CHAPTER XXI

THE ERAS OF THE FLOWERING PLANTS

THE AGE OF COMPOSITÆ.

The islands of the tropical Pacific as the homes of new genera and new species.—The significance of a large endemic element.—Synopsis of the eras.—The era of endemic genera.—The endemic genera of Compositæ.—Their affinities and mode of dispersal.—The mystery of the suspension of the dispersing agencies.—Mr. Bentham’s views.—The remnant of an ancient Composite flora in the tropical Pacific.—The dispersion of the Compositæ antedates the emergence of the island-groups of the Fijian region at the close of the Tertiary period.—Summary.

_The Endemism of the Pacific Island Floras._

AS far as the production of new species is concerned, the Hawaiian group presents the same contrast with the Fijian and Tahitian groups in respect of the flowering plants that it does as regards the ferns and lycopods. The proportion of endemic species, after excluding all introduced plants, is in Hawaii 80 per cent., in Fiji about 50 per cent., and in Tahiti 35 per cent. (see Table A). The same contrast is also displayed in the number of peculiar genera. In Hawaii there are, according to Dr. Hillebrand, 37 or 38, and in Fiji Dr. Seemann discovered 16; whilst, as we learn from Drake del Castillo, there are only 3 or 4 in the Tahitian Islands. (As will be pointed out later on, these numbers for Fiji and Hawaii have to be reduced, but the general inference to be drawn from them is not materially affected; see Table B.)

But if we look at the accompanying table (Table B) we notice that the flora of Hawaii is sharply contrasted with those of Fiji and Tahiti not only in the large proportion of endemic genera, but also in the large number of non-endemic genera with peculiar species, and in the small proportion of genera possessing no peculiar species. There is an endemic element of greater or less degree in about 70 per cent. of the Hawaiian genera, whilst in Fiji only about 53 per cent. and in Tahiti as few as 34 per cent. of the genera contain to a varying extent peculiar species. Another feature brought out in this table is the relative poverty of genera in the Hawaiian Islands. Fiji, though about the same size as Hawaii, contains nearly half as many genera again, whilst the islands of the Tahitian region, which in the aggregate amount to only one-third or one-fourth of the area of the islands of Hawaii, possess nearly as many genera.

TABLE A (FLOWERING PLANTS).

_Proportions of Endemic Species in the Hawaiian, Fijian, and Tahitian floras, with those for Samoa, Tonga, and Rarotonga added._

+---------+------------------+----------------+----------------+ | Groups. |Number of species.| Number of | Percentage of | | | |endemic species.|endemic species.| +---------+------------------+----------------+----------------+ | Hawaii | 686 | 546 | 80 | | | | | | | Fiji | {S. 617 | {288 | {47 | | | {H.1086 | {620 | {57 | | | | | | | Tahiti | 315 | 112 | 35 | +---------+------------------+----------------+----------------+ | Samoa | 326 | 110 | 34 | | | | | | | Tonga | 285 | 17 | 6 | | | | | | |Rarotonga| 140 | 17 | 12 | | Island | | | | +---------+------------------+----------------+----------------+

_Remarks._—The materials for this table have been obtained from the works of Hillebrand for Hawaii, Seemann and Horne for Fiji, Drake del Castillo for Tahiti, Reinecke for Samoa, Hemsley and Burkill for Tonga, and Cheeseman for Rarotonga. The two estimates for Fiji are marked S. for Seemann and H. for Horne, the last being a rough preliminary computation made by Horne himself.

The results given are only to be considered as approximations liable to emendation, but as regards the proportion of endemic species in the several groups they no doubt illustrate fairly well the relative degree of endemism in the various archipelagoes. The results for Samoa, Tonga, and Rarotonga are merely added in order to enable a comparison to be made with sub-groups of a region and with solitary islands, the Hawaiian, Fijian, and Tahitian groups being regarded as the three principal centres of plant-life in the open Pacific.

All plants introduced by the aborigines and the white man are excluded. In so doing, I have mainly followed Seemann, a safe guide in all matters relating to weeds and to cultivated plants. The flora of a Pacific island thus treated undergoes serious diminution in its extent. In the case of the Rarotonga flora, for example, which according to Cheeseman includes about 260 flowering plants, the number of truly indigenous plants, in the sense here implied, is only 140. Though this is an extreme case, it will serve to illustrate the principle here followed.

TABLE B (FLOWERING PLANTS).

