B U L L E T I N O F T H E T O R E E Y B O T A N I C A L C L U B

VOL. 94, No. 3, pp. 129-162 M A T - J U N E 1967

T h e biota of long-distance dispersal. V. Plant dispersal to Pacific Islands

Sherwin Carlquist

Claremont Graduate School and Raneho Santa Ana Botanic Garden, i 'laremont. California

CARLQUIST. Snr.p.wix. (Claremonl Graduate School, Claremont, California). The biota of long-distance dispersal A*. Plant dispersal to Pacific islands. Bull. Torrey Bot. Club 94: 129-162. 1967.—The native Hawaiian flora is analyzed by genera for probable method of arrival of ancestral immigrants, and these data are listed in detail and dis­cussed. Fruit and seed morphology are used as evidence for mode of dispersal interpreted in the light of known dispersal methods. Modes of arrival on the Hawaiian islands are computed as percentages, and similar percentages are presented for angiospcrms of other Pacific oceanic islands and archipelagos. These data show that ecology of recipient island is much more important than distance in determining modes of successful transport. On high islands, internal transport by birds is of prime importance, and decreases only slightly with increased distance. On atolls, oceanic drift is of overwhelming importance. Adherence to feathers by barbed or bristly seeds or fruits is of maximum importance on dry volcanic islands. Transport of seeds by air flotation decrease-, sharply with distance, and thus is responsible for only a few genera on the Hawaiian Islands. To account for floras of oceanic islands, no factors not currently operative seem necessary. Shore birds are cited as a particularly likely vector for transport of fleshy-fruited plant species. Rea­sons are given for absence of South Pacific drift species in the North Pacific. The great similarity between floras and faunas of the Pacific oceanic islands and those of Mascarene oceanic islands derives from the fact that islands in both the Pacific and Indian oceans have been populated by those groups from Iudo Malaysia capable of long-distance dis­persal. The implications of the present data for inter-continental and intracontinental dispersal are discussed, and tlie implications for hypotheses of continental drift are explored.

The majority of biologists hold that islands east of the Andesite line in the Pacific are oceanic islands, and that these islands have acquired their floras and faunas by means of long-distance dispersal. Some of the reasons for this supposition lie in floristics and faunistics, lines of evidence reviewed

129

130 BULLETIN" OF THE TORREY BOTANICAL CLUB [VOL. 94

by Thorne (1963). Other lines of evidence or inference relate to frui t and seed morphology. Contributions by Guppy (1906) and Ridley (1930) in this connection are noteworthy. Much more remains to be learned about mechanisms of long-distance dispersal. Without such new evidence, argu­ments, reviews, and reassessments of this phenomenon will contribute little. However, certain aspects seem worthy of examination at this time by way of indicat ing what types of new evidence might be most meaningful. I n this study, therefore, information and speculations concerning plant dispersal in the Pacific are placed in a context somewhat different from those em­ployed by other workers. The modes of dispersal likely to have resulted in the present native floras of Pacific islands are analyzed here, and unex-pectcd and somewhat surpr is ing correlations become visible.

By calculating the proportions of a flora which have presumably ar­rived by air flotation, internally in birds, etc., and by comparing these per-centages with those categories in other floras, we can see that ecology of the island receiving immigrants is of tremendous importance. Long-distance dispersal probably provides many times the number of disseminules which can actually establish. Inferentially. this might suggest that a species pre-adapted to establishment on an island might well be introduced by more than a single seed. Distance from the source area is apparent ly of secondary importance in determining which met hods of dispersal are effective in pop­ulating a par t icular island. Such problems as so-called ' ' success ion" on lava flows prove related to differential dispersal ability. Dispersibility is the key to the surpr is ing floristic and faunistic affinities between Polynesian and Mascarene islands. The realities of lung-distance dispersal to islands, when compared with distribution problems on continents, are both applicable and illuminating. Long-distance dispersal proves the most satisfying ex­planation for certain instances of inter- and infra-continental disjunction.

Data on dispersal methods have been drawn from various sources, such as Guppy (1906), Ridley (1930), and my own observations. I n using frui t and seed morphology as an indicator of dispersal method, one can hypothe­size a certain degree of loss of dispersibility (Carlquist , 1966b, 1967), but only in part icular taxonomic groups and in part icular islands, such as the Hawaiian Islands. Direct observations of plant dispersal events to islands (other than via oceanic drif t ) are virtually impossible, so Indirect lines of evidence must be used. In plants, certain experimental techniques are un­fortunately of limited value. One might have expected thai aerial t r app ing experiments, such as those which proved so useful to J . Linsey G ress.it t in s tudy of insect dispersal, might be useful to botanists. This is not t rue , be­cause aerial flotation of seeds appears, according to data belnw. to be insig­nificant except ra ther close to source areas.

Observations by ornithologists do give useful hints about dispersal of

1967] CARLQUIST: THE BIOTA OF LONG-DISTAXCE DISPERSAL 131

plants by birds. These are often of particular value because such data are not collected by ornithologists with the intention of stressing or ' ' proving' ' long-distance dispersal of plant disseminules by birds. Such observations are an excellent source of indications as to which means of dispersal by birds might usefully be investigated further. The importance attributed in the present study to dispersal via birds does suggest to me that botanists inter­ested in dispersal problems must turn intensively to observational ami exper­imental work with birds if methods of long-distance dispersal are ever to be "demonstrated." "We must be realistic in such investigations, however. Ob­viously there are many kinds of birds which probably have virtually no function in long-distance dispersal.

Processes of long-distance dispersal are thoroughly credible, but only if unsupportable hypotheses can be eliminated. We can no longer accept Guppy's idea that different taxonomic groups dispersed at various times (e.g., "age of ferns," "age of Compositae," etc.). Nor can we agree with the full implications of Guppy's claim that "agencies of dispersal . . . have ceased to be active'' in the Pacific. High islands in the Pacific may well have once existed where only atolls or seamounts now stand, and such former high islands may once have served as subsidiary source areas or "stepping-stones" for populating more remote islands. Nevertheless, the birds which now function in dispersal are probably much the same as those which have done so in the previous areas of the relatively recent geological past, and Guppy's suggestion that now-vanished fruit-eating pigeons were vectors in dispersal of plants must be largely dismissed as rather visionary. While not­ing the improbability of some of Guppy's ideas, one must concede that his work provides most of the information we have on dispersal of Pacific plants, and the contributions since his are relatively few. Guppy's survey of the Hawaiian flora with respect to methods of dispersal is understand­ably incomplete. If the Hawaiian flora arrived entirely via long-distance dispersal, however, we must ultimately be able to account for each and every arrival with respect to dispersal. Similarly, we must be able to account for other Pacific floras. The geographic sources of insular floras are reasonably well understood, thanks to the work of Fosberg (1948), Van Balgooy, (1960), and Thorne (1963). We need similar information (which must, similarly, be speculative) on the probable means of arrival of the plant immigrants to oceanic islands.

Methods and acknowledgments. Some observations relative to dispersal were made in the Hawaiian Islands, both main islands and the Leeward chain, during the summer of 1966. This field work was aided by a National Science Foundation Grant, GB-4977X. Observations in southern Polynesia during 1962 were aided by a similar grant. NSFG-23396. Individuals who

132 BULLETIN OF THE TORRET BOTANICAL CLUB tVoL- 9 4

kindly helped in field activities in the Hawaiian islands include Mr. Edwin C. Bonsey, Mr. Norman C. Carlson, Mr. Charles Christensen, Mr. Ralph Daehler, Dr. George W. Gillett, Mr. Ronald Hurov, Mr. Eugene Kridler , Mr. James "W. Larson, Mr. Herber t Shipman, Dr. A. C. Smith, Dr. F r a n k Tabrah, and Mr. Bunichi Usagawa. Observations in the Leeward chain were made during a voyage of the U.S. Coast Guard Cut ter I ronwood; the leadership of Mr. Eugene Kridler is gratefully acknowledged. Thanks are also due Dr. Robert F . T h o m e for his s t imulat ing discussions and for read­ing this manuscript .

The tabulations and computations presented for the Hawai ian flora were influenced by Fosberg 's (1948) tabular summary of Hawaiian angiosperms. Fosberg ' s list presents geographical sources, numbers of immigrants , and numbers of contemporary species. Using his immigrant number estimates (wi th certain al ternat ions) , percentages for seven methods of arrival were computed. Only certainly native taxa have been included. Lists similar to Fosberg 's were then prepared for other Pacific floras, and estimates of per­centages of the methods of arr ival were made for each of those floras. Ad­mittedly, some floras are better known than others, and my speculations on mode of arrival may be incorrect in par t icu lar cases. E r ro r s may or may not tend to balance out within a flora, but I doubt that markedly different figures for par t icular modes of ar r ival would be obtained by other workers.

The following floristic works were used as sources for making the esti­ma tes : San Clemente I . : Raven (1963a) ; Revillagigedo I s . : Johnston (1931) ; Cocos I . : S tewar t (1912) ; Galapagos Is . : Stewart ( 1 9 1 1 ) ; Des-venturadas I s . : Skottsberg (1937, 1963) ; J u a n Fernandez I s . : Skottsberg (1922b, 1928, 1951) ; Easter I . : Skottsberg (1922a, 1927, 1951) ; Oeno I . : St. J o h n and Philipson (1960) ; Rapa I . : Brown (1931, 1935) ; Society I s . : Drake del Castillo (1893), Moore (1933, 1963) ; Rarotonga: Cheeseman (1902) ; Tonga : Yuncker (1959) ; Samoa: Reinecke (1898) , Christophersen (1935, 1938) ; Marquesas I s . : Brown (1931, 1935) ; Equator ia l I s . : Christo­phersen (1927) ; Laysan I . : Christophersen and Caum (1931), Lamour<ux (1963) ; Hawaiian Is . : Hi l lebrand (1888), Christophersen and Caum (1931), Fosberg (1948). Certain floras were not used if too few species to provide accurate pictures of modes of arrival (e.g., Clipperton I . ) .

Analys is of the Hawa i i an flora. Concepts of means of arr ival arc pre­sented in the following tabular summary. This is offered in order tha t my decisions on probable arr ival method for each introduction to the Hawaiian flora can be examined. Only native seed plants are considered. Vascular cryptogams may be assumed to have arr ived via a i r flotation of spores with a few exceptions, such as Marsilea (Malone and Proctor, 1965).

The desirability of using the Hawaiian biota as a source for s tudy of biological phenomena related to oceanic islands has been emphasized by the

1967] CARLQUIST: THE BIOTA OF LONG-DISTANCE DISPERSAL 1 3 3

writer on previous occasions. In the present study, the fact that the Hawaiian Islands are exceptionally remote yet have ecological diversity broad enough to support a wide variety of immigrants is of significance. That the Hawaiian Islands have acquired their native flora by means of long-distance dispersal has not been seriously challenged. While now-van­ished archipelagos may have aided dispersal from Indo-Malaysian source areas, no such former islands have been hypothesized between the Hawaiian Islands and America. American sources must be claimed for no fewer than 50 successful arrivals in the Hawaiian angiosperm flora. If long-distance dispersal can explain these 50 (which represent, in fact, all seven modes of arrival below), then dispersal over shorter distances can surely be explained also.

