, which was the length of the smallest sardine measured in the release of May, 1958. Subsequently, sardines smaller than 4.3 cm were taken in Keehi Lagoon and Kaneohe Bay in July, September, October, November, and December, 1959 (Fig. 3). In I960, small sardines were caught at Keehi La- goon, Kaneohe Bay, and Haleiwa on Oahu, and at Port Allen, Kauai, in May, June, July, and September. Marquesan Sardine— -HlDA and MORRIS Nakamura and Wilson (ms) have reported that in the Marquesas 8.4 cm, plus or minus 1.7 cm, is the mean length of females at sexual maturity. They considered that year-round oc- currence of ovarian eggs of 0.6 mm or larger diameter indicated that sardines spawn through- out the year in Marquesan waters. Although our Hawaiian samples were inadequate for deter- mination of the spawning season of the intro- duced fish, the occurrence of ova 0.43 to 0.74 mm in diameter in sardines taken in May, June, July, September, and October and the appear- ance of small sardines from May through De- cember lead us to believe that the species also has a prolonged spawning period in Hawaii. SUMMARY Eight releases of the Marquesan sardine, Har- engula vittata, have been made around the island of Oahu in the hope that it would become estab- lished and sufficiently abundant to increase the tuna bait fish supply. An estimated 144,000 individuals were released from 1955 through 1959. Recoveries by skipjack fishermen have been made from all of the major islands except two. The habitat occupied by the sardine seems to coincide with that of the most important native bait fish, the nehu ( Stolephorus pur - 437 pureus ) . The occurrence of young sardines since 1958 indicates that they have spawned success- fully, but the species is not yet making any significant contribution as a bait fish. The sar- dines feed primarily on copepods, and also eat gastropod larvae, larval and adult shrimps, crab larvae, fish larvae, amphipods, and polychaetes. This diet is similar to that of the nehu. REFERENCES Brock, Vernon E. I960. The introduction of aquatic animals into Hawaiian waters. Int. Rev. Hydrobiol. 45(4) : 463-480. Hiatt, Robert W. 1951. Food and feeding habits of the nehu, Stolephorus purpureus Fowler. Pacif. Sci. 5(4) .347-358. Murphy, Garth I. I960. Introduction of the Marquesan sardine, Harengula vittata (Cuvier and Valenciennes) , to Hawaiian waters. Pacif. Sci. 14(2): 185-187. Nakamura, Eugene L., and Robert C. Wil- son. ms. The distribution and biology of the Marquesan sardine. Bu. of Comm. Fish. Bio. Lab., Honolulu. Hawaiian Records of Folliculinids (Protozoa) from Submerged Wood1 Donald C. Matthews2 ABSTRACT: To folliculinids in Hawaii, taken from naturally submerged objects ( Halofolliculina annulata, Ascobius simplex, Metafolliculina andrewsi) and from submerged glass-plate panels ( Metafolliculina nordgardi, Parafolliculina violaceae) , are added those from Douglas fir panels riddled by subsistent teredines and Limnoria: Eufolliculina lignicola, Mirofolliculina limnoriae, and Lagotia viridis. Variations in loricae and moniliform nuclear components are illustrated and discussed. Thus far five species of folliculinids embracing four genera have been recorded for Hawaii (Matthews, 1962). Of these, Halofolliculina annulata, Lagotia simplex, and Metafolliculina andrewsi were taken from submerged naturally- occurring objects, whereas Metafolliculina nord- gardi and Parafolliculina violaceae were taken from submerged glass-plate panels. Douglas fir ( Pseudotsuga taxi folia?) frames supporting these panels (Matthews, 1962, fig. 1), subsequently honeycombed by subsistent teredines and Limnoria, were broken apart; the tortuous burrows and their sequestered organ- isms revealed the folliculinids recorded here. Folliculina lignicola, Faure-Fremiet 1936, La famille des Folliculinidae. Mem. Mus. d’Hist. Nat. de Belg. (Ser. 2), 3: 1129-1175. Eufolliculina lignicola, Hadzi 1951. This slender folliculinid (reassigned by Hadzi [1951] to Eufolliculina lignicola [Faure-Fre- miet}) was abundant in tracheids opened by burrowing gribbles. In riddled wood frames examined November 21, 1961 (corroborated January 11, 1962) E. lignicola was the most abundant, although not the most sequestered, folliculinid. In fact, just outside the burrows, 1 Contribution No. 196, Hawaii Marine Laboratory, University of Hawaii, Honolulu. Manuscript received March 28, 1962. a Department of Zoology, University of Hawaii, Honolulu. attached especially to calcareous tubes of ser- pulid worms ( Spirobis sp. and Mercierella sp.) were numerous folliculinids whose sacs, necks, and moniliform nuclei fell well within those limits prescribed for E. lignicola. Thus, as sug- gested by Mohr (1959: 86) E. lignicola is not restricted to tracheids. Mention should be made, however, that although they were abundant on calcareous tubes of serpulids, none was attached to calcareous walls of abandoned teredo burrows, shells, or pallets, although these apparently af- forded similar attachment potentials. As frequently observed in other folliculinids, sac length, width, and height often varied with site conditions. In young colonies, composed of few folliculinids, ample space resulted in normal sac formation; whereas in old colonies, com- posed of many folliculinids, limited space re- sulted in "abnormal” sac formation. Thus, sacs lying contiguously were usually longer, whereas those laid one on another were usually shorter, etc. Despite their occurrence on serpulid shells, certain restrictive sites seemed preferred. A com- mon restrictive site is illustrated in Figure 1. Although the neck (c) was free and always ex- tended at an angle from the opened portion of a tracheid (b), the sac (/) usually was con- fined, at least in part, in the unopened portion (g). This condition, which resulted in a long narrow sac (up to 1 66g), affected in no way either the length of the neck (c) or the number of its spiral whorls (d) . Thus, regardless of the 438 Folliculinids from Submerged Wood — Matthews 439' i- 165 p Fig. 1. Eufolliculina lignicola, as viewed from the right side showing: a, extended peristomal lobes; h, distal opening of neck; c, neck; d, spiral whorls; e, nuclear components; f, sac; g, unopened portion of tracheid; h, opened portion of tracheid. site chosen, neck lengths varied only between 132/x and l67g with the mean at approximately 159 g. Spiral whorls ( d ) usually numbered six, although occasionally five and seven were ob- served. The diameter of the neck (49g) was approximately that of the unrestricted sac. In reflected light the color of the extended body resembled that of Parafolliculina violaceae (Matthews, 1962), except that perhaps in E. lignicola the wine color was somewhat more intense. However, in transmitted light the body and lorica appeared bottle-green, a characteristic of most folliculinids. As previously reported (Matthews, 1962), both size and number of nuclear components varied in accordance with body contraction resulting from fixation. Thus, in contracted bodies, nuclear components were large (5g) and often numbered as few as 6, whereas in more relaxed bodies, nuclear com- ponents (Fig. le) were small (3g) and often 440 PACIFIC SCIENCE, Vol. XVII, October 1963 numbered as many as 12. The peristomal lobes (a) were long (approximately 11 6(x) and in feeding specimens, were carried high above the distal opening of the neck ( b ). No swimmers were observed. Mirofolliculina limnoriae (Giard 1883) Dons 1927 {vid. Fig. 2 A and B) Freya limnoriae Giard, Bull. Scientific, t.XV: 264-265, 1883. Folliculina limnoriae (Giard) Fragments bi- ologiques XIII. Sur les genres Folliculina et Pebrilla. Bull. Sci. Nord. 3:310-317, 1888. Since Giard s publication of Folliculina lim- noriae, this bizarre folliculinid has been charac- terized both by its habitat (dorsal surface of Limnoria pleotelson or adjacent segments) and by the lateral outpocketings of its lorica (Figs. 2 A,B; 3 A,B; 5 A, B; 6). Although both Mohr (1959: 86), working on Limnoria lignorum of the Pacific coast of North America (Friday Harbor), and Giard (1888:314), working on Limnoria (sp.) of the a b FIGS. 2 A, B. Tracings of Giard’s plates of Fol- liculina limnoria taken at Wimereux showing: a, spherical nucleus; b, extent and region of pouches. Magnification not given. I 1 , : 1 215 )i 1 80 p FIGS. 3 A, B. Mirofolliculina limnoriae, showing: a, empty loricae with lateral pouches. Atlantic coast of France (Wimereux), consider M. limnoriae an abundant species, rarely was it observed on L. ( Paralimnoria ) andrewsi (Cole- man) of Hawaii (Menzies, 1959:10). In fact, not until November 29, 1961, after hundreds of gribbles had been examined, were even the empty loricae (Fig. 3 A, B) observed. This paucity may in part be due to the fact that M. limnoriae temporarily had lost its habitat to another, as yet undetermined, organism; for, almost covering the dorsal surface of the pleotel- son of every third or fourth gribble examined, was a large brown hemispherical cyst (Fig. 4a) which, when opened, liberated two minute un- pigmented dorsoventrally flattened worms. This impression is further supported by the fact that during January, 1962, while the numbers of these cysts decreased, the numbers of M. lim- noriae increased. However, even in old estab- lished gribble colonies (such as those from the riddled hull of the "Seth Parker" at Coconut Island) only one M. limnoriae was taken for every 21 L. (P.) andrewsi examined. Of 25 M. limnoriae taken at random, the length of the sac varied from 133ft to 21 6g, with the mean at approximately 149 ft. The sac at its greatest width, varied from 83 ft to 132g, with the mean at approximately 11 6/x. The neck, which was completely devoid of spiral thicken- ings or proximal lip, varied in length from 49ft to 66ft, with the mean at approximately 59g. Folliculinids from Submerged Wood-— MATTHEWS 441 The diameter of the neck was almost uniformly 50 fiy except for those of certain empty loricae whose necks were ellipsoidal; in such instances the greater diameter ( at right angle to the longi- tudinal axis) was roughly 58 /x and the lesser diameter 2 5 /a. The greatest variation was observed in the number, arrangement, and size of the lateral pouches (Figs. 2 A, B; 3 At B; 5 A, B; 6). Al- though smaller loricae usually exhibited fewer pouches, this was not always the case. For ex- ample, a lorica (Fig. 5 A) with but 6 pouches ( the smallest number observed ) was larger than a lorica (Fig. 5 B) with 16 pouches. (Nineteen was the largest number observed. ) Although each lorica was somewhat swollen at its base, the pouches did not necessarily ex- tend, as did those shown by Giard (Fig. 2 A, B ) , in finger-like protrusions from the swollen base, nor were they bilaterally arranged. Ar- rangement of pouches varied from as few as 2 pouches on one side and 4 on the other (Fig. 5 A) to as many as 9 on one side and 10 on the other (Fig. 6). Figure 5 B is interesting for not only does it show a bilaterally symmetrical arrangement of the 16 pouches, but the anterior 2 (a) are dearly seen through the neck when viewed dorsally; hence similar outpocketings (as .66 mm. Fig. 4. Pleoteison of L. ( Paralimnoria ) andrewsi, showing: a, large hemispherical cyst on dorsal surface. FIGS. 5 A, B. Mirofolliculina limnoriae as viewed from the dorsal surface, showing: a, pouches visible through neck; b, lobes of moniliform nucleus; c, large, lateral pouch. in Fig. 6a) may have been mistaken for the so- called valves mentioned by earlier investigators. Even more variable was the size of the lateral pouches. These ranged from only a few microns (Fig. 6b) to half the width of lorica (Fig. 5 A, c ). And, although large pouches were usually located posteriorly near the base of the lorica, occasionally they were located anteriorly near the neck. Because to my knowledge the process of lorica formation in this species has not been observed, neither pouches nor their variation in number, arrangement, and size can at present be explained. Unfortunately, too many observers have been concerned with the unique lorica and too few with the living organism. Even actual descrip- tions are meager. Giard '( 1888:315), who at- tributed to this species a non-moniliform nu- cleus (Fig. 2 By a) merely states . . [it] is a little smaller than the preceeding species { Fol - liculina abyssorum Giard], of a darker greenish blue with the striation of the integument less apparent. . . Kahl ( 1932 ) is concerned that 442 PACIFIC SCIENCE, Vol. XVII, October 1963 folliculinids of quite varied nuclear components have been assigned to M. limnoriae. This is understandable. Since many different species of Limnoria have evolved in various regions of the world (Menzies, 1959) one would not expect the folliculinids, with which they are so inti- mately associated, to have remained constant. In M. limnoriae, as was previously pointed out for E. lignicola, nuclear components varied with the degree of body contraction following fixation. With peristomal lobes well contracted into the lorica, the nucleus appeared as little more than a dark spherical mass near the point of the body attachment. However, with peristomal lobes re- laxed and carried high above the distal opening of the neck, the nucleus appeared as a string of beads whose size and numbers of components h -i 130 p Fig. 6. Mirofolliculina limnoriae as viewed from the dorsal surface, showing: a, pouch viewed through neck; b, an extremely small pouch. 180 p Fig. 7. Lagotia viridis, as viewed from the right side, showing: a, neck; b, three spiral whorls; c, non- flattened sac; d, spherical nucleus; e, non-spatulate foot. varied. The largest number of nuclear compo- nents (Fig. 5 A, a) observed was nine (and this was not in a completely relaxed specimen). The question of whether variations in nuclear com- ponents and/or variations in number, arrange- ment, and size of lateral pouches have specific value must await further and more extensive investigations. However, as mentioned by Mohr (1959) some correlation seems to obtain be- tween the latitude in which M. limnoriae is found and the degree to which its lorica is branched (compare Giard’s figure of M. lim- noriae from France {Fig. 2 A, B] with my figure of M. limnoriae from Hawaii {Fig. 6] ) . Lagotia viridis, T. S. Wright 1858. Description of New Protozoa. Edinb. New Phil. Journ, New Series, 7:276-281. While not directly attached either on or in riddled wood panels, L. viridis was abundant on and in the empty osseous valves of the prio- nodesmacean mollusk Ostria virginica, which was present on all panels observed. Although, in an earlier publication (Matthews, 1953) Lagotia simplex (Dons, 1917) (reassigned by Hadzi, 1951, to Ascobius simplex ) was also credited with occupying this habitat, there is no Folliculinids from Submerged Wood-— MATTHEWS 443 good reason to confuse these related but quite distinct species. Although both species are de- void of vestibule and closing apparatus ( valves ) and possess spherical nuclei, Hadzi is justified in his reassignment of L. simplex to Ascobius simplex on the basis of its flattened lorica and spatulate foot; whereas in Lagotia viridis , the lorica (Fig. 7c) is not flattened nor is the foot spatulate (e) . For representatives of L. viridis in Hawaii an even more obvious difference is at once apparent. Whereas in Ascobius simplex the neck is exceedingly short (Matthews, 1952: 344), in Lagotia viridis (Fig. 7) the neck (a) extends from the sac and is strengthened by two or three spiral whorls (b). Moreover, while A. simplex is comparatively small, L. viridis is comparatively large, often possessing sac lengths of over 180/a and neck lengths of over 100/a. Thus, to date, folliculinids embracing eight species and seven genera are known from Ha- waii. Accepting Hadzi ’s reassignments, these now stand as: Halofolliculina annulata (which I have not yet taken), Ascobius simplex , Meta- folliculina andrewsi, Metaf olliculina nordgardi (which may prove to be a new species), Para- f olliculina viola ceae, Euf olliculina lignicola, Mi - rof olliculina limnoriae , and Lagotia viridis . It is interesting to note that, whereas Fol- liculinopsis gunneri Dons 1927, reassigned by Hadzi (1951) to P achy f olliculina gunneri, is present on the ventral surface, of Limnoria lig- nomm from the Pacific coast of North America (Mohr, 1959:86), not one specimen has been observed on the hundreds of L. (P.) andrewsi which I have examined. There is every good reason to believe that new records of folliculinids from Hawaii will continue to appear. Re-examination of rather an extensive pagurid collection made over the past 10 years discloses many folliculinids which previously had been overlooked. Although these possess undamaged loricae, their still pigmented but highly contracted bodies are not suitable for specific determinations. REFERENCES Dons, G 1927. Neue und wenig bekannte Pro- tozoen. KgL Norske Yidensk. Selsk. Skrifter. 1927(7): 1-17. Faure-Fremiet, E. 1936. La famille des Fol- liculinidae (Infusoria-Heterotricha). Mem. du Mus. d’Hist. Nat. de Belg. (Ser. 2) Fasc. 3: 1129-1175. Giard, A. 1888. Fragments biologiques XIII. Sur les genres F olliculina et Pebrilla. Bull. Sci. Nord. 3:310-317. Hadzi, J. 1951. Studien fiber Follikulinider. Academia Scientiarum et Artium. Slovenica Biology. 2:1-390. Kahl, A. 1932. Urtiere oder Protozoa. 1, Wimp- ertiere oder Ciliata Spirotricha. P. 474 in F. Dahl, Die Tierwelt Deutschlands, 25:399- 650. Matthews, D. C. 1953. New Hawaiian Records of Folliculinids (Protozoa). Trans. Amer. Micros. Soc. 72:344. 1962. Additional Records of Folliculi- nids (Protozoa) in Hawaii. Pacific Sci. 16(4): 429-433. Menzies, R. T. 1959- The Identification and Distribution of the Species of Limnoria. In: Marine Boring and Fouling Organisms. Dixy Lee Ray, ed. University of Washington Press, Seattle. P. 14. Mohr, J. L. 1959. On the Protozoan Associates of Limnoria. In: Marine Boring and Fouling Organisms. Dixy Lee Ray, ed. University of Washington Press, Seattle. Pp. 84-9 L Wright, T. S. 1858. Description of New Pro- tozoa. Edinb. New Phil. Journ. New Series, 7:276-281. Investigations of Rhinoceros Beetles in West Africa1 Charles P. Hoyt2 In April, 1959, the author was sent by the South Pacific Commission to the countries of Sierra Leone and Nigeria, West Africa, to con- duct an investigation of the natural enemies of the various species of Oryctes (Coleoptera, Scarabaeidae, Dynastinae) occurring there. The object of this study was to find and introduce to the islands of the South Pacific suitable para- sites and predators of these beetles in order to establish a biological control over the introduced Oryctes rhinoceros Linn, which has become a serious pest of coconut palms in the area. In Sierra Leone and Nigeria, coconut palms are confined mostly to village sites; the groves are from 15 to 50 palms, depending on the size of the village. Because of this, nearly all the work was carried out on oil palms ( Elaeis gui- neensis Jacq. ) which, together with the Raphia palms of the swamps, are the most important hosts of the species of Oryctes found. The oil palms occur in an extensive belt that follows the coast of West Africa and extends inland nearly 100 miles in some places. There are two distinct climatic seasons in West Africa, and these greatly affect the insect populations. The wet season begins generally at the end of April and lasts until sometime in September. By the end of November the dry season has set in, and from then until April little or no rain falls. At times the rains will cease in August and commence again in Sep- tember, but during I960 this type of rainfall did not occur in Nigeria. Insect activity begins with the coming of the rains in April or May and gradually ends in November or December. There are, of course, many exceptions to this; some species seem to be found only during the dry season and others congregate along streams and in swamps, giving a false impression of their absence. 1 Presented at the Tenth Pacific Science Congress, Honolulu, Hawaii, 1961. 2 Technical Officer, Pe$ts & Diseases, South Pacific Commission, Noumea, New Caledonia. Manuscript received March 19, 1962. During the dry season in Sierra Leone and in parts of Nigeria, it is the practice to burn off the bush to clear the land for planting. The oil palms in these areas are usually not affected by the fires, and the larvae of Oryctes and other beetles within the standing rotten trunks are not harmed. The low-lying land of the southern part of eastern Nigeria is divided by numerous slow- moving streams which give rise to large swampy areas as they wind their way towards the sea. Because of this and the heavy rainfall, extensive burning is not possible. In the swamps are dense stands of Raphia palms which flower and die, providing a continuous supply of breeding sites for Oryctes. The higher ground between the streams and swamps supports large stands of oil palms, both cultivated and wild, from which come the main export of the area in the form of palm oil and kernels. PREDATORS AND PARASITES ENCOUNTERED Neochryopus savagei Hope (Coleoptera, Cara- boidea, Scaritidae) The large scaritid beetle Neochryopus savagei was first found in Raphia palms growing in the swamps at Umudike, near Umuahia, eastern Nigeria. In April, I960, an adult beetle of this species was found inside a standing rotten Raphia palm trunk where it was actually en- gaged in feeding on a larva of Oryctes ohausi Minck. A search of other rotten palm trunks in this area turned up two large larvae of Neo- chryopus. Afterwards, additional adult speci- mens were recovered from the debris which normally collects in the old dead leaf bases just below the crowns of oil palms left by workers engaged in cutting the bunches of fruit in a grove of palms on the Agricultural Station. This accumulation of rotten organic matter provides a breeding site for Oryctes sjostedti Koble, and Neochryopus breeds here and preys on these dynastid grubs. 