_Comparison of the Hawaiian, Fijian, and Tahitian genera._ (_All genera containing introduced plants entirely are excluded._)

+----------+------------------------------------------+---------+-----------+ | | Non-endemic genera. | | | | +------------+---------------+-------------+ | | | Group. | No endemic | Some species | All species | Endemic | Total. | | | species. | endemic, some | endemic. | genera. | | | | | not. | | | | +----------+------------+---------------+-------------+---------+-----------+ |Hawaii | 70(31)| 30(13)| 95(43)| 28(13)| 223(100)| | | | | | | | |Fiji {| S. 150(47)| S. 74(23)| S. 87(27)| S. 10(3)|S. 321(100)| | {| H. 162(47)| H. 80(23)| H. 94(27)| H. 10(3)|H. 346(100)| | | | | | | | |Tahiti | 125(66)| 21(11)| 40(21)| 4(2)| 190(100)| |(Eastern | | | | | | |Polynesia)| | | | | | +----------+------------+---------------+-------------+---------+-----------+

_Remarks._—The figures in brackets are percentages. S. = Seemann, H. = Horne and Seemann.

In the construction of this table, Hillebrand, Seemann, and Drake del Castillo have been mainly followed, except with regard to the endemic genera for Hawaii and Fiji. In this respect the _Index Kewensis_ has been largely consulted as well as Engler’s publications, as indicated in the text. Hillebrand’s total of nearly forty Hawaiian peculiar genera and Seemann’s total of sixteen for Fiji have thus been considerably reduced. The two results given for Fiji are those of Seemann alone and with Horne superadded. Horne discovered, according to Hemsley, no new genera, but several genera from outside regions were added to the Fijian flora. Taking them as twenty-five (two-thirds of his own computation), I have apportioned them as in Seemann’s results. The Tahitian region here includes Eastern Polynesia.

It is necessary before proceeding further to obtain a correct idea of the significance of a large endemic element in the phanerogamic flora of a Pacific archipelago. We have therefore at the outset to inquire whether it is indicative of isolation or of antiquity. If the number of peculiar genera is to be regarded as the test of the relative antiquity of different Pacific floras and, by implication, of the islands to which they belong, these three groups, as shown in Table B, would arrange themselves in the following order, namely, Hawaii, Fiji, Tahiti. This test might be reliable if the several groups were in the same condition of isolation. Since, however, as we have previously seen, the Fijian Islands still enjoy a fairly free communication with the islands westward, whilst the Hawaiian group is largely cut off, it is apparent that the tendency to generic differentiation in Fiji might have been often swamped by immigration, and that Fiji with its much smaller number of endemic genera may even be older than Hawaii. This objection does not apply quite as forcibly to a comparison between Hawaii and Tahiti, yet for reasons before given it may be regarded as sufficient to negative any inferences concerned with relative antiquity.

On account, therefore, of the great differences in the degree of isolation of these three groups, we cannot be guided in our estimation of the relative antiquity of their floras by their number of peculiar genera. With the evidence at our disposal we are compelled to accept the view, which indeed a single glance at a map would suggest, that the number or proportion of endemic genera is to be connected with the degree of isolation. Whether a parallelism can be traced in the original stocking of these groups with their earliest flowering-plants is a matter that can only be elucidated by a further analysis of the peculiar genera.

SYNOPSIS OF THE ERAS OF THE FLOWERING PLANTS IN THE TROPICAL PACIFIC.

A. _The Era of the Endemic Genera._—Mostly American in their affinities. Represented particularly by Compositæ and Lobeliaceæ.

B. _The Era of Non-Endemic Genera._

(1) The mountain genera, either cosmopolitan in temperate latitudes or derived from the New Zealand or the Antarctic flora. Mostly represented in Hawaii.

(2) The genera forming the low-level flora of Hawaii below 4,000 or 5,000 feet and composing almost the entire floras of the Fijian and Tahitian regions. Predominantly Indo-Malayan.

(_a_) The age of general dispersal over the tropical Pacific, the genera with only peculiar species being first treated, and afterwards those possessing a non-endemic element.

(_b_) The age of local dispersal over the tropical Pacific.

THE FIRST ERA OF THE FLOWERING PLANTS, BEING THE AGE OF THE ENDEMIC GENERA.

With the above preliminary remarks I pass on to the next stage in the history of the stocking of these islands with their plants. The age of the ferns and lycopods is left behind, and it is assumed that the next era is mainly indicated by those genera of phanerogams that are now peculiar to their respective groups. In this connection by far the most interesting of the three regions, the Hawaiian, the Tahitian or East Polynesian, and the Fijian, is that of Hawaii, which, as before observed, is distinguished from the groups of the Fijian and Tahitian regions, or, in other words, from all the oceanic archipelagoes of the tropical Pacific, by its large number of endemic genera.