For each genus, interpretations are given in parentheses. The first fig­ure in each case indicates the number of successful introductions needed to account for the native species now resident. This may be considered as the minimum number of introductions necessary, and a somewhat larger num­ber is conceivable in some cases. Next in each case is an abbreviation indi­cating presumed mode of arrival. These abbreviations are:

A = Air flotation BB • Birds—seeds or fruits attached to feathers mechanically by

means of barbs, bristles, awns, trichomes, etc. BI = Birds—seeds or fruits eaten and carried internally

BM = Birds—seeds or fruits embedded in mud on feet. BV = Birds—seeds or fruits attached to feathers by viscid sub­

stances, or viscid quality of fruit parenchyma surrounding seeds.

DF = Oceanic drift—frequent or repetitive, as suggested by dem­onstrable ability of fruits or seeds to float for prolonged period.

DR = Oceanic drift—rare, on one or only a few occasions; seeds or fruits probably resistant to seawater, but unable to float for prolonged periods and thus likely to have arrived via some kind of rafting.

The last item within each parenthesis is the estimated number of con­temporary species plus subspecies and varieties, both endemic (e) and wide-ranging, native to one or more areas in addition to the Hawaiian Islands (w).

In some eases, more than one type of dispersal occurs within a genus. For example (contrary to Fosberg, 1948), I hold that at least three intro­ductions rather than one are necessary to account for the Hawaiian species of Scaevola (Goodeniaeeae). Oceanic drift has brought the common beach Scaevola, variously known as S. scricea or S. taccada (fig. 8). Scaevola

134 BULLETIN OF THE TORREY BOTANICAL CLUB [VOL. 94

' ySSreS '

Fig. 1-7. Hawaiian plants, photographed in circumstances relevant to their di — Pig. L—2. Plants OH the 1955 lava flow, Puna District. Hawaii (photographs taken in 1965 . -F ig . 1. Nephrolepis exaltata growing in cracks in lava flow.—Pig. -- Metrosideroa [loh/mor/iiiii in depression iii lava Bowj this plant has flowered and is approximately & years old.— Fig. :'.. Horrhtiriu diffusa growing on coral rubble on East bland, French F r i g a t e Shoa l s .— Fig. 4. Juven i l e sooty t e rn , cap tu red and pho tographed a f t e r the bird had run through Boerhavia plants and collected several Boerhavia fruits (three visible below eye on leathers.—-Fig. f>. Various seeds (presumably all Mucuna spp.) in the drift accumulations of Southeast Island. Pear] and Hermes Reef.—Fig. »>. Mucuna seedling uprooted where it grew, on coral beach of Southeast Island, Pearl and Hermes Reef. This seedl ing has died a t t ip . and will not surv ive . N o seedlings of Mucuna survive longer than this on Pearl and Hermes Beef.— Fig. 7. Branch of I'orliiliicu htha. collected on G a r d n e r Pinnacles. Branch is shown Hoating in water in white enamel tray, indicat ing

ability to float.

19071 < AKI.QUIST: THE BIOTA OF LONG-DISTANCE DISPEBSAL 135

Fig. B—12. Hawaiian plants, photographed to Bhow fruit morphology.— Fig. 8. Scaevola taecada (8. terioea a widespread Pacific species distributed by oceanic drift. — Pig. !'. Sea* volt "• an endemic Hawaiian species belonging to a series of montane purple-fruited Pacific i irobably distributed by frugivorous birds. Speci-men from Mr. Konahaanui, Oahu.— Fig. 10. Clermontta arborescens (Campanulaeeae: Lobelioidene . mature fruit Bhowing the natural dehiscence typical of mature Clermontia fruits. Small seeds embedded In the orange fleshy parenchyma are risible; also note latex (white) on broken surfaces. Specimen from summit ridge. Lanai.—Fi«j. 11. Phyllostegia vestita. The fruit consists of drupelets, bright green or dark purplish at maturity. Speci­men collected in kipuka on Saddle Road. Hawaii.— Fig. 12. F.xocarpns {/audichaudii. Fruiting specimen, from summit region of Ilualalai. Hawaii. Receptacle is white, fleshy at maturity

136 BULLETIN OP THE TORREY BOTANICAL CLUB t V o r " 9 4

glabra (including S. kavaiensis) represents an early and rather distinctive immigrant (regarded as a segregate genus, Camphusm, by some) in which seed and fruit gigantism has occurred autochthonously (Carlquist, 1967), as have other morphological changes. The remainder (e.g., S. gaudichaudi-ana, fig. 9) of the Hawaiian species stem from a third stock.

If a clear decision as to mode of arrival could not be reached, a pref­erence is expressed by indicating one method in upper case, the alternative in lower case. If two arrivals are claimed to have come by different means within a genus, both choices are given in upper case. Undoubtedly some poor choices have been made.

Mode of arrival was computed not only for the entire native angiosperm flora, but also for those segments from each of the source areas suggested by Fosberg. By this means I hoped, for example, to see if immigrants from America came predominantly by a method different from those from Indo-Malaysia. In fact, no such differences emerged, and the methods of long­distance dispersal seem relatively alike in different geographical areas. Among the few such figures which seem to show significant deviation are the relatively high number of arrivals in mud on birds' feet (6) and the low number internally in birds (1) from Boreal regions. Among the Pan-tropical species, repetitive oceanic drift rates notably high (11 introduc­tions out of 27), a fact which merely shows that many of the Pantropical species have achieved their wide distribution by virtue of the great effi­ciency for wide dispersal characteristic of oceanic drift.

Another type of computation proved somewhat more significant, and these results are given at the end of the compilation of genera. For each of the seven modes of arrival recognized, the number of contemporary Hawai­ian species was divided by the number of immigrants hypothetically neces­sary to account for those species. Thus, a ratio denoting degree of speciation related to each mode of arrival could be obtained. This figure is discussed below under various dispersal headings.

Pandanaceae: Freycinetia (1, BI, le) ; Pandanus (1. DF, lw) . Potamo-getonaceae: Potamogeton (2, BI, 2w). Ruppiaceae: Ruppia (1, BM, lw) . Xaiadaceae: Naias (1, BM, lw) . Hydrocharitaeeae: Halophila (1, DF, lw) . Gramineae: Agrostis (2, BB, 2e, l w ) ; Andropogon (1, BB, l w ) ; Calama-grostis (2, BB, 2e) ; Cenchrus (1, BB, 3e) ; Deschampsia (1, BB, 3e) ; Digi-taria (1, BB, lw) ; Dissochondrus (1, BB, le) ; Eragrostis (1, BI or bb, 12e); Festuca (1, BB, l e ) ; Garnotia (1, BB, l e ) ; Heteropogon (1, BB, l w ) ; Isachne (2, BB, le, l w ) ; Ischaemum (1, BB, lw) ; Lepturus (1, DF, lw) ; Microlaena (1, BB, lw) ; Oplismenus (1, BB, lw) ; Panicum (3, BI, BB, BM, 23e); Paspalum (2, BB, 2w); Poa (1, BB, 4e) ; Sporobolus (1, BV, l w ) ; Trisetum (1, BB, 2e). Cyperaeeae: Carex (6, BM, BI, 6e, 3w); Cladium (and derivatives Baumea and Yincentia) (3, BI, 2e, lw) ; Cyperus

1907] CABLQUIST : THE BIOTA OF LONG-DISTANCE DISPERSAL 137

(8, BI, BM, 16e, 7w); Eleocharis (1, BM, l w ) ; Fimbristylis (1, BM, or bi, le, l w ) ; Gahnia (2, BI, BM, 6e); Oreobolus (1, BB, le) ; Rhynchospora (2, BI, or bb, le, 2w); Sdtpua (5, BM, or BI, le, 4w); Selena (1, BM, l e ) ; Uncinia (1, BB, lw) . Palmae: Pritchardia (1, BI, or dr, 2—30e). Flagellari-aceae: Joinvillea (1, BI, l e ) ; Juncaceae: Luzula (1, BM, 3e). Liliaeeae: Astelia (1, BI or bv, 12e); Dianella (1, BI, 3e); Dracaena (Pleomele) (1, BI, 2e) ; Smilax (1, BI, 2e). Iridaceae: Sisyrinchium (1, BM, le ) . Or-chidaceae: Anoectochilus (1, A, 2e) ; Habenaria (1, A, l e ) ; Liparis (1, A, l e ) .

Piperaceae: Peperomia (3, BV, 48e, 2w) ; Ulmaceae: Pseudomorus (1, BI, lw) ; Trema (1, BI, lw) . Urticaceae: Neraudia (1, BI, 9e) ; Pilea (1, BM, l w ) ; Pipturus (1, BV, 13e); Touchardia (1, BI, l e ) ; Vrera (2, BI, 3e). Santalaceae: Exocarpus (1, BI, 3e) ; Santalum (2, BI, 6e). Lorantha-ceae (Viscaceae) : Korthalsella (2, BV or BI, 6e, 2w). Polygonaceae: Poly­gonum (1, BI, lw) ; Rumex (1, DR or bi, 2e). Chenopodiaeeae: Chenopo-dium (1, BM, le ) . Amaranthaceae: Achyranthes (1, BB, 2e) ; Aerva (1, BB, le) ; Charpentiera (1, BI or bin, 3e) ; Nototrichum (1, BB, 5e). Nyctaginaceae: Boerhavia (3, BV, 3w) ; Pisonia (3, BV, 3e). Phytolacca-ceae: Phytolacca (1, BI, le) . Aizoaceae: Sesuvium (1, DF, lw) , Portulaca-ceae: Portulaca (3, DF, 4e, 2w). Caryophyllaceae: Sagina (1, BM, le) ; Schiedea (and Alsinodendron) (1, BM, 32e) ; Silene (1, BI or BM, 5e). Ranunculaeeae: Ranunculus (1, BM or bb, 2e). Menispermaceae: Cocculus (1, BI, 2e). Lauraceae: Cassytha (1, DF, lw) ; Cryptocarya (1, BI, l e ) . Capparidaceae: Capparis (1, DR or dr, lw) . Cruciferae.- Lepidium (2, BV, 4e). Droseraceae: Drosera (1, BM, lw) . Saxif ragaceae: Broussaisia (1, BI, 2e). Pittosporaeeae: Pittosporum (1, BV, 30e). Rosaceae: Acaena (1, BB, 2e); Fragaria (1, BI, l w ) ; Osteomeles (1, BI, l w ) ; Rubus (1, BI, 2e). Leguminosae: Acacia (1, DR, 3e) ; Caesalpinia (2, DF, 2w) ; Canavalia (1, DF or DR, 2e) ; Cassia (1, DR, l e ) ; Erythrina (1, DR, l e ) ; Mucuna (2, DF, 2w); Sophora (1, DR, 4e) ; Sesbania (1, DR, l e ) ; Strongijlodon (1, DF, lw) ; Vicia (1, BI, le) ; Vigna (2, DF, 2e, lw) . Geraniaceae: Ger­anium (1, BB, 6e). Zygophyllaceae: Tribulus (1, BB, lw) . Rutaceae: Fagara (1, BI, 14e) ; Pelea (1, BI, 40e) ; Platydesma (1, BI, 5e). Euphor-biaeeae: Antidesma (1, BI, 4e) ; Claoxylon (1, BI, 2e) ; Drypetes (1, BI, le) ; Euphorbia (1, BI, 30e) ; Phtjllanthus (1, BI, 2e) ; Aquifoliaeeae: Ilex (1, BI, l e ) . Celastraeeae: Perrottetia (1, BI, le) . Anacardiaceae: Rhus (1, BI. l e ) . Sapindaceae: Alectryon (1, BI, 2e) ; Dodonaea (1, DF or DR, 3e, lw) ; Sapindus (1, BI, l e ) . Rhamnaceae: Alphitonia (1, BI, l e ) ; Colubrina (2, DF, le, l w ) ; Gouania (1, DR, 3e). Tiliaeeae: Elaeocarpus (1, BI, l e ) . Malvaceae: Abutilon (2, DF, 2e, l w ) ; Gossypium (1, DR, l e ) ; Hibiscadelphus (1, DR, 4e) ; Hibiscus (4, DR, lOe); Kokia (1, DR, 4e) ; Sida (2, DF, 2w) ; Stereuliaceae: Waltheria (1, DR, lw) . Theaceae: Eurya