444 Rhinoceros Beetles in West Africa — Hoyt 445 Data obtained from the examination of field- collected larvae of N. sawagei indicated that there were five instars. The length of the fourth instar period of a single larva provided with a constant supply of small cetonid and Oryctes grubs was slightly over 2 weeks. The pupal period was 10 days. Several of these scaritid larvae lived 3 weeks without food. Considerable difficulty was experienced in hatching the eggs laid by captive adult beetles and rearing their larvae in the laboratory because both stages were frequently attacked by fungus. In the case of the larvae this fungus appeared to be a Metar- rhizium. The adult beetles mated readily in small ciga- rette tins (14 cubic inches in capacity), which were partly filled with moist rotten palm fiber. The eggs of these scaritids were found in the me- dium, there being one egg per beetle. In a single instance a second egg was laid 3 days after the first had been discovered. The eggs were oval in shape, about 2 mm long and V2 mm wide. When laid they were sticky, and quickly became covered with bits of frass. This made them hard to find and, as no special attempts were made to discover them, no doubt many were missed. Adult female beetles which were dissected con- tained a maximum of four large eggs. Both the adults and larvae of N. savagei readily attacked, killed, and consumed cetonid, Oryctes, and other dynastid grubs of a size up to 10 g in weight (the size of an early third instar Oryctes larva). The larvae of the palm weevil, Rhynchophorus phoenicis F., were of- fered to the adult beetles and while these were usually killed, the beetles did not seem to feed on them to any extent. The adult scaritids survived for long periods without food, the maximum period noted being about 4 weeks. The adult beetles are winged and are strong fliers. Occasionally they are attracted to lights at night. The sexes are readily distinguishable. The mandibles of the male are elongated, and the distal portion is without well-developed hori- zontal teeth. Those of the female are slightly broader and shorter, and the left mandible has a broad, flattish, horizontal inner tooth which forms a subapical notch. The mandibles of the males are apparently adapted for grasping the female around the back of the head during copulation. However, both sexes feed readily on scarabaeid grubs. The sex ratio was found to be 1 to 1. Usually one or two adult beetles were found in the debris of an oil palm, but on one occasion four were taken in a single site. No figures were obtained on the sexes collected from individual palms; the ratios were calculated from total field collections. The majority of the beetles were collected from the oil palm debris; however, a fairly large number of both adults and larvae were also found in the rotten Raphia trunks. The fauna of the oil palm trash and of the standing rotten trunks was, with one exception, identical. The exception was Oryctes sjostedti, which was only found breeding in the oil palms. However, it seems safe to regard this site as a sort of elevated trash heap rather than an unique environment. At Umudike the beetles were kept in round, 50-cigarette-size tins. At first only one adult was put in each tin, but later, to conserve space, two were placed in each container. Rotten palm fibre was added to each tin to give the beetles something in which to dig. After about 1,400 specimens had been collected at Umudike, they were taken to Ibadan and there packed for ship- ment. Plywood trays divided by partitions into 2 -inch cubes were used to hold the beetles in transit. Each cube was partly filled with moist wood wool. A single scaritid was placed in each compartment, and the tray was covered with -inch-mesh wire screen held down with staples. Three or more trays were then packed into a heavy polythene bag to prevent loss of moisture. The trays in their bags were packed into heavy cardboard boxes which were wrapped with paper. From 4 to 6 boxes made up a ship- ment. The insects were sent from the Ikeja air- port via Paris to Fiji by air freight. Cigarette tins were used as containers for the first two or three shipments, but these were heavier than the plywood trays. During the shipments, there were always more beetles than containers. Usually, - enough males were discarded to bring the sex ratio to 1 male to 3 females instead of the normal 1 to 1. An ideal container would have been an alumi- 446 num box of about 8 cubic inches capacity, and had these been available in quantity many more insects could have been sent. An attempt was made to keep N. savagei adults together in a large tin, but this was un- successful due to cannibalism even when 8 cubic inches were allowed per beetle. At first it was thought desirable to provide food for the insects during the time they were held in Nigeria. Later it was found that they fared much better if they were not fed. The grubs put into the tins were only partly con- sumed, and the rotting remains in the small containers evidently created a most unfavorable environment. Dorylus ( Anomma ) nigricans 111. (Hymenoptera, Formicidae, Dorylinae) Throughout the search for parasites and pred- ators of Oryctes in West Africa it was noted that the larvae of the rhinoceros beetles occurred in large numbers in standing rotten palm trunks but not in fallen logs. In Sierra Leone this situa- tion was often observed, but no experiments were made to determine why the fallen logs were not used by the beetles. It was thought that possibly the principal species encountered, Oryctes owariensis Beauv., unlike other mem- bers of the genus, did not favour such sites. On Moor Plantation, Ibadan, Nigeria, an experimental grove of coconut palms was found to be infested by a fair number of Oryctes monoceros Ol. This species was found breeding inside the bases of old coconut palm stumps and in standing rotten coconut trunks, the products of several severe lightning strikes in the grove. However, in the numerous rotten trunks which had been felled, there were no Oryctes larvae nor, for that matter, any larvae of any of the numerous species of cetonids and small dynastids which usually occur with Oryctes in West Africa. There were clear signs that, before the trunks had been felled, a considerable amount of breed- ing by Oryctes had taken place within them. It was also noted that the hollowed interiors of these fallen trunks contained much less frass than would have been expected. Specimens of adult rhinoceros beetles from this grove were sent to the British Museum (Natural History) with a request that they be PACIFIC SCIENCE, Vol. XVII, October 1963 compared with specimens of Oryctes monoceros from East Africa. This was done, and the Ibadan species was found to be identical with the East African 0. monoceros . To determine whether the fallen trunks were suitable as breeding sites, four sections were selected and set upright. After these had been in position for about 3 weeks, they were split open. Seven adult beetles of 0 . monoceros and 14 eggs were collected from them. Five sections of the fallen trunks were care- fully split lengthwise, and new hollows were either carved out by hand or the existing hollows were utilized. These prepared sections were stocked with Oryctes larvae, and frass was packed into the hollows around them. The two halves were then brought together and fastened in place with bands of wire. Some sections were left lying on the ground, others were raised slightly at one end. In addition to the logs, a sawdust pile was established and stocked with larvae of O. boas F. and a few O. owariensis . For all of these trials only third-instar larvae were available. The following events took place: 1. 15 November 1959. One log was entered, and the larvae were eaten by a small greyish rat. 2. 30 November 1959. All the logs were en- tered, the frass was removed, and the larvae were eaten by a swarm of "driver” ants, Dorylus ( Anomma ) nigricans. 3. 14 December 1959. The logs, having been restocked with Oryctes larvae, were again en- tered, the frass was removed, and the larvae were eaten by a very large swarm of "driver” ants, Dorylus ( Anomma ) nigricans. The saw- dust pile was also entered by the ants and about one-third of the grubs was eaten. 4. 10 January I960. Additional logs were added on 3, 4 January and a small section of log which could be moved about with ease was prepared. This was placed in the dry bed of a stream, about 50 m distant, where "driver” ants had been observed. On the evening of 10 Janu- ary the ants invaded this log and also entered all the other logs in the experimental area. One of these logs was constructed so that the larvae were in a central frass-packed hollow which had no openings at either end. The only means of access to this hollow were the holes left in the Rhinoceros Beetles in West Africa- — Hoyt 447 trunk by the workings of xylocopid bees, a natural feature of all of the rotten logs in the grove. The ants entered through these holes, removed the frass, and ate the Oryctes larvae, leaving only the head capsules. 5. 11 January I960 . The logs which had been invaded by the "driver” ants on the night of the 10th were broken up to determine if any of the Oryctes larvae had escaped. In three of the logs in the upper breeding site experimental area a few intact larvae and pupae were found, but in the log located in the stream bed nothing remained. The following results were obtained from the logs in the upper area: One log which had been stocked with 7 larvae contained 1 liv- ing larva and 1 pupa; one log which had been stocked with 10 larvae contained 2 living larvae and 1 pupa; one log with no access from either end which had been stocked with 6 larvae con- tained 1 living larva and 1 pupa. The area was visited daily and the logs were inspected at intervals of 4 to 5 days to determine if the larvae were alive. The last stocking of the logs with larvae was on 16 December 1959. In November, the end of the wet season in Nigeria, the ants were active throughout the area. However, after the end of the rains the ants moved into stream beds; by January, only invasions of short duration were made into nearby areas. The invasion of the experimental site on 30 November lasted about 14 hr. The last sortie on 10 January was only about 7 hr long. Another species of Dorylus ( Anomma ) was found in the coconut grove; this ant was never observed to enter the logs. No attempts were made by either species of Dorylus ( Anomma ) to enter the tops of the stumps or to climb up the standing trunks. The ants were able to climb vertical surfaces but did not seem to do any work, such as removing frass, while on them. When D. nigricans entered the rotten logs, they did so by choosing holes which were more or less horizontal. They always used these holes to bring out the frass from the interior. Ochryopus gigas Schio. (Coleoptera, Caraboidea, Scaritidae) Occasionally the very large scaritid Ochryopus gigas was found both in the debris accumulated below the crowns of oil palms and in standing rotten palm trunks. During April and May, I960, adults of this species were taken around lights at night. However, this species was never found to be common and its larva was not dis- covered. In the laboratory O. gigas attacked, killed, and consumed both larvae and adults of Oryctes. Scolia sp. ( Hymenoptera, Scoliidae) In the oil palm debris and in rotten Raphia trunks large Scolia cocoons were frequently found. Many of those taken from these sites had already hatched. Sometimes the head capsule of the host grub would be found entangled in the silk of the cocoon, and from this it was de- termined to be parasitic on Oryctes ohausi and O. sjostedti. The adult wasps were never observed in the field, but two males and two females were bred out in the laboratory. The female specimens were sent to the British Museum (Natural His- tory), where they were identified only as Scolia sp., "not in British Museum collection.” Judging from the number of empty cocoons, this parasite was thought to be rather common, but large-scale collecting efforts resulted in only about 65, nearly half of which contained dead pupae. Nevertheless, the cocoons were sent to Fiji where an attempt was made to rear the wasps. Unfortunately, this has proved unsuccess- ful, and the wasps which did hatch died in the laboratory. OTHER PREDATORS AND PARASITES 1. Alaus ? sp., Calais sp. (Coleoptera, Ela- teridae) 2. Morio guineensis Imh. (Coleoptera, Caraboidea, Carabidae) 3. Genus ?, species ? (Diptera, Tachinidae) 4. Platymerus higuttata Stal (Hemiptera, Reduviidae ) 5. Genus? species? (Diptera, Tabanidae) 6. Genus? species? (Araneida, Aviculari- idae) 1. Large elaterid larvae were found inside the standing rotten oil palm trunks in close proxim- ity to Oryctes larvae in both Sierra Leone and 448 PACIFIC SCIENCE, Vol. XVII, October 1963 Nigeria. They were never common, and in Sierra Leone only 24 were collected from over 500 rotten oil palm trunks. Specimens were de- termined at the British Museum to be perhaps a species of Alaus. A smaller elaterid larva found in a rotten oil palm was reared on Oryctes and cetonid larvae, the adult beetle being identified as Calais sp. 2. Occasionally larvae and adults of a small carabid beetle, Morio guineensis, were found in the frass in the rotten palm trunks. In the laboratory these readily consumed first instar Oryctes larvae. 3. Large numbers of the larvae of Oryctes boas F., collected from a manure heap on the farm of the University College, Ibadan, were reared in captivity. These were dissected when- ever they showed signs that something was wrong with them. One, a prepupal stage larva, died suddenly and was cut open. In the abdomen was a very large larva of a tachinid fly which had also perished. Following this discovery, some 1,000 Oryctes boas grubs were collected from this manure heap, but no other parasites were recovered. 4. Neanides and adults of the large reduviid, Platymerus biguttata, were found fairly fre- quently in the open hollow tops of the standing rotten oil palm trunks in Sierra Leone during April, May, and June. Laboratory trials were conducted to see if this bug would prey on adult Oryctes owariensis, but without result. Even adults which had been held without food for nearly 2 weeks showed no interest in this large rhinoceros beetle. During May and June, 1959, Platymerus biguttata came to lights at night on several occasions. No other Coleoptera than Oryctes owariensis were tried as prey for this reduviid.2 5. In a standing rotten coconut palm trunk on Moor Plantation, Ibadan, three tabanid larvae 2 The author is at present engaged in a study of the allied species, Platymerus rhadamanthus Gerst., which does attack the adults of Oryctes monoceros Ol. How- ever, old adults which have accumulated large fat deposits usually refuse to attack anything. At the time Platymerus biguttata was investigated it was not known that the fifth-instar neanides would also at- tack adult rhinoceros beetles and, perhaps because of their more rapid metabolism, are generally easier subjects with which to work. were found in the frass along with the grubs of Oryctes monoceros. These readily attacked and killed young scarabaeid larvae in captivity. A week or so later all three pupated and after 2 weeks the adult flies emerged. Unfortunately, the specimens sent to the British Museum were badly broken in transit and were not able to be identified. The tabanids were bright orange with a large purple band across each wing. 6. In Sierra Leone, both inside the hollow tops of standing rotten oil palm trunks and in the crowns of living oil palms, a large aviculariid spider was frequently encountered. This fear- some-looking arachnid evidently attacked the oil palm climbers, who had their own remedy for the effects of its bite. Experiments were made to see if this spider would kill adult Oryctes owariensis, but the results were negative. In Nigeria this palm tarantula seemed to be unknown to the people and no specimens were found. Coleolaelaps sp. (Acarina, Laelaptidae) Several investigators have reported that mites feed on the eggs of Oryctes. Venkatraman re- ported a Coleolaelaps sp. which "apparently fed on Oryctes eggs” in Ceylon. Surany reported several instances of mites feeding on Oryctes eggs and stated that they were important under certain circumstances in its control. Unfortu- nately, Surany did not identify the mite in question in his report.3 Clusters of mites around eggs of Oryctes owariensis were frequently observed in Sierra Leone from June onward throughout the rainy season in 1959. These mites were found on eggs enclosed in balls of frass formed by the adult beetles as they oviposited in the rotten wood inside the standing rotten oil palm trunks. It was noted at the time that eggs which had mites on them never hatched, while eggs in the vicinity without mites hatched readily. These mites were collected and sent to the British Museum where they were identified as Coleolaelaps sp. Adult mites of this species were common on the larvae of Oryctes owariensis , O. boas, and O. monoceros, but were encountered only rarely on the adult beetles. 3 Diseases and Biological Control in Rhinoceros Beetles. South Pacif. Com. Tech. Pap. 128. I960. Rhinoceros Beetles in West Africa — Hoyt 449 In Nigeria, Coleolaelaps was again found on the eggs of Oryctes boas in March, I960, follow- ing a few early rains. The Oryctes eggs were taken in a large manure heap on the farm of the University College, Ibadan. Mite-infested eggs were collected and placed for observation in covered shallow glass dishes filled with a manure and frass mixture. At the same time newly laid eggs were obtained from captive specimens of Oryctes boas. Mite eggs and im- mature and adult mites were transferred from the originally infested, field-collected eggs to the newly laid eggs and to uninfested, field-collected eggs. It was noted that Oryctes eggs which had mites or mite eggs on them were all slightly discoloured and at times showed brownish spots. When mite eggs were transferred to non- discoloured, field-collected eggs, or to newly laid eggs, the mites hatched and in a short time made their way throughout the medium to the original egg from which they had been trans- ferred. Here they gathered and fed on the yolk of the now partly collapsed egg. The adult Coleolaelaps were never observed to feed on either nondiscoloured, field-collected eggs or on the newly laid eggs, but they fed readily on the discoloured ones and on a non- discoloured egg which was punctured with a needle. None of the discoloured eggs hatched and, with the exception of the punctured, nondis- coloured egg, both newly laid and field-collected ones produced normal larvae. No further tests were carried out with Coleo- laelaps due to the discovery of other predators which appeared to be more promising. OTHER MITES ASSOCIATED WITH Oryctes In all, some 14 species of mites (Acarina, Uropodina, and Canestrinidae ) were found as- sociated with the larvae and adults of Oryctes in Nigeria. Two of these, a canestriniid and a species of the Uropodina, were found commonly on the adults of Oryctes monoceros and O. owariensis. The canestriniid was found in large numbers on the first three abdominal tergites beneath the elytra. The uropodinid was taken wandering about both on the venter of the beetle and under the wing cases. At times both species of mites were found heavily infested with a fungus resembling closely some of the Laboulbeniales. The fungus did not appear to harm them, and gravid females were frequently observed to produce normal eggs in spite of heavy infestations. The species of Uropodina found on Oryctes was also collected in numbers from the adults of Neochryopus savagei. SPECIES OF Oryctes ENCOUNTERED ( Coleoptera, Scarabaeidae, Dynastinae) 1. Oryctes monoceros Olivier Both in Sierra Leone and in Nigeria Oryctes monoceros was always found in places where there were coconut palms in the immediate vi- cinity. On Moor Plantation, Ibadan, this species was taken breeding in standing rotten coconut trunks and in the bases of coconut stumps. With the exception of Moor Plantation, however, O. monoceros was rare in all of the places visited. 2. Oryctes boas Fabricius In Sierra Leone Oryctes boas was commonly encountered breeding in village compost heaps, but in similar sites in Nigeria it was rare. How- ever, in Nigeria a large manure pile was found at the farm of the University College, Ibadan, in which there were thousands of larvae, pupae, and adults of this species. In March and early April numerous adults and eggs were found in this site, but at other times mainly larvae of various instars were present. These findings seemed to indicate that, while there was some overlap of generations, nevertheless there was a definite yearly cycle. Recently an experimental planting of oil palms in Northern Nigeria was reported to be rather severely attacked by an Oryctes , and specimens extracted from the crowns of these palms proved to be O. boas. 3. Oryctes owariensis Palisot de Beauvois Oryctes owariensis is a large species, the well- developed adults being larger than the other Oryctes species in West Africa and larger than 0. rhinoceros of Southeast Asia and the Pacific 450 region. Probably because of its size, the species has been confused with O. gigas of Madagascar. Oryctes owariensis breeds, as far as is known, only in rotten wood and prefers rotten palm logs. It was found commonly in Sierra Leone and Nigeria in standing rotten oil palm trunks, standing rotten coconut trunks, and in rotten Raphia palm trunks. Usually this was the only species of Oryctes present in these sites, but sometimes the larvae of O. monoceros or O. ohausi occurred with it. In Sierra Leone an adult specimen was extracted from the crown of an oil palm. Observations in both Sierra Leone and Nigeria seemed to indicate that this beetle did not prefer coconut palms as a host plant. 4. Oryctes sjostedti Kolbe As has been stated previously, this species was only taken breeding in the debris accumu- lated in the dead leaf bases just below the crowns of oil palms. In this site the maximum number of larvae found was six, all third in- stars. Pupae were not infrequently encountered, usually in a cavity within one of the old leaf bases; the adult beetles were also found both within these and in the debris enclosed by them. Adults were taken at mercury vapour lights fairly7 frequently, but never in numbers; one or two per night was the largest catch. This species was most common at Benin, Nigeria, on the West African Institute for Oil Palm Research station and at Umudike in eastern Nigeria. 