Tuatara: Volume 3, Issue 3, November 1950
Evidences of Trans-Oceanic Dispersal of Plants to New Zealand
Evidences of Trans-Oceanic Dispersal of Plants to New Zealand
Purpose and Introduction
The following is entirely preliminary and is aimed at the problem of the origin of the New Zealand flora and vegetation solely for the purpose of drawing attention to different lines of evidence for transoceanic dispersal of plants in order to stimulate more effort to assemble information about this neglected subject. There has been much speculation and some debate among plant geographers and ecologists about the modes of origin of the New Zealand flora and vegetation. From studies of the distribution of the various floras of Pacific regions, particularly the islands, a number of schools of thought regarding the causes of the present distribution have developed. One of these schools has emphasized the long distance migration of plants across bodies of water by such natural agencies as ocean currents, winds, and some animals, particularly birds. Another school has put emphasis on formerly continuous, or nearly continuous land areas over which plants migrated in the past, or parts of the land areas themselves migrated by drifting apart. This land migration school postulates two main types of land area changes (1) that a series of land bridges formerly existed with narrow water gaps between areas that are now far apart, and (2) that there has been some drifting apart of continental areas or fragments of such areas. The first, or land-bridge theory, was formerly held by more persons than at present. The second, or continental-drift theory, is now more popular and advocated by such plant geographers as Good (1947).
The relative merits of these three possible explanations of the present distribution of plants can not be fully discussed in this short paper but some of relative values of evidences for these will be briefly considered. A number of writers, Cain (1944), Oliver (1925), Gordon (1949) and Fleming (1950) have discussed these three theories. In general, the exponents of either the land-bridges or the continentaldrift theories have heavily discounted the effectiveness of dispersal of various elements of the floras across oceans and seas. They have, as a rule, used the patterns of the present and some of the past distribution of species, genera, families and other categories of plants to show how former land areas were arranged or of different size. This mode of postulating former land areas has much less geologic evidence to support it than is desirable and many conclusions based on such hypotheses are open to question until the evidences from geophysical investigations, paleobotany, and paleogeography more conclusively support these theories. Some fragile constructions of logic have resulted from attempts using distribution to prove land bridges and continental drift without more direct evidences.page 88
It is probable that the most satisfactory explanations of present plant distribution will eventually involve both land area changes and trans-oceanic dispersal, and until a better consistency of theories is evolved from the slowly accumulating evidences for each it is very appropriate to examine some of the sources of evidences of transoceanic dispersal. New Zealand is a group of islands particularly well suited by the nature of its flora and its geographical position to studies of the dispersal of plants, their establishment in new localities, and their incorporation into different types of vegetation. The ecology of vegetation, plant communities, needs to be considered in the evaluation of modes of origin of the floras.
General Ecological Considerations
The normal circumstances involved when a particular plant species expands its distribution into new areas not previously or recently occupied by it are usually very definitely associated with the nature of the vegetation this species in invading. There is usually much competition unless the vegetation is of the open type of a primary succession. In general, a region with well established climax or relatively fixed types of successional plant communities tends to keep its flora fixed or consistent as regards the abundant species. Therefore the migration of plants is not the end of the process of establishment and a whole complex of factors concerned with successful establishment (Weaver and Clements, 1938) is involved.
Long geologic periods may provide for species, genera, families and even larger category changes in the flora on a given area. However, for rapid changes of the flora numerous additions from outside the existing vegetational complex are most effective. This has been the case through the agency of man in modern times, particularly in New Zealand (Clark, 1949). During the long geologic past additions from the outside have had a great cumulative effect on many floras and vegetational complexes even though the rate and consistency of long distance dispersal may have been slow and hazardous.
Agents and Plants Involved in Dispersal
Activities of many animals, birds particularly as regards New Zealand because mammals were almost non-existent, help disperse plants. Ridley (1930) in his monograph on the dispersal of plants, Darwin (1845 and 1859) and Guppy (1906 and 1917) considered these agents and gave many examples of plants dispersed by them, showing in numerous cases a very high degree of efficiency with which some plants are dispersed. Much, but not perhaps enough evidence has been accumulated to show how many plants are dispersed by them. Oliver (1925) states the problem thus as regards New Zealand and South America: “The conclusion seems to be inevitable that plants have been carried from New Zealand to South America by agencies comparable page 89 to those which populate the remote islands of Polynesia. Exactly what these agents are may require long and close observance in inhospitable climates.”