138 BULLETIN OF THE TORREY BOTANICAL CLUB [ V o L - 9 4

(1, BI, 2e): Thymeleaceae: Wikstroemia (1, BI, 20e). Flacourtiaceae: Xylosma (1, BI, 2e). Cucurbitaceae: Sicyos (1, BB or dr, lOe). Myrtaceae: Eugenia (2, BI, 4e) ; Metrosideros (1—2, A, 5e). Begoniaceae: Hittt brandia (1, BM, le ) . Haloragaceae: Gunnera (1, BI, 7e). Araliaceae: Cheiro-dendron (1, BI, 5e) ; Reynoldsia (1, BI, l e ) ; Tetraplasandra (1, BI, lOe). Umbelliferae: Peucedanum (1, DR, 3e); Sanicula (1, BB, 4c). Ericaceae: Vaccinium (1, BI, 8e). Epacridaceae: Styphelia (1, BI, 2e). I'rhuulaeeae: Lysimachia (2, DF, BM, 12c, Iw;. Myrsinai-eae: Embelia (1, BI, l e ) ; Myrsine (1, BI, 25e). Sapotaeeae: Nesoluma (1, DF or bi, l w ) ; Plancho-nella (1, BI, 6e). Ebenaceac: Diospyros (Maba) (1, BI, 7e). Plumbagina-ceae: Plumbago (1, BV, lw). Loganiaceae: Labordia (2, BI, 40e). Oleaceae: Osmanthus (Negrestis) (1, BI, le) . Apocynaceae: Alyzia (1, BI, le) ; Ochrosia (1, DR, l e ) ; Pteralyxia (1, DR, 2e); Rauwolfia (1, BI, 7e). Con-volvulaceae: Breweria (1, BV or dr, 2e) ; Cressa (1, DF, lw) ; Cuscuta (1, DF or BI, 2e) ,• Ipomoea (7, DF, 4e, 5w); Jacquemontia (1, DR, le) . Hy-dropbyllaeeae: Narna (1, BB, le) . Verbenaceae: HeUotropium (2, DF, le, lw) . Labiatae: Lepechinia (1, BV, lw) ; Phyllostegia (1, BI, 40e) ; Steno-gyne (1, BI, 40e); Haplostachys (1, DR, 3e) ; Plectranthus (1, BB, lw) . Solanaceae: Lycium (1, DF, lw) ; Nothocestrum (1, BI, 6e) ; Solatium (1, BI, 7e). Scrophulariaceae: Bacopa (1, BM, lw) . Myoporaceae: Myo-porum (1, BI, le) . Ge.sneriaceae: Cyrtandra (1—2; BI, 130e). Plantagina-ceae: Plantago (2, BV, 19e). Rubiaceae: Bobea (1, BI, 4e) ; Canthium (1, BI, lw) ; Coproama (3, BI or bv, 27e) ; Gardenia (2, BI, 2c) ; Hedyotis (and Gouldia) (1, BM or BI, 50e); Morinda (1, DR, le, lw) ; Nertera (1, BI, l w ) ; Psychotria (1, BI, lw ; Straussia (1, BI, lOe). Campanulaeeae: Brighamia (1, BI, l e ) ; Clermoniia (1, BI or BV, 30e); Cyanea, Delissea, and Rollandia (1, BI or BV, 48e); Lobelia (1, BM, 15e); Trematolobelia (1, BM. 3e). Goodeniaceae: Scaevola |:{. BI and DF, 9e, lw) . Compositae: Adenostemma (1 , BV, l w ) ; Argyroxipkium, Dubautia (including Ruilliar-dia) and Wilkesia (1, BV or BB, 37e) ; Artemisia (1, BV, 3e); Bidens (1, BB, 30e) ; Hesperomannia (1, BB, 5e) ; Lagenophora (1, BV, 3e) ; Lipochacta (1, BB, 35e); tfemya (1 , BB, 2e) ; Tetramolopium (1, BB, lOe).

Ratios of numbers of contemporary species in the Hawaiian flora to the numbers of postulated immigrant antecedents for the various dispersal types are as follows:

A: 2.3 BV: 7.4 BB: 4.7 Birds (all means) : 5.7 B I : 6.0 D F : 1.4

BM: 4.9 DR: 2.5 Ratio for angiosperm flora as a whole.- 4.8

By way of footnotes, the following comments can be offered. A number of species included in Fosberg's list have been omitted because I feel they

19G7] CARLQUIST: THE BIOTA OP LOXG-DISTAXCE DISPERSAL 139

are dubiously native. The Dubautia—Argyroxvphiwm—Wilkesia complex can now be clearly regarded as American in affinities. Lipochaeta is clearly also American rather than Indo-Pacific in origin. Bidens is probably better regarded as American than "Austra l ," despite its occurrence in the South Pacific. Trematolobelm probably does not stem from the same event of in­troduction as do the Hawaiian species of Lobelia. Several genera are prob­ably in a stage of transition with respect to loss of dispersibility. These were not discussed in my survey of this phenomenon (1967) because only the more obvious instances were considered there. According to Guppy (1906), Hawaiian specimens of Dodonaea vary in ability of seeds to float. Similarly, he notes that some Hawaiian species of Sivyos have hooked fruits, while some do not. Jacquernontia sundwicrnsis has mm-floating seeds. This was interpreted by Guppy as indicative that seeds arrived in crevices of logs whch rafted to the Hawaiian Islands. Ridley (1930) reached what appeals to me as the more logical conclusion, namely that J. sandwia nsis has been derived from a littoral species from America. These species do, in fact, have floating seeds. A possible instance of partial loss of dispersibility has recently been claimed for Plantago (Mr. M. Tessene, pers. eommun.). The species from the Waialeale summit area (e.g., P. krajmae) apparently lack the mucilaginous seed coat characteristic of other Hawaiian montane species, as well as of most Plantago species elsewhere in the world.

Comments on modes of dispersal in the Hawaiian flora.—AIR FLOTA­

TION. In the Hawaiian flora, this means of transport appears to have been of minimal importance. The only seed plants which appear likely to have arrived by this means are Metrosideros (Myrtaceae) and the three genera of orchids. Even these could conceivably have arrived in mud on birds' feet. The complexity of Metrosideros in the Hawaiian Islands might be more readily explained if there have been more than a single introduction. Hybridization between earlier and later immigrants could account for the patterns which exist. Of course there is little difference between re-intro­duction of a stock from elsewhere and juxtaposition by a dispersal event of two Hawaiian stocks which have been isolated from each other within the archipelago.

Of the three genera of orchids, two, Liparis and Habc-narui, have reached many islands of the world. One can hypothesize that these two have attained a wide distribution because of simple pollination require­ments, among other reasons.

Compositae are often thought to be well adapted for wind dispersal, but the Hawaiian composites do not seem to have arrived in this manner. Those Compositae best fitted for wind-dispersal (Cichorieae, Senecioneae, Cy-nareae) are the groups notably absent. Those three tribes are, however, present in the Juan Fernandez Islands, which are much nearer to mainland

1 4 0 BULLETIN OP THE TORREY BOTANICAL CLUB tVoi. 94

source areas. Wind-dispersal may be possible for composites with plumose pappus , but the limit even for such types is probably much less than the distance between a source area and the Hawaiian Islands. Even species with plumose pappus can be envisaged a,s capable of t ranspor t while enmeshed in bird feathers. Travel of fruits of Compositae in bird feathers has, in fact, been observed under ordinary circumstances (Ridley, 1930).

Ferns in the Hawaiian flora seem suited for air flotation of spores ex­cept for Marsilea and possibly Selaginella. Marsilea seem ideally suited to dispersal by water birds, either internally or by adherence of the gelatinous sorophores to feet or feathers. One feature of the Hawaiian fern flora which is exceptional is the very high ra te of endemism (70 .8%) . Fosberg (1948) gives it as 64.9%, but division of his total for endemic species plus varieties by his total for all native fern species plus varieties yields 70.8%. Although 70.8 is a percentage lower than for Hawaiian angiosperms (94.4% accord­ing to Fosberg, 194-S'i, it is high by comparison with rates of endemism for ferns in other areas, insular or otherwise. Evident ly t ranspor t to the Ha­waiian Islands by air flotation is not as easy or as frequent for fern spores as has generally been supposed. The fact that so few seed plants appear to have arrived by this means in Hawaii is indicative. Possibly spores of wet-i'oresl ferns are more easily killed by desiccation than those of other ferns. and tin* long route to the Hawaiian Islands would prove fatal for many forest speeies. If arrival events of Hawaiian ferns have in fact been infre­quent and non-repetit ive (only a single introduction of a species), then isolation leading to speciation could occur.

The significance of air flotation within the present-day Hawaiian flora is probably different from the relatively small role played by air flotation in br inging angiosperms to the islands. This is demonstrated by pa t te rns of vegetation which appear on new lava flows. Despite suggestions by various workers that "success ion ' ' in the sense of Clements is involved, the entire mat ter appears to relate to differential dispersal ability, [ndeed, the sup­posed sequence from lichens to trees does nol exist. Lichens, many ferns (fig. 1) , and a scat ter ing of angiosperms appear almost simultaneously. Among the earlier angiosperms visible on flows are Metrosideros (fig. 2 ) , Dubautia (Bailliardia) cUiolata or other Dubautia species, Bumex gigan-/> us, and Dodonaea spp. The pappus of Dubautia permits it to cross rela­tively short distances by wind, while Rumex and Dodonaea have winged fruits. These probably arrive on new lava flows via wind-flotation, despite the probability that they probably arr ived ancestrally in the Hawai ian Islands in other ways, except for ferns and Mi Irosidi ros.

Hawaiian genera with fleshy at tract ive fruits, such as Stypheha, Coprosma, and Vaccinium also appear early on lava flows. These three genera are typically dispersed by frugivorous birds. Both native and intro-

1 9 6 7 ] CARLQUIST: THE BIOTA OF LONG-DISTANCE DISPERSAL 141

dueed birds are quite active in dispersal of seeds as is shown by the excep­tionally rapid spread of such weeds as Psidmm and Schinus within the Hawaiian Islands. Consequently, species with fleshy frui ts may reach new lava flows only a short time later than do the wind-dispersed species. Both certain wind-dispersed and certain bird-dispersed species can grow well on recent lava, so the concept of " success ion" on lava flows appears to be merely the artifact first of dispensability and second of ability to grow in relatively dry sites—but not a phenomenon related to development of soil profiles. I'll'. Indeed, the Kona forest on the island of Hawai i are the most magnificent in the Hawai ian Islands, and grow on minimal soil of relatively undecomposed lava flows.

The Hawaiian genus Trematolobelia is evidently an autochthonous de­rivative of Lobelia. Frui t and seed morphology appear to have altered in the direction of greater efficiency at wind dispersal, over short distances at least (Carlquist , 1962). Circumstances suggest tha t Lobelia probably reached the Hawaiian Islands in mud on b i rds ' feet, however.