Peculiar genera of shrubby and arborescent Compositæ and of arborescent Lobeliaceæ form the most striking characteristics of the endemic genera, and therefore of the ancient flora of Hawaii. It is in this connection of singular interest to remark that of the three endemic genera of the Tahitian flora one is an arborescent genus of the Compositæ, and the other two are shrubby genera of the Lobeliaceæ. There are, therefore, indications here of an ancient insular flora of the Pacific, characterised mainly by the prevalence of Compositæ and Lobeliaceæ. It is, however, remarkable that not only are no endemic genera of these orders known from Fiji or from the adjacent groups of Samoa and Tonga, but that the Lobeliaceæ are not represented at all, whilst amongst the Fijian Compositæ, with the exception of Lagenophora, the genera display no endemic element as far as the data at my disposal indicate.

The problem we are brought face to face with is clearly stated by Mr. Hemsley in the _Introduction to the Botany of the Challenger Expedition_ (p. 68). “In Polynesia as elsewhere,” he remarks, “the Compositæ more particularly are perplexing to the botanical geographer, for although they have their greatest affinities in America, as well as the sub-arboreous Lobeliaceæ, so numerous in the Sandwich Islands, yet the bulk of the vegetation seems to have been derived from the Australo-Asiatic region.”

In attempting to approach this problem I do so from the standpoint of dispersal. There are so many intricate questions bound up with the systematic position of these genera that in dealing with them the student of plant-distribution would require the capacities and opportunities of the eminent botanist who dealt with the distribution of ten thousand species of Compositæ. On such ground, therefore, and only under the guidance of others, I will lightly tread.

THE ENDEMIC GENERA OF COMPOSITÆ.

On account of their endemic character the peculiar genera of Compositæ are regarded as belonging to the oldest era of the flowering plants of the island-groups lying in the tropical latitudes of the open Pacific. This is the view of Bentham, but it is, of course, the opinion that most botanists would arrive at with the facts before them. With the exception of the solitary Tahitian genus Fitchia, they are all restricted to the Hawaiian Islands, and nearly all are either shrubby or arborescent, the greatest height of 25 to 30 feet being attained in the Tahitian genus and in Hesperomannia of Hawaii.

Nine Hawaiian genera are included in this era, though, strictly speaking, we ought only to concern ourselves with the six genera, Remya, Argyroxiphium, Wilkesia, Dubautia, Raillardia, and Hesperomannia, since the other three, Tetramolopium, Lipochæta, and Campylotheca, are only on the borderland of generic distinction. It is, however, necessary that we should include these three genera in our treatment of the Hawaiian endemic genera, more especially because they appear to have been the last arrivals of the early Compositæ. They still display, as shown below, a very suggestive connection with the land of their birth, a circumstance that is of much importance in finally determining the source of the other strictly endemic genera, where the links with their original homes have been in most cases largely severed.

It would, however, be quite out of place here to enter into any details into the affinities of these Hawaiian genera of Compositæ, and I will limit myself here to such general conclusions as may be derived from the pages of Bentham, Hillebrand, Hemsley, and other writers, and such as are in accordance with the facts of distribution given in the _Index Kewensis_. Most ancient of all are the genera Remya, Argyroxiphium, Wilkesia, and Hesperomannia, which, although belonging to tribes that only occur on the American continent, as in the Mexican region, stand quite isolated, and, as Dr. Hillebrand remarks, probably belong to the oldest denizens of the Hawaiian Islands. It is noteworthy that these four ancient genera only contain two species apiece, a circumstance that favours their priority in point of age.

The American affinities, however, are not always of the character that we might have expected. Thus, it was remarked by Mr. Bentham that although the tribe Mutisiaceæ attains a great development in South America, and especially in Chile, its only representative in the Pacific islands is the very rare arboreous Hesperomannia of Hawaii.