Among the plants many parts are dispersed, such as spore, seed, fruit or vegetative part, or even the whole plant. The edible, fleshy fruits are often bird-carried, the pappus fruited species of the Compositae and the small spores of ferns are often wind-carried, and the floating fruits of some shore plants, and the floating seedlings of some mangrove swamp plants are transported by water. Hundreds of special adaptations of plants that aid their dispersal are considered by Ridley (1930) and others, and many of these adaptations indicate that no doubt they have depended on certain agents to increase their range of distribution.
General Evidence of Trans-oceanic Dispersal
The reliable factual data of long distance dispersion are very few because any complete observation of disseminules migrating long distances requires elaborate observations. Davis (1940) experimentally showed how mangrove, Rhizophora mangle, seedlings were efficiently aquivectant for nearly a hundred miles, and Ridley gives some instances of similarly accurate observations. Future research to establish direction, rate and distance of dispersal should be done and probably can be done with more modern techniques such as the airplane, better observations of bird migrations, and more data about hurricanes and other violent winds.
In spite of this lack of direct evidence, however, there are many indirect evidences among which are (1) the observed establishment of floras, and some inferred establishment, on new land areas, especially new islands; (2) the invasion of species into a vegetational complex, which species were not observed previously; and (3) related geological and paleobotanical evidences, some of which may be used to postulate certain floristic changes on islands.
There are also some usually neglected ecological evidences which can be considered. Among these are (1) an abundance of epiphytes in the vegetation; and (2) the occurrence of the same species or similar species near or on the shore areas and far inland on the high mountains with few or no such species in areas between these localities. These two will be stressed in this brief discussion because they apply particularly well to New Zealand.
Floras Established on New or Disturbed Areas
The volcanic islands Krakatau, near Java, and Rangitoto, near Auckland, New Zealand, and some of the islands of the Florida Keys west of Key West (Davis, 1942) furnish examples of the establishment of floras on new islands. These islands are not far way from land masses and there are few if any good records from islands distant from page 90 continents or larger islands. However many useful observations of additions to the floras have been made on the Hawaiian, Galapagos, and Cocos Islands in the Pacific, and the Bermudas and St. Helena in the Atlantic. The few instances of records of new floras on islands near land masses are indicative of the modes of dispersal even though not affording direct evidence of long distance dispersal.
In relation to the long spans of time man has had very little time to accumulate information about migrations of plants. Doubtless many records of flora changes lie buried in accounts of explorations and other documents. Exact information will accumulate slowly and for the present much reliance must be placed on indirect evidence. Later experiments may be undertaken to measure rates and distances of dispersal, especially by winds and water currents.
Invasion of Species into Vegetation
Many of the common plant successions are the results of the successful invasion of new species into the sere communities. The primary successions on coastal and other physiographically young areas follow a course of development that is frequently related to the relative ability of the new species to migrate. At times bare mountain areas have received plants from a great distance by wind, and coastal areas often receive plants that came in by water or wind currents. Secondary successions, or successions after fire or other devastation, are partly the results of new plants invading the areas concerned. Some seeds, or other reproductive parts, of species of the former flora may remain after fire but in many cases the fire followers are species that may have migrated a great distance.
These are but a few of the examples of the ability of plants to migrate into non-climax vegetation. On islands such successional and sub-climax communities are often in greater proportion to the total vegetation than on continents. This seems to be particularly the case in New Zealand where so many of the lowland, non-Nothofagus types of forest vegetation are in a state of flux and very few true climax communities exist.
The invasion of a climax vegetation by new species is much more difficult to accomplish because, in general, the survival and establishment of the new species are relatively less than in successional communities due to the facts that most space is occupied and the species of the climax are so well established that few new species can displace them.