BntDs. Adherence of seeds or fruits to feathers by means of hairs, barbs, bristles or similar appendages is one of four methods suggested as a source of t ransport by birds. Indeed, it is the most distinctive of the four, and prob­able instances can be recognized with less possible chance for confusion. In the Hawaiian flora, most grasses appear to have arr ived in this fashion. Other very likely candidates include T'ncinia (Cyperaceae) , 8anicida (Ura-bell iferae), Acaena (Rosaeeae), and Bidens (Composi tae) . Endemism is notably high in groups dispersed in this manner (see above). I n fact, of the genera just listed, all species native to the Hawaiian Is lands are also endemic. This suggests that t r anspor t of d ry fruits by catching of their appendages in b i rds ' feathers probably tends to operate on an infrequent or non-repetitive basis, and that the Hawaiian species in each of these genera may derive from a single dispersal event each. B y "s ingle dispersal even t , " however, we can mean more than a single seed—and a group of seeds seems a better mode of introduction it' genetic variability sufficient for long-term survival of a par t icular stock is to be achieved (Carlquist , 1966a, 1966d). A group of seeds brought at the same time is possible: for example, on feathers of a flock of birds migra t ing from the same source area. In t ro-duet ion of several seeds of a species over a period of years with consequent formation of an interbreeding population, then followed by no fur ther introductions of tha t species in subsequent years would also qualify as a "s ingle dispersal even t . "

I n the case of supposed internal t ranspor t of seeds and fruits by birds, tliere are possibilities of al ternat ive interpretat ions. "Where seeds are small and fruit parenchyma is even moderately sticky, the possibility definitely exists that seeds could be at tached to head feathers when a bird eats por-

142 BULLETIN OF THE TORREY BOTANICAL CLUB tVoL- 9 4

tions of a fruit. Such a small-seeded fruit is seen in Clermont la (fig. 10) ?

where the baccate fruit breaks open at matur i ty . Concomitantly with this rup tu re , latex is released, providing an additional means of adhesion. Also in this category are other baccate Hawaiian lobelioids (Cyanea, Rollandia, Delissea, and possibly Brighamia). Other groups in the Hawaiian flora in which seeds are very small, and thus possibly t ransported externally as well as internally, include Vactinium (characteristically a food of the Hawaiian goose according to Guppy, 1906), Labordia (and its South Pacific relative. Geniostoma), Cyrtandra, and Pipturus. In addition, some groups with rela­tively large seeds have fruit parenchyma which is notably viscid: Korth-alsella and Peperomia, for example. Both of these genera have a very wide distribution in Polynesia, perhaps because of this feature.

Large-seeded fruits in the Hawaiian flora likely to be eaten by birds are either relatively small, if single-seeded, or else a gigantism related to loss of -lispersibility may be assumed to have taken place (Carlquist , 1966b, 1967). No single-seeded fruit longer than 1 cm. need be postulated as ancestral for any angiosperm in the native Hawaiian flora. The seed con­tained in such a frui t need be no longer than about 8 mm. F ru i t s which are vir tual ly black in color, as in Scucvola (fig. 9 ) , fruits which are green ami seemingly inconspicuous (fig. 11), and fruits in which accessory struc­tures form fleshy at tract ive organs (fig. 12) are among the many ways in which quite effective dispersal by frugivorous birds is assured. In fact, acquaintance with the Hawaiian flora convinces one that adaptat ion to dis­persal by frugivorous birds is abundant ly present. Even it' some small-seed. ';1 fruits are assumed to have experienced external t ranspor t on birds (and some are so assumed here) , the number of immigrants brought in­ternally in birds still remains the largest single category of dispersal method basic to the Hawaiian flora. The percentage in the Hawaiian flora (38.9, represent ing at least 101 arr ivals) may seem high to the reader, but this percentage is, in fact, lower than the percentage of internal ly- t ransported immigrants necessary to account for other Polynesian floras (fig. 13) . The fact tha t internal t ransport in birds is the only way in which one can ac­count for two J u a n Fernandez natives, Coprosma and Santalum, which have thereby spanned almost the ent i rety of the Pacific demonstrates the tremen­dous capability of this dispersal method.

Tha t events of internal t ranspor t of seeds by birds are infrequent seems indisputable, but one need not believe for that reason that they are non­existent. Nor should or need one depend on very improbable means of t rans­por t via birds. Visits by typical frugivorous land birds to the Hawaiian Islands are undoubtedly very rare. Among the stragglers listed by Munro (1960) are no typical frugivorous birds, such as pigeons. The stragglers include mostly marine birds, water birds, or birds of prey. However, many

19GT] CARLQU1ST: T H E BIOTA OF LONG-DISTANCE DISPERSAL 143

ducks, teals, etc. are in Munro's list, and these have been sighted with great regularity according to ornithologists who have worked in the Hawaiian Islands (Eugene Kridler, pers. commun.). Ducks could presumably carry a wide variety of fruits or seeds—particularly from boreal source areas. Ducks would be less adequate for transport of plants from Indo-Pacific regions, however. How have the 53 hypothetical immigrants by internal transport in birds arrived from Indo-Malaysia ? On the answer to this question literally hinges the credibility of long-distance dispersal as a whole, for if these introductions can be satisfactorily explained, others certainly can also. We can, to be sure, imagine atolls which were once high islands and which therefore enhanced dispersal possibilities from Indo-Malaysia to the Hawaiian Islands. However, such islands are meaningless as stepping-stones in dispersal unless suitable bird vectors exist. As a hint of possible birds, we may consider Peale's observation (reported in Guppy. 1906) that the principal food of the bristle-thighed curlew (Numerrius tuhitensis) is, at least at certain times and in certain localities, fruits of Canthium odora-tum (Rubiaceae). Many fruits in the Hawaiian flora are comparable to those of Canthium in size and morphology. No single-seeded fruits larger than those of Canthium need be postulated as ancestral to the Hawaiian flora.

Shore birds such as the curlew would probably not be expected by many botanists to eat fruits and seeds to any appreciable extent. Sink birds, how­ever, can easily be omnivorous, and the above report is not an isolated one. To select a single reference, one may cite Munro's (1960) report of con­sumption of seeds by such shore birds as the sanderling, the ruddy turn-stone, and the stilt. Shore birds such as these are wide-ranging. The bristle-thighed curlew, for example, ranges between Alaska and Tahiti; other shore birds exceed that range.

Some will recall pessimistic estimates, such as those of Ridley (1930), on the brief length of time which elapses between ingestion and excretion of fruits by birds. On more than one occasion, however, workers have empha­sized mitigating circumstances: the entire content of a bird's digestive tract may not be completely emptied, and a few seeds may remain for longer pe­riods ; birds may not always fly on empty stomachs, as some claim they do; storms may alter habits of digestion, excretion, and preening from what normally is the case; and a few seeds may by chance adhere to anal regions of birds. Those who are dubious about the potential of internal transport of sciMIs by birds stress the rarity of its occurrence, not realizing that they are thereby supporting an argument for likelihood of such transport. If this— or any other—kind of dispersal were frequent, isolation leading to specia-tion and endemism would be impossible. A means of transport does not need to be frequent to be effective.

144 BULLETIN OP THE TORKEY BOTANICAL CLl'B i Vol,. 04

LAVSAN I . N » 2 6

4 . 0 . HAWAIIAN IS.

8.5

N ' 2 5 6

63.6«

12.8 > *

38.9

EQUATORIAL I S . N = 33

.C

»40.0

REVILLAGI6ED0 I S . N - 8 2

3.7 . . 8 . 5 24.4«

7 * 15.1

^fFT'

MARQUESAS IS . N-132

3 L .3.6

C0C0S L N « 4 7 0 ^ 4 . 3

12.6 • « * 15.0 1U

8.5

5 0 3

.24.2

TONGA N ' 2 9 6

21.C

50.8

KEY Drift,Rare (Rafting) v ^ A i r f | 0 ( a t i o n

Drift, Frequent •

Birds- Viscid fru'rts,seeds

Birijjy Mud on feet

!375

GALAPAGOS IS . N -308

13 .4 .3 21.8.

6.5 fv I3T X ,

22.8

•27.7

DESVENTURADAS IS. N ' 2 0

20.0

)QD

ISLO * 10.0

•Birds- Barbs, bristles, etc,

JUAN FERNANDEZ I S . N-IOO

Birds-Internal

40.0

Fig. 13. Diagrammatic representation of modes of dispersal responsible for populat­ing the islands of the eastern Pacific and Polynesia with seed plants. Diagrams are ar­ranged geographically, although distances are distorted. For each flora, the number of immigrants presumed necessary to account for the contemporary native species has been calculated (N) . Of these, the percentages which are supposed to have arrived according to the seven methods shown in the key have been calculated and presented graphically. Only native species are included. The figures for the Hawaiian Islands include the Lee­ward chain; figures for Laysan alone also are presented.

The most elusive category of dispersal by birds is that of inclusion of seeds in mud on birds' feet. Main lines of inference leading one to claim this method of dispersal for a particular species include small seed size and the tendency for plants to grow in marshes, mud-flats, or streamside habitats. Seeds which could become embedded in mud might, to be sure, just as easily

1967] CARLQUIST: T H E BIOTA OP LONG-DISTANCE DISPERSAL 1 4 5

be eaten by birds and distributed internally. For this and other reasons, ar­r ival in mud on b i rds ' feet is difficult to identify with certainty. This method has been estimated for 12.8% of the arr ivals to the Hawaiian Islands, how­ever. This percentage is higher than claimed on most other Polynesian islands (fig. 13) . The importance of this means of t ranspor t to the Hawai ian Islands may be related to the likelihood tha t ducks and other water birds assume—if only by default—a greater role in long-distance dispersal there than they do elsewhere. As a corollary, one might infer that t r anspor t in­ternal ly in birds is somewhat more difficult here because of remoteness of the Hawai ian Islands.

Wi th respect to t r anspor t by birds of seeds and fruits which have viscid surfaces, one can cite some clear examples such as those of Boerhavia (fig. 3, 4) and Pisonia. Other examples are more subtle, or perhaps show a lesser degree of efficiency for this means of travel. The fact tha t such a species as Boerhavia diffusa or Pisonia it mh< II if era ranges over wide stretches of the Pacific without so much as subspeeific differentiation between ilistant islands testifies to the excellence of this dispersal device. Repetitive re-introduc­tion leading to swamping of distinctive characteristics in incipient speeia-tion evidently occurs. Indeed, Boerhavia could probably not persist on some remote atolls where it grows if re-introduction did not provide new genetic material . According to the considerations of MacAr thur and Wilson (1963), small remote islands foster a very high ra te of extinction. A species must therefore be re-stocked frequently. Distribution of Boerhavia by marine birds is evident (fig. 4 ) . Similar evidence could be cited concerning Pisonia. A relationship between sooty terns and Pisonia on Rose Atoll is mentioned by Mnnro (1960). On Heron Is land (Queensland) , marine birds are claimed to become so thickly covered with Pisonia fruits that they are immobilized and become prey for rats (R. F . Thome, pers. commun.) . Another p lant which seems to owe a wide distribution into the Pacific to viscidness of seeds is Lepidium. This accounts for the occurrence of Lepidium owaihense on small and remote islands such as the islets of Pear l and Hermes Reef. On this atoll, it was observed as one of the first colonists on an islet, " S o u t h Nor th I s l a n d , " which was formed recently and has (as of 1966) acquired only four species of angiosperms. Lepidium owaihense is also present in the Line Islands. Mature seeds of this species become gelatinous when wetted. This characteristic also occurs in other Hawaiian plants, such as Euphorbia a n d Plantago. Plantago is not as wide-ranging within the Is lands as is Euphorbia, probably because Plantago has narrower ecological require­ments, ra ther t han poorer dispersibility.