5. Oryctes ohausi Minck Perhaps this species has been confused with O. monoceros in the past; anyway when speci- mens were sent to the British Museum for de- termination, it was surprising to learn that no examples of this species were in the collection. 0. ohausi was commonly found breeding in the standing rotten trunks of Raphia palms. Adults, larvae, and pupae were found in fair numbers in these sites, and not infrequently in company with O. owariensis. There was an average of eight larvae per trunk during April, I960, at Umudike in eastern Nigeria. A single adult male was taken in a rotten oil palm stump on Moor Plantation at Ibadan. PACIFIC SCIENCE, Vol. XVII, October 1963 6. Oryctes erehus Burmeister This species has been recorded a number of times from West Africa but was not found during this investigation. NOTES CONCERNING ADULTS OF Oryctes All of the investigations conducted in West Africa on the adults of the various species of Oryctes were carried out with the objective of finding parasites of the mature beetles. The main difficulty in searching for parasites of the adults is to get large numbers of beetles which have been exposed in nature. When ma- ture beetles are taken from rotten logs, a fairly high proportion of them are recently emerged and have never been outside the breeding site. If Oryctes flew readily to lights, the problem would be simple. However, it is not particularly attracted by light, although recently some suc- cess has been reported with a special ultra-violet lamp used in the Pacific. In Western Samoa split logs laid flat on the ground are used to trap adult rhinoceros beetles. Each trap consists of four to eight logs each about 4 ft long. A similar type of trap was tried in Nigeria without success. Another type of rotten log trap was devised which gave fairly good results in Nigeria. This was made of a foot-tong section of coconut trunk through which a hole was bored from end to end. This section was placed on top of a No. 10 can which was either sunk into or rested on a coconut stump. The beetles landed on the top of the log, crawled down the hole and fell into the can underneath. The trap caught quite a few Oryctes monoceros , the largest single catch being three. Efforts were made to use captive beetles to attract others. No positive results were obtained along this line, but it was found that 0. owari- ensis had two distinct flight times, one in the evening and the other in the early morning. Furthermore, beetles which did not fly in the evening flew in the morning, and beetles which did fly after sundown did not fly later on. No parasites were found in the adults of Oryctes, but the numbers examined were prob- ably not high enough to give any reasonable hope of success. Rhinoceros Beetles in West Africa — HOYT 451 DISCUSSION AND CONCLUSIONS In both Sierra Leone and western Nigeria predators and parasites or Oryctes larvae were not at all common. The only controlling factor of this nature seemed to be periodic invasions of the fallen logs by "driver” ants. Standing rotten palm trunks contained large numbers of Oryctes in all stages of development. It should not be supposed, however, that there were nor- mally large numbers of these breeding sites in any one small area. Further, not all of the trunks contained Oryctes. It was noted that rotten trunks with long black fibers in their interiors seldom supported anything at all. Perhaps the form of decay, fungal or bacterial, made these unattractive to the beetles. In areas like one reported in western Nigeria where an old oil palm plantation had been poi- soned off and the dead palms left standing, serious damage was caused to the young palms planted between them by Oryctes, which bred in the rotten trunks. Parasites and predators were found readily only in the regions of the deltas of the Niger and Cross rivers of eastern Nigeria. While the same species of these are wide-spread through- out West Africa, they were common in this particular area. It is thought that the practice of burning off the bush, which is done on a large scale in western Nigeria and Sierra Leone, plays an important role in limiting the numbers of predators and parasites. The land of the delta areas is too swampy for burning. Oryctes is not affected by bush fires on account of its habits. It is, of course, logical to assume that the long dry season, together with the absence of numerous streams and swamps, probably also plays a major role in limiting populations of parasites and predators which have fairly short life cycles. No doubt it does, but this and burn- ing complement each other, and it is difficult to say which is the more important. Under West African conditions it appears that many of the populations of Oryct^-%. re limited by the number of available and suitable breeding sites, and that their predators and parasites are limited by unfavorable environ- mental conditions which are determined largely by cultural practices and climatic conditions. Characteristic Features of the Volcanism of the Siberian Platform V. S. Sobolev1 Widespread volcanism is characteristic of the Siberian platform, and was especially intense during uppermost Paleozoic and lower Mesozoic time. The Siberian traps, which occupy an area of more than 1,500,000 km2, are best developed there. These are effusive and hypabyssal rocks of basalt-dolerite type, closely resembling trap rocks in other parts of the globe, especially the Karroo dolerites of South Africa. The rise of trap magma began in the upper Paleozoic (Permian or even as early as Upper Carboniferous) time, and reached its climax in the Lower Triassic period. It was accompanied by the ejection of much pyroclastic material, which formed a thick series of tuffs. Lava sheets and hypabyssal intrusions of various kinds and sizes were formed. The process of volcanism was rather com- plicated, and at present M. L. Lurie and V. Z. Masaitis distinguish five volcanic phases and 13 separate intrusive complexes, each having its own specific features and pattern of develop- ment in various parts of the platform. In spite of this, however, the magma had some charac- teristic features over the entire area, notably an iron content somewhat higher than is usual and an especially rapid increase of relative iron con- tent during the process of crystallization dif- ferentiation. The increase of the iron content of the femic minerals in the process of crystalliza- tion prevails over the conventional reaction series of Bowen. For example, olivine of early formation contains about 20% fayalite, that characteristic of the usual type of traps contains about 40% fayalite, and the iron content of olivine in pegmatoid veins is as high as 80%. The residue of the differentiation is as a rule micropegmatite, either in the mesostasis, or in some cases forming veinlets of granite compo- 1Siberian Division, Academy of Sciences, Novosi- bizsk, U.S.S.R. Manuscript received April 19, 1962. sition. However, such veinlets are quantitatively very small. Much rarer is the formation of alka- line rocks, such as teschenite, in the last stages of differentiation. The above characteristics of crystallization differentiation (rapid change of the iron con- tent of femic minerals and the subordinate role of the discontinuous reaction series ) are typical not only of the trap formations, but also of the deeper-seated magmatic complexes of the plat- form. The gabbro-anorthosite-granite complex of the margin of the Russian platform, with its characteristic granites of Rapakivi type, be- longs to this group. These characteristic features distinguish complexes of this type from the typical granodiorite complexes of orogenic zones, in which crystallization largely corres- ponds to the well-known Bowen series. Relative increase of the iron content of femic minerals is there much slower, as is clearly seen by com- paring it with the change in piagioclase com- position. Along with the trap formation, in part simul- taneously and in part a little later, another type of volcanism developed widely on the margin of the platform, with the formation of ultrabasic and alkaline rocks. Differentiated effusive and intrusive complexes were formed in some re- gions, kimberlites in others. A typical example of the differentiated com- plexes can be seen in the northern part of the Siberian platform in the area of the so-called Gulinski intrusion. The effusives range from meimechite, the closest extrusive analogue of true intrusive ultrabasite, to different kinds of alkaline basaltoids containing nepheline and piagioclase. Among the intrusives are all kinds of rocks from dunite to various alkaline rocks rich in nepheline. Carbonatite is also present. E. L. Butakova (1956) and Y. M. Sheinman (1955) have shown that the volcanic rock for- mation here was largely simultaneous with erup- 452 Volcanism of the Siberian Platform — Sobolev tion of the trap magma. There are traps both older and younger than the alkaline rocks. Kimberlite has been formed extensively in the northeastern border of the platform, more than 100 pipes and dikes being known there at present. For some kimberlite bodies the same sort of relationship to the traps has been es- tablished as for the alkaline basaltoids. Some of the kimberlite is not younger than Permian in age, since pyrope and diamonds from it occur in the Upper Permian deposits. However, there undoubtedly are younger kimberlites also, for a fragment of a belemnite of Upper Jurassic or Lower Cretaceous age was found in one of the pipes. Kimberlite, as a magmatic rock, belongs to the ultrabasic group, its composition ranging from a form nearly devoid of alumina and alka- lies to one rather high in AI2O3 and especially high in K20 in mica-rich varieties. The prin- cipal mineral is always olivine, containing up to 10% Fe2Si04, and is present in at least two generations: large crystals, commonly partly re- sorbed, and small idiomorphic microphenocrysts. Phlogopite occurs in idiomorphic plates and ranges widely in quantity. It is unquestionably magmatic. Pseudomorphs of pyroxene microlites are sometimes seen in the vitreous matrix. The latter is always -altered. In the northern regions fine-grained monticellite has been found in kim- berlite for the first time, chiefly in dikes. Nephe- line also is supposed to be present. Pyrope, and probably picroilmenite and chrome spinel, com- monly belong to the first generation of pheno- crysts. Perovskite is a later accessory. As in South Africa, the kimberlite has a brec- ciated structure and is contaminated by frag- ments of various sorts of rocks. There are, on the one hand, typical pyroclastic rocks filling explosion pipes, and, on the other, magmatic breccias with various contents of xenoliths. Since the rocks have been altered ( serpentinized and carbonatized ) , it is not always possible to prove the presence of magmatic cement. The fragments of other rocks may be sub- divided into: 1) Fragments of ultrabasites and eclogites whose origin is in some way connected with the origin of the kimberlite itself; 2) Fragments of rocks picked up by the 453 magma from (a) the crystalline basement for- mations, and (b) the sedimentary cover. Fragments of the first type include various ultrabasites — olivinites, peridotites, and others - — often containing pyrope as well as typical eclogites. The discovery of diamond-bearing ec- logites, resembling the well-known eclogite xenolith found by Bonney (1899) in South Africa, is of particular interest. Together with such eclogites brought up from great depth, there are eclogites and eclogite-like rocks (con- taining plagioclase) picked up from the crystal- line basement and formed by eclogitization of hypersthene schists. The fragments of rocks picked up by the magma vary widely in quantity and composition. Xenoliths of gneisses and schists are abundant in several of the pipes. This can be taken as proof that the "explosion” that formed the pipe took place at a level lower than the base of the sedimentary cover. Allowing for this, and taking into consideration the geophysical data on the depth of the crystalline basement in the area and also the thickness of the rocks since removed by erosion, we can say that the explosion took place at a depth somewhere between 2 and 4 km. The depth is greater than in the case of the formation of the trap necks, which was 0.5 to 1 km. As in South Africa, xenoliths of rock forma- tions that occur at much higher levels (some- times several hundred meters higher ) are found among the fragments, proving that there was not only an ascending but also a descending movement of the material in the pipe. The synchroneity of their formation has led to the hypothesis of a genetic connection be- tween the ultrabasic rocks and kimberlites and the trap magma of the Siberian platform. Petro- graphic data, however, do not support this hy- pothesis. The olivine of the kimberlites and the meimechites contains only 10% fayalite, and this is proof enough that these rocks could not have originated as a result of differentiation of the trap magma. The author quite agrees with Y. M. Sheinman’s (1955) suggestion of the formation, in this case, of magmatic chambers at much deeper levels than those of the trap magma. The development of trap volcanism on such 454 PACIFIC SCIENCE, Vol. XVII, October 1963 an enormous scale leads us to the conclusion that regional melting of the basalt layer took place here, probably in its upper part. However, no- where in the platform did the magma chambers reach the sialic shell, and the small granitoid veinlets were formed wholly as a result of local crystallization differentiation. Taking into con- sideration the existing data on the structure of the earth’s crust in the platform, the depth of such magma chambers appears to have been about 25 km, and the geothermal gradient at the time of volcanism appears to have reached 40 C per kilometer. Some differences in composition of the traps might have resulted from differences in depth of the magma chambers in different parts of the platform and resultant differences in the differentiation phenomena. The presence of effusives of ultrabasic com- position ( meimechites, kimberlites) is definite proof of the existence of a corresponding magma. This magma could be formed only by the remelting of ultrabasic rocks, which in turn is proof of the existence of rocks of corres- ponding compositions below the Mohorovicic discontinuity. In the case of the formation of differentiated complexes, there is no doubt of the presence of big magma chambers and a relatively slow rise of magma either to the earth’s surface or to the corresponding intrusion chambers. A compli- cated evolution of the rocks takes place as a result of involvement of the higher levels of the earth’s crust in melting and, perhaps, as a result of assimilation and differentiation. In the case of the kimberlites the quantity of rising magma is very small. This can hardly be the result of the low penetrability of the earth’s crust. Rather, it is a proof of the forma- tion of very small magma chambers in which remelting was partial, and a magma which rose very rapidly up the deep fissures containing many suspended crystals that formed not only as a result of crystallization in the chamber but also that remained as a result of the partial melting. Not only theoretical calculations but also ex- perimental data now show that if diamonds were formed at a temperature of about 1200 C, the pressure must have been more than 40 kilo- bars. The notion that diamonds were brought by the magma from great depth, and not formed at the time of explosions near the earth’s surface, can be considered valid. Sometimes, on the basis of the above data, an attempt is made to de- termine the depth of the magma chamber from the implied hydrostatic pressure of the overlying rocks. This approach we cannot agree with, since the pressure in the earth’s crust, within the zone of metamorphism, can be as high as 15 kilobars, which is several times the pressure resulting from load at that depth, the difference between the pressure at the time of the mineral forma- tion and the calculated pressure due to the load being as much as 10 kilobars. Such zones of higher pressure may extend into the depths of the mantle, and it is to them that the regions of kimberlite development are likely to have belonged. The depth of formation of the magma chambers, in this case, may be less than calcu- lated— that is, not 150 km but 70 to 100 km. As a result, when magma rises to a higher level, pressure still remains high, though it falls by a quantity corresponding to the weight of the vertical column of magma. In places at a depth of 3-4 km (see above), a breaking of the earth’s crust occurred, accompanied by the formation of peculiar pipes and a sudden pressure decrease, constituting a kind of explosion. A great quan- tity of pyroclastic and xenogenic material rushed into the pipe, part of it being thrown up and then sucked back into the pipe again. The frag- ments filling the pipe may later be cemented by the rising magma. The pressure before the explosion is not only below that shown by the equilibrium curve of diamonds but also below that shown by the curve of pyrope, since the kelyphite rims around grains of the latter must have formed before the explosion. The fact that the diamonds are neither completely resorbed nor graphitized is due to the rapid rise of the magma and its com- paratively low temperature. The temperature of the magma must be lower than that shown by a curve corresponding to the region of metastable existence of diamonds (V. Sobolev, I960), which begins at 1200 C at normal surface pres- sure and rises to 2200 C at 30 kilobars. As is known, the diamonds show only traces of graphitization, which appear as graphitic ro- settes near some inclusions. Volcanism of the Siberian Platform — Sobolev Recently, in connection with the discussion of the composition of the subcrustal substratum, great attention is once more being paid to the eclogite problem. The suggestion has been made (Fermor, 1914; Lovering, 1958) that the sub- stratum below the Mohorovicic discontinuity is eclogite. Some authors are of the opinion that this eclogite layer extends to a depth of 900 km (V. V. Belousov, I960), i.e., to the base of the Galitzin layer. The only valid data available for discussion of this matter have been derived from study of the kimberlites — a fact that makes it desirable to treat this problem here. As is known, a great quantity of xenoliths of ultrabasic rocks, in some cases directly related to eclogites, are to be found in many kimberlite pipes. This fact suggests that such xenoliths are at least in part the remains of the partially- melted substratum, the more so as the compo- sition of the olivine in them resembles that of the first-generation olivine of the kimberlites. The question is, however, still open to discus- sion. There is some probability that these rocks were picked up by the kimberlites during their rise toward the surface, not only in the sub- stratum but also at much higher levels. The ultrabasic magma chambers are likely to have revived several times, and the formation of ultrabasite intrusions may have taken place dur- ing the first stages, further intrusions taking place later with the rapid movement of new portions of magma in new geologic conditions. Such intrusive massifs could have consisted of pyrope peridotite, such as that in Czechoslovakia. The absence of diamonds (at least in appre- ciable quantity) in the ultrabasite xenoliths speaks against the supposition that the xenoliths were brought directly from the deep magma chamber. Although the presence of diamonds in pyrope peridotites has been asserted by some workers, neither a xenolith with a diamond, nor its photograph, nor a detailed description of it, can be found anywhere. Many attempts to obtain diamonds by grinding and concentration of con- siderable quantities of ultrabasite xenoliths and eclogites have resulted in failure. This fact, how- ever, cannot altogether disprove the hypothesis of the subcrustal origin of these rocks, since the distribution of diamonds in the substratum may be nonuniform. Also, they may have, for the 455 most part, crystallized directly with the forma- tion of kimberlite magma. On the other hand, two findings of diamond- bearing eclogites, which are subject to no doubt and have been described in detail, are proof of the existence of subcrustal eclogites. Diamond formation in the zone of metamorphism is im- possible. The pressure there has never reached even that of coesite crystallization, which is lower than that of diamond formation. It is in- teresting to point out that the Jakutian diamond- bearing eclogite in its ratio of FeO to MgO is nearer to basic rocks than to ultrabasic rocks; it undoubtedly was not brought directly from the deep magma chamber, but was picked up from higher levels in the substratum. Comparing all the above-mentioned data, we come to the following conclusions regarding the constitution of the upper mantle and its rela- tionship to the earth’s crust: 1. At comparatively small depths, probably less than 50-70 km, the subcrustal substratum is of peridotite composition, corresponding ap- proximately to the composition of meimechite or kimberlite, the latter being a magmatic rock. 2. In the region of kimberlite distribution, higher than the peridotite layer but below the Mohorovicic discontinuity, the substratum is eclogite with chemical composition very near that of basalt. 3. Pressures in the zone of metamorphism can vary greatly at one and the same depth, reaching at least 15 kilobars but never 20-25 kilobars. 4. There is some reason to believe that higher pressures are characteristic of large parts of the earth’s crust, especially the border of the plat- forms. Pressure higher than simple hydrostatic pressure is also characteristic of the upper part of the subcrustal substratum in these same areas (probably down to a depth of about 150-200 km). 