Paleobotanical Evidence, Past Vegetation and Flora
Only a very sketchy outline can be made of this indirect line of evidence because little is known of the past floras of New Zealand and almost nothing of the past vegetation. Most of the interpretations are speculations and inferred from possible analogies with other areas page 91 where palaeobotanical evidence is greater. However, there are some reasonable probabilities. Thus it may be supposed that before the arthipelagic area of the Cretaceous was developed the pre-Cretaceous land areas were the home of many conifers and perhaps other Gymnosperms, some of the species or descendants of same persist as the Taxads and Araucarians of the present. The early Angiosperms invaded this emerging and changing Cretaceous archipelago either by crossing seas, or by some connections which existed with other land areas. But soon after the Cretaceous, during the Tertiary, the New Zealand land areas were again small and land bridges, if any, became increasingly far apart until possibly the high mountain period of the Pliocene. Some of the possible Cretaceous and post-Cretaceous Pacific connections of land masses have been outlined by Benson (1923), and Oliver (1950) Fleming (1950) and most writers agree that during the Tertiary there was a breaking down of most of the Cretaceous connections until the waters between New Zealand and other large islands or continents were about as wide as at present. The modern Angiosperm flora came to or developed on these islands during upper-Cretaceous and Tertiary times.
The affinities of many genera and families lead to the assumption that some plants migrated from Malaya, some from Australia, some from other Pacific island regions, and possibly some from the Antarctic areas, these latter being assumed to have been larger and warmer than at present.
Perhaps early in the Cretaceous the genus Nothofagus was established and some of the climax beech forests developed. These forests probably have varied in size and composition since then but there is little doubt that the main elements, species of Nothofagus, have persisted and because of the climax nature of the community few species of modern floras have invaded these forests.
The other dicotyledonous forest trees and herbs and grasses now constituting the flora of the mixed Podocarp and Dicotylous forests, and the various grasslands or scrub vegetation probably did not reach their maximum number until mid-Tertiary times and maybe now are increasing in number by both additions from the outside and the development of new species on these islands, endemism and hybridization now being relatively abundant.
Cockayne (1928) and others have emphasized the endemic nature of the New Zealand flora and the extensive hybridization. These two conditions may be interpreted to indicate a state of flux in the vegetation types as well as the flora, perhaps because of the relatively late arrival of additions to the flora, intensive competition, selection, and other factors. Much of the endemism is of the initial type, rather than the relic, which indicates that some plants are migrating into new habitats, physical and social. A number of the new flora developed since the Tertiary began, and probably after mid-Tertiary times, seem page 92 to have affinities with Australia. This Australian element has a number of genera and a few species that may have recently migrated across the Tasman Sea by air and birds. Among these are possibly Olearia, Senecio, Epacris, and Cyathodes.
Land connections of the past may explain some of the Tertiary additions to the flora of New Zealand but care has to be exercised in making such interpretations from maps of the present distribution of species, genera, and families, or from the scanty fossil members of similar groups. A notable case in point is the genus Nothofagus which has been assigned a subantarctic origin because of its then known distribution. However Nothofagus forests have been observed recently in New Guinea (Archbold, 1942, and Landon, 1947) within a few degrees latitude of the equator, and on this island it is not probably related to any antarctic land mass. It is now just as plausible to consider Nothofagus a genus of Malayan origin as antarctic origin.
Migrations of plants across wide oceans may not have occurred consistently during the long geologic periods but it is extremely likely that even more intensive and extensive migration occurred during certain times. With the advent of flying animals, possibly reptiles and certainly birds, which migrate in large numbers animals may have moved countless disseminules in nearly all directions. Birds are now very effective in carrying fleshy fruits and hard indigestible seeds or fruits, or fruits or seeds that become attached to them. Some New Zealand plants now readily carried by birds are Rhopalostylis, eaten by parrots, Astelia and Hedycarya, the latter eaten by pigeons, and probably many species of Coprosma and Podocarpus. Perhaps over 150 of the present species are consumed by birds and transported.
Ocean currents of the past may have been even more effective than at present. The present ones are fairly well known (Davies, 1947) and from their direction of movement some correlations with dispersal of plants is sufficiently exact to indicate transportation by these currents.
In general insular areas are converging points for many surface ocean currents, some air currents, and migratory routes of birds. It seems probable that islands during a long period of time have therefore received more plants from afar than would continental areas of similar size and location. Even a small amount of observed migration of plants during the present has significance because the interval of observation has been very small compared to geologic ages and the long past has multiplied enormously the effectiveness of the occasional migration.