Pittosporum has seeds covered with a viscid, almost oily, substance. The distr ibution of this genus, which includes some ra ther remote islands, may be related to this quali ty. However, Pittosporum seeds are black and may

146 BULLETIN OP THE TORREY BOTANICAL CLUB [VOL. 9 4

therefore be attractive to seed-eating birds. Ptttosporum rarely grows close to where marine birds nest—although a relationship cannot be ruled out. In any case, dispersal events would be expected to be relatively infrequent compared with those of Boerhavia, so the higher endemism in Pittosporum species in the Pacific is understandable.

Readers may note that shore and water birds do migrate and do eat plant material and are otherwise suitable as vectors, but the question of efficacy of marine birds remains. Do marine birds ever characteristically undertake interisland travel or do they, as is often averred, invariably re­turn to a home island? Recent work shows that considerable interisland travel occurs. For example, on Laysan Island, 69,900 sooty terns were banded. Of these, 611 returns were discovered on Laysan, 7 on Johnston, 4 on Lisianski, and 2 on Kure. Similarly, of red-footed boobies banded on French Frigate Shoals, 225 returns were noted. < >f these, all were on French Frigate Shoals except for 18 on Johnston, 8 on Laysan, 3 on Midway, 2 on Oahu, 2 on Kauai, 1 on Pearl and Hermes Reef, and 1 on Wake (Richard Crossin, unpub.) Interestingly, some of those individuals observed on Sand Islet of Johnston were later seen back on French Frigate Shoals again. The fact that in a single year so much interisland movement (of which these figures surely represent a modest sampling) occurs in marine birds means that these birds are entirely adequate as potential vectors. The white-tailed tropic bird must range widely, judging from the observations offered by Munro (1960). Moreover, it nests in cliff faces, such as those on northern Kauai, where it could contact a great variety of plants. Because these are windward north-facing cliffs, wet forest species (e.g., Pittosporum, Metrosi-deros range much closer to sealevel than one would expect. Thus, there is considerable likelihood that a seabird, the white-tailed tropic bird, might come into contact with upland plants. The dark-rumped petrel, a Hawaiian endemic, can nest well inland, and at various altitudes, from 1,500 to above 5,000 feet. Suggestions that marine birds can nest well inland and may ex­plain some peculiar distribution patterns in New Zealand are offered by Falla (1960).

Although marine birds would not be expected to eat much vegetable matter, they evidently sometimes do. According to Ridley (1930), gulls eat "al l sorts of berries, and especially those of Empetrum nigrum.11 Interest­ingly, Empetrum is one of the plants which shows the noted amphitropical distribution (Constance, 1963). The Hawaiian species of Bidens which show the least loss of dispersibility, and which therefore suggest mode of arrival best, are those which live on bluffs where marine birds nest (Carlquist, 1966b). Various instances in which marine birds have been observed to carry seeds and fruits are given by Ridley (1930).

Shore birds and water birds may, however, be responsible for transport

1967] CARLQUIST: THE BrOTA OF LONG-DISTANCE DISPERSAL 147

of more numerous plant species. To claim that such birds always preen off nil so.-ds nf evacuate nil material before flying is. in my Opinion, highly un­realistic. One must remember that "regular ," " repeated" or even " l ikely" dispersal via birds is not only not to be expected, it does not need to be ex­pected to account for distribution of plants by birds vectors. One can do little better than to quote again Ridley's opinion. "There is at first the endeavor in the absence of precise knowledge to disregard the bird and to look for a land connection. "With, the increase in our acquaintance with the efficacy of bird-agency in seed distribution there is abandonment of such a view." Regrettably, the 60 years since the publication of that remark have seen relatively few attempts to gather information on " the efficacy of bird-agency." The need for investigation of birds as vectors in plant dispersal remains one of the more needed aspects of biological study. If anything, we have less study of these phenomena now than previously. The absence of such investigations, however, should not be mistaken for inability of birds to transport seeds and fruits.

OCEANIC DRIFT. Oceanic drift is considered here to include two phe­nomena: frequent or repetitive drift on the one hand, rare or infrequent drift (rafting, for example), on the other. This distinction was suggested by the fact that there are some plants, such as Ipomoea pes-caprae, which are pantropical and show virtually no variation from one island to another, whereas in the Hawaiian Islands, there are lowland or littoral plants much higher in endemism but, as it happens, also poor at flotation, such as Malvaceae and Leguminosae (Stephens, 1966; Carlquist, 1967). Experi­mental proof that there can be species with seeds resistant to seawater but poor at flotation (either individually or within the fruit) has been shown in several eases (e.g., Stephens, 1958, 1966 • Whitaker and Carter, 1961).

In cases of repetitive drift, lack of endemism would be expected. Species with obvious adaptation to flotation, such as Mucuna (fig. 5, 6) offer no problems in this regard. Less obvious are species which range widely in the Pacific without much variability, but which are not known to have floatable seeds or fruits. In some cases, as with Portulaca lutea (fig. 7) and Seswrium portulacastrum, vegetative portions of plants prove to be buoyant, and one can suspect that this faculty is responsible for introduction of these species to remote islands. Similarly, littoral species with wide ranges may be sus­pected of floatability in some manner. Lycium sandwichense (Rapa, Easter, Tonga, Hawaii) and Capparis sandwichiana (Hawaii, Pitcairn, Tuamotus, Rarotonga, Cook Is.) are notable in this regard. Vegetative portions of Lycium sandwichense can, in fact, float. Fresh fruits and seeds of Capparis sandwichiana are reported by Guppy (1906) not to float. However, my ex­periments show that dried fruits do float indefinitely.

An interesting phenomenon apparently not yet appreciated is the fact

148 BULLETIN OF THE TORREY BOTANICAL CLUB [ V r " : "

tha t seeds in the drif t flora are not forced to grow only on the spots where the surf deposits them. On Necker Island. I repeatedly observed seeds of Mucuna and Aleurites (the former viable, the latter dead) in and around nests of boobies, nests far above the surf. Boobies apparen t ly habitual ly pick up these objects and deposit them (apparent ly carried in beaks but not eaten) at nest ing sites. A similar phenomenon was observed in the case of Laysan albatrosses on Pear l and Hermes Reef and other islands of the Ha­waiian Leeward chain visited by the wr i te r and other members of the Iron-wood scientific pa r ty . Mr. Kar l Kenyon noted numerous carcases of Laysan albatrosses in which floating objects—plastic, floating seeds, pumice, char­coal, etc.—remained within the body cavity. These objects were evidently picked up at sea, conspicuous by vir tue of their flotation. They may have been fed by regurgitat ion to albatross chicks, which died as a result. I n any case, albatrosses can act conjunctively with oceanic drif t to abet ar r ival of floating objects on islands.

Rafting—or similar infrequent dispersal via oceanic drift—is much more difficult to assess. The species in the Hawaiian flora which suggest arrival by this means are generally coastal or lowland species. These some­times have relatives adapted to dispersal by flotation. Guppy (1906) found tha t in Hawai i a very few instances of ecological shift of l i t toral elements into the forest can be claimed to exist. To me, there seem more numerous instances, and I would add virtually all Hawaiian legumes and Malvaceae, as well as certain other elements in the dry lowland forest. The Hawaiian flora is, however, the only Polynesian flora in which this ecological shift can be claimed to have occurred to any appreciable extent, as noted by the writer elsewhere (1967). Why only in the Hawaiian Island-. ' Apparent ly the drif t flora of the Pacific is primari ly a South Pacific drift flora, which ex-changes easily among soul hern archipelagos, but apparent ly does not traverse the transverse equatorial currents frequently. I Inly in this fashion could one explain (1) tha t the drif t flora has yielded so many endemics in the Hawaiian I s lands ; and (2 ) , that many common South Pacific genera are absent. In the former category, endemic species in the following genera a re notewor thy: Ochrosia, Dodonaa. Wezoneurum, Canavalia, Erythrina, Acacia, Sophora, Colubrina, etc. In the latter category one notes the absence in Hawai i of such genera as Barringtonia, Suriana, Sonneratia, Brugiera, and Bhizophora. Tha t at least some of these are ecologically suited for es­tablishment in the Hawai ian Islands is shown by the fact t ha t they have been introduced in recent times and have natural ized (e.g., Bhizophora and Brugiera on Oahu) .

Compar isons be tween the Hawai ian Is lands and o the r is lands of t he eas tern Pacific. I n fig. 13, the dispersal mechanisms by which the native floras of Polynesia or the offshore Pacific islands of the Americas a re sup-

19G7J ( AI.I .QUIST: T H E BIOTA OF LONG-DISTANCE DISPERSAL 149

posed by the writer to have arrived are shown in a schematic fashion. These comparisons are useful because they show some unexpected correlations with respect to dispersal. They also place in relief certain characteristics of particular islands or archipelagos. These figures were developed in the same way as those for the Hawaiian Islands. Flora of archipelagos (e.g., Society Islands) were used in some cases. To be sure, the floras of archi­pelagos may not be entirely comparable to floras of individual islands, such as Easter I., but comparisons only of single islands would have resulted in more distorted data.

SAN CLEMENTE I. Generally conceded to be an oceanic island, San Cle­mente is near enough to the mainland to show essentially continental pat­terns in many respects. As an island, San Clemente has acquired a very large percentage of its plants via air flotation—a much higher percentage than any of the other islands considered here. My estimate of groups claimed to be brought internally in birds (40.0%) would be too great if concepts applied to the Hawaiian flora were used. In the San Clemente flora, most of the seeds in this category are small round, often black, seeds borne in dry capsules. Such seeds might be brought in various ways, but because such seeds are often sought by birds and because San Clemente is close enough to the mainland to have been visited frequently by land birds, internal transport in birds is suggested here. There are, however, a few genera with fleshy fruits (Ribes, Rhus, Heteromeles). Many small seeds borne in dry fruits might be claimed to be brought in mud on birds' feet, but this alterna­tive seems somewhat less likely in many cases because muddy sites (except for streams and marshes) are relatively infrequent or short-lived in the California climate.

REVILLAGIGEDO IS. The effect of a dry subtropical climate in influencing the types of plants which establish is shown clearly by the pattern of arrival types in the Revillagigedo flora (fig. 13). The proportion of dry fruits which have arrived via barbs, hairs, etc., caught in feathers is high. The propor­tion of seeds carried internally in birds is relatively low. Dry climates would be expected to be related to more numerous dry fruits, fewer fleshy ones, as compared with rain forests. The percentage of plants claimed to have arrived via air flotation (8.5%) seems small, but it is large compared with other islands.

Cocos I. The Cocos flora contrasts with the floras of the Rovillagigedos and Galapagos Islands in being a notably mesic flora. This is reflected in the very high percentage of plants alleged to have arrived internally in birds. This mode of arrival is easier on Cocos than on other very mesic is­lands (e.g., Juan Fernandez) because of the close proximity of Cocr-- f> the Central American mainland.