5. Higher pressure persisted in certain zones through considerable periods of geologic time, as is proved by the finding of ancient eclogitized schist in kimberlite of both Upper Paleozoic and Mesozoic age. Examining the above statements, we come to the conclusion that the hypothesis of a subcrustal eclogite layer (Fermor, 1914) has been con- firmed, but only in part. The author quite agrees 456 PACIFIC SCIENCE, Vol. XVII, October 1963 with the opinion of J. F. Lovering ( 1958 ) , V. V. Belousov ( 1960b), and others, that a change of conditions (chiefly pressure, rather than tem- perature) leads to a shift of the Mohorovicic discontinuity, with the formation of eclogites at the expense of gabbroic rocks of the "basalt” layer. The total thickness of the basic layer, however, is probably not more than 30-50 km. Thus the eclogite layer is not to be found all over the globe, but only in the zones of higher pressure. In some cases, the basalt layer has been fully eclogitized and has entirely disappeared, and the sialic layer has come into direct contact with the Mohorovicic discontinuity. We can ap- proach the problem of distribution of the sub- crustal eclogite layer by comparing geophysical and geologic data: the distribution of kimberlite, the appearance of eclogite inclusions in effusives, and, partly, the distribution of rocks formed at high pressures in the zone of metamorphism, such as kyanite schists, eclogites, jadeite (taking into consideration possible changes over periods of time). In the zones of normal or lower than normal pressure, the pressure of about 15 kilobars, which is necessary for the formation of eclogites, is reached at a depth of about 60 km, which is, as a rule, below the boundary separating the basic and ultrabasic rocks. Garnet peridotites, or some interlayers of eclogites that are close to ultrabasic rocks in composition, may be present there. In such cases the Mohorovicic discon- tinuity evidently corresponds to the true com- positional border between the basalt and the peridotite layers, not to a phase transformation. The isobar of the limit of possible diamond formation (40 kilobars) is, of course, well below that level. In the areas of normal or lower pres- sures, it must be below 120 km. In the authors opinion, the penetration of magma from such a depth is unlikely. Still less likely is the preser- vation of the diamond, even if magma chambers have formed at such depths. In connection with this problem it is inter- esting to compare the data on the finding of diamonds in meteorites. As far as is known, dia- monds have been discovered in stone meteorites of ureilite type (first in the Novourei meteorite) and in some iron meteorites. The author quite agrees with the opinion of Urey (1954, 1957) that the presence of diamonds there is evidence of the formation of these meteorites by the breakdown of some celestial body, as big as the moon or bigger, in which the pressure was high enough for the formation of diamonds. Thus, we cannot agree with A. P. Vinogradov’s (1959) opinion that achondrites were formed by the breakdown of small celestial bodies. The para- genetic associations characteristic of eclogites, and specifically pyrope itself, have not been found in the meteorites, however. This shows that in the basalt shell of the disintegrated body pressures nowhere reached 15 kilobars. This fact, together with the absence of meteorites of granitic composition, suggests that the body probably was smaller than the earth and less differentiated. This agrees with ideas which have already been developed on other grounds by A. N. Zavaritski ( 1943 ) . The above-mentioned data do not, however, prove that all meteorites have had the same origin and resulted from the breakdown of one planet. Various meteorites are still being searched for diamonds. This search is certainly of great in- terest, but there is little likelihood that diamonds will be found in other types of meteorites, par- ticularly in chondrites. Without going into de- tails on the hypotheses of the formation of chondrites, we are quite certain that in the later stages of existence of these meteorites the tem- perature was high enough so that diamonds would have turned into graphite even if they had existed. If, however, we should succeed in finding pseudomorphs of diamonds in chon- drites, as we have in some iron meteorites, this would be a direct proof of the formation of chondrites by the breakdown of some big celes- tial body. On the other hand, the discovery of graphite pseudomorphs of diamonds in iron meteorites shows that, after the breakdown of the parent planet, the temperatures of the meteorites were greater than 1200 G The preservation of dia- monds in the Canyon Diablo octahedrite sug- gests that the temperature of that meteorite at the time of the breakdown was below 1200 C, which means that it was not melted. The general questions discussed here are, of course, still open to argument. Already, however, on the basis of available petrographic and min- Volcanism of the Siberian Platform— -Sobolev eralogic data, we can be more certain of the thermodynamic conditions in and the constitu- tion of the upper mantle of the earth, and of the conditions of meteorite formation. The forma- tion and alteration of diamonds are of particular importance in these considerations. There is no doubt that further mineralogic investigations in general, and the investigation of diamonds in particular, will open new approaches to the study of the composition of the earth and aid in penetrating the secrets of the solar system. REFERENCES Belousov, V. Y 1960a. Development of the earth and tectonics. Sovetskaja Geologia 7. — — 19601c Development of the earth and tectogenesis. J. Geophysical Res. 65:4127- 4146. Bgbrievich, a. P, M. N. Bondazenko, et al. 1959'. Diamond deposits in Yakutiya. U.S.S.R., Min. Geol. i Okhrany Medv., Moscow. 527 pp. — — G. J. Smirnov, and V. S. Sobolev. 1959. Eclogite xenoliths with diamond. Dok- lady Akad. Mauk., U.S.S.R.: 126(3). BONNEY, T. G. 1899. The parent-rock of the diamond in South Africa. Roy. Soc. London Proc. 65:235-236. 457 Butakova, E, L. 1956. On petrology of ultra- basic and alkalic rocks of Meimecha-Kotuockii complex. Tr. Niga 86:6. Fermor, L. L. 1914. The relationship of isostasy, earthquakes, and volcanicity to the earths infraplutonic shell. Geol. Mag. 51:65-67. Lebedev, A. P 1957. Some problems of the petrology of original diamond rock in U.S.S.R. Akad. Nauk., U.S.S.R., Izvestiya, ser. Geol. 11. Lovering, J. F. 1958. The nature of the Mo- horovicic discontinuity. Am. Geoph. Union Tr. 39:947-955. Sheinman, Y. M. 1955. Some petrologic char- acteristics of ultrabasic rocks, and ultrabasic activity in platforms. Zapiski Ysesoyuznogo Mineralogicheskogo Obshchestva. Sobolev, V. S. 1936. Petrology of traps in the Siberian platform. Arctic Institute, Trans. 43 : 224. — — — 1960a. Role of high pressure in meta- morphism. Int. Geol. Cong., 21st Session, Rep. 1:14. — I960 A Conditions of the formation of diamonds. Geol. and Geoph. 1. Urey, H. C. 1954. Diamonds, meteorites, and origin of the solar system. Astroph. 129. — - 1957. Diamonds in stone meteorites. Geochim. et Cosmochim. Acta 13: 1-4. Gibsmithia hawaiiensis gen. n. et sp. n.1 Maxwell S. Doty2 ABSTRACT: A new genus, Gibsmithia, is described and tentatively placed in the Dumontiaceae of the red algal order Cryptonemiales. Its diagnostic features are: possession of auxiliary cells in specialized filaments separate from those bearing the carpogonia; the known sexual structures occurring in sori at the tips of soft, gelatinous branches which arise from perennial round stems so as to form a hemi- spherical head; cruciate tetraspores borne on filaments protruding from the surface of the branches and these same branch filaments often bearing terminal seirospores. The type species of the genus is G. hawaiiensis , known only from the island of Oahu in the Hawaiian Islands. THE ALGA described here was first collected by Dr. J. T. Conover in 1943, and from 1951 on has been found by eight other individuals col- lecting in Hawaii. It seems to comprise a new genus of the Cryptonemiales in the Rhodo- phyta, and we take pleasure in naming it in honor of the late Prof. G. M. Smith, of Stan- ford University, whose publications and teach- ings have probably done more to facilitate work with the algae in America than has the work of any other phycologist. Gibsmithia hawaiiensis genus et species nova Figs. 1-17 DESCRIPTIO TYPI: Thalli ramis gelatinosis fasciculatis, e stipite erecto fere cartilaginose lignoso egredientibus, formam hemisphaericam simulantes. Frondes e filamentis uniseriatis, sicut in genere Callithamnio, compositae. Fila- menta interdum pilis uniseriatis, multicellulari- bus e superficie gelatinosa ramorum exstanti- bus terminata. Tetrasporangia cruciformiter di- visa, subsphaerica, in pilis supra descriptis uni- 1 Contribution No. 190 from the Hawaii Marine Laboratory. Received for publication February 26, 1962. This study was made possible by funds from the U. S. Atomic Energy Commission (contract AT-f04- 31-15) and the Graduate Research Committee of the University of Hawaii. Dr. Hannah Crossdale prepared the Latin diagnosis. 2 Department of Botany, University of Hawaii, Honolulu 14, Hawaii. lateraliter in cymulis fortasse velut in genere Spermothamnio producta, Cystocarpia mo do Cryptonemialium simili crescentia. The holotype (Fig. 1) was found drifting in the water at Waikiki, Honolulu, Oahu, Ha- waii (21° 161 min N, 157° 49.5 min W), on January 12, I960; 19263.3 It is cystocarpic. The type material, consisting of a liquid-preserved isotype specimen and a dried holotype her- barium specimen, is being deposited in the Ber- nice P. Bishop Museum in Honolulu, Hawaii. Tetrasporic material has been collected from the vertical surface of ■ a dredge-cut in the reef at Waikiki, first on January 5, 1961, by Dr. D. W. Strasburg. This specimen (8525) is also being deposited in the Bishop Museum her- barium. Other collections are listed below. DESCRIPTION OF GENUS AND SPECIES: Thalli (Figs. 1, 2), as generally collected, consisting of 'woody’ stems and a number of soft, but not slippery, gelatinous ultimate branches. Forming bright-pink, hemispherical tufts often 8 cm high, the stems are up to 3 mm in diameter, blackish when dried, and they bear rings be- lieved to be indicative of successive crops of the ultimate branches. The stems are cylindrical or, when large, compressed toward the apex and sometimes branched once or twice. When the stems are fresh they are colored approxi- mately Eugenia Red or darker to Acajou Red 3 Such 4- and 5 -digit numbers are the collection or herbarium numbers of the author. 458 Gibsmithia hawaiiensis — Doty 459 Oi N> Ol o 3 FIG. I E o CVI FIGS. 1-2. Gross aspects of Gibsmithia hawaiiensis (type specimen 19263). I, Whole thallus illustrating the rounded stem with growth rings. 2, A few of the soft gelatinous branches enlarged to show asymmetrical placement of the sexual sori and the superficial filaments which, in tetrasporic thalli, would bear the tetra- spores and seirospores. Below is shown some of the stem material of very different, almost woody, nature and the origins and rebranching of the ultimate branches of the thallus. (Ridgway, 1913). The soft, gelatinous, ultimate branches are lighter red in the same series or their bases are lighter and their apical portions darker. The branches are often up to 4 cm long, and they gradually enlarge apically until they are club-shaped and up to 1 cm in diam- eter near their tips. These ultimate branches rarely arise singly from the parent stem; more often (Fig. 2) they arise from a disk that is a flattened expanse of the same gelatinous na- ture and which divides somewhat dichoto- mously into the branches. The gelatinous branches are made up of cal- lithamnioid filaments (Fig. 3). The exact na- ture of the central axis of the thallus has not been determined, but the alga is believed to be multiaxial in structure. The major axes in the medulla bear, most often, opposite branches which seem to arise near the apical cell. Toward the exterior the branching of the filaments is pseudodichotomous, and the surface of a gela- tinous branch is generally covered densely, espe- cially toward the tip, with free tufts of such filaments. Branches of the axial filaments in the medulla often give rise to recurrent branches (A in Figs. 3-5) or rhizoids, which anastomose with a particular cell of another branch (Fig. 3), the supporting cell of the branch (Fig. 4), or with the next cell below in the same branch (Fig. 5). Pit connections appear between all cells concerned in these cases. In content, the cells of such anastomosing branches, like those reported and figured by Feldmann-Mazoyer (1940: 142, 393, fig. 48) in Pleonosporium horreri, are not particularly different from the cells of similar nearby filaments. Reproductive structures of three kinds have been found in the collections studied. These are interpreted as seirospores, as tetrasporangia, or as various stages and structures leading toward carpospore production. The same thal- lus may produce both tetraspores and seiro- spores, a situation I have seen in Seirospora, and which is reported ( Feldmann-Mazoyer, 1940) in Dohrniella. The evidences of sexuality have been seen positively in only the type and two other collections. Unfortunately the critical stages involving the primary and secondary connecting filaments between the carpogonia 460 PACIFIC SCIENCE, Vol. XVII, October 1963 and auxiliary cells have not been found. Thus, though for the present Gibsmithia seems cer- tainly to be a member of the Dumontiaceae as this family is circumscribed by Kylin (1956), the description of the sexual reproduction and discussion of the systematic disposition within the family are of necessity brief. The seirospores are borne in unbranched chains at the tips of the exterior filaments in the asexual material. At some place along a branch bearing seirospores (B in Fig. 5) the chromatophore and central vacuole disappear rather abruptly and the cytoplasm ( after killing in Karpetchenko’s seawater fluid) is then more densely granular and with a refractory body becoming conspicuous as the cells age. The spores, which are about 9 ^ in diameter, appear to become more spherical with maturity, as in Seirospora. Their ultimate fate is unknown, but they are broken off easily in preserved material and presumably in nature. The subspherical tetrasporangia appear on the more basal portions of scattered external filaments similar to those bearing the seiro- spores. They are, for the most part, in acro- petalous series (Fig. 6) on the adaxial side of the filaments on which they are produced. In a few cases immature tetrasporangia were seen below an otherwise acropetalous series. In some instances what appeared to be chromo- some groups undergoing meiosis were distin- guished within immature tetrasporangia. The tetrasporangia are apically dehiscent, as in Antith amnion, and at least the terminal sporan- gia were pedicellate. The first division of the cytoplast in the tetrasporangium (Fig. 7) is near the equatorial plane. Usually the tetra- spore pairs are at right angles to each other, i. e., decussately cruciate (Figs. 8, 9); only rarely is a somewhat tetrahedral (Fig. 10) or other ( Fig. 11) arrangement seen. The tetra- spore groups in the collection 12343 are often as long as 25 or 30 /i in their major dimension. The maturing cystoearps, when stained, ap- pear as dots in raised asymmetrically placed welts (Figs. 1, 2) near the tips of the branches. The largest thalli collected, 10386, were be- lieved to be sterile until smears made of the branch tips were examined. When this was done what are accepted as carpogonial branches (Figs. 12-14) bearing nonfertilized carpogonia were found. The carpogonial branch of Gibsmithia is the terminal portion of a modified member of the two lateral branch systems arising (Fig. 13) from a cell in a medullary filament. Upon oc- casion both lateral branch systems bear carpo- gonia; less frequently only one of the mem- bers is carpogonial, and the other is vegetative but reduced in its development. Only mono- carpogonial branch systems have been seen. Branching from the cells below the special cells of the carpogonial branch is variable. In some cases there are no such branches (Fig. 12), and in others (e. g., Figs. 13, 14) up to three branches of one to six small cells each occur. The cells of these branches, except for size, are of the same appearance as the homol- ogous cells in vegetative branches nearby. The carpogonial branch itself is consistently composed of five cells ( cp, a, b, c, and d in Figs. 12-14), though the terminal carpogonium is (Fig. 14) not always present. The antepenulti- mate cell (b in Figs. 12, 13) is smaller than the other hypogynous cells. Its contents are often of different appearance and it is flattened, so that its axial length is noticeably less than its diameter. The other three hypogynous cells are more spherical, similar in content, and the largest one (c in Figs. 12, 13) is usually about 9 to 13 v in diameter. The remaining two (a, d in Figs. 12, 13) are usually similar in size. The carpogonium is borne on the first4 hypo- gynous cell excentrically, often by a displace- ment from the axis of the carpogonial branch of about 45 degrees. The carpogonium {cp in Figs. 12, 13) is usually relatively small and does not stain strongly in aniline-blue, in con- trast to the other cells of the carpogonial branch. In the material available, stages were not found which would enable a satisfactorily com- plete description of sexual reproduction. How- ever, my impression is that the heavily aniline- blue stained condition of almost all carpogonial branches seen is a result of development be- yond the fertilizable point. Most such branches consist of either only the cells a through d ( Fig. 14) or have in addition what are interpreted as remnants of a carpogonium. Such carpo- gonial branches are relatively straight and not Gi bsmithia hawaiiensis — Doty 461 FIGS. 3—11. Vegetative and asexual reproductive microscopic structural details of Gibsmithia hawaiiensis (collection No. 12343). 3, An axial filament from the medullary region of a gelatinous branch of the thallus showing the opposite ramification of the medullary filaments and, A , characteristic reconnecting rhizoidal branches. 4, A more simple reconnecting rhizoidal branch, A, from one of the subdichotomous cortical fila- ments. 5, A seirospore-bearing ultimate branchlet on a subdichotomous cortical filament with, A, a very simple reconnecting filament or rhizoid. Such lateral nonrebranched filaments, B, with a terminal series of seirospores were found arising from almost every second cell in the vegetative axis of this part of the filament system. The origins of but three such seirosporic branches are shown here. 6, A tetrasporangium-bearing cortical fila- ment system. Divisions within the tetrasporangia are not shown and no attempt was made to represent all pit connections. 7—11, Tetrasporangia in various stages of maturity. The arrangement of spores is always at first cruciate but, as indicated by Figure 10, in age they may not be in a strictly cruciate arrangement. IQOjj 5QJJ FIG. 16 FIGS. 12-15,17 FIGS. 12-17. Microscopic sexual reproductive features of Gibsmithia hawaiiensis. The specialized cells, a-d, are similar in nature whether in a carpogonial or auxiliary-cell branch. On the latter, the piliform exten- sion, h, shows as a multicellular hair, the cells of which are in staining reaction and appearance, rather like vegetative cells of similar size and relative terminal position elsewhere. 12-14 , Carpogonial branches with or without nonfertilized carpogonia, cp. The hypogynous cells a-d were always densely granular and their contents took up aniline-blue from aqueous solution much more than did the other nearby cells. Cell b was usually smaller and the contents were usually different from the rest. Cell c was almost always the largest. The num- bered cells, 1, 2, 3, etc., represent cells that looked like the strictly vegetative cells of similar size and position nearby and connected the carpogonial branch to a medullary cell, med. The cytologically similar cells of the sterile branchlets are indicated as la, lb, lc, etc. (Collection No. 10386.) 15, A nonfertilized auxiliary-cell branch arising from one of two opposite branch systems of the same medullary cell, med, the one vegetative branch system, veg, being of only three cells, the other more extensive and bearing an auxiliary-cell branch. Gibsmithia hawaUensis—DoTY 463 strongly curved apically, as is usual in the Du- montiaceae. In a very few cases carpogonium- bearing branches that are more curved are found with hypogynous cells that are less highly differentiated in staining quality and in form. In these cases the hypogynous and carpogonial cells are more nearly similar in staining quality and are smaller; such (Fig. 12) are judged to be more nearly normal. Primary connecting filaments4 were not seen, and no clues as to the origin of the abundant, presumed secondary, connecting filaments that connect to the auxiliary cells were found. In a very few cases the carpogonium as well as the cell beneath it were seen to have some indica- tion of protuberances extending toward one another or toward the smallest of the hypogy- nous cells. This could have been the effect of smearing the material for microscope examina- tion. The auxiliary-cell branches (Fig. 15) develop 4 The author is attempting to apply terms to the post-fertilization morphological phenomena in such a way that they are consistent with general biological usage, and so that the frequent special cases and varia- tions from the normal do not require a special no- menclature. Thus, the term "zygote” is used for the post-fertilization carpogonium and its contents, and "primary connecting filament” for the connection be- tween the zygote and the auxiliary cell without dis- tinguishing whether a "carpogonial” or "generative” auxiliary cell is concerned. "Secondary connecting filament” is used to refer to any connecting filament ("ooblast” of the older nomenclature) arising from an auxiliary cell. Similarly, the term "hypogynous” is used for any of the specialized cells of the carpogonial branch other than the carpogonium itself; these are referred to as a, b, c, etc., or the "first,” "second,” etc., proceeding away from the carpogonium. farther from the apex than the carpogonial branches. It may be that initiation, or at least maturation, of an auxiliary-cell branch system is dependent upon fertilization, for in 10386, where many nonfertilized carpogonial branches were found, especially of the forms illustrated in Figures 13 and 14, branches bearing mature auxiliary cells were not observed. They are usually in the same positions as carpogonial branches, or, at times, appear to have developed (Fig. 16) after the lateral branch system had grown further. While the complement of cells in the auxil- iary-cell branches is similar in number, form, arrangement, and appearance to the hypogynous cells in a carpogonial branch, there are two major differences: in place of the carpogonium there is a reduced terminal branch or row of vegetative cells, and there are more sterile fila- ments developed laterally from and near the base. The auxiliary cell itself is an intercalary cell, and, contrary to the condition illustrated in Figures 15 and 16, is usually the cell there labeled as b. The presumed secondary, connecting fila- ments ( cn in Fig. 16) were seen fused with the auxiliary cells (c in Fig. 16). The tissue of the weltlike sori (Fig. 2) in which cysto- carps develop is the toughest of all parts of the branch surface and this toughness seems to be related to the great abundance of sterile branches appearing in the fertile system, as well as to numerous, presumed secondary, connect- ing filaments. In addition, multicellular rhizoi- dal branches from the bases of the branch filaments (rh in Figs. 15, 16, and possibly A This particular auxiliary-cell branch would seem to be peculiar, for usually one of the two cells labeled b and c is about half the size of the cells labeled a and d, which are similar in size. The other cell, c or b, is usually distinctly the largest of the four. A rhizoidal branch, rh, arises from the single vegetative cell, l, con- necting this auxiliary-cell branch system to the medullary system. (Collection No. 19263.) 