Summary of Three Lines of Evidence
These three lines of evidence are greater than presented here in so brief a paper, and they are not enlarged upon because they have been more fully considered in other publications, especially about other regions. Each evidence is significant but the concerted value of all page 93 three has more weight because a generalized estimate of trans-oceanic dispersal usually involves all three lines of evidence, and the two evidences now to be considered. Students may gain much more information about particular modes of dispersal by reference to Ridley (1930) and Guppy (1907) who have considered the subject much more exhaustively. Many more careful observations and some experimental data will be needed to finally substantiate the significant role of trans-oceanic dispersal. Such plants as the coconut, Cocos nucifera might well be studied and experimental migration tests made. Oceantographers, ornithologists and botanists with such training can do much to contribute more information.
Additional Ecological Evidence of Dispersal
The above lines of evidence have been based to some extent on principles of plant ecology, the two following points of evidence are particularly related to ecological concepts and observations. They involve mainly (1) the significance of epiphytes as these plants imply certain modes of dispersal, and (2) the singular manner of distribution of certain genera, and species either along the coasts or on the high mountains, or at both places. In both instances disseminules are “precipitated” from the air; epiphytes often being perched in trees because the winds or some bird deposited them there, and coastal or high mountain plants occurring where they are mainly because winds are frequently reduced in speed or otherwise changed by coastal features, or the mass of the high mountains. Thus the two agents, birds and winds, are in a sense precipitating from the air many disseminules onto the forest canopy, or onto the relatively unpopulated areas of the coasts or high mountains. The fruit eating birds of the past and present have helped lodge disseminules above the ground. The winds have acted to carry many light weight disseminules to the coasts and to the high mountains. In New Zealand at present the most constant winds imoinge upon much of the western coasts from the northwest thus breaking their force against the high mountains of the Southern Alps and seasonally the Fohn wind blows off these mountains onto the Otago and Canterbury plains.
Evidence from the Epiphytes
Fortunately Oliver (1930) has very thoroughly considered the kinds and nature of the epiphytes of New Zealand, His description of them also stresses their possible mode of origin as regards affinities to species of other land areas. He did, in some cases, particularly stress their means of dispersal by winds and birds, which may be considered the sole means of dispersal in numerous instances. Classifying the epiphytes as to modes of dispersal he noted: 35 species of Pteridophytes dispersed by minute spores, winds; 7 species of orchids by minute seeds, winds; one common Senecio species by pappus fruits, winds; 2 species page 94 of Pittosporum, by viscid coated seeds, birds mostly; and 5 species with fleshy fruits, birds. These are but the consistently epiphytic species. If the occasionally epiphytic or ephemeral epiphytic species are also considered the list is much longer and includes some of the important forest trees as Dacrydium cupressinum, Podocarpus spp. and many species of Coprosma with fleshy fruits. Over 10 other species of shrubs and trees occasionally epiphytic have fleshy fruits eaten by birds. Droppings from birds often contain seeds not digested and many of these may sprout and grow upon limbs and trunks of trees. Windcarried seeds of tree species as the Metrosideros species are similarly deposited on limbs and trunks and ephemeral epiphytes develop from them.
The epiphytes of all types, including the lichens, mosses, and other non-vascular plants are very numerous in some of the New Zealand forests. In terms of number of species these epiphytes may not be many more than found on some other areas of similar size and conditions elsewhere, but in terms of number of individuals they are probably more numerous than in many other similar areas. This abundance of perched plants, many of which were precipitated from air currents or were dropped by birds signifies an efficient mode of dispersal by these agents. Cockayne (1928) and others have related the abundance of the epiphytes in general to the “rain-forest” condition and the subtropical nature of the vegetation. There is little doubt that the abundant precipitation over large areas and the high humidity account for some of the epiphytes, but there is some doubt as to the subtropical nature of the forests. The relatively great abundance of epiphytes may be a reflection of the ease of dispersal of the species involved, their ability to survive and other ecological considerations, as well as the climatic conditions.
Possibly the habit of epiphytism is a partial result of their mode of dispersal. Some species normally not epiphytic, or of genera that are non-epiphytic, upon migrating into a dense type of vegetation depend upon their only survival by their ability to remain perched above the soil and away from the intense competition of the forest floor. The genus Pittosporum illustrates this possible mode of development of epiphytism. The genus is distributed over Malaya and, with the exception of 2 species that are epiphytic in New Zealand, all so far found are non-epiphytic. Griselinia may be another example. It is a genus of 8 known species with all but 2 species in Chile, and some in Chile and in New Zealand are epiphytic. Similarly the genus Astelia furnishes some evidence of trans-oceanic dispersal. It is represented by about 19 species, some occur in dense lowland forests and some as subalpine mat plants. Significantly, species in New Zealand are both epiphytic and terrestrial. Epiphytes of this genus also occur in Fiji, Samoa, Society Islands and New Caledonia. Winds and birds may have dispersed these widely. It is interesting to note that the most fleshy fruited page 95 species of Astelia are the most widely distributed indicating that birds may have aided in their dispersal.