150 BULLETIN OF THE TORRET BOTANICAL CLUB fVor" 0 4

GALAPAGOS I S . The Galapagos flora is the largest of the floras of offshore islands in the eastern Pacific. Notable in this flora is the high percentage of disseminules claimed here to have arrived caught in feathers by means of barbs or hairs. The proport ion of seeds and fruits thought to be brought internally in birds is notably low. and the dr i f t element is ra ther large. These percentages are remarkably similar to those in the R<?villagigedo flora, reflecting an ecological situation oriented primari ly toward xerophytic plants.

Desventuradas Is. The number of hypothetical introductions to these islands (20) is so small tha t the profile of probable arr ival methods is very likely to be misleading. However, the exceptionally low proportion of types brought internally in birds and the high proportion of plants in the drif t flora are notable in these low. dry, and ecologically unvaried islands.

J U A N FERNANDEZ I S . The distance of the J u a n Fernandez Islands from the mainland is somewhat greater than for other offshore islands. This, com­bined with the distinctly mesic ecology seems to favor a ra ther high propor­tion of immigrants brought internally in birds, a ra ther low proport ion of arrivals via air flotation. The fact tha t the percentage supposedly brought internally in birds is very close to that on the Society Islands and Hawaiian Islands is probably due to similar ecology. Attract ive fleshy fruits are largely correlated with wet forest habitats . The percentage of the flora claimed to have arr ived in mud on b i rds ' feet is very large, and may be questionable. I t is the largest percentage for this mode of arr ival in any of the floras studied here. The J u a n Fernandez flora does contain some marsh plants , but the correlation with highly mesic habitats may be more signifi­cant in this regard. The drif t flora is exceptionally poor. This seems related to the minimal amount of s t rand habitat on these islands. J u a n Fernandez shores are essentially rocky, and the li t toral zone is very narrow.

EASTER I. Unexpectedly, Easter has a large proportion (21.4%) of ar­rivals supposed to have come in mud on b i rds ' feet. This would seem quite unlikely for a low dry island unless one remembers tha t there is a sizable crater lake with muddy margins on Easter . The large percentage of barbed or hairy seeds likely to have been brought in feathers seems related to the d r y conditions—especially grasslands. Easter also has a ra ther high pro­port ion of drift flora species. This would be expected for a relatively low, remote island. The low proportion of fruits at tract ive to seed-eating birds betokens the d ry ecology of the island. The low number of air-dispersed seeds may be at t r ibuted primari ly to remoteness, secondarily to dryness.

ATOLLS. Three sets of figures for islands in this category were obtained: ( 1 ) , Laysan, in the Hawaiian Leeward chain; ( 2 ) , Pacific Equator ia l is­lands, roughly synonymous wdth the te rm " L i n e I s l a n d s " and following

1907] CARLQUIST: T H E BIOTA OF LONG-DISTANCE DISPERSAL 151

the area included in Christophersen's (1927) flora; and (3 ) , Oeno I. Despite the different locations of these floras and their differing proximities to source areas, their profiles of supposed modes of arrival of p lant immigrants a r e remarkably similar. Most conspicuous is the scarcity of fleshy fruits brought internal ly in birds. This does not seem related to lack of vectors. Shore birds, mentioned above as likely carriers of such fruits, are certainly present on these atolls, and may have carried some of the seeds ancestral to flowering plants now resident there (e.g., Solatium nelsonii, Santalum ellip-ticum). Rather, the ecological correlation between fleshy fruits and the forest habit is much stronger, and the absence of appropr ia te ecological niches is of overriding importance. The large percentage of the atoll floras traceable to oceanic drif t comes as no surprise. Disseminules of dr i f t p lants a r e not only brought in large quantit ies to these islands, they are mostly ideally suited ecologically for growth on atolls, which can often be consid­ered as little more than beaches. An exception is formed by species of Mucuna (fig. 5 ) . Seeds of this genus arrive in large numbers, occasionally germinate (fig. 6 ) , but have never been known to survive on the Hawaiian Leeward chain and in other localities where they are known to be deposited on beaches. The case of Mucuna does serve to emphasize the pre-eminent role of ecology of the recipienl island. Mucuna probably requires less sandy soils, more rainfall and shade than characterize many atolls.

SOUTHERN POLYNESIA. The high volcanic inlands of the South Pacific can be discussed together because of thei r similarities. One feature common to all of these islands is a relatively high percentage of plants brought internally in birds, according to my estimates. Of the islands of southern Polynesia, the lowest percentage for this method of arrival is a t t r ibuted to Rarotonga, the highest (50.8) is shared by Samoa and Tonga. The low per­centage on Rarotonga seems related to the fact that Rarotonga is a decadent volcanic island with broad lowlands but very limited montane area. Fleshy-fruited species thus have only a moderate area suited to their ecology, whereas the drif t flora is advantaged on the broad coastal plain. Except for this distinction, there are no major differences among the South Pacific high islands with respect to mode of arrival of angiospcrms. N'otable on these islands is the small proportion of d ry fruits carried in bird feathers. (>ne may note in this connection that grasses form a small proportion of the native floras of these islands. The drift flora is least conspicuous in Rapa— which has a limited l i t toral zone. The similarity in dispersal profile (fig. 13) between Samoa and Tonga is somewhat surprising, for these two island groups differ topographically. In par t , this resemblance derives from the fact that Tonga and Samoa are close together, and about equally close to Indo-Malaysian source areas—part icular ly Fij i . Very notable is the rela­tively high proportion of air-dispersed species on all of the South Pacific

152 BULLETIN OF THE TORREY BOTANICAL CLUB 1VOL- 9 4

high islands except Rapa. Rapa ' s low percentage of arr ivals by this means is probably related to its greater distance from source areas. Orchids (all of which are regarded here as dispersed by air flotation) are pr imari ly re­sponsible for the percentages in this category in the various floras, and orchids seem to decrease markedly with increasing distance from Indo-Malaysia.

MARQUESAS I S . AND HAWAIIAN IS . The pat tern of dispersal types in these two island groups is not unlike those of southern Polynesian high islands. The resemblance between Rapa and Hawaii is striking. The main reasons for this would appear to be similarity in ecological spectrum and the simi­lar ly great distances from either Indo-Malaysian or American source areas. A very low proport ion of air-Heated species characterize Rapa, the Mar­quesas, and the Hawaiian Islands.

Long-d is tance dispersal to regions other than Polynes ia .—MASCARENE ISLANDS. If, as I have stated earlier (1966a), disharmony is an indication tha t certain plants and animals are positively adapted to long-distance dis-persal, these ought to be more conspicuous on oceanie islands. A phenomenon which might not have been predicted, however, is the tendency of the Mas­carene flora to be " P o l y n e s i a n " in character. Computations demonstrate this. Despite the fact that the Seychelles, Reunion, Marit ius and Rodriguez are close to Madagascar and Africa, their floras reveal many of the char­acteristic genera of Polynesia. The islands named are included in I laker 's (1877) flora. Of the 408 genera of native flowering plants in this flora, 158 occur also in Polynesia (Samoa to Eas ter on the east. Hawaii on the north, Rapa on the south) . Comparing smaller floristic units, this becomes even more striking. If one regards 210-216 genera of flowering p lants as basic i aii'-'-strnl'' in the Hawaiian flora. 11"). or o.'i-fWi', of these also occur in the Mascarene flora. To be sure, a number of such genera represent elements of the drift flora which might be expected to be widespread on islands. Bu t many of the genera in common between the two areas are not drift p lants a t a l l : DianeUa, Dracaena, Ash I in. Pipturus, Peperomia, Diospyros (in-cluding Maba), Myrsine, Psychotria, Canthium, Pittosporum, Elaeocarpus, Fagara, etc. One would certainly not expect American and Boreal elements on the Mascarene Islands. These comprise a little over 2 0 % of the hypo­thetical immigrants to the Hawaiian flora. Likewise, one would not expect African elements in the Hawaiian flora. If one deducts these, the element of similarity between the Mascarenes and Polynesia is really quite high, and is an expression of that portion of the Indo-Malaysian flora adapted for long-distance dispersal. Interestingly there is a marked zoological counter­part to this phenomenon. Gressitt I 1956) shows a Pacific-Mascarene dis­t r ibut ion for several insect group-, such as Chrysomelidae—Ilispinae and Cerambycidae. Mascarene-Polynesian affinity can even be shown in the

19G7] CARLQUIST: THE BIOTA OP LONG-DISTANCE DISPERSAL 1 5 3

case of land shells. Elasmias (Tornatel l inidae) is an insular Pacific genus which has an extension on the Mascarene islands (Cooke and Kondo, 1960). The family Tornatell inidae seems strongly adapted for long-distance dis­persal, from the distribution maps these authors offer.

INTERCONTINENTAL DISPERSAL. This topic is potentially very important , for the degree and na tu re of intercontinental dispersal bears a direct rela­t ionship to the evidence on which the probabili ty of continental drif t rests. Also in question are theories of dispersal within continents. One area which has been of critical significance is the disjunction between western North America and the Chile-Peru region. This has been the subject of a review and individual papers by Chambers (1963), Constance (1963), I leckard

L963 I, <»rnduff (1963) and Raven (1963). Difficulties to interpretat ion of this disjunction as the product of dispersal by bird agencies have been hinted by Constance (1963) and Cruden (1966). Although Cruden claims not to be motivated by disbelief in the efficacy of birds, he evidently wishes us to consider an alternative hypothesis, tha t of a series of mountain-tops as stepping-stones across the tropics, br idging the two temperate regions. Al­though the amphitropical disjunction needs a more thorough examination with respect to whether seeds and fruits are adapted to long-distance dis­persal by birds, the following comments are offered. These comments are intended to support the concept of birds as dispersal agents. The back­ground of studies in Polynesia and in the Mascarcnos seems an essential ingredient , for distant oceanic islands form compelling criteria of the reali­ties of long-distance dispersal.

( 1 ) . The distances of the North America-South American disjunction mentioned above are of the same level of magnitude as those of the America-Polynesia disjunction. If 20% (at least 52 immigrants) of the establish­ments ancestral to the Hawaiian flora arrived (mostly via birds) from American sources, then the amphitropical American disjunction does not seem highly improbable on the basis of long-distance dispersal. Moreover, the North America-Hawaii or North America-Polynesia disjunction is a one-way dispersal, whereas both North America and Chile-Pern have been able to serve as source areas for the amphitropical disjuncts.

( 2 ) . The list of plants showing amphitropical disjunction is very similar to the list of migrants from American or Boreal regions to Hawai i : Lepi-dium. Ranunculus, ('uranium, Asttr, Artemisia, l.ir.ula, Corex, Scirpus, S/ipa, I'oa, Yaccinium, Rub us, Fruijaria, Viola, Ca< snlpinio, and Ka nicula are among the amphitropical disjuncts one might cite in this regard. More­over, f rui t morphology is very similar to tha t of the America-Hawaii dis­juncts . The list of genera showing both disjunctions indicates clearly excep­tional ability for long-distance dispersal. Some genera in either disjunct

1 5 4 BULLETIN OP THE TORREY BOTANICAL CLUB IV o L- 9 4

pattern could be said to have "no obvious dispersal mechanism," but this would be a mistake. Anything that has been able to cross 2,000 or more miles of ocean to Hawaii must have an effective dispersal mechanism. "We must remember that birds may eat and cany internally fruits other than those which are fleshy and brilliantly colored. For example, ducks eat and trans­port a wide variety of fruits which are inconspicuous, often greenish or brownish and tough and woody in texture (Ridley, 1930). Moreover, lack of a dispersal median ism in the amphitropieal disjuncts might, in a few in­stances, reflect a loss of dispersibility (Carlquist, 1966b, 1967).