16, A lateral branch system arising as one of a pair, B and C, from a single medullary cell, med, and producing vegetative filaments, fil, and rhizoidal filaments, rh, in a normal manner, and showing (at c) a fertilized auxiliary cell. Two other types of filaments are illustrated here: what may be labeled as an accessory rhizoid, arh, of multicellular nature arising near a fertilized auxiliary cell and, ac, an accessory connecting filament. The auxiliary cell, c, is shown with a connecting filament, cn, associated. The actual relationships of cells, c, cn, and ac, to one another were not entirely clear at the highest magnifications. (Collection No. 19263.) 17, A young gonimoblast showing the apparent carposporogenic cells heavily stippled. Cell 1, seemed to correspond to cells B and C in Figure 16. Cell d bore three vegetative branches in addition to the auxiliary-cell branch. No connecting fila- ment could be distinguished. Seemingly, the auxiliary cell, b, had produced the gonimoblast to the left at the tip of the expansion, p, from but one gonimoblast initial. The six cells to the right seemed unattached and were interpreted as having been broken free from the mass at the left. (Collection No. 19263.) 464 PACIFIC SCIENCE, Vol. XVII, October 1963 in Fig, 3) are abundant, and undoubtedly con- tribute to the strength of the otherwise weak, gelatinous, branch substance. The surface of the auxiliary cell (b in Fig. 17), beneath the connecting filament, appar- ently develops a protrusion that moves the con- necting filament away from the auxiliary-cell branch axis. From this protuberance of the auxiliary cell (p in Fig. 17) gonimoblast ini- tials arise. While the gonimoblast, upon smear- ing, reveals its fundamentally filamentous na- ture (Fig. 17), in the undisturbed condition it is a dense almost spherical mass. Older gonimo- blasts have one (or a few) very large cells at the center: all the other cells appear to be uni- form, as though all could become carposporan- gia in time. MATERIALS EXAMINED (all from the island of Oahu, Hawaii; the collection numbers are the authors, except the first): J. T. Conover, 200 Berkeley,5 tetrasporic, found about 550 ft along a line extending about 113° from the old marine laboratory at Waikiki on underside of a coral head about halfway out on the reef at a depth of 1.2 m, VIII-43; D. W. Strasburg , 8525 (bishop) tetrasporic and seirosporic, University of Hawaii old beach laboratory, Wai- kiki, Honolulu (21° 16.2' N, 157° 49.6' W), 1-51; M. Doty, 8562 (Stanford), another branch of the same thallus as 8525, but col- lected 25 days later; George Ikeda, 10385 (AUSTRALIA) tetrasporic, Lae o Kaoio, Kua- loa, Kaaawa, one of two clumps seen in 5 ft of water, 11-52; George Ikeda, 10386 (Stan- ford, BERKELEY) carpogonial, in 4-5 ft of water on limestone bottom (same area as Ikeda collection of 11-52, above), V-52; Jack Randall, 17020 (BERKELEY) seirosporic, drifting freely below the surface off Oahu, Hawaii, VI-52; 6 Names appearing in this position refer to the fol- lowing repositories of the particular material : bishop, B. P. Bishop Museum, Honolulu, Hawaii; STANFORD, Hopkins Marine Station of Stanford University, Pa- cific Grove, Calif.; BERKELEY, Herbarium of the University of California, Berkeley, Calif.; PARIS, Laboratoire de Cryptogamie du Museum National d’Histoire Naturelle; AUSTRALIA, Herbarium of the University of Adelaide, Adelaide, S. Australia; HOK- KAIDO, Herbarium of the Faculty of Science, Hokkaido University, Sapporo, Japan; VANCOUVER, Herbarium of the University of British Columbia, Vancouver, British Columbia. M. Doty, 12343 (Stanford) tetrasporic, floating freely at the surface just north of the mouth of Koloa stream south of Laie, 1-54; M. Doty, 12286 (hokkaido) first of a series of collections from the same tetrasporic thallus growing about 7 ft below the surface along the inshore edge of the swimming channel in front of the new University of Hawaii beach labora- tory at Waikiki, Honolulu, XII-52; M. Doty, 12652 and 12647 (PARIS) tetrasporic, both same thallus as 12286, about X-54; 17019 (bishop) sterile, same thallus as 12286, IH- 55; Jan New house and Malvern Gilmartin, 12671 (VANCOUVER) tetrasporic, beyond the reef, Waikiki, Honolulu, XI-54; HOLOTYPE, M. Doty and B. C. Stone, TYPE 19263 (BISHOP) cystocarpic, drifting at the surface, Waikiki, Honolulu, (near 21° 16' N, 157° 49.5' W), 1-60; Elvin Fong, 19478 (Stanford) cysto- carpic and tetrasporic, on a 3 by 4 ft rock in 15 to 20 ft of water 150 yd off Haleiwa Army Beach near the channel at Haleiwa, VI-55. This is one of the less common algae occur- ring along the shores of Oahu in Hawaii. It has been found adrift in a few cases, but for the most part it is found growing L to 2 m below low tide on reef flats. The hemispherical pale thalli were often found growing in a concavity or on a vertical surface. One individual thallus, near the University’s aquarium and beach lab- oratory at Waikiki in Honolulu, was watched and pieces were harvested from it at irregular intervals for about 3 yr. The deepest collection was that made by Jan Newhouse and Dr. Mal- vern Gilmartin from the base of a coral patch on rock bottom in 7 m of water about 1000 m beyond the reef opposite the Natatorium at Waikiki. The stem, the perennial nature of the thallus and some of the reproductory features of Gib- smithia show some resemblance to the much recopied small portion of Gmelin’s specimen (1768: pi. 5, fig. 2) of Constantinea 6 (e. g., Kylin, 1956: fig. 102). It is to be noted that at least three collections of Gibsmithia, including the type, were of free- floating branches. It would appear that the w The figure given by Okamura (1912: pi. 77, figs. 1, 2) conveys a rather different conception of this alga— a more correct one it would seem. Gibsmithia hawaiiensis — Doty 465 branches are deciduous at maturity, as is the disklike peltate blade of Constantinea simplex. However, the many quite different features of Constantinea, especially as revealed by Masaki (1952), indicate that Constantinea could be, at best, but remotely related to Gibsmithia. Some of the anatomical features, the seiro- spores, and tetrasporangia are reminiscent of the Ceramiaceae. Yet I believe Gibsmithia should be placed among the primitive Crypto- nemiales for several reasons, most notable of which is the occurrence of the carpogonia and the auxiliary cells in separate but similar spe- cial branch systems. In their general morphology and in having a small differentiated cell among the enlarged hypogynous cells, the carpogonial branches of Gibsmithia (if, indeed, we have seen normal carpogonial branches) recall both the carpogonial and auxiliary-cell branches of Dudresnaya crassa , as I have seen them in ma- terial collected by Dr. A. J. Bernatowicz in Bermuda (his number 51-580), and as they are illustrated by Taylor (1950: figs. 5-7, 38) in his splendid account of the reproduction of that species. The terminal row of sterile cells distinguishes an auxiliary-cell filament from a carpogonial branch with its terminal trichogyne. While the terminal row of sterile cells is distinct from the terminal part of a normal vegetative branch, it is less specialized than the "hair” to be seen in Dudresnaya crassa and which Taylor ( 1950: figs. 36, 37 ) refers to as a "piliform extension.” In other features Gibsmithia also recalls such a genus as Dudresnaya rather than any of the polycarpogonial forms discussed by Norris (1957), which are far more complex. It is like Thuretellopsis (Kylin, 1925: 14) in that the carpogonial and auxiliary-cell branches are alike; but in Gibsmithia they are less compact and less specialized. In having a terminal row of sterile cells and in having a larger number of cells, the auxiliary-cell filaments of Gib- smithia would seem to be more primitive than those of Thuretellopsis. A point of some systematic interest is the development of low, wartlike sori containing the cystocarps. The tetrasporangia are borne quite differently, being exposed as in the Cera- miaceae. These two features seem to set this genus apart from most of the other genera now placed in the Dumontiaceae. The possession of cruciately divided tetrasporangia further distin- guishes Gibsmithia from the type genus of the Dumontiaceae and likens it to families such as the Squamariaceae of the Cryptonemiales which have cruciate tetrasporangia and the reproduc- tive structures in nemathecia. Until the simpler cryptonemiaceous families are better known, it seems best to place Gibsmithia tentatively among the Dumontiaceae. REFERENCES Feldmann-Mazoyer, G. 1940. Recherches sur les Ceram iacees de la Mediterranee Occiden- tale. Imprimerie Minerva, Alger. 540 pp. Gmelin, S. M. 1768. Historia fucorum. Typo- graphy Academiae Scientiarum, Petropoli. 12 + 239 + 6 pp. + illus. Hollenberg, G. J. 1959. Smith ora, an inter- esting new algal genus in the Erythropelti- daceae. Pacif. Natural. 1(8): 3-11. Kylin, H. 1925. The marine red algae in the vicinity of the biological station at Friday Harbor, Washington. Lunds Universitets Arsskrift, n. f. Avd. 2, Bd 21., Nr 9., 87 pp. 1956. Die Gattungen der Rhodophy- ceen. Gleerup, Lund. Xv + 673 pp. Masaki, T. 1952. Studies on the reproductive organs of the red algae. I. Constantinea rosa- marina (Gmel.) Post, et Rupr. and C. subu- lifera Setchell. Bull. Jap. Soc. Sci. Fish. 18: 30-38. Norris, R. E. 1957. Morphological studies on the Kallymeniaceae. Univ. Calif. Publ. Bot. 28:251-333. Okamura, K. 1912. leones of Japanese algae. Privately published by the author, Tokyo. Vol. II, 191 pp. + illus. + index. R IDG WAY, R. 1913. Color standards and color nomenclature. Private publication by the au- thor, Washington, D. C. 43 pp. and 53 color pi. of 115 different colors. (Though the book is dated 1912, according to D. H. Hamly [Science 109:605], it appeared in early 1913.) Taylor, W. R. 1950. Reproduction of Dudres- naya crassa Howe. Biol. Bull. 99:272-284. Revision of the Genus Pandanus Stickman, Part 16 Species Discovered in Thailand and Vietnam Harold St. John1 The following new species are all but one based upon specimens collected by others and preserved in Bangkok, either in the herbarium of the Department of Agriculture at Kasetsart University or in the herbarium of the Royal Forestry Department. SECTION Aero stigma Pandanus biplicatus sp. nov. (sect. Aero stigma) Fig. 192 DIAGNOSIS HOLOTYPI: Licet acaulescens, foliis 1.01 m longis 3.6 cm latis parte lA basali sub- coriacea parte ultima crassiter chartacea supra viridibus infra pallidis et licet glaucis 1-sulcatis acute 2-plicatis in sectione mediali M-formatis et cum 41 nervis parallelis secundariis in quoque dimidio, nervis tertilis in parte 2A ultima con- spicuis et reticulis anguste oblongis formantibus, laminis ligulatis subiter in apice trigono subu- late 3 cm longo 1-2 mm lato diminuentibus basi amplexicauli et inermi sed ex 14-15 cm marginibus cum aculeis 1.5-2 mm longis 2-8 mm separatis subulato-serratis complanatis lu- teis, midnervo infra ex 20 cm cum aculeis 2-3.5 mm longis 8-15 mm separatis subulatis reflexis, in sectione mediali marginibus cum aculeis 1.5-2 mm longis 5-10 mm separatis subulatis proxime adscendentibus, midnervo infra cum aculeis 2.5-3 mm longis 15—22 mm separatis subulatis adscendentibus, in apice subulato mar- ginibus et midnervo infra cum serrulis 0.2-0.3 mm longis 1.5-3 mm separatis, in pagina superi- ori ad apicem plicis cum serrulis 0.2-0. 3 mm longis 1-4 mm separatis, inflorescentia foeminea erecta et recta, pedunculo 22-30 cm longo gracili clavato 9-12 mm diametro in apice folioso brac- teato, bracteis herbaceis coloratis sed in sicco 1 B. P. Bishop Museum, Honolulu 17, Hawaii, U.S.A. Manuscript received September 28, 1961. nigro-brunneis infima 19 cm longa 23 mm lata lineari-lanceolata ad basem integra parte XA ul- tima proxime subulato-serrata, syncarpio 9.5 cm longo 7.2 cm diametro obovoideo-ellipsoideo cum drupis numerosis, eis 32-34 mm longis 6-7.5 mm latis 4-6 mm crassis anguste oblanceo- fusiformibus 1-carpelatis parte lA supera libera corpore 24-26 mm longo oblanceoloideo lateri- bus planis, pileo 13-14 mm longo cartilagineo basi 5-6 mm alta rotundato-pyramidali 6-angu- lata, stylo 6-8 mm longo crassiter subulato anguloso divergent! subsinuoso, stigmate 3-4 mm longo distali lineari brunneo sulcato papil- loso in apice extenso, endocarpio in parte }A> infera obovoideo cartilagineo stramineo in apice truncato lateribus 0.1 mm crassis, semine 8-9 mm longo obovoideo, mesocarpio apicali 7 mm longo cum medulla alba delicata completa, meso- carpio basali 5-6 mm longo fibroso et carnoso. diagnosis of holotype: Perhaps acaules- cent; leaves about 1.01 m long, 3.6 cm wide, the lower lA subcoriaceous, the outer 2A thick chartaceous, green above, pale and apparently glaucous below, 1 -furrowed above the midrib, sharply 2 -pleated, at midsection M-shaped in section and with 41 parallel secondary veins in each half, tertiary cross veins conspicuous in outer 2A’s , forming narrow oblong meshes, the blade ligulate, abruptly narrowed to a trigonous subulate apex 3 cm long, 1-2 mm wide, the base amplexicaul and unarmed, but beginning at 14-15 cm the margins with prickles 1.5-2 mm long, 2-8 mm apart, subulate-serrate, flat, yel- lowish; the midrib below beginning at 20 cm with prickles 2-3.5 mm long, 8-15 mm apart, subulate, reflexed; at midsection the margins with prickles 1.5-2 mm long, 5-10 mm apart, subulate, closely ascending; the midrib below with prickles 2.5-3 mm long, 15-22 mm apart, subulate, ascending; on the subulate apex the margins and midrib below with serrulations 0.2-0. 3 mm long, 1.5-3 mm apart; on upper 466 Fig. 192. Pandanus biplicatus St. John, from holotype. a, Syncarp profile, X 1; b, drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; ^ drupe, apical view, X 1; e, drupe, lateral view, X 4; j, drupe, longitudinal median section, X 4; g, style and stigma, distal view, X 4; h, leaf base, lower side, X 1; h leaf middle, lower side, X 1; j, leaf apex, lower side, X 1; k, leaf apex, upper side, XI \ U young pistillate inflorescence, lateral view, X Vl- 468 PACIFIC SCIENCE, Vol. XVII, October 1963 side towards the apex the pleats with serrula- tions 0.2-0. 3 mm long, 1-4 mm apart; pistillate inflorescence erect, straight; peduncle 22-30 cm long, slender clavate, 9-12 mm in diameter at apex, leafy bracted; floral bracts herbaceous, col- ored, drying blackish brown, the lowest one 19 cm long, 23 mm wide, linear-lanceolate, entire below, the outer p3 closely subulate-serrate; syn- carp 9.5 cm long, 7.2 cm in diameter, obovoid- ellipsoid, with numerous drupes, these 32-34 mm long, 6-7.5 mm wide, 4-6 mm thick, nar- rowly oblance-fusiform, 1 -celled, upper free, the body 24-26 mm long, oblanceoloid, the sides flat; pileus 13-14 mm long, cartilaginous, the base 5-6 mm high, rounded pyramidal, 6-angled, tapering upwards; style 6-8 mm long, stout subulate, angled, divergent, slightly sinuous, but nearly straight; stigma 3-4 mm long, distal, linear, brown, creased, papillose, running to the apex; endocarp in lower obovoid, the apex truncate, cartilaginous, stramineous, the walls 0.1 mm thick; seed 8-9 mm long, obovoid; api- cal mesocarp 7 mm long, filled with a delicate white pith; basal mesocarp 5-6 mm long, fi- brous and fleshy. HOLOTYPUS: Thailand, Lower Siam, Kopah, Jaujau, 9 Dec. 1917, Md. Haniff & Md. Nur 2,703 (SING). discussion : P. biplicatus is a member of the section Aero stigma, as is its closest relative, P. pseudofoetidus Martelli of Burma, a species which has the syncarp subglobose; drupes 20 mm long, 5 mm wide; endocarp 10 mm long; and the apical mesocarp 4 mm long. P. bipli- catus has the syncarp obovoid-ellipsoid; drupes 32-34 mm long, 6-7.5 mm wide; endocarp 13 mm long; and the apical mesocarp 7 mm long. The new epithet is from the Latin, bi, twice, plicatus, folded, in reference to the two longi- tudinal pleats of the leaves. Pandanus monotheca Martelli, Soc. Bot. Ital., Bui. 303, 1904; Webbia 4(2): pi. 41, fig. 18-20, 1914. THAILAND: Bachaw, Pattani, evergreen forest, 600 m. alt., July 14, 1923, A. F. G. Kerr 7J91 (bk). This species, originally described from Malaya, is here recorded for the first time for Thailand. Pandanus Toei sp. nov. (sect. Acrostigma) Figs. 193, 194 NOM. vern.: "toei.” DIAGNOSIS HOLOTYPI: Acaulescens, foliis 1.3 m longis in media 20 mm latis proxima basem 15-18 mm latis supra midervum sulcatis 2-plicatis chartaceis supra viridibus infra palli- dioribus ligulatis in apice 9 cm longo 0. 3-0.8 mm lato subulato trigono in sectione mediali cum 26-29 nervis secundariis parallelis in quo- que latere, nervis tertilis non evidentis basi amplexicauli et inermi sed ex 4. 5-6. 5 cm mar- ginibus cum aculeis 1.3-2 mm longis 3-7 mm separatis subulatis adscendentibus luteis, mid- nervo infra ex 6-8.5 cm cum aculeis 1.5-2. 5 mm longis 6-15 mm separatis crassiter subulatis arcuatis reflexis, in sectione mediali marginibus cum serrulis 0.2-0. 6 mm longis 3-7 mm sepa- ratis, midnervo infra cum aculeis 2 mm longis remotis subulatis arcuatis reflexis vel adscenden- tibus, in apice subulato marginibus et midnervo infra cum serrulis 0.2 mm longis 0.5-2 mm se- paratis, inflorescentia foeminea terminali erecta cum syncarpio unico, pedunculo 15 cm longo clavato trigono in apice 5 mm diametro in di- midia ultima cum bracteis foliosis paucis eis su- peris 40-50 cm longis 15-18 mm latis, syncarpio 3.5 cm diametro globoso viridi cum circa 160 drupis, eis 15-18 mm longis 4-5 mm latis 3-4 mm crassis subfusiformibus parte 2A supera li- bera corpore oblanceo-ellipsoideo obtuso 10-12 mm longo, pileo 9-11 mm longo basi 2-3 mm alto oblato-pyramidali-hemispherico laevi 5-7- anguloso angulis in stylo continuis, stylo 5-7 mm longo valido 5-7-anguloso subulato proxime arcuato parte exteriori osseosa lucida lutea, stig- mate 2.5-3 mm longo distali sublineari sulcato brunneo papilloso paene ad apicem extento, en- docarpio in tertia infera cartilagineo luteo muris 0.1 mm crassis intra lucidis, semine 6-7 mm longo 2-2.5 mm diametro oblanceoloideo trun- cato, mesocarpio apicali 2 mm longo hemisphe- rico vel oblato-hemispherico et cum membranis medullosis albis, mesocarpio basali fibroso et car- noso. DESCRIPTION OF ALL SPECIMENS EXAMINED: Acaulescent, or decumbent with the stem 20 cm tall, 12-22 mm in diameter, brown, smooth; leaves 0.96-1.3 m long, 20 mm wide near the middle, 15-18 mm wide near the base, 1-fur- Fig. 193. Pandanus Toei St. John, from holotype. a , Peduncle and syncarp, X 1; ^ drupe, lateral view, X 1; c, drape, apical view, X 1; ^ drape, longitudinal median section, X 1; drupe, lateral view, X 4; f, drape, longitudinal median section, X 4; g, style and stigma, distal view, X 10; h, leaf base, lower side, X 1; i, leaf middle, lower side, X 1; h leaf apex, upper side, X 1; k, leaf apex, lower side, X 1. 5 »nnv. FIG. 194. Pandanus Toei St. John, from Smitinand 1,391. a, Staminate inflorescence, X 1; b, axis and stamens, X 10; c, leaf base, lower side, X 1; ^ leaf middle, lower side, X 1; e> leaf apex, upper side, X 1; f, leaf apex, lower side, X 1; g> secondary longitudinal veins and tertiary cross veins, at midsection, lower side, X 4. Page 309: Revision of Pandanus, 16. Thailand, Vietnam— St. John 471 rowed over the midrib, 2 -pleated, chartaceous, green above, paler green below, ligulate, taper- ing to a 9 cm subulate, trigonous apex 0.3-0.8 mm wide, at midsection with 2,6 -29 secondary parallel veins in each half, the base unarmed, but beginning at 4. 5-6, 5 cm the margins with prickles 1.3-2 mm long, 3-7 mm apart, subulate, ascending, yellowish; the midrib below begin- ning 6-8.5 cm up with prickles 1.5-2. 5 mm long, 6-15 mm apart, stout subulate, fully re- flexed arcuate; at midsection the margins with serrulations 0.2-0.6 mm long, 3-7 mm apart, the midrib below with remote prickles 2 mm long, subulate, arcuate, reflexed or ascending; on the subulate tip the margins and midrib below with serrulations 0.2 mm long, 0.5-2 mm apart; pistillate inflorescence terminal, erect, with 1 syncarp or rarely with a small secondary one; peduncle 15 cm long, clavate, trigonous, 5 mm in diameter at apex, with a few leafy bracts on upper half, the upper bracts 40-50 cm long, 15-18 mm wide; syncarp 2. 