Species and genera normally epiphytic wherever found are well represented in the flora and many are important vegetational components, particularly the many ferns and a few orchids. These autoepiphytes do not suggest special adaptability to the New Zealand conditions but their great abundance may indicate that they were easily dispersed, and precipitated.
The numerous facultative or ephemeral epiphytes such as Podocarpus, Dacrydium, Olea, Metrosideros and Coprosma are suggestive by the adaptability of their young that they may be readily dispersed by winds and birds. Metrosideros with many species in New Zealand is of very wide distribution, as far west as South Africa, but only in New Zealand are the species ephemeral epiphytes or climbers. The climbing or vine habit is in some ways much like the epiphytic habit and the abundance of this growth form of plants is also a notable feature of many New Zealand forests.
Oliver concludes from his study of epiphytes thus: “The evidence from geographical distribution thus points to the derivation of most of the epiphytes of New Zealand from the element of Malayan tropical facies in the flora. The method of dispersal is important, for obviously for the dissemination of epiphytes from tree to tree it is necessary for the seeds or fruits to be carried to suitable spots by such agencies as birds and air currents.”
Coastal and Mountain Distribution an Evidence
As pointed out above the winds often converge upon island areas and these are frequently in two main strata, the low winds and the high winds. Both these may be in the same direction or may be in different directions. They normally are reduced in speed by obstructions provided by the physiographic features of the land mass. In many cases the winds are reduced at the coasts; in some other cases, especially the high elevation winds, the high mountains act as a barrier to them. At both localities there probably would be more deposition or precipitation of objects from the wind currents than at places between the coasts and the high mountains. Birds are affected by winds and may also drop more disseminules at these localities.
If, therefore, there are species similar or the same along the coasts and in the mountains there presence might indicate these modes of origin of the species. Frequency of such species is difficult to determine from the floristic literature and many more observations are needed to get enough data to give proper weight to this possible line of evidence for trans-oceanic dispersal. However, some data has been accumulated and a few samples of it will be considered.
There are a number of genera of the Compositae with pappus fruits which are distributed over the two main islands and some of the islands off the New Zealand group in the manner we are considering. Olearia, as an example has many endemic species such as O. page 96 semidentata, O. traversii, O. allomii and O. traillii on the small offshore islands and along the coasts of the main islands are species such as O. virgata, O. paniculata, O. operina and O. angustifolia. In the mountains, often on the high subalpine parts, are species of Olearia such as O. macrondonta, O. avicenniaefolia, O. moschata and O. ilicifolia. Some species of Olearia, as O. colensoi, occur at the coasts and in the mountains.
Another genus with similar pappus fruits and coastal and mountain distribution is Celmisia. On islands or the coastal areas are C. rigida, C. major, C. lindsayi, C. vericosa and C. graminifolia, and in the subalpine areas are C. alpina, C. gracilenta, C. spectabilis, C. coriacea, C. argentea, C. lyalii and many others.
The genus Senecio has S. lyallii, S. bellidioides and S. eleagnifolius in the mountains and S. rotundifolius, S. glaucophyllus and S. greyii on islands and along the coasts. Other pappus fruited Com-positae genera as Helichrysum and Gnaphalium are similarly coastal and montane. Some genera, notably Haastia, are entirely montane and Raoulia is chiefly montane but has some species along the coasts.
A number of genera of the Epacridaceae, particularly Dracophyllum are common near the sea and on the high mountains. Of Dracophyllum the species D. rosmarinifolium and D. menziesii are subalpine and scoparium is typically coastal, but D. longifolium is both coastal and subalpine. Some of the other species spread over areas from the coast into the mountains.
Other families have genera of similar distribution, as Ourisia of the Scrophulariaceae, which are herbs with small seeds in capsules. The species O. colensoi and O. sessilifolia are high montane and O. macrocarpa and O. macrophylla are distributed from the coastal sounds to the mountains.