Rather less isolated in character, and we would presume therefore of somewhat less antiquity, are the two closely allied genera of Raillardia and Dubautia, which have a close relative in Raillardella of the Sierra Nevada in California. Then we come to the three genera, Tetramolopium, Lipochæta, and Campylotheca, that, being still in touch with the world outside, may be regarded as the latest arrivals of the early genera of the Compositæ. Tetramolopium, concerning which botanists were unable to agree, would seem, according to the _Index Kewensis_, to possess Mexican and Ecuadorian as well as Hawaiian species. Lipochæta, nearly related to other American genera, contains a dozen species, of which eleven are found only in Hawaii, whilst the twelfth occurs, according to the _Index Kewensis_, in California, and, according to Dr. Hillebrand, in the Galapagos group. Of the generic value of Campylotheca there seems a doubt, and its distinctness is scarcely recognised in the _Index Kewensis_. It is, however, closely allied to Coreopsis, an American genus represented, according to Drake del Castillo, in the Marquesas.

In the Tahitian region, that is to say in Eastern Polynesia, the genus Fitchia alone belongs to the early age of the Compositæ, so characteristic of Hawaii. Indications of the former widespread range of the genus over this region of the South Pacific are afforded by its being now represented by two species in Tahiti and by one species in Rarotonga, localities nearly 700 miles apart. It was thus regarded by Bentham, who saw in it a solitary remnant of the ancient South Pacific flora. Like the Hawaiian genera, as shown below, it is often restricted to the higher levels. Botanists differ about its affinities, and a discussion of the subject will be found on pages 20 and 66 of the _Introduction to the Botany of the Challenger Expedition_.

The restriction of these ancient genera of the Polynesian Compositæ to the upland regions is of some interest. “The preponderance of Compositæ among the high-level plants obtains almost throughout the world.” This observation was made by Mr. Hemsley in connection with the flora of the highlands of Tibet (_Journ. Linn. Soc. Bot._ vol. 35, 1902), where the Compositæ constitute about 19 per cent. of the flowering plants; and I may remark in passing that, according to Mr. Ball, one of the most conspicuous elements in point of frequency in the higher flora of the Great Atlas is presented by the Compositæ which make up between 12 and 13 per cent. of the whole flora (Hooker and Ball’s _Marocco and the Great Atlas_). This feature of alpine floras is brought into great prominence in Schimper’s recent book on Plant Geography.

Some of the most lasting reminiscences that the naturalist will bear away with him from the highlands of Hawaii are connected with the Compositæ. Those who have ascended the mountains of Mauna Kea and Mauna Loa, will remember that amongst the last plants occurring above the forest zone, and scattered about on the ancient lava fields at elevations exceeding 10,000 feet above the sea, are species of Raillardia and the beautiful “Ahinahina” (Argyroxiphium). It is, however, in the open, scantily wooded region, elevated 6,000 to 9,000 feet, and lying between the true forest zone below and the bare lava slopes above, that the shrubby and arborescent Compositæ of the large island of Hawaii are most at home. Such regions, as Hillebrand well describes (p. xxiv), are characterised by stunted trees, chiefly Sophora, Cyathodes, Myoporum, and others, associated with arborescent Raillardiæ of the order of Compositæ. Between them luxuriate other shrubby Compositæ of the genera Raillardia, Dubautia, Campylotheca, and Artemisia, together with Strawberries, Raspberries, and species of Vaccinium.

Botanists have not given us much account of the associates of the interesting genus Fitchia on the uplands of Tahiti. We learn, however, from Nadeaud that in his time these Composite trees and shrubs were spread over the higher region of the island of Tahiti above 800 and 1,000 metres. Cheeseman, to whom we are indebted for the discovery and the description of the Rarotongan species, tells us that this tree, which attains a height of 25 feet in the sheltered valleys, and is much dwarfed on the exposed ridges and hill-tops, often forms the greater part of the forest above 500 feet, and reaches the highest peaks of the island (2,250 feet).

In discussing the probable mode of dispersal of these early Composite plants of the Pacific we shall be treading on somewhat debatable ground. We will, however, point out that the mere possession of structures that could be utilised for dispersal of the seeds is not the only important question here involved. If we could demonstrate that all these genera possess exceptional capacities for distribution over the ocean, we should prove too much, since the process has been in the main suspended for ages. If, on the other side, it could be shown that their fruits are not at all suited for such dispersal, we should prove too little, since the ancestors of these genera must have been transported to these islands in some fashion or other. This clearly indicates that other important factors have also come into play in determining the distribution of the early Compositæ of the Pacific islands.