5-4.5 cm in diame- ter, globose to ellipsoid, green, bearing about 160 drapes, these 15-18 mm long, 4-5 mm wide, 3-4 mm thick, subfusiform, upper 2A free, the body oblance-ellipsoid, obtuse, 10-12 mm long; pileus 9-11 mm long, the base 2-3 mm high, oblate pyramidal-hemispheric, smooth, 5-7-angled, the angles continuing well up the style; style 5-7 mm long, stout, 5-7-angled, su- bulate, proximally arcuate, the outer part bony, shining, yellowish; stigma 2.5-3 mm long, distal, sublinear, creased, brown, papillose, running al- most to the apex; endocarp in lower 14, carti- laginous, yellowish, the walls 0.1 mm thick, the inside shining; seed 6-7 mm long, 2-2.5 mm in diameter, oblanceoloid, truncate; apical meso- carp 2 mm long, hemispheric or depressedly so, containing white, medullary membranes; basal mesocarp fibrous and fleshy. STAMINATE PLANTS: Prostrate, stems 7-11 mm in diameter, very short; leaves 80-90 cm long, 18 mm wide at the middle and base, the subulate apex 3-4 cm long, at midsection with 24-27 parallel secondary veins in each half; staminate inflorescence almost erect, 18 cm long, fragrant; peduncle 7 cm long; flowering part 11 cm long, with conspicuous yellow bracts, the lowest one 9.5 cm long, 16 mm wide, linear lanceolate, chartaceous, serrulate only on the short subulate tip; median bract 6 cm long, 14 mm wide, lanceolate, the margins of the outer half and the midrib below near the tip with pale, fragile serrulations 0.1 mm long; spikes about 7 and 1.5-2 cm long, 8-15 mm in diame- ter, dense; stamens numerous, attached singly, directly to the axis; filaments 0. 1-0.2 mm long; anthers 6-7 mm long, linear, bearing at the apex a projection of the connective 0.2-0.3 mm long, sometimes subulate but more commonly flat, lanceolate, obtuse. HOLOTYPUS: Thailand, Peninsular, Krabi, 83 km. from Huay Yawt, very common in moist localities by roadside, 100 m. alt., 28 Jan. 1958, T em Smitinand 4,1 43 ( BKF ) . SPECIMENS EXAMINED: Thailand, Trat, Koh Chang, shrub in evergreen forest by stream, 40 m. alt., 19 Feb. 1954, ( BKF ) ; erect shrub 40 cm. tall, common in swampy ground in evergreen forest, 1 June 1952, Bunnak ( Sang Kachand ) 333 (BKF); Trat, Huayraent, Tern Smitinand 1,392 (BKF); Peninsula, Trang, Kantang, Khuan nang hong, less than 100 m. alt., common in evergreen jungle, 26 Aug. 1955, Tem Smitinand 3,012 (BKF); S. E. Part, Trat, Kaw Kut, Ao Salat, common in evergreen forests, 5 m. alt., 5 April 1959, Tem Smitinand 3,702 , staminate, ( BKF ) ; S. E. Part, Trat, Huang Raeng, Dong Tapit, prostrate, common in swamp by path in evergreen jungle, under 50 m. alt., 21 June 1952, Tem Smitinand 1,391, staminate, (bkf no. 7,276); Kasvon, Nov. 1896, C. Curtis 3,247 (SING). MALAYA: Penang, near Tanjong Toking, July 1889, C. Curtis 1,821 (SING); Perak, Lahat, Ipoh, Oct. 1904, H. K.^Riidley'], (sing),, one of the original paratypes of P. perakensis RidL DISCUSSION: P. Toei is a member of the sec- tion Aero stigma, as is its closest relative, P, Kin- gianus Mar tell i of Malaya, a species which has the drupes 11-13 mm long; style 5 mm long; pileus rounded; and the leaves 40-50 cm long, 14 mm wide. P. Toei has the drupes 15-18 mm long; style 5-7 mm long; pileus prominently angled; and the leaves 96-130 cm long, 20 mm wide. The new epithet is the vernacular name of the plant in its native region in Thailand. Fig. 195. Pandanus calcis St. John, a-e, h-j, from holotype; f—g from Kerr 13,358. a, Syncarp, X 1; drupe, lateral view, X 1; c, drupe, longitudinal median section, X 1; ^ drupe, apical view, X 1; ^ drupe, median transverse section through horizontal plate, X 4; f, drupe apex and stigma, proximal view, X 4; g, i drupe apex and stigma, lateral view, X 4; h, leaf base, lower side, X 1; i, leaf middle, lower side, X 1; j, leaf apex, lower side, X 1. Page 311: Revision of Pandanus , 1 6. Thailand, Vietnam -St, John 473 section Micro stigma Pandanus catch sp. nov. (sect. Micro stigma) Fig. 195 nom. verm.: "lang kai.” DIAGNOSIS HO'LOTYPI : Truncus 3 m alms ra- mosus eo ramisqne cum "spinis brevibus arma- tis,” foliis 75 cm longis 2-2.5 cm latis tenuibus sed subcoriaceis infra subpallidioribus supra sub catis marginibus subrevolutis in sectione rnediali cum 32 nervis secumdariis in quoque latere ligu- latis sed sensim in apice subulate longe diminu- entibus eo 10 cm ex apice 4 mm lato basi exarmatis sed ex 4 cm marginibus cum serris fortibus arcuatis subulatis 1-2 mm longis 5-15 mm distantibus toto obscure brunneis, midnervo infra inermi, in sectione medialis marginibus cum dentibus 0.2-0. 5 mm longis 6-8 mm dis- tantibus subulatis toto adpressis adscendentibus brunneis, midnervo infra cum dentibus simu- lantibus 3-8 mm distantibus, circa apicem mar- ginibus cum serris 0.2-0.3 mm longis 4-9 mm distantibus brunneis, midnervo infra cum ser- rulis simulantibus 13-15 mm distantibus, pedunculo 7 cm longo 7 mm diametro trigono folioso, syncarpio subgloboso 6 cm diametro ter- minali solitario erecto cum 26 drupis eis 3. 2-3. 4 cm longis 1.7-2 cm latis 1.7- 1.8 cm crassis (sed in gemino 2.6 cm latis) ellipsoideis in sicco brunneis obovoideis sed basi cuneata lateribus laevibus sublucidis 5-anguiatis angulis subrotun- datis parte Vz supera libera, apice rotundato sed biangulatis, stigmate 3 mm longo excentrico obdeltoideo cordato sulcato obscure brunneo pa- pillose valde oblique et proximo, endocarpio supramediali osseoso obscure brunneo 12-13 mm longo lateribus 2 mm crassis, semine 6 mm longo adamantine sed trilobate, mesocarpio apical i plerumque medulloso sicco, mesocarpio basali fibroso et medulloso. DESCRIPTION OF ALL SPECIMENS EXAMINED: Trunk 3-5 m tall, branching, it and the branches "armed with short spines”; leaves 75-84 cm long, 2-2.7 cm wide, rather thin but subcoria- ceous, slightly paler beneath, above with a cen- tral furrow and the margins gently down-rolled, at midsection with 32 secondary veins in each half, ligulate but gradually long tapering to the subulate apex which 10 cm down is 2-4 mm wide, the base unarmed for 4 cm then the mar- gins with heavy arcuate subulate serrations 1-2 mm long, 5-15 mm apart, wholly dark brown, the nearby midrib below unarmed; at the mid- section the margins with the' teeth 0.2-0. 5 mm long, 6-8 mm apart, subulate, fully appressed ascending, brown; the midrib below with simi- lar teeth 3-8 mm apart; near the tip the margins with brown serrations 0.2-0.3 mm long, 4-9 mm apart, the midrib below with similar ser- rations 13-15 mm apart; peduncle 7 cm long, 7 mm in diameter, 3 -angled, leafy bracted; lower bracts 25-33 cm long, 15 mm wide, distended and paler at base; syncarp subglobose 6-7 cm in diameter, terminal, solitary, erect, bearing 20-26 drupes, these 3.2-3 .8 cm long, 1. 7-2.1 cm wide, 1.7—1. 8 cm thick, (but in a twinned one 2.6 cm wide) , ellipsoid, when dried brown, obo- void but the base cuneate, the sides smooth, somewhat shiny, 5-6-angled, the angles more rounded than sharp, upper Vz free, the apex rounded, but two lateral angles produced into the sharp ridges curving and leading to the ex- centric stigma; stigma 2. 5-3. 5 mm long, obdel- toid cordate or oblong cordate or lance-cordate, creased, dark brown, papillose, sharply oblique and proximal; endocarp supramedian, bony, dark brown, with two vertical columns near the cen- ter, 12-13 mm long, a narrow transverse median plate connecting to the marginal vertical flanges 2 mm thick; seed 6 mm long, diamond-shaped but with 3 ascending lobes; apical mesocarp - mostly of dry, membranous pith; basal mesocarp fibrous and pithy. HOLOTYPUS: Thailand (Siam): Kao Tala, Ranawng, on limestone rocks, 200 m. alt., Feb. 3, 1927, A. F. G. Kerr 11,796 (bk). SPECIMENS EXAMINED: Thailand, Kaw Wieng, Chumpawn, limestone rocks, 50 m. alt., Jan. 11, 1927, A. F. G, Kerr 11,373 , nom. vern. "chan deng,” (BK); Ban Kawp Kep, Surat, form- ing pure growth on top of limestone hill, 100 200 m. alt., Aug. 16, 1927, Kerr 13,338 (bk); Sapli, Chumpawn, Sept. 8, 1927, Put, A. F. G. Kerr 1,018 (bk) ; Kantuli, Surat, Sept. 10, 1931, Put, A. F. G. Kerr 4,189 (bk). DISCUSSION: P. edicts is a member of the section Microstigma, as is its nearest relative, P. utilissimus Elmer, of the Philippines, a species which has the syncarp 60 cm long, 20 cm in diameter; drupes 4.5-7 cm long, linear-oblan- ceoloid, the sides semiviscid, the upper VS free; FIG. 196. Pandanus bifdus St. John, from holotype. a, Drupe, proximal view, X 1; b, drupe, longitudinal median section, X 1; c, d, lateral drupes, apical view, X 1; e, lateral drupe and stigma, proximal view, X 4; /. apical drupe, lateral view, X 1; g, h, apical drupes, apical view, XI \ h apex of apical drupe and stigma, oblique view, X 4; j, leaf base, lower side, X 1; ^ leaf middle, lower side, X 1; h leaf apex, lower side, X 1;». margins of base of mature leaf, of Kerr 20,084, X 1. Page 313: Revision of Pandanus , 16. Thailand, Vietnam- St. John 475 endocarp inframedian, lacking distinct marginal flanges; bark grayish, unarmed; and the leaves 5 m long, 8-20 cm wide, the apex abruptly short acuminate. P. calcis has the syncarp 6-7 cm in diameter, subglobose, drupes 3. 2=3. 8 cm long, not viscid, the upper Vl free; endocarp supra- median, with large lateral flanges; bark armed with short spines; and the leaves 75-84 cm long, 2=2.7 cm wide, gradually long tapering to the subulate apex. The new epithet is from the Latin, calx , lime, and is given in reference to the habitat of the species, the side cliffs or summits of precipitous limestone ridges. SECTION Rykia Pandanus bifidus sp. nov. (sect. Rykia) Fig. 196 nom. vern.: "t6i noi.” DIAGNOSIS HOLOTYPI: Planta subacaulis, fo- lds 1.3 m longis 3.3 cm latis subcoriaceis gladiatis in sectione mediali cum 32 nervis secundariis parallelis in quoque latere in apice subulato breve diminuentibus et 10 cm ex apice 14 mm latis proxima basem marginibus cum aculeis 3.5=4 mm longis 25-35 mm distantibus subulatis adpressi-adscendentibus stramineis, midnervo infra cum aculeis 2.5-3 mm longis 50-58 mm distantibus simulantibus sed valde reflexis, in sectio mediali ; marginibus et mid- nervo infra cum aculeis 2.5-3 mm longis 15-48 mm distantibus adscendentibus, proxima apicern marginibus et midnervo infra cum serris 0.8-1. 5 mm longis 2=7 mm distantibus, syncarpio soli- tario terminal! 12 cm longo 10 cm diametro ap- parente ellipsoideo cum drupis numerosis eis 3-3.2 cm longis 8-10 mm latis 8-11 mm crassis oblongo-oblanceoloideis 5-7-angulosis lateribus subplanatis, parte Va supera libera, pileo 4-5 mm alto semiorbiculari 5-7-anguloso sublaevi tarde deciduo, stylo 3.5-5 mm longo valido osseoso corniformato eis lateralibus proximo- curvatis irregulariter bifidis vel in eis apicalibus integris et subulatis, stigma d 3-4 mm longo brunneo papilloso proximo lanceolate vel bifido, endocarpio mediali osseoso obovoideo acumi- nate pallido sed intra obscure brunneo lucido lateribus 1 mm crassis, seminibus 12 mm longis obovoideis, mesocarpio apicale cavernoso, meso- carpio basali fibroso et carnoso. DESCRIPTION OF ALL SPECIMENS EXAMINED: Trunk from almost none to 3 m tall; leaves 1.3-3 m long, 3.3-5 cm wide, subcoriaceous, gladiate, at midsection with 32 secondary veins in each half, tapering to a short subulate apex which 10 cm down is 14 mm wide, near the base the margins with prickles 3.5-4 mm long, 20-35 mm apart, subulate, appressed ascending, stramineous; the nearby midrib below with prickles 2.5-3 mm long, 50-58 mm apart, similar but sharply reflexed; at the middle the margins and midrib below with prickles 2.5-3 mm long, 15-48 mm apart, ascending; near the apex below finely reticulate, and the margins and midrib below with serrae 0.8-2 mm long, 2-17 mm apart; syncarp solitary, terminal, 12 cm long, 10 cm in diameter, apparently ellip- soid, bearing very numerous drupes, these 3-3.2 cm long, 8-10 mm wide, 8-11 mm thick, ob- long-oblanceoloid, 5-7-angled, the sides almost plane, upper Va free; pileus 4-5 mm high, semi- orbicular, 5-7-angled, nearly smooth, finally de- ciduous; style 3.5-5 mm long, stout, bony, horn- like, the lateral ones sharply curved towards apex of syncarp, irregularly bifid, or on the apical ones entire and subulate; stigma 3-4 mm long, brown, papillose, proximal, lanceolate or bifid; endocarp median, bony, obovoid, with an apical acumination, pale except for the dark brown, shiny inner surface, the lateral walls 1 mm thick; seed 12 mm long, obovoid; apical mesocarp cavernous; basal mesocarp fibrous and fleshy. HOLOTYPUS: Thailand (Siam): Pak Tong Chai, Nakawn Rachsima (=Korat), marshy ground, 200 m. alt., Dec. 27, 1923, A. F. G. Kerr 8,132 (bk). SPECIMENS EXAMINED: Thailand (Siam), Kao Krading, Loi, evergreen forest, 1,200 m. alt., Feb. 11, 1931, A. F. G. Kerr 20,084 (bk). DISCUSSION : P. bifidus St. John is a member of the section Rykia. Its closest relative is P. furcatellus Martelli from Tonkin. This has the endocarp in the lower Vdl apical mesocarp cav- ernous, 10-11 mm long; and the style forks ovate, acute. In contrast P. bifdus has the endo- carp median; apical mesocarp cavernous, 4 mm long; and the style forks subulate. The new epithet is the Latin, bifdus, two- cleft, in reference to the cleft style. “T to cm. 0 3 cm. 1 » i t_ i i Fig. 197. Pandanus bipollicaris St. John, from holotype. a, Drupe, lateral view, X 1; b, drupe, longitudinal median section, X 1; c~i> lateral drupes, apical view, X 1; k—n, apical drupes, apical view, X 1; o, drupe, lateral view, X 4; p, drupe, longitudinal median section, X 4; q, apex and stigma of lateral drupe, proximal view, X 4; r, apex of lateral drupe, apical view, X 4; s, apex and stigma of apical drupe, proximal view, X 4; t, leaf base, lower side, X 1; u, leaf middle, lower side, X 1; ^ leaf apex, lower side, X 1; w, in upper third of leaf, lower side, secondary and tertiary veins, X 4; z, in upper third, secondary and tertiary veins, upper side, X 4. Page 315: Revision of Pandmus , 16. Thailand, Vietnam— St. John 477 Pandanus bipollicaris sp. nov. (sect. Rykia ) Figs. 197, 205 a, b DIAGNOSIS HOLOTYPI: Acaulescens caespi- tosus 80- cm diametro, foliis 2-3 m longis 4-5 cm latis supra obscure viridibus lucidisque infra viridibus vel luteo-viridibus supra sulcatis 2- plicatis in sectione mediali cum 36 nervis secundariis parallelis in quoque latere, nervis tertiis transversis multis et in sicco conspicuis et reticulis quadratis vel oblongis formantibus, laminis ligulatis sensim per 40 cm ultimas in 10 cm apice subulato trigono contract is eo circa 10 cm ex apice 2 mm la, to, basi amplexi- cauli inermf luclda cupracea et ex 12-15 cm marginibus cum aculeis 3-6 mm longis 10- 20 mm separatis crassiter subulatis subarcuato- adscendentibus stramineis vel nigris, midnervo infra ex 14-17 cm cum aculeis 4-6.5 mm longis 10-20 mm separatis crassiter subulatis arcuatis reflexis, in sectione mediali marginibus cum aculeis 2.5-3 mm longis 14-30 mm separatis arcuato-subulatis valde adscendentibus basibus incrassatis, in apice contractescentibus margini- bus undulatis et cum subulato-serris 1.5-2 mm longis 1.5-5 mm separatis, midnervo infra an- gusto et cum serris simulantibus, in apice mar- ginibus et midnervo infra cum serrulis 0. 5-0.8 mm longis 1.5-3 mm separatis, inflorescentia foeminea terminali erecta cum syncarpio solita- rio, pedunculo 20 cm longo 12 mm diametro trigono folioso-bracteato, syncarpio ex bracteis tandem siccis et papyraceis subclauso, syncarpio 11.5 cm longo 6.7 cm diametro ellipsoideo sub- trigono viridi cum circa 672 drupis, eis 20-25 mm longis 5-9 mm latis 4-8 mm crassis anguste cuneatis apice hemisphaerico vel pyramidali- hemisphaerico 5-7-anguloso, corpore 19-22 mm longo, basi pile! 5-6 mm alto viridi glauco, stylo 3-4 mm longo brunneo lucido osseoso proxime curvato lato compresso cornuformi bi- fido lobis 0.5-2 mm longis deltoideis (raro inte- gro), stigmatibus 1.5-2. 5 mm longis late ovato- lanceolatis proximis paene ad apicem productis papUlosis brunneis, endocarpio in parte lA in- feta oblongo-ellipsoideo apice truncate intra pallidi lucido lateribus lateralibus 0.2 mm cras- sis osseosis brunneis exteriori sillonati, sernini- bus 7-8 mm longis 3-3.5 mm diametro, meso- carpio apicali albi medullosi in media sine fibris. mesocarpio basali 2-3 mm longo in lateribus fibroso in media carnoso. DIAGNOSIS OF holotype: Stemless herb with several shoots, forming a clump 80 cm in di- ameter; leaves 2-3 m long, 4-5 cm wide, above dark olive green and shining, below green to yellowish green, above with V-shaped furrow over the midrib and two lateral pleats, at mid- section with 36 secondary parallel veins in each half, tertiary cross veins numerous and visible, but on drying conspicuous, forming squarish or oblong meshes, blade ligulate, gradually con- tracted in the last 40' cm to a 10 cm subulate, trigonous apex, this about 10 cm down 2 mm wide, the base amplexicaul, unarmed, shining, dull copper-colored, beginning at 12-15 cm the margins with spines 3-6 mm long, 10-20 mm apart, thick subulate, slightly upcurved, stramin- eous to blackish; the midrib below beginning at 14-17 cm with spines 4-6.5 mm long, 10-20 mm apart, heavy subulate, arcuate, reflexed; at midsection the margins with prickles 2.5-3 mm long, 14-30 mm apart, arcuate subulate from a thickened base, strongly ascending; the midrib below unarmed; in the contracting part near the tip the margins undulate and with subulate- serrae 1.5-2 mm long, 1.5-5 mm apart; the nar- row midrib below with similar serrae; on the apex the margins and midrib below with ser- rulations 0. 5-0.8 mm long, 1.5-3 mm apart; pistillate inflorescence terminal, erect, bearing one syncarp; peduncle 20 cm long, 12 mm in diameter, 3 -sided, leafy bracted; syncarp sur- rounded and nearly hidden by bracts that be- come dry and papery; syncarp immature but ap- parently about full sized, 11.5 cm long, 6.7 cm in diameter, ellipsoid, slightly 3 -sided, green, bearing about 672 drupes, these 20-25 mm long, 5-9 mm wide, 4-8 mm -thick, narrowly euneate, the apex hemispheric or pyramidal-hemispheric, 5-7 -angled, the body 19-22 mm long; p ileus with its base 5-6 mm high hemispheric or pyra- midal-hemispheric, green, glaucous (immature); style 3-4 mm long, brown, shining, bony, proxi- mally curved, broad, flattened hornlike, equally or unequally bifid, the lobes 0.5-2 mm long, deltoid (rarely unbranched this mostly on ter- minal drupes); stigmas 1.5-2. 5 mm long, broadly ovate-lanceolate, proximal, following each lobe almost to the tip, papillose, brown; endocarp in lower oblong-ellipsoid, the apex 478 PACIFIC SCIENCE, Vol. XVII, October 1963 truncate, the inner surface pale, shining, the walls 0.2 mm thick, bony, brown, and the outer surface with sharp longitudinal ridges; seed 7-8 mm long, 3-3.5 mm in diameter; apical meso- carp of white pith, without fibers except up the sides; basal mesocarp 2-3 mm long, fibrous up the sides, fleshy within. HOLOTYPUS: Vietnam, Bach Ma, Province de Thua-Thien, on forested stream bank, 1,400 m. alt., Jan. 26, 1961, H. St. John 26,379 ( BISH ) . DISCUSSION: P. bipollicaris is a member of the section Rykia , as is its closest relative, P. bicornis Ridl. of Malaya, a species which has the peduncle 5-10 cm long; syncarp 7-8 cm long, 5.6 cm in diameter, bearing 100-120 drupes, these 28-34 mm long, 8-10 mm wide; style 2-3 mm long; endocarp central, with a central apical prolongation, the lateral walls 2 mm thick; leaves 0.7-2 m long, at midsection with 24 sec- ondary veins in each half, the blade quickly tapering into a 5 cm subulate apex, and near the base the margins with prickles 2.5-4 mm long and 2-7 mm apart. P. bipollicaris has the pe- duncle 20 cm long; syncarp 11.5 cm long, 6.7 or more cm in diameter, bearing about 672 drupes, these 20-25 mm long, 5-9 mm wide; style 3-4 mm long; endocarp in lower V3, the apex trun- cate, the lateral walls 0.2 mm thick; leaves 2-3 m long, at midsection with 36 secondary veins on each side, very gradually tapering into a 10 cm subulate apex, the margins near the base with spines 3-6 mm long, 10-20 mm apart. The epithet is the Latin adjective, bipollicaris, length of two thumbs or two inches, and is chosen with reference to the leaves which are about two inches wide. Pandanus magnifibrosus sp. nov. (sect. Rykia) Fig. 198 DIAGNOSIS HOLOTYPI: Planta sine caule, foliis "3.65 m. longis,” 4.8 cm latis coriaceis ligulatis in sectione mediali cum 34 nervis secundariis in quoque latere in apice subulato longe diminuen- tibus et 10 cm ex apice solum 16 mm latis triner- vatis et in sectione M-f ormatis, in regione mediali marginibus cum aculeis 2-2.2 mm longis 8-15 mm distantibus arcuato-subulatis subluteis, midnervo infra cum aculeis 2-2.5 mm longis 9-17 mm distantibus salientibus sed api- cibus reflexis, proxima apicem marginibus et midnervo infra cum serris 0. 5-0.8 mm longis 1-4 mm distantibus et nervulis secundariis supra cum serris simulantibus 3-8 mm distantibus, "pedunculo 0.38 m. longo, syncarpio 0.34 m. am- bitu,” drupis numerosissimis 27-30 mm longis 3-4 mm latis et crassis anguste oblanceo-fusi- formibus 5-6-angulatis, parte 3/7 supera libera, pileo 12-13 mm longo arcuato lineari-lanceo- loideo firmo papilloso rugosoque stramineo, stylo 5-6 mm longo subulato arcuato ad apicem vel ad basem syncarpii cartilagineo laevi et lucido, stigmate 6-7 mm longo sublineari brunneo pa- pilloso proximo, endocarpio in parte 2A infera et 12 mm longo pallido, seminibus non evidentis forsan sterilibus, mesocarpio apicali et laterali conjunctis valde fibrosis et medullosis, mesocar- pio basali fibroso et carnoso. DIAGNOSIS OF HOLOTYPE: Stemless; leaves "3.65 m. long,” 4.8 cm wide, coriaceous, ligulate, at midsection with 34 secondary veins in each half, long tapering to a heavy subulate apex which 10 cm down is 16 mm wide; the base not seen; 3 -nerved and in section M-shaped; at the middle the margins with prickles 2-2.2 mm long, 8-15 mm apart, arcuate-subulate, yellow- ish; the nearby midrib below with prickles 2-2.5 mm long, 9—17 mm apart, salient but the apex reflexed; near the apex the margins and midrib below with serrae 0.5-0. 