Astelia, a genus of the Liliaceae has berry fruits and may be transported by birds. Some of the coastal species of this genus are A. banksii and A. sublata, and some of the high mountain species are A. linearis, A. petriei and A. cockaynei.
Many more genera could be cited to illustrate this coastal and high mountain distribution which may reflect the mode of dispersal and indicate something of the effectiveness of such dispersal. Some species with members both along the coasts and on the high mountains similarly, and perhaps more effectively, also indicate the wind and bird types of dispersal. A few have been cited above, and some more are Raoulia australis, Celmisia cordata, Olearia avicenniaefolia, Senecio bellidioides, and some grasses as Danthonia semiannularis and Poa caespitosa.
Persons more familiar with New Zealand vegetation and flora can no doubt cite many more examples of this disjunct type of distribution than given here, and possibly some of the species cited are not good examples as the writer has relied mainly upon distribution notes from Cheeseman (1925) and a few observations and conversations.
Relative values of different modes of origin of the New Zealand flora and types of vegetation are briefly discussed and the possibly important mode of dispersal across oceans and seas is emphasized. Among the different lines of evidence supporting the importance of this trans-oceanic dispersal of plants the evidences that may be obtained from the abundant epiphytic flora and the ecology of the condition of epiphytism are stressed. Similarly the disjunct distribution of species of some genera and of some species are considered as significant evidence of dispersal, possibly across oceans, the disjunct distribution being near the coasts and on the high mountains.
Archbold, R. et al., 1942. Summary of the 1938-39 New Guinea Expedition. Amer. Mus. Nat. His. Bull. 79, Art. 3, p. 197-288.
Benson, W. N., 1923. Paleozoic and Mesozoic Seas in Australasia. T.N.Z. Inst. 54, p. 1-62.
Cain, S. A., 1944. Foundations of Plant Geography, New York, Harper and Brothers.
Camp, W. H., 1947. Distribution patterns of modern plants and problems of ancient dispersals. Ecol. Mon. 17, p. 123-126, 159-183.
Campbell, D. H., 1919. The derivation of the flora of Hawaii. Leland Stanford Jr. Univ. Pub. Ser. 1919.
Cheeseman, T. F., 1919. The vascular flora of the Macquarie Islands. Rep. Aust. Ant. Exp. 7, p. 1-63.
Clark, A. H., 1949. The invasion of New Zealand by people plants and animals. Rutgers Univ. Press. Brunswich, N. J.
Cockayne, L., 1928. The vegetation of New Zealand. Leipzig.
Darwin, C., 1845. Journal of researches—during the voyage—of H.M.S. Beagle. London.
Davis, J. H., 1940. The ecology and geologic role of mangroves in Florida. Carnegie Inst. Wash. Pub. No. 517, p. 303-412.
1942. The ecology of the vegetation and topography of the Sand Keys of Florida. Carnegie Inst. Wash. Pub. No. 524, p. 113-195.
Fleming, C. A., 1949. The Geological History of New Zealand. Tuatara 2, p. 72-90.
Good, R., 1947. The geography of flowering plants. Longmans Green, London.
Gordon, H. D., 1949. The problem of sub-antarctic plant distribution. Prs. Address. Bot. Sec. Austr. and N.Z., A.A.S.
Guppy, H. B., 1906. Observations of a naturalist in the Pacific, II Plant dispersal. London.
1907. Plants, seeds and currents in the West Indies and Azores. London.
Landon, L. M., 1947. The comparative morphology of Fagaceae, I. the genus Nothofagus. Bot. Gaz. 108, p. 350-371.
Oliver, W. R. B., 1925. Biogeographical relations of the New Zealand region. Jour. Linn. Soc. 47, p. 99-140.
1930. New Zealand epiphytes. Jour. Ecol. 18, p. 1-50.
1950. The fossil flora of New Zealand. Tuatara 3, p. 1-11.
Ridley, H. N., 1930. The dispersal of plants throughout the world. London, L. Reeve and Co.
Setchell, W. A., 1935. Pacific insular floras and Pacific paleogeography. Amer. Naturalist 69, p. 289-310.
Skottsberg, C. 1925. Juan Fernandez and Hawaii, a phytogeographical discussion. Bishop Museum Bull. 16, Honolulu, Hawaii.