(3). No "mountain-top stepping-stones" have been proposed for the North America-Hawaiian Islands disjunction, nor are they necessary to explain the dispersal which has occurred. The proposal of intervening mountain-tops to serve as way-stations in the amphitropieal disjunction seems unnecessary. At best, it merely cuts long-distance dispersal into a series of shorter segments, and thus still demands the operahility of long­distance dispersal while at the same time making the process less simple for reasons given in the following item.

(4). Mountain-tops in the tropics are not equivalent to temperate cli­mates, montane or otherwise. This holds particularly with regard to sea­sonal changes in temperature and day-length, but also applies to rainfall patterns, soils, etc. To accept a mountain-top hypothesis, we must expect temperate zone plants to make numerous adaptations to tropical mountains. Then, after they have so adapted, one must imagine they can re-adapt to temperate conditions (as in the opposite hemisphere), and simultaneously disappear in the tropics. If the roster of amphitropieal disjuncts did adapt. to tropical mountain-tops, why have they disappeared in the tropica so completely that a pattern which can, in fact, be termed amphitropieal dis­junction occurs? The data from Polynesia show just the reverse of the mountain-top hypothesis. They show that intervening way-stations e.ir.. Samoa as a point midway between Fiji and Tahiti) have retained more immigrants than the end-points. Moreover, relatively little ecological change has occurred in most angiosperm stocks on islands (the Hawaiian Islands are the only major exception in this regard). The various authors of the group of papers cited on the amphitropieal herbaceous disjuncts emphasize the genetic closeness of Chilean plants to Californian, or vice verse, evidenc­ing remarkably little evolutionary change since the time of disjunction. However, if we accept tropical mountain-tops as way-stations, we must imagine considerable physiological change, first progressive then retrograde, no evidence of which is now visible. Moreover, morphology of the amphi­tropieal disjuncts reveals no hint of a period of residence on tropical moun­tain-tops.

The islands of Polynesia seem to have received a relatively abundant

1907] CAKI.QUIST: T H E BIOTA OF LONG-DISTANCE DISPERSAL 155

quantity of waif disseminules; what has been severely limiting seems not lack of transport, but ecological conditions and sufficient land area suitable for establishment and then survival over a number of generations. The pat­terns of fig. 13 show that Pacific islands have "selected ou t" from a flow of immigrants those which are exactly suited ecologically to recipient is­lands. The disjunction between temperate Xorth America and temperate South America is a disjunction precisely because ecological conditions are s arly identical in the two regions, yet unlike in intervening areas. If plants had to adapt to equatorial alpine peaks, some relicts of this sup­posedly recent migration ought still to be there. Instead, they appear not to be, and the floras of equatorial mountain tops are relatively different.

(5). For successful migration, the genetic conditions of an immigrant stock must have or must be capable of manufacturing sufficient heterozygos­ity for survival over many generations (or else the stock vanishes, and is thus outside the question). This principle has been enunciated by Mac-Arthur and Wilson (1963) and Carlquist (1967), but is apparently not taken into account by Cruden (1966). If an introduced stock lacks hetero­zygosity, it would be expected to disappear in a few generations, or else attain it by re-introduction. At any rate, homozygous species would be expected to have such a short expectancy that virtually none would be on hand at any given time. Introduction of more than one disseminule of a species would seem the best means of achieving heterozygosity in an immi­grant population, and this does not seem unlikely.

(6). The evidence concerning bird groups offered by Cruden (1966) has been assessed by him in a very conservative fashion. To one acquainted with Polynesian floras and potential bird vectors in the Pacific, birds seem more than adequate to explain transport of the amphitropical disjuncts.

Other problems which have been subject to explanations other than long-distance dispersal include certain South America-Africa disjuncts. Rapatcaceae is an interesting example: a single genus, Maschalocephalus, occurs in Liberia, whereas the remainder of the family occurs in northern South America and Panama—centering especially atop the massifs of the Guayana Highland. Interestingly, Maschalocephalus is a highly specialized genus in the family, close to Windsorina and Potarophytum (Maguire, 1958; Carlquist, 1961, 1966c). This suggests recent long-distance dispersal from South America to an ecologically similar site in Africa. Rapateaeeae can be called marsh plants. As noted above, marsh plants seem related to many excellent bird vectors, such as ducks, teals, rails, etc. Among other South America-Africa disjuncts which seem best explained by long-dis­tance dispersal are the families Mayacaceae and Velloziaceae.

Plants with very small seeds are likely to become widespread and to

156 BULLETIN OF THE TORREY BOTANICAL CLUB lVoL- 9 4

show disjunctions: Coriaria (Coriariaceae) and Pilostylcs (Rafflesiaceae) can be cited in this regard.

While the possibilities of continental drift, etc., at some (at best, prob­ably remote) period cannot be categorically denied, the citation in that re­gard of distribution patterns which can perfectly well be explained as the result of long-distance dispersal (e.g., Del Corro, 1964) seems a quite dubious procedure. On the contrary, one can find examples where long-dis­tance dispersal has occurred within or among continents, even where dis-persibility would not seem great (e.g., Gillett, 1965).

Conclusions and summary. The purpose of this paper is analysis of the floras of the oceanic islands of Polynesia and the eastern Pacific with respect to mode of arrival of the pre-human immigrant angiosperms. In order to do this, the number of original immigrants was estimated for each flora, and the probable mode of dispersal chosen for each such immigrant. The results of this analysis and examination of allied phenomena may be sum­marized as a series of principles:

(1). All elements in the floras of Polynesia can be said to be adapted for long-distance dispersal by one of the following methods of long-distance transport: air flotation; barbed, bristly or hairy dry seeds or fruits caught in feathers; seeds or fruits eaten by birds and carried internally; seeds or fruits embedded in mud on birds' feet; seeds or fruits capable of flotation and carried at frequent intervals by oceanic drift; seeds or fruits which can withstand contact with seawater but which cannot ordinarily float, and must therefore be carried by rare events of drift, such as connoted by the term "raf t ing."

(2). The percentages of species which arrive by these respective meth­ods are affected by (a) distance from source area; and (b) ecology of the recipient area. Ecology is of overwhelming significance, and there is reason to believe that dispersal provides a rather great number of events of trans­portation compared to the number of successful establishments.

(3). Air flotation is responsible for arrival of a relatively small pro­portion of the pre-human plant immigrants to Pacific oceanic islands. This proportion decreases sharply with distance. The Hawaiian islands have received a very small number of immigrants in this fashion, despite the favorable ecological conditions there. The absence in the Hawaiian Islands of wind-distributed plants of South Pacific oceanic islands—Weinmannia, Balanophora, and many genera of orchids—is very suggestive. If one com­pares the number of orchid species on islands with the distance from Indo-Malaysian source areas, one notes a sharp decrease with distance.

(4). Patterns of appearance of vegetation on new lava flows seem ex­plainable by dispersal ability and method primarily, ecological require-

19G7] CARLQUIST: T H E BIOTA OF LONG-DISTANCE DISPERSAL 157

ments secondarily. On new lava flows in the Hawaiian Islands, early occu­pants of lava flows, whether trees, herbs, or lichens, are wind-dispersed organisms.

(5). Insular floras in which a larger proportion of species arrive by virtue of dry fruits bearing barbs, hairs, or bristles suitable for attaching to feathers are generally more xeric floras. Easter Island and the Galapagos Islands fall into this category. The abundance of grasses and Compositae on islands is related to drier climates, although they may evolve into wetter situations.

(6). Internal transport of seeds and fruits in birds is the most impor­tant mode of arrival on high islands. Although this means of transportation may seem unlikely to some, it has been responsible for 38 to 50% of the arrivals on the major high islands of Polynesia. The link between this method of arrival and the wet forest, habitat is rather strong. That this method of arrival is slightly less efficient where very great distances are in­volved is shown by the Hawaiian flora, but the decrease even there is not very great. Migratory frugivorous birds (such as Guppy's hypothetical vanished pigeons) are not needed to explain the presence of seeds and fruits attractive to birds in the Hawaiian or other Polynesian floras. There is considerable circumstantial evidence that certain shore birds which regu­larly migrate throughout Polynesia are mixed feeders and are, in fact responsible for many events of seeds transported internally in birds. This avenue of dispersal has received little effective investigation since Guppy's time, and is very much in need of renewed examination, if the hints from various ornithological works are any criterion.

(7). Dispersal in mud on birds' feet accounts for only a small percentage of arrivals on oceanic islands. It may be most abundant on islands close to continents, where flyways can be said to exist. Otherwise, this mode of dis­persal is limited largely to marsh plants or land plants of very moist habitats.

(8). Attachment of seeds or fruits to bird feathers by means of viscid materials accounts for a small but appreciable proportion of plant arrivals on oceanic islands. Several possibilities seem involved under this heading: (a) , plants of atolls, readily and repetitively dispersed on the feathers of marine birds, such as Boerhavia and Pisonia. (b), plants of wet forest with viscid seeds, carried to islands (perhaps by shore birds) by relatively infre­quent successful dispersal events, and therefore giving rise to endemic species on high islands: for example, Pittosponim and Plant ago. (c), plants with very small seeds contained in fleshy fruits may be attached to bird feathers at rare intervals by means of the sticky parenchyma and carried externally by birds rather than internally. This is a possibility in such

158 BULLETIN OF THE TORREY BOTANICAL CLUB £VOL- 9 4

genera as Clermontia, Cyanea, Yaccinium, Gcniostoma, Labordm, and pos­sibly Cyrtandra.

(9). Dispersal by means of repetitive drift characterizes plants with a demonstrable capability for seawater flotation. These plants are possibly the only ones which can be said to have a definitive adaptation for long­distance dispersal. Such plants bulk very large in atoll floras, both because of dispersal capacity and because there is a strong link between plants with this means of dispersal and their ability to grow on low, sandy, rather dry islands. Some plants with excellent ability at seawater flotation have failed to reach the Hawaiian Islands, although they are frequent in South Pacific floras (Rhizophora, Barringtonia). Ecological situations suitable for these plants are probably present in the Hawaiian Islands, and their absence is best explained by the rarity with which fruits or seeds cross the trans­verse equatorial currents from the Southern to the Northern Hemisphere. The tropical drift flora, in fact, may be essentially a Southern Hemisphere flora in the Pacific.

(10). Some plants which have fruits and seeds with poor or no flotation ability may nevertheless have seeds relatively resistant to seawater. Such plants may achieve long-distance dispersal by some form of rafting. These plants would be expected to be mostly lowland or dry forest species. The Hawaiian flora contains more species traceable to rafting or similar infre­quent arrival via oceanic drift than do other Pacific floras. Lack of re-intro­duction or occurrence of very long intervals prior to re-introduction would permit isolation and thereby the formation of endemic species. Endemic species in Gossypium, Sida, Erythrina, Acacia, etc. or some of these may owe their origin to this phenomenon.

(11). While there is the tacit assumption by many workers that success­ful establishment may stem from a single successfully introduced seed, this may well not be true in many cases. Several seeds introduced simultaneously or a series of introductions followed by no further arrivals would provide populations with sufficient heterozygosity to permit long-term survival, yet either of these hypothetical situations would still constitute a "single dis­persal event.'' In the drift flora the high degree of autogamy characteristic of littoral plants seems detrimental to attainment of levels of heterozygosity suitable for long-term success. Occasional outcrossing may occur in these plants, but frequent re-introduction of new disseminules from another area by means of oceanic drift is a way of compensating for autogamy or cleisto-gamy.