8 mm long, 1-4 mm apart, and the secondary nerves above with simi- lar serrae 3-8 mm apart; "peduncle 0.38 m. long; syncarp 0.34 m. in circumference”; drupes very numerous, 27-30 mm long, 3-4 mm wide and thick, narrowly oblance-fusiform, 5-6-angled, the upper 3/7 free; pileus 12-13 mm long, ar- cuate linear-lanceoloid, firm papillose and ru- gose, stramineous; style 5-6 mm long, subulate, curved and often recurved, cartilaginous, smooth and shining; stigma 6-7 mm long, almost linear, brown papillose, proximal; endocarp in lower 7 6 and 12 mm long, pale; seeds not seen, per- haps sterile, though apparently mature; apical and lateral mesocarp continuous, strongly fibrous and pithy; basal mesocarp fibrous and fleshy. HOLOTYPUS: Thailand (Siam), Kao Soi Dao, Patalung, evergreen forest, 300 m. alt., April 29, 1930, A. F. G. Kerr 19,227 (bk). DISCUSSION : Pandanus magnifibrosus St. John is a member of the section Rykia. Its closest Oj FIG. 198. Pandanus magnifihrosus St. John, from holotype. a, b, Drupe, lateral view, X 1; c, drupe, longi- tudinal median section, X I; d, drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe, apical view, X 4; g, drupe apex and stigma, proximal view, X 4; h, drupe, median transverse section, X 10; i, leaf middle, lower side, X 1; j, leaf apex, lower side, X 1; 4 leaf apex, upper side, X 1- 5 cm.. Fig. 199. Pandanus obconicus St. John, from the holotype. a, Syncarp and bracts, X 1; b, drupe, distal view, X 1; c, drupe, apical view, X 1; ^ drupe, longitudinal median section, X 1; lateral drupes, apical view, X 1; o, apical drupe, lateral view, XI \ P, drupe, proximal view, X 4; q, drupe, longitudinal median section, X 4; r, apex of apical drupe and stigma, distal view, X 4; s, stigma of apical drupe, proximal view, X 4; t, apex of lateral drupe, distal view, X 4; u, apex of lateral drupe and stigma, proximal view, X 4; v, leaf base, lower side, X 1; ^ leaf middle, lower side, X 1; x, leaf apex, lower side, X l;^ lower side of leaf in outer third, with secondary and tertiary veins, X 4. Page 319: Revision of Pandanus, 16. Thailand, Vietnam — St. John 481 relative is P. Sarasinorum Warb. of the Celebes and Minhassa, a species having the syncarps 3.5 cm in diameter, racemose; drupes 12-14 mm long, 2 mm in diameter; style 4 mm long; and the leaf spines becoming black. In contrast, P. magnifibrosus has the syncarps 11.3 cm in di- ameter, solitary; drupes 27-30 mm long, 3-4 mm in diameter; style 5-6 mm long; and the leaf spines yellowish. The new epithet is from the Latin, magnus, large, fibrosus, with fibers, in allusion to the very large, longitudinal fibers in the drupes. Pandanus obconicus sp. nov. (sect. Rykia ) Fig. 199 DIAGNOSIS HOLOTYPI: Planta acaulescens, fo- liis 96 cm longis in media 3.6 cm latis proxima basem 2.7 cm latis coriaceis late sulcatis 2- plicatis in sectione mediali AA-formatis et cum 37-39 nervis secundariis parallelis in quoque dimidio nervis tertiis ad basem et ad apicem conspicuis et reticulos breviter oblongos for- mantibus laminis gladiformatis et in apice 8-12 cm longo trigono subulato eo 10 cm ex apice 3 mm lato basi amplexicauli et inermi sed ex 5-13 cm marginibus cum aculeis 2.5-5 mm longis 7- 15 mm separatis crassiter arcuatis subulatis stramineis adscendentibus, midnervo infra ex 8- 15 cm cum aculeis 5-6 mm longis 12-25 mm separatis subulatis subcurvatis valde reflexis in sectione mediali marginibus cum aculeis 2-3 mm longis 6-9 mm separatis crasse subulatis adpresse adscendentibus, midnervo infra cum aculeis 3-4 mm longis remotis subulatis arcuatis basi incrassati reflexis vel adscendentibus, in apice subulato marginibus cum serris 0.5-1 mm longis 2-4 mm separatis, midnervo infra cum aculeis simulantibus 4-6 mm separatis, infructe- scentia terminali erecta cum syncarpio unico, pedunculo 17 cm longo 1 cm diametro trigono folioso-bracteato, bracteis superis 9-10 cm longis 4 cm latis minute serrulatis dentibus 0.2 mm longis, syncarpio 7.5 cm longo 4.5 cm diametro ellipsoideo cum circa 496 drupis eis 16-19 mm longis 7-9 mm latis 4-7 mm crassis eis termi- nalibus majoribus crassiter cuneiformatis com- pressis 5-7-angulatis parte Ve supera libera corpore 14-17 mm longo lateribus planis, pileo 3-4 mm alto basi depresse obtuso 5-7-anguloso minute papilloso, stylo 2-3 mm longo osseoso brunneo lucido compresso proxime curvato (vel V5-V2 ) bifido lobis plerumque divergen- tibus, stigmatibus 1.5-2 mm longis apicalibus vel in parte proximalibus brunneis papillosis, endocarpio infra mediali osseoso pallide brun- neo lateribus 0.6-0.7 mm crassis intra obscure brunneis lucidis apice cum acumine fragili, se- mine 7 mm longo anguste obconico, mesocarpio apicali grandi cavernoso cum membranis medul- losis albis, mesocarpio basali fibroso et carnoso. DIAGNOSIS OF HOLOTYPE: Plant acaulescent; leaves 96 cm long, 3.6 cm wide at the middle, 2.7 cm wide near the base, coriaceous, broadly furrowed above the midrib, 2 -pleated, in section AA-shaped, at midsection with 37-39 parallel secondary veins in each half, the tertiary cross veins marked near the base and the tip, forming short oblongs, and from these the outer third of the blade has a reticulate appearance, the blade sword-shaped, narrowed to an 8-12 cm trigonous subulate apex which 10 cm down is 3 mm wide, the base amplexicaul, unarmed, but beginning at 5-13 cm the margins with prickles 2.5-5 mm long, 7-15 mm apart, stout arcuate subulate, as- cending, stramineous; the midrib below begin- ning at 8-15 cm with prickles 5-6 mm long, 12-25 mm apart, subulate, slightly curved, sharply reflexed; at the midsection the margins with prickles 2-3 mm long, 6-9 mm apart, stout subulate, appressed ascending; the midrib below with remote prickles 3-4 mm long, subulate, arcuate, with an incrassate base, reflexed or as- cending as here the change-over occurs; on the subulate apex the margins with serrations 0.5-1 mm long, 2-4 mm apart; those of the midrib similar but 4-6 mm apart; infructescence termi- nal, erect, bearing one syncarp; penduncle 17 cm long, 1 cm in diameter, trigonous, leafy bracted, the upper bracts 9-10 cm long, 4 cm wide, minutely serrulate, the teeth 0.2 mm long; syncarp 7.5 cm long, 4.5 cm in diameter, ellip- soid, bearing about 496 drupes, these 16-19 mm long, 7-9 mm wide, 4-7 mm thick, the apical ones perceptibly the larger, thick cunei- form, compressed, 5-7-angled, upper Ve free, the body 14-17 mm long, the sides flat; pileus 3-4 mm high, the base low obtuse, 5-7 -angled, minutely papillose; style 2-3 mm long, bony, brown, shining, flattened, curved proximally, bifid usually 2A way, rarely ]/$ or V2 way, the lobes usually divergent; stigma 1.5-2 mm long, 482 PACIFIC SCIENCE, Vol. XVII, October 1963 on the apex or partly on the proximal side of each lobe, brown, papillose; endocarp subme- dian, bony, pale brown, the lateral walls 0.6-0. 7 mm thick, the inside surface dark brown, shiny, crowned by a weak apical acumination; seed 7 mm long, narrowly obconic; apical mesocarp a large cavern divided by white medullary mem- branes; basal mesocarp fibrous and fleshy. HOLOTYPUS: Thailand, s. e. part, Chanburi, Makham, 100 m. alt., common along edge of savannah in scrub forest, 13 Jan. 1958, Tem Smitinand 4,054 (SING). DISCUSSION: P. obconicus is a member of the section Rykia, as is its closest relative, P. hori- zontalis St. John, of Vietnam, a species with peduncle 45 cm long; syncarp 10.5 cm long, 7-8 cm in diameter, with about 720 drupes, these 26-31 mm long, 9-11 mm wide; pileus base high rounded conic; endocarp in lower P3; seed 12 mm long; stem 1.5 m tall; leaves 2.7 m long, at midsection with 33 secondary veins in each half, and near the base the midrib below with thorns 3-4 mm long, 17-35 mm apart. P. ob- conicus has the peduncle 17 cm long; syncarp 7.5 cm long, 4.5 cm in diameter, with about 496 drupes, these 16-19 mm long, 7-9 mm wide; pileus base low obtuse; endocarp submedian; seed 7 mm long; acaulescent; leaves 96 cm long, at midsection with 37-39 secondary veins in each half, and near the base the midrib below with prickles 5-6 mm long, 12-25 mm apart. The new epithet is Latin, ob, inversed, coni- cus, conical, given in reference to the obconic shape of the endocarp. Pandanus obovatus sp. nov. (sect. Rykia ) Fig. 200 DIAGNOSIS HOLOTYPI: Arbor ”6 m. alta ra- mosa, trunco cum spinis parvis,” foliis 1.1 m plus longis 3.1 cm latis subcoriaceis in sectione mediali cum 39 nervis secundariis in quoque di- midio gradatim ad apicem longe diminuentibus supra secundum midnervum canaliculatis mar- ginibus reflexis, basi integra amplexicaulique sed ex 5-7 cm marginibus cum aculeis pallidis 2-2.5 mm longis 5-9 mm distantibus subulatis adscendentibus, midnervo infra per 15 cm inermi deinde cum aculeis reflexis 0.5-1 mm longis 15-18 mm distantibus subulatis, in sec- tione mediali marginibus cum aculeis 0.8-0. 9 mm longis 7-10 mm distantibus subulatis ad- scendentibus, midnervo infra cum aculeis 0.3- 0.5 mm longis 8-18 mm distantibus simulanti- bus adscendentibus, ad apicem marginibus mid- nervoque infra cum serrulis 0.2 mm longis 5-15 mm distantibus, syncarpio "cum 15-18 carpelis,” drupis 7. 5-8. 3 cm longis 4.9 cm latis 3-7 cm rrassis obovoideis deltoideis laevissimis, lateri- bus superis laevibus brunneo-lucidis quasi cera- tis et pallide multirimosis deltoideis, parte supera libera, parte V2 infera pallide brunnea et cum dorsis et valleculis angustis longitudinalibus intercbseratis, apice rotundato sed cum dorsis duobis ad stigmatem connectis, stylo cornicu- lato compresso divergento, stigmate 8 mm longo late obovato exsulcato subverticale brun- neo, endocarpio submediali parte centrali 4.3 cm longa osseosa obscure brunnea ellipsoidea muris lateralibus 2-3 mm crassis et cum humeris fortibus pallidis lateralibus submediale affixis eis oris 3. 5-4. 5 cm longis marginalibus crassis, semine unico 25 mm longo 9 mm diametro ellipsoideo mesocarpio laterali apicalique con- tinuo sicco et medulloso sed cum fibris longitu- dinalibus, mesocarpio basali dimidio supero simulanti, dimidio infero fibroso et carnoso. DIAGNOSIS OF HOLOTYPE: Tree ”6 m. tall, branching; stem with small spines”; leaves more than 1.1 m long, 3.1 cm wide, subcoriaceous, at midsection with 39 secondary veins in each half, gradually long tapering towards the apex, (the tip missing in ours), channeled above the midrib, the margins reflexing, the base am- plexicaul and entire, but beginning 5-7 cm up the margins with pale prickles 2-2.5 mm long, 5-9 mm apart, subulate, ascending; the midrib below unarmed for 15 cm, then with reflexed prickles 0.5-1 mm long, 15-18 mm apart, subulate; at midsection the margins with prickles 0. 8-0.9 mm long, 7-10 mm apart, su- bulate, ascending; the midrib below with similar appressed ascending prickles 0.3-0. 5 mm long, 8-18 mm apart; towards the apex the margins and midrib below with serrations 0.2 mm long, 5-15 mm apart; syncarp "with 15-18 carpels”; drupes 7. 5-8. 3 cm long, 4.9 cm wide, 3.7 cm thick, obovoid, 3 -sided, very light, the upper sides smooth, brown and shiny as if varnished and with many pale longitudinal cracks, 3- angled, upper lA free, lower half pale brown and with several sharp longitudinal ridges and r \ y \ y \ y \ \ , FIG. 200. Pandanus ohovatus St. John, from holotype. a, Drupe, lateral view, X 1; b, drupe, longitudinal median section, XI \ c, drupe, apical view, X 1; 4 stigma, proximal view, X 1; e, apex of drupe, lateral view, X 1; /> leaf base, lower side, X 1; g, leaf middle, lower side, X 1; h, leaf apex, lower side, X 1- 484 PACIFIC SCIENCE, Vol. XVII, October 1963 valleys like mortising angles, the apex rounded but from each of two angles a ridge mounts to support the stigma on a flattened, overhanging process or style; stigma 8 mm long, broadly obo- vate, without crease, nearly vertical, brown; en- docarp submedian, the central part 4.3 cm long, bony, dark brown, and ellipsoid, its lateral walls 2-3 mm thick, a little below the middle of this central endocarp are heavy, pale, lateral shoul- ders bearing heavy marginal flanges 3. 5-4. 5 cm long; seed solitary, 25 mm long, 9 mm in di- ameter, ellipsoid; lateral and apical mesocarp continuous, dry and pithy but with longitudinal fibers, and this pithy mesocarp also occupies the upper half of the area below the endocarp, the remaining basal area fibrous and fleshy. HOLOTYPUS: Thailand (Siam): Menam Pasak, Keng Koi, on rocky limestone hill, Dec. 9, 1923, A. F. G. Kerr 7,669 (bk). DISCUSSION: P. ohovatus is a member of the section Rykia, though with its large, dry, pithy drupes it is unique and without a close relative. It can be compared to P. borneensis Warb. which has the syncarps spicate, 6-8 cm long; drupe apex 10-12 mm broad; and the style 2 mm long. In P. obovatus the syncarp is solitary, about 23 cm long; drupe apex 4.9 cm broad; and the style 1 1 mm long, horn-like, flattened. The type specimen consists of a leaf minus the tip, and three fruit halves, yet so unique is the species that this much suffices. There are no really close relatives known. The new epithet is the Latin obovatus, refer- ring to the obovate profile of the drupes. Pandanus reticulosus sp. nov. (sect. Rykia ) Fig. 201 NOM. VERN.: "toei yai” (— Pandanus big). diagnosis holotypi: Licet acaulescens, foliis 2 m longis 5 cm latis in medio 3.8 cm latis circa basem coriaceis reticulatis in sectione mediale cum 33 nervis secundariis parallelis fortibus in quoque dimidio, nervis tertiis transversis fortibus elevatis et reticulam quadratorum vel oblongo- rum brevium formantibus, laminis ligulatis sed subabrupte in apice circa 10 cm longo caudato trigono subulato diminuentibus eo circa 10 cm ex apice 5 mm lato, basi amplexicauli et inermi pallida sed ex 7-15 cm marginibus cum aculeis 3-5 mm longis 12-22 mm separatis subulatis curvatis fortibus pallidis adscendentibus, mid- nervo infra ex 12.5 cm cum aculeis 3-6 mm longis 32-65 mm separatis arcuato-subulatis reflexis, in sectione mediali marginibus cum aculeis 2.8-3 mm longis 14-24 mm separatis arcuatis subulatis adscendentibus, midnervo infra inermi, in apice subulato marginibus cum ser- rulis 0.6-1 mm longis 1-3 mm separatis, mid- nervo infra cum serrulis 0. 3-0.8 mm longis 3-4 mm separatis, pedunculo "brevi,” syncarpio soli- tario 7 cm longo 5.5 cm diametro latiter ellip- soideo purpurascenti cum 113 drupis, eis 22-26 mm longis 9-14 mm latis 9-11 mm crassis ob- lanceoloideis basi truncata 6-angulosis parte Va supera libera, corpore 18-22 mm longo later ibus planis, pileo 6-8 mm alto basi 4-5 mm alto semiorbiculari vel depresse semiorbiculari rugosa exlaevi minime 6-angulosa, stylo 2.5-5 mm longo osseoso brunneo lucido compresso proxime curvato eis druparum lateralium plerumque breviter bifidis sed aliquis basalibus subulatis, stigmate 1-1.5 mm longo cordato vel late cor- dato brunneo papilloso in apice proximali caver - noso affix!, endocarpio mediali 11 mm longo osseoso extus pallid! lateribus 0.8 mm crassis intra obscure brunneis et lucidis, seminibus 10 mm longis 8 mm diametro obovoideis, meso- carpio apicali aerenchymati cum membranis latis transversis albis, mesocarpio basali fibroso et carnoso. DIAGNOSIS OF holotype: Apparently acau- lescent, gregarious on ground among under- shrubs; leaves 2 m long, 5 cm wide at the mid- dle, 3.8 cm wide near the base, coriaceous, reti- culately veined and conspicuously so in outer half, at midsection with 33 parallel secondary veins in each half, the tertiary cross veins heavy, raised, forming a reticulum of squares or short oblongs, blade ligulate, rather abruptly narrowed to a caudate trigonous subulate apex about 10 cm long, this about 10 cm down 5 mm wide, the base amplexicaul and unarmed, pale, but be- ginning 7-15 cm up the margins with spines 3-5 mm long, 12-22 mm apart, stout curved subulate, ascending, pale; the midrib below be- ginning at 12.5 cm with spines 3-6 mm long, 32-65 mm apart, arcuate subulate, reflexed; at midsection the margins with prickles 2.8-3 mm long, 14-24 mm apart, arcuate subulate, ascend- t o 1 o cm/. Fig. 201. Pandanus reticulosus St. John, from holotype. a, Drupe, lateral view, X 1; b, drupe, longitudinal median section, X 1; c— h drupe, apical view, X 1; (d and / are the most common kinds; e is basal, rarely lateral); m, apex of apical drupe and stigma, lateral view, X 4; n, apex of lateral drupe, oblique view, X 4; o, apex of lateral drupe and stigma, proximal view, X 4; p, leaf base, lower side, X 1; leaf middle, lower side, XI \ leaf apex, lower side, X 1; s, leaf middle, upper side, secondary and tertiary veins, X 4; t, leaf middle, lower side, secondary and tertiary veins, X 4. 486 PACIFIC SCIENCE, Vol. XVII, October 1963 ing; the midrib below unarmed; on the subulate apex the margins with serrations 0.6-1 mm long, 1-3 mm apart; the midrib below with serrations 0. 3-0.8 mm long, 3-4 mm apart; pe- duncle "short”; syncarp single, 7 cm long, 5.5 cm in diameter, broadly ellipsoid, purplish, bear- ing 113 drupes, these 22-26 mm long, 9-14 mm wide, 9-1 1 mm thick, oblanceoloid from a trun- cate base, 6-angled, upper Va free, the body 18-22 mm long, with flat sides; pileus 6-8 mm high, its base 4-5 mm high, semiorbicular or oblately so, the surface roughened, dull, weakly 6-angled; style 2.5-5 mm long, bony, brown, shining, flattened, proximally curved, the lateral ones mostly shortly bifid, but some of the basal ones with a single, subulate tip; stigma 1-1.5 mm long, cordate or broadly so, brown, papil- lose, in the hollow proximal tip; endocarp me- dian, 1 1 mm long, bony, pale without, the walls 0.8 mm thick, within dark brown and shining; seed 10 mm long, 8 mm in diameter, obovoid; apical mesocarp an aerenchyma with broad transverse, white membranes; basal mesocarp fi- brous and fleshy. HOLOTYPUS: Thailand, n. e. region, Loei, Phu Krading, Tham Saw, common in evergreen for- est, 1,300 m. alt., Tem Smitinand 406 (bkf no. 11,999). DISCUSSION: P. reticulosus is a member of the section Rykia, as is its closest relative, P. bicornis Ridl. of Malaya, a species which has the syncarp 15 cm long; drupes 34 mm long; style forks spreading and with the tips recurved; seed 6 mm long, cylindric; and the leaf margin near the base with spines 2.5-4 mm long, and 2-7 mm apart. P. reticulosus has the syncarp 7 cm long; drupes 22-26 mm long; style forks as- cending or diverging; seed 10 mm long, obo- void; and the leaf margin near the base with spines 3-5 mm long, and 12-22 mm apart. The new epithet is the Latin adjective, reticu- losus, with a network, and is given in reference to the reticulate venation of the leaves. Pandanus thailandicus sp. nov. (sect. Rykia ) Figs. 202, 203 nom. vern.: "toei nam.” diagnosis holotypi: Frutex 1.5 m altus ramosus, foliis 35-37 cm longis 6-8 mm latis subcoriaceis ligulatis plana tis vel marginibus paene involutis in sectione mediali cum 13 nervis secundariis parallelis in quoque dimidio parte Vi supera sensim in apice subulato dimi- nuenti 10 cm ex apice 4 mm lata in sicco palli- dis et infra subglaucis in basi amplexicaulibus et integris proxima basem marginibus cum acu- leis 1.8-2. 2 mm longis 8-15 mm separatis ad- scendentibus arcuato-subulatis stramineis, mid- nervo infra cum aculeis 2.2-2. 5 mm longis 8-25 mm distantibus simulantibus adscendentibus, in sectione mediali marginibus cum aculeis 1. 3-1.8 mm longis 4-7 mm separatis simulantibus sed subadpressis, midnervo infra cum aculeis 1.7-2 mm longis 6-15 mm separatis simulantibus ad- scendentibus, proxima apicem marginibus cum serris 0. 1-0.3 mm longis 1-4 mm separatis ad- pressis cum apicibus subulatis, midnervo infra cum serris 0.2-0.6 mm longis 2.5-6 mm sepa- ratis simulantibus, pedunculo terminali 7 cm longo 3-4 mm diametro folioso-bracteato, syn- carpio solitario pendenti 36 mm longo, 35 mm diametro elliptico-subgloboso cum circa 230 drupis, eis 14-1 6 mm longis 4-4.2 mm latis 3 mm crassis anguste oblanceoloideis 5-7-angu- losis lateribus subcurvatis parte Va supera libera, pileo conico sed in sicco cum rugis longitudi- nalibus, stylo 2.5-3 mm longo crasse subulato cartilagineo pallido proximo-curvato, stigmate 2-2.5 mm longo lanceolato proximo brunneo papillose, endocarpio centrali 6 mm longo an- guste obovoideo cartilagineo obscure brunneo intra lucido lateribus 0.1 mm crassis, mesocar- pio apicali lateralique conjuncto sicco et medul- loso, mesocarpio basali fibroso et carnoso. DESCRIPTION OF ALL SPECIMENS EXAMINED: Branching shrub 1.5 m tall; stem 8-10 mm in diameter, yellowish, unarmed; prop roots not mentioned; leaves 35-50 cm long, 6-10 mm wide, subcoriaceous, ligulate, flat or the margins slightly involute, at midsection with 13 parallel secondary veins in each half, the upper Vi gradu- ally tapering to the acute tip, this 10 cm from the apex 4 mm wide, when dried pale and below somewhat glaucous, the very base amplexicaul and unarmed, near the base the margins with prickles 1.8-2 .2 mm long, 8-15 mm apart, ascending, arcuate subulate, stramineous; the nearby midrib below with similar ascending prickles 2. 2-2. 5 mm long, 8-25 mm apart; at midsection the margins with prickles 1.3-1. 8 mm long, 4-7 mm apart, similar but subap- Fig. 202. Pandanus thailandicus St. John, from holotype. a, Syncarp, XI \ b, drupe, lateral view, X 1; c, drupe, longitudinal median section, XI \ d, drupe, lateral view, X 4; e, drupe, longitudinal median section, X 4; f, drupe, apical view, X 4; g, b, apex of drupe and stigma, proximal view, X 10; /, leaf base, lower side, >< 1; j, leaf middle, lower side, X 1; ^ leaf apex, lower side, X 1* Fig. 203. Pandanus thailandicus St. John, from Dee Bun Phong 914. a, Staminate inflorescence, X 1; b, column and anthers, X 10; c, leaf base, lower side, X 1; d, leaf middle, lower side, X 1; e, leaf apex, lower side, X 1* Page 327: Revision of Pandanus, 16. Thailand, Vietnam — St. John 489 pressed; the midrib below with similar prickles 1.7-2 mm long, 6-15 mm apart, ascending; near the apex the margins with subulate-tipped ap- pressed serrations 0. 1-0.3 mm long, 1-4 mm apart; those of the midrib below similar but 0.2-0.6 mm long, 2.5-6 mm apart; peduncle terminal, 7-12 cm long, 3-4 mm in diameter, with many leafy bracts, the median ones 21 cm long, 8 mm wide, the upper ones 7 cm long, 17 mm wide, foliaceous, white; syncarp single, pendent, elliptic-subglobose, 36-55 mm long, 35-42 mm in diameter, bearing 230-592 drupes, these 14-19 mm long, 4-4.2 mm wide, 3 mm thick, narrowly oblanceoloid, 5-7 -angled, the sides gently curving, upper lA free; pileus conic, but when dry with longitudinal wrinkles, the base connate with adjacent ones, leaving no visible suture; style 2-3.5 mm long, stout su- bulate, pale, cartilaginous, curved upward to- wards apex of syncarp ( or the basal ones curved distally); stigma 1.5-2. 5 mm long, lanceolate, proximal, brown papillose; endocarp central, 6 mm long, narrowly obovoid, cartilaginous, brown, shiny within, the walls 0.1 mm thick; apical and lateral mesocarps continuous, dry pithy; basal mesocarp fibrous and fleshy. STAMINATE PLANT: Stem 8 cm or a little more in height, 7-10 mm in diameter, dark brown, striate, shining; leaves numerous in terminal tuft, 46-48 cm long, just above the base 12-13 mm wide, at the middle 7-8 mm wide, green, secondary parallel veins prominent throughout and at midsection 14 in each half, the base amplexicaul, unarmed, softer, beginning 3.5-4 cm up the margins with prickles 2. 3-3. 