It was long ago pointed out by De Candolle that the possession of a pappus does not, as a rule, increase the area of a Composite plant, although as regards hooks and barbed appendages, such as occur in Bidens, the greater areas of the plants thus provided may be, as he thought, in some measure explained. Even in respect to hooks and barbs it would be easy to point to cases where, as Bentham remarks, unusual powers of adherence are by no means indicative of wide dispersal in all cases. In any event it will be also incumbent on us to explain why these genera no longer possess facilities for distribution. This suspension of the means of dispersal is not, however, peculiar to the age of the endemic genera of the Pacific islands. It is a character but in a less degree of the succeeding age, the age of genera found outside the group, but represented within it by endemic species; and from this we may suspect that we have had in operation in the Pacific an influence, far-reaching both in time and space, to which the agencies of dispersal have been compelled to adapt themselves, an influence which has acted as a distributor of the distributing agencies.

Coming to the fitness for dispersal of the achenes of the early Composite genera of the Pacific islands, it will be assumed that they have been, as a general rule, transported in birds’ plumage. The fruits are usually 2·5 to 12 millimetres (1/16 to 1/2 inch) in length, and are provided either with a pappus of soft or stiff bristles, or with awns or teeth, but these appendages vary much in size in the different genera and in different species of the same genus. The instance of Lipochæta is especially significant as indicating the alterations which the appendages of the achene may have undergone in the cases of other genera. With most species there are usually two or three teeth or short awns, but in some species these are obsolete, and in others they are long and stout.

Bearing these facts in mind we should hesitate to rely too much on the present condition of the achenes in the other genera as an indication of the fitness for dispersal of the fruits of their ancestors. In one genus, Campylotheca, which may be regarded as among the youngest of the genera, the achenes are provided with barbed or hooked awns which cause them to adhere as tenaciously to one’s clothes as in the case of those of Bidens, an allied genus. In Fitchia, the Tahitian genus, which may be looked upon as one of the oldest of the Pacific genera of Compositæ, the achene is furnished with two long awns or setæ, which, as Drake del Castillo observes, recall those of Bidens. The achenes of the other Hawaiian genera, as regards their fitness for dispersal in plumage, may be said to give less definite indications. In some, as in Dubautia and Raillardia, there is a typical pappus of ten to twenty long hair-like bristles. In others again, as in Wilkesia and Argyroxiphium, the pappus is much reduced, and in some species of Lipochæta it is, as above remarked, quite obsolete.

The chances of the achenes of the parent plants having in some cases been originally transported to the islands in the plumage of birds would be increased by a bird making its nest of the plant-materials or amongst the plants themselves, or by its pecking at the fruit-heads. In our own time different species of the grouse family on the slopes of the Californian and Columbian mountains make their nests on the ground under the shade of Artemisia bushes and find a portion of their sustenance in their fruits. Artemisias also form one of the features of the vegetation of the Hawaiian uplands; but since they present only specific differentiation they are referred to a later era. Yet it will be on the slopes of the Rocky Mountains and of the Californian Sierra Nevada, amongst the “sage-brush” and the grouse, that we may have to stand when we look in thought across the Pacific towards far distant Hawaii and ask ourselves whence came its tree-like Raillardias, its shrubby Dubautias, its tall Wilkesias, and the silvery Ahinahinas (Argyroxiphium).

It is possible that in some genera the achenes have, or had, a means of adhering to plumage through a “sticky” secretion, such as is sometimes found with Lagenophora, an Hawaiian genus of the next era, and also with the weed-plant Adenostemma viscosum; but this is a point that has not yet been investigated. Nor can we altogether exclude the chance of the achenes having in some cases been transported unharmed to Hawaii in a bird’s stomach. The possibility of this has been above implied in the case of Artemisia; and it is pointed out in Chapter XXXIII. that pigeons in Hawaii feed sometimes on the achenes of Compositæ. The Hawaiian goose (Bernicla sandwicensis) lives, according to Mr. Dole, on Sonchus asper, an introduced plant, as well as on berries (Wilson’s _Aves Hawaiiensis_). There are numerous references of this nature in books about birds, and it should always be remembered that birds in pecking at the fruit-heads scatter the seeds on their feathers. (See Note 67.)