(12). Figures from the Hawaiian flora show that endcmism is highest in groups of plants which owe their arrival to transportation by birds. This is probably true in other insular floras as well. A plant species introduced to

1907] ( A R L Q U I S T : T H E BIOTA OF LON"G-DISTANCE DISPERSAL 159

an island by birds probably experiences the longest intervals between in­troductions, or is successfully introduced only a single time. Such isolation from parental populations permits ecological shift and therefore more radia­tion into a variety of habitats. Adaptive radiation may be expected more in bird-distributed stocks than in others, therefore. The lowest number of species per original introduction is found in those plants brought by oceanic drift where such drift is on a frequent or repetitive basis. In the drift flora, isolation from populations elsewhere would be expected to be minimal.

(13). Hawaiian ferns have a rate of endemism which is notably high. This endemism is not as high as that of Hawaiian flowering plants, but it is exceptionally high for a fern flora. High endemism in Hawaiian ferns probably derives from the fact that air flotation over long distances is very difficult (see (3) above), and successful introductions of particular ferns to the Hawaiian Islands are unique or are repeated only after long intervals.

(14). If dispersal possibilities to an island are relatively small, any one of several requirements may be highly effective in barring establishment. If particular pollinators are required, as for an orchid for example, simul­taneous introduction of both an orchid species and its insect pollinator is much less likely on distant islands. This may help explain the paucity of orchids in the Hawaiian Islands, or the absence there of such genera as Ficus.

(15). Plants which may be thought to have "no obvious adaptation" for long-distance dispersal but which are native to an oceanic island (or comparable area) may prove, upon study, to have some unappreciated adaptation. For example, a fruit or seed which is not attractive but is eaten by migrating ducks, not destroyed by digestion, and carried to a new locality can certainly be said to have adaptation for long-distance dispersal. Auto­chthonous loss of dispersal mechanisms on insular areas may be suspected in some instances.

(16). The flora of the Mascarene islands contains an unexpectedly high number of genera in common with the flora of Polynesia. This is interpreted as due to independent acquisition by these two areas of those Indo-Malaysian genera well adapted for long-distance dispersal. This phenomenon is true not merely in the drift flora, but in bird-dispersed plants as well. Similar Polynesian—Mascarene affinity occurs in insects and landshells, probably for similar reasons. One of the interesting features about the Mascarene— Polynesian similarities is that some genera which one might guess not ex­ceptionally well adapted to long-distance dispersal actually must be (e.g., Elaeocarpus, Claoxylon, Diospyros).

(17). Evidence from the nature of dispersal on oceanic islands can be applied to questions of dispersal within continents and to problems of inter-

160 BULLETIN OP THE TORRET BOTANICAL CLUB [VOL. 94

continental dispersal. Arguments are presented to suggest that the noted floristic disjunction in many herbaceous groups between western North America and the Peru—Chile region can be most easily explained as the product by long-distance dispersal by birds. Similarly, some of the inter­continental patterns of disjunction which are sometimes attributed to con­tinental drift (e.g., Rapateaeeae, Mayacaceae, Restionaceae, Coriariaceae, Mutisieae, Cunoniaceae) are probably best explained as the result of long­distance dispersal.

Literature Cited

BAKER, J. G. 1877. Flora of Mauritius and the Seychelles. L. Reeve & Co., London. BROWN, F . B. H. 1931. Flora of southeastern Polynesia. I. Monocotyledons. B. P .

Bishop Mus. Bull. 84: 1-194. . 1935. Flora of southeastern Polynesia. I I I . Dicotyledons. B. P . Bishop Mus. Bull. 130: 1-386.

CARLQUIST, S. 1961. Pollen morphology of Rapateaeeae. Aliso 5 : 39-66. . 1962. Trematolobelia: seed dispersal; anatomy of fruits and seeds. Pac. Sci. 16: 126-134. . 1966a. The biota of long-distance dispersal. I. Principles of dispersal and evolution. Quart. Bev. Biol. 4 1 : 247-270. . 1966b. The biota of long-distance dispersal. I I . Loss of dispersibility in Pacific Compositae. Evolution 20: 30-48. . 1966c. Anatomy of Rapateaeeae—roots and stems. Phytomorphology 16: 17-38. . 1966d. The biota of long-distance dispersal. IV. Genetic systems in the floras of oceanic islands. Evolution 20: 433-455. . 1967. The biota of long-distance dispersal. I I I . Loss of dispersibility in the Hawaiian flora. Brittonia.

CHAMBERS, K. L. 1963. Amphitropical species pairs in Microseris and Agoseris (Com­positae: Cichorieae). Quart. Rev. Biol. 38: 124-140.

CHEESEMAN, T. F . 1902. The flora of Rarotonga, the chief island of the Cook group. Trans. Linn. Soc. ser. 2, Bot., 6: 261-313.

CHRISTOPHERSEN, E. 1927. Vegetation of Pacific equatorial islands. B. P. Bishop Mus. Bull. 44: 1-79. . 1935. Flowering plants of Samoa. B. P. Bishop Mus. Bull. 128: 1-221. . 1938. Flowering plants of Samoa. I I . B. P . Bishop Mus. BulL 154: 1-77. , and E. L. CACM. 1931. Vascular plants of the Leeward Islands, Hawaii. B. P . Bishop Mus. Bull. 8 1 : 1-41.

CONSTANCE, L. 1963. Amphitropical relationships in the herbaceous flora of the Pacific Coast of Xorth and South America: a symposium. Introduction and historical review. Quart. Rev. Biol. 38: 109-116.

COOKE, C. M., J R . , and Y. KONDO. 1960. Revision of Tornatellinidae and Achatinellidae (Gastropoda, Pulmonata). B. P. Bishop Mus. Bull. 221: 1-303.

CRUDEN, R. W. 1966. Birds as agents of long-distance dispersal for disjunct plant groups of the temperate western hemisphere. Evolution 20: 517-532.

DEL CORRO, G. 1964. La Gondwania: el antigue continente austral. Mus. Argentina de Ciencias natur. "Bernardino Rivadavia" Pub. no. 12: 1-90.

DRAKE DEL CASTILLO, E. 1893. Flore de la Polynesie Francaise. Paris, G. Masson. FALLA, R. A. 1960. Oceanic birds as dispersal agents. Proc. Roy. Soc. London, B, 152:

655-659.

1967] CARLQUIST: THE BIOTA OF LONG-DISTANCE DISPERSAL, 161

FOSBEEG, F . R. 1948. Derivation of the flora of the Hawaiian Islands. In E. C. Zimmer­man, ed., Insects of Hawaii : Introduction. Univ. Hawaii Press, Honolulu.

GRESSITT, J . L. 1956. Some distribution patterns of Pacific island faunae. Syst. Zool. 5 : 11-47.

GUPPY, H. B. 1906. Observations of a naturalist in the Pacific between 1896 and 1899. Vol. I I . Plant Dispersal. MacMillan & Co., London.

HECKAED, L. R. 1963. The Hydrophyllaceae. Quart. Rev. Biol. 38: 117-123. HILLEBRAND, W. 1888. Flora of the Hawaiian Islands. Williams & Norgate, London. JOHNSTON*, I. M. 1931. The flora of the Revillagigedo Islands. Proc. Calif. Acad. Sci,

4th ser., 20: 9-104. LAMOUREUX, C. H. 1963. The flora and vegetation of Laysan Island. Atoll Res. Bull.

97: 1-14. MACARTHUR, R. H. and E. O. WILSON. 1963. An equilibrium theory of insular zoogeog­

raphy. Evolution 17: 373-387. MAGUIRE, B., ET AL. 1958. The botany of the Guayana Highland. Par t I I I . Mem. N.Y.

Botan. Gard. 10: 1-156. MALONE, C. R., and V. W. PROCTOR. 1965. Dispersal of Marsilea mucronata by water

birds. Am. Fern Journ. 55: 167—170. MOORE, J . W. 1933. New and critical plants from Raiatea. B. P . Bishop Mus. Bull. 102:

1-53. . 1963. Notes on Raiatean flowering plants. B. P. Bishop Mus. Bull. 226: 1-36.

MUNRO, G. C. 1960. Birds of Hawaii. Ed. 2. Ridgeway Press, Rutland, Vermont. ORNDUFF, R. 1963. Experimental studies in two genera of Helenieae (Compositae) :

Blennosperma and Lasthenia. Quart. Rev. Biol. 38: 141—150. RAVEN, P . H. 1963a. A flora of San Clemente Island, California. Aliso 5 : 289-347.

. 1963b. Amphitropical relationships in the floras of North and South America. Quart. Rev. Biol. 38: 151-177.

REINECKE, F . 1898. Die Flora der Samoa-Inseln. LT. Bot. Jahrb. 25: 578-708. RIDLEY, H. N. 1930. The Dispersal of Plants Throughout the World. L. Reeve & Co.,

Ashford, England. S T . J O H N , H., and W. R. PHILIPSON. 1960. List of the flora of Oeno Atoll, Tuamotu

Archipelago, South-Central Pacific Ocean. Trans. Roy. Soc. New Zealand 88: 401-403.

SKOTTSBERG, C. 1922a. Catalogue of Easter Island phanerogams. Nat. Hist. Juan Fer­nandez and Easter I. 2 : 63-84.

. 1922b. The phanerogams of the Juan Fernandez Islands. Nat . Hist. Juan Fernandez and Easter I. 2 : 95-240. . 1927. The vegetation of Easter Island. Nat. Hist. Juan Fernandez and Easter I . 2 : 487-502. . 1928. Pollinationsbiologie und Samenverbreitung auf den Juan Fernandez-Inseln. Nat. Hist. Juan Fernandez and Easter I. 2 : 503-547. . 1937. Die Flora der Desventuradas-Inseln. Acta Horti Gotoburg. 5 : 3-88. . 1951. A supplement to the pteridophytes and phanerogams of Juan Fernandez and Easter Island. Nat. Hist. Juan Fernandez and Easter I. 2 : 763-792. . 1963. Zur Naturgeschichte der Insel San Ambrosio (Islas Desventuradas, Chile). Ark. for Bot., ser. 2, 4 : 465-488.

STEPHENS, S. G. 1958. Factors affecting seed dispersal in Gossypium and their possible evolutionary significance. North Carolina Agr. Exp. Stat. Teclui. Bull 131: 1-32. . 1966. The potentiality for long-range oceanic dispersal of cotton seeds. Amer. Nat. 100: 199-210.

162 BULLETIN OF THE TOBBEY BOTANICAL CLUB | V O L . '.'1

STEWART, A. i ; n i . A botanical survey of the Galapagos Islands. Proc. Calif. Acad. Sci., 4th ser., 1: 7-288. . 1912. Notes on the botany of Cocos [aland. Proc. Calif. Acad. Sci., 4th ser., 1 : 375-404.

THORNE, R. F. 1963. Biotic distribution patterns in the tropical Pacific. In J. L. Gressitt, ed., Pacific Basin Bioge< pp. 311-350). Bishop Museum Press, Hono­lulu.

VAN BALGOOY, M. M. J. 1960. Preliminary plant-geographical analysis of the Pacific. Blumea 10: 385-430.

WHITAKER, T. W., and G. F. CARTER. 1961. A unit? on the longevity of seed of Lagenaria siceraria (Mol.) Standi, after Moating in sea water. Bull. Torrey Botan. Club 88: 104-106.

YUNCKER, T. G. 1959. Plants of Tonga. B. P. Bishop Mus. Bull. 220: 1-283.