5 mm long, 5-15 mm apart, arcuate, heavy subu- late, ascending, yellow; the nearby midrib below beginning at 5-5.5 cm with prickles 2-2.5 mm long, 6-15 mm apart, similar but reflexed, at the point of reversal of direction often paired; at midsection the margin and midrib below with similar prickles 1-2.5 mm long, 4-6 mm apart, ascending; staminate inflorescence terminal; pe- duncle 6 cm long, 1.5 mm in diameter, leafy bracted; floral part 11 cm long; lowest floral bract about 45 cm long, the base 15 mm wide, soft, white, veiny, unarmed, narrowly lanceolate, the upper part 6 mm wide, green, subcoriaceous, at midsection the margins with prickles 0.5-1 mm long, 2-3 mm apart, arcuate subulate, as- cending, yellow; the midrib below with similar ones 0.3-0. 5 mm long, 1.5-3 mm apart; the 7 spikes fragrant, 2.3-3 cm long, 6-8 mm in di- ameter, narrowly ellipsoid, densely flowered; stamens numerous, in fascicles; staminal column 2-2.5 mm long, stout, but tapering upwards; free filament tips 0.5 mm long; anthers 0.7-0.8 mm long, lanceolate, bearing on its apex a su- bulate projection of the connective 0.2 mm long. HOLOTYPUS: Thailand (Siam): Pii Wieng, Kawn Ken, by dry stream bed, 300 m. alt., Feb. 7, 1931, A. F. G. Kerr 20,016 (bk). specimens EXAMINED: Thailand (Siam), Bung, Ubon, by stream in open forest, 100 m. alt., Jan. 29, 1924, A. F. G. Kerr 8,365 (bk); n. e. part, Loei, Phu Luang, by stream, 1,000 m. alt., 24 Sept. 1957, Dee Bun Pheng 914, stami- nate and pistillate (bkf no. 16,154). DISCUSSION: P. thailandicus is a member of the section Rykia, as is its closest relative, P. immersus Ridl. of Malaya, a species with the syncarp (immature) 9.5 cm long, 6 cm in di- ameter, ellipsoid, with about 1,440 drupes, these 5-6 mm wide; pileus 7-8 mm long; style 4-6 mm long; plant aquatic; leaves "many feet long," 10.2 cm wide, caudate subulate apex 3.0. , cm long, at midsection with 70 secondary veins in each half, and the margins with prickles 2.5-4 mm long, 9-18 mm apart. P. thailandicus has the syncarp 3.6-5. 5 cm long, 3. 5-4.2 cm in diame- ter, elliptic-subglobose, with 230-592 drupes, these 4-4.2 mm wide; pileus 3.5-6 mm long; style 2-3.5 mm long; plant terrestrial; leaves 35-50 cm long, 6- 10 mm wide, the subulate apex 7 cm long, at midsection with 13 veins in each half and the margins with prickles 1.3-1. 8 mm long and 4-7 mm apart. The new epithet is geographic, referring to the native country of the species. Pandanus unicornutus sp. nov. (sect. Rykia) Fig. 204 NOM. VERN.: "chawng 11”; "toi yai.” DIAGNOSIS HOLOTYPI: Arbor ad 8 m alta ramosa, foliis "ad 3.5 m. longis” 10-11 cm latis proxima basem crassiter coriaceis in apice subu- lato 8 cm longo abrupte contractis eo 10 cm ex apice 7 mm lato in sectio late M-formatis in puncto 30 cm ex basi cum 70 nervis secundariis in quoque latere, basi integra et amplexicauli sed ex 15 cm marginibus cum spinis 4.5-6 mm longis Fig. 204. Pandanus unicornutus St. John, from holotype. a, Infructescence, X Vdl h, drupe, lateral view, X 1; c, drupe, longitudinal median section, XI \ d, drupe, apical view, X 1; e> drupe apex and stigma, proximal view, X 4; f, rare drape apex with bifid stigma, apical view, X 1; S> apex of rare drupe with bifid stigma, proximal view, X 4; h, leaf base, lower side, X h leaf middle, lower side, X 1» is lea^ apex, lower side, X 1- Page 329: Revision of Pandanus, 16. Thailand, Vietnam — St. John 491 Fig. 205 a, Pandanus hipollicaris St. John, from holotype. Branch, leaves, and syncarp. 7- 15 mm distantibus basi crassa apice subulato subarcuatis divergentibus vel subadscendentibus pallide brunneis, midnervo infra cum spinis 4-5 mm longis 10-41 mm distantibus crasse conico- subulatis reflexis, in sectio mediali foliis 12.5-13 cm latis in paginis ambis rugoso-reticulatis marginibus cum spinis 4-6 mm longis 21—38 mm distantibus adscendentibus basi crassa apice crassiter subulato, midnervo inermi, proxima apicem marginibus et midnervo infra cum serris subulatis 0. 5-0.8 mm longis 2-4 mm distantibus paginis ambis rugoso-reticulatis, laminis in sec- tio M-formatis fere ad apicem, spicis cum 7 syncarpiis, "syncarpio maximo 15 cm. longo” 11 cm diametro, drupis multis 3. 5-3. 9 cm longis 8- 10 mm latis crassisque anguste ellipsoideis lateribus laevibus planis vel subcurvatis 5-angu- latis parte 1/7 supera libera, pileo oblato-semi- orbiculari firmo quoque solum dehiscenti, stylo 6-8 mm longo osseoso corniformi integri ( rari- ter bifido) proxime curvato pallide brunneo, stigmate 3-4 mm longo lanceolate brunneo pa- pillose, endocarpio submediali 25 mm longo la- teribus 0.1 mm crassis cartilagineo stramineo pagina interiori lucido apice lanceoloideo pro- ducto, seminibus 15-16 mm longis 5-6 mm FlG. 205 b. Pandanus hipollicaris St. John, from holotype. Branch, leaves, and syncarp, and longitudinal median section of syncarp-. diametro ellipsoideo, mesocarpio apicali in mar- ginibus fibroso intra cavernoso, mesocarpio basali fibroso et carnoso. DESCRIPTION OF ALL SPECIMENS EXAMINED: Tree 7-15 m tall, 17-20 cm in diameter, branch- ing; bark pale, with short spines; "prop roots to 1 m. long, 5-8 cm. in diameter, armed with short spines”; leaves '"3.2-3.85 m. long,” 8.4-11 cm wide near the base, thick coriaceous, abruptly acuminate to a subulate apex 8 cm long, the tip 10 cm down from the apex 7 mm wide, very wide M-shaped in section, at 30 cm from the base with 70 secondary veins in each half, the base entire and amplexicaul, but beginning 15 cm up the margins with thorns 4.5-7 mm long, 7-15 mm apart, heavy based, subulate-tipped, somewhat arcuate, divergent or slightly ascend- ing, pale brownish; the midrib below with thorns 4-5 mm long, 10-14 mm apart, heavy conic-subulate, reflexed; at midsection 12.5-13 cm wide, both surfaces rugose reticulate, the margins with thorns 4-6 mm long, 21-38 mm apart, ascending, the base heavy, the tip thick subulate; the nearby midrib prominent but un- armed; near the apex the margins and midrib below subulate-serrate, the teeth 0. 5-0.8 mm 492 PACIFIC SCIENCE, Vol. XVII, October 1963 long, 2-4 mm apart, both the upper and lower surfaces rugose-reticulate from the heavy cross veinlets, the blade M-shaped in section and the pleats running almost to the tip; peduncle 70 cm long; spike with 5-7 syncarps, the largest 14-18 cm long, 11-12 cm in diameter, ovoid but distinctly 3 -sided; drupes numerous, these 3. 5-4.3 cm long, 8-10 mm wide and thick, nar- rowly ellipsoid, the sides smooth, plane or gently curved, 5 -angled, upper 1/7 free; pileus oblate- semiorbicular, firm, shedding singly, ending in a style 6-8 mm long, bony, horn-like, entire ( rarely bifid ) , strongly curved towards the apex of the syncarp, pale brownish; stigma 3-4 mm long, lanceolate, brown papillose; endocarp slightly submedian, 25 mm long, the lateral walls 0.1 mm thick, cartilaginous, stramineous, the inner surface shining, the apex produced upwards into a long lanceoloid tip; seeds 15-16 mm long, 5-6 mm in diameter, ellipsoid; apical mesocarp fibrous around the margin, cavernous within; basal mesocarp fibrous and fleshy. holotypus: Thailand (Siam), Kaw Tao, Surat, in high evergreen forest, under 5 m. alt., April 16, 1927, A. F . G. Kerr 12,772 (bk). specimens EXAMINED: Thailand (Siam), Kaw Tao, Surat, evergreen forest, 300 m. alt., Sept. 21, 1928, A. F. G. Kerr 16,052 (bk); Wangka, Kanburi, by stream in evergreen forest, 700 m. alt., Feb. 3, 1926, A. F . G. Kerr 10,432 ( BK ) ; Ta Ngaw, Chumpawn, common along streams in savannah evergreen {forest}, Jan. 22, 1927, A. F. G. Kerr 11,601 (bk). DISCUSSION : Pandanus unicornutus is a mem- ber of the section Rykia, as is its closest relative, P. penangensis Ridley, a Malayan species which has the drupes 42-47 mm long, 6-1 mm thick; endocarp 32-33 mm long, the lateral walls 1-1.5 mm thick; and the leaves 12 cm wide, near the base the margins with prickles 1.3-1. 5 mm long, 4-7 mm apart. P. unicornutus has the drupes 35-43 mm long, 8-10 mm thick; endocarp 25 mm long, the lateral walls 0.1 mm thick; and the leaves 10-11 cm wide, near the base the margins with spines 4.5-6 mm long, 7-15 mm apart. The epithet is from the Latin unus, one, cor- nutus, with a horn, in allusion to the usual single hornlike style. Index to Volume XVII Author Index Banner, A. H. : On Malayan Shores, (review) , 374 Barlow, George W. : Species Structure of the Gobiid Fish Gillichthys mirabilis from Coastal Sloughs of the Eastern Pacific, 47-72 Bonham, Kelshaw, and Edward E. Held : Ecological Observations on the Sea Cucumbers H olothuri&'atra and H. leucospilota at Rongelap Atoll, Marshall Islands, 305-314 Bowers, Darl E. : Field Identification of Five Species of Californian Beach Hoppers (Crustacea: Amphipoda), 315- 320 Carlquist, Sherwin, and Martin L. Grant: Studies in Fitchia (Compositae) : Novelties from the Society Islands; Anatomical Studies, 282-298 Chang, Jen-hu: Role of Climatology in the Hawaiian Sugar-Cane Industry: An Example of Applied Agricultural Climatology in the Tropics, 379-397 CHUANG, S. H. : On Malayan Shores, (review of) , 374-375 Doty, Maxwell S.: Gihsmithia haw alien sis, gen. n. et sp. n., 458-465 Emery, K. O. : Aerial Study of Hawaiian Wave Patterns, 255—260 Gosline, William A.: Notes on the Osteology and Systematic Position of Hypoptychus dybowskii Steindachner and Other Elongate Perciform Fishes, 90-101 Hida, Thomas S., and Robert A. Morris: Preliminary Report on the Marquesan Sardine, Harengula vittata, in Hawaii, 431-437 Hobson, Edmund S. : Feeding Behavior in Three Species of Sharks, 171- 194 Hoyt, Charles P. : Investigations of Rhinoceros Beetles in West Africa, 444-451 Hsiao, Sidney C, Walter K. Fujii, and Helen H. Fine : Simple Device for Making Successive Photomicro- graphic Records of Large Groups of Developing Organisms, 321-328 Inman, D. L., W. R. Gayman, and D. C. Cox: Littoral Sedimentary Processes on Kauai, a Sub- tropical High Island, 106-130 Lamoureux, Charles H. : Additional Plants from the Midway Islands, 374 Lewis, Alan G. : Life History of the Caligid Copepod Lepeophtheirus dissimulatus Wilson, 1905 (Crustacea: Caligoida), 195-242 Martin, Edgar J.: Toxicity of Dialyzed Extracts of Some California Anemones ( Coelenterata ) , 302—304 Matthews, Donald C. : Hawaiian Records of Folliculinids (Protozoa) from Submerged Wood, 438— 443 Nizamuddin, Mohammed : Studies on the Green Alga, Udotea indica A. & E. S. Gepp, 1911, 243-245 Okutani, Takashi : Preliminary Notes on Molluscan Assemblages of the Submarine Banks Around the Izu Islands, 73-89 Paramonov, S. J. : Lord Howe Island, A Riddle of the Pacific, Part III, 361-373 Pemberton, C E. : Important Pacific Insect Pests of Sugar Cane, 251- 252 Pequegnat, Willis E. : Population Dynamics in a Sublittoral Epifauna, 424-430 PLUCKNETT, D. L., J. C. Moomaw, and C. H. Lamoureux: Root Development in Aluminous Hawaiian Soils, 398-406 St. John, Harold : Revision of the Genus Pandanus Stickman, Part 14. New Species from Malaya and Singapore, 3-46 Part 15. Malayan Species Described by H. N. Rid- ley, 329-360 Part 16. Species Discovered in Thailand and Viet- nam, 466-492 Schafer, Rita D. : Effects of Pollution on the Amino Acid Content of Mytilus edulis, 246-250 SCHWABL, MATHILDE : Solenogaster Mollusks from Southern California, 261-281 Sobolev, V. S.: Characteristic Features of the Volcanism of the Siberian Platform, 452-457 493 494 Stone, Benjamin C. : Studies in the Hawaiian Rutaceae, IV. New and Critical Species of Pelea A. Gray, 407-420 Tester, Albert: The Role of Olfaction in Shark Predation, 145-170 Tomlinson, Jack T. : Lithoglyptes hirsutus (Cirripedia: Acrothoracica) , A New Burrowing Barnacle from Hawaii, 299— 301 PACIFIC SCIENCE, Vol. XVII, October 1963 Wilbur, Robert L. : A Prior Name for the Hawaiian Gouldia terminalis (Rubiaceae) , 421-423 Winkler, Lindsay R., and E. Yale Dawson: Observations and Experiments on the Food Habits of California Sea Hares of the Genus Aplysia, 102-105 Yoshida, Tadao, Takeo Saw ada, and Masahiro Higaki: Sargassum Vegetation Growing in the Sea around Tsuyasaki, North Kyushu, Japan, 135-144 Subject Index adoption of the metric system and Celsius scale, 131 aerial study of wave patterns, 255-260 aluminous Hawaiian soils, root development in, 398— 406 amino acid content of Mytilus edulis, effects of pollu- tion on, 246—250 amphipoda, Californian, identification of, 315-320 anemones, toxicity of, 302—304 Aplysia, food habits of, 102-105 applied agricultural climatology in the tropics, 379- 397 California anemones, toxicity of, 302-304 beach hoppers, identification of, 315-320 sea hares of the genus Aplysia, 102-105 solenogaster mollusks from, 261—281 Celsius scale, adoption of, 131 characteristic features of the volcanism of the Siberian platform, 452-457 cirriped, new species of, from Hawaii, 299—301 climatology in Hawaiian sugar-cane industry, 379-397 copepod, caligid, life history of, 195-242 device for making successive photomicrographic rec- ords of developing organisms, 321-328 eastern Pacific, species structure of Gillichthys mirabilis from, 47-72 effects of pollution on the amino acid content of Mytilus edulis , 246—250 feeding behavior in three species of sharks, 171-194 Fitchia, studies in, 282—298 folliculinids from submerged wood, 438-443 food habits of California sea hares of the genus Aplysia, 102-105 Gibsmithia hawaiiensis, gen. n. et sp. n., 458-465 Gillichthys mirabilis, species structure of, 47-72 Gouldia terminalis, a prior name for, 421-423 Harengula vittata in Hawaii, 431—437 Hawaii, new species of burrowing barnacle from, 299- 301 records of folliculinids from submerged wood, 438- 443 Rutaceae, studies in, 407-420 wave patterns, aerial study of, 255-260 Holothuria atra, 305-314 Holothuria leucospilota, 305-314 Hypoptychus dybowskii, osteology and systematic posi- tion of, 90—101 identification of Californian beach hoppers, 315—320 important Pacific insect pests of sugar cane, 251-252 insect pests of sugar cane, 251-252 investigations of rhinoceros beetles in West Africa, 444-451 Izu Islands, molluscan assemblages of the submarine banks around, 72-89 Japan, Sargassum vegetation in the sea around, 135— 144 Kauai, littoral sedimentary processes on, 106—130 Lepeophtheirus dissimulatus, life history of, 195-242 life history of Lepeophtheirus dissimulatus, 195-242 Lithoglyptes hirsutus n. sp., 299—301 littoral sedimentary processes on Kauai, 106-130 Lord Howe Island, Part III, 361-373 Malaya, new species of Pandanus from, 3-46 Malayan species of Pandanus described by H. N. Rid- ley, 329-360 Marquesan sardine in Hawaii, 431—437 Marshall Islands, sea cucumbers at Rongelap Atoll, 305-314 metric system, adoption of, 131 Midway Islands, additional plants from, 374 molluscan assemblages of submarine banks around the Izu Islands, 73-89 mollusks, solenogaster, from southern California, 261- 281 Index to Volume XVII 495 Mytilus edulis, effects of pollution on the amino acid content of, 246—250 Notes : Additional Plants from the Midway Islands, 374 Adoption of the Metric System and Celsius Scale, 131 On Malayan Shores: A Review, 374 olfaction, role of, in shark predation, 145-170 osteology and systematic position of elongate perci- form fishes, 90—101 Pacific insect pests of sugar cane, 251-252 Pandanus, Malayan species described by H. N. Ridley, 329-360 new species from Malaya and Singapore, 3-46 revision of genus, 3-46; 329—360; 466— 492 species discovered in Thailand and Vietnam, 466- 492 Pelea, new and critical species of, 407-420 perciform fishes, osteology and systematic position of, 90-101 photomicrographic records, device for, 321-328 pollution, effects of, upon amino acid content of Mytilus edulis, 246-250 population dynamics in a sublittoral epifauna, 424- 430 predation by sharks, role of olfaction in, 145—170 prior name for the Hawaiian Gouldia terminals, 421- 423 rhinoceros beetles in West Africa, 444-451 Rongelap atoll, sea cucumbers at, 305-314 root development in aluminous Hawaiian soils, 398— 406 Sargassum vegetation around Tsuyazaki, Japan, 135- 144 sea cucumbers at Rongelap Atoll, 305-314 sedimentary processes on Kauai, 106-131 sharks, feeding behavior in three species of, 171-194 role of olfaction in predation, 145—170 Siberian platform, volcanism of, 452-457 Singapore, new species of Pandanus from, 3-46 Society Islands, Pitchia of, 282—298 soils, aluminous, and root development, 398—406 solenogaster mollusks from southern California, 261- 281 studies on the green alga, Udotea indica, 243—245 sublittoral epifauna, population dynamics in, 424-430 submarine banks around Izu Islands, molluscan as- semblages of, 73-89 sugar-cane industry, role of climatology in, 379-397 sugar cane, Pacific insect pests of, 251-252 Thailand, Pandanus species from, 466-492 toxicity of California anemones, 302-304 Udotea indica, 243-245 Vietnam, Pandanus species from, 466-492 volcanism of the Siberian platform, features of, 452- 457 wave patterns, aerial study of, 255-260 West Africa, investigations of rhinoceros beetles in, 444-451 Manuscript Form. Manuscripts should be typed on one side of standard-size, white bond paper and double-spaced throughout. Pages should be consecu- tively numbered in upper right-hand corner. Sheets should not be fastened together in any way, and should be mailed flat. Inserts should be either typed on separate sheets or pasted on proper page, and point of insertion should be clearly indicated. Original copy and one carbon copy of manuscript should be submitted. The author should retain a car- bon copy. Although due care will be taken, the editors cannot be responsible for loss of manuscripts. introduction and Summary. It is desirable to state the purpose and scope of the paper in an introductory paragraph and to give a summary of results at the end of the paper. Dictionary Style. It is recommended that authors fol- low capitalization, spelling, compounding, abbrevia- tions, etc., given in Webster’s New International Dic- tionary (unabridged), second edition; or, if desired, the Oxford Dictionary. Abbreviations of titles of pub- lications should, if possible, follow those given in World List of Scientific Periodicals. Footnotes. Footnotes should be used sparingly and never for citing references (see later). When used, footnotes should be consecutively numbered by supe- rior figures throughout the body of the paper. Foot- notes should be typed in the body of the manuscript on a line immediately below the citation, and sepa- rated from the text by lines running across the page. Citations of Printed Sources. All references cited should be listed alphabetically by author at the end of the paper, typed double-spaced. References to books and to papers in periodicals should conform to the following models: Batzo, Roderick L., and J. K. Ripkin. 1849. A Treatise on Pacific Gastropods. Rice and Shipley, Boston, vii + 326 pp., 8 figs., 1 map. Crawford, David L. 1920 a. New or interesting Psyllidae of the Pacific Coast (Homop.). Proc. Hawaii. Ent. Soc. 4(1): 12-14. 1920&. The sandalwoods of Hawaii. Proc. Hawaii. Ent. Soc. 4(2): 374-375, 13 pis. In the text, sources should be referred to by author, date, and page, as follows: "It was noted (Rock, 1916: 18) that . . or "Rock (1916: 21-24) says . . .” Quotations. Quoted matter of fewer than five printed lines (about 200 characters) should be given in the text in the usual form, using double quote marks. Longer quotations should be set flush with left mar- gin. The author is responsible for the accuracy of quoted material. Numbers. Decimals, measurements, money, percent- ages, time; enumerations in which any figure is 10 or over; and isolated enumerations of 10 and over should be given in Arabic figures, rather than spelled out, except when the number begins a sentence. ILLUSTRATIVE MATTER Only the minimum number of illustrations required to supplement the text will be accepted by the editors. Reproduction costs of illustrations in excess of the number allowed by the editors will be paid by the author. Artwork for illustrations should be 8 V2 x 11 inches or smaller, and it should accompany manuscript, on separate sheets. Often more valuable than a photo- graph is a good line drawing. Figures and Graphs. Copy for figures and graphs should always be drawn large enough to allow for at least one-third reduction by the engraver. Copy should consist of carefully prepared line drawings in one color only, drawn in India ink on plain white draw- ing paper or tracing cloth. Co-ordinate paper with lines in light blue (a color which will not show in a photograph) may be used; but co-ordinates which should appear in the finished graph must be drawn in India ink. If original figures may not be conven- iently submitted with manuscript, duplicate rough sketches or photographic prints may be furnished to aid the editors in their decisions. It is strongly urged that an indication of scale be incorporated as a part of all drawings in which mag- nification and size are critical considerations. Photographs. Photographs should be chosen for clarity in portraying essential information. They should be printed for contrast, on glossy paper, and should be sent unmounted. They should be identified with serial number written in soft pencil on the back to corre- spond with list of captions. Illustrations will be returned to the author. Tables. Tabular matter should be kept to a minimum. Each table, prepared to conform with Pacific Science style, should be typed on a separate page, and its posi- tion indicated on the manuscript. Mathematical Formulas. Complicated formulas cannot be set by the printers. Authors should submit them as illustrations. Captions. Readily identifiable captions for figures, graphs, photographs, and other illustrative matter should be supplied on a separate page. PROOF Proof should be corrected immediately and returned at once to Robert Sparks, assistant to the editors. Authors are reminded that the editors will allow only a minimum number of corrections on galley proof. Additions to the printed text and changes in style and content are not permitted. All queries on proof should be answered. If cor- rected proof is not received within four weeks after being sent to the author, author’s changes cannot be accepted. REPRINTS Reprints or separates should be ordered on the form provided and returned with author’s proof. All correspondence concerning separates must be directed to the printer, Star-Bulletin Printing Company, 420 Ward Avenue, Honolulu 14, Hawaii. * 'i I