From the foregoing remarks it may, I think, be inferred that the achenes of the ancestors of the original Composite genera of the Pacific islands were in all probability not unfitted for transport by birds, more especially in their plumage. Some of my readers, however, may express a doubt as to whether birds likely to disperse seeds would be found in any numbers at the great heights where some of the continental Compositæ occur. But it is well known that birds of the grouse and partridge family frequent high levels in continental regions over much of the globe. Arborescent Compositæ are found at heights of 10,000 to 14,000 feet on the mountains of Central Africa; and it should be noticed that Sir Harry Johnston observed “francolins” on the slopes of Ruwenzori up to 13,000 feet (_Uganda Protectorate_, vol. 1; _Trans. Linn. Soc. Bot._, Ser. II. vol. 2). Sir Martin Conway in the Bolivian Andes found geese, ducks, gulls, snipe, &c., numerous in suitable places up to 17,000 feet (_Journ. Roy. Geogr. Soc._, 1899); whilst geese and teal were noticed by Sir Joseph Hooker and others at elevations of 17,000 feet in the mountains of Tibet (Hooker’s _Himalayan Journals_; _Journ. Linn. Soc. Bot._, vol. 35, p. 147). These are all birds, as shown in Chapter XXXIII., that are likely to disperse plants, and probably none more effectually than the goose, of which Hawaii possesses a particular variety or species. It may be remarked that geese, ducks, gulls, and other birds use Cotula plumosa in Kerguelen for making their nests (Dr. Kidder quoted by Mr. Dixon in his book on Birds’ Nests).

Sea-birds were probably the principal agents in carrying the achenes of the early genera of the Compositæ to Hawaii. Dr. Hillebrand attached importance to the tropic-bird (Phaethon) in the distribution of species (Introd., p. 30); and since these birds breed at the crater of Kilauea in Hawaii, 4,000 feet above the sea, and also high up in Tahiti (Moseley), its agency is not unlikely, I am inclined to think, however, that birds like the petrels and puffins, that in nesting burrow in the ground, choosing places where the vegetation is thickest, and where they would be likely to get seeds on their feathers, would be more efficient agents. This is the view expressed by Prof. Moseley in Wallace’s _Island Life_, p. 250. He considered that albatrosses, petrels, and puffins have played a great part in the distribution of plants, and to some degree especially account for the otherwise difficult fact that widely distant islands in tropical seas have similar mountain plants. Birds, he says, that in high latitudes, as at Tristan da Cunha and Kerguelen, often burrow near the sea-level, in the tropics choose the mountains for their nesting-place; and he refers to a puffin that nests on the top of one of the high mountains of Viti Levu at an altitude of 4,000 feet, to a petrel nesting among ferns at Tahiti at an elevation of 4,400 feet, and to another petrel breeding in like manner in the high mountains of Jamaica at a height of several thousand feet above the sea. He gives point to these interesting remarks, which might be supplemented by data from other parts of the world, by observing that it is not necessary that the same species should now cover the range of the plants concerned. The ancestor of the species might have carried the seeds, and the range of the genus is alone sufficient. It may be added that, as I have shown in Chapter XXXIII., sea-birds have been far more active agents in the distribution of plants than many people might imagine. The more recent observations of Ekstam in Spitzbergen have thrown considerable light on this subject.

Having in the first place formed the opinion that the achenes of the early Hawaiian Compositæ are suited for dispersal by birds, and then shown that sea-birds were probably the principal agents, we are met with the curious difficulty that in the case of the early Hawaiian genera of Compositæ the complete suspension for ages of the means of dispersal is involved in the circumstances that these genera are confined to the Hawaiian group. We can attribute to the agency of existing sea-birds the occurrence of the genus Lagenophora in the uplands of Hawaii, on the mountain-tops of Fiji, and in Australia and New Zealand; but the agency of birds as at present in operation does not assist us except indirectly in the case of the genera restricted to Hawaii or to Tahiti. Is it possible, we may inquire, to penetrate this mystery? Why, we may ask with Mr. Hemsley, has the agency ceased acting, and why have its operations been confined to the conveyance of seeds _to_ the islands and not _from_ the islands as well (_Intr. Bot. Chall. Exped._, p. 66)? I need scarcely add that the same question presents itself with all the other peculiar genera of these islands, and in fact with endemic genera all over the world. What can be stranger, it may be remarked, than the limited distribution of the Pandanaceous genus Sararanga in the Western Pacific, although suited for dispersal by frugivorous birds. This is not, indeed, a special difficulty connected with oceanic islands; it applies to the whole plant-world; yet it is possible that, as it is exhibited by the Compositæ in these islands, we may be in a better position to grapple with the problem. But before doing so it will be requisite to look a little closer at these early Hawaiian genera of the Compositæ.

The distribution within the archipelago of the genera and species of the early Compositæ of Hawaii is worthy of notice from the light it throws, not only on the relative antiquity of the genera, but also on the subsequent conditions of isolation. Of the nine genera here referred to five are distributed over most of the islands of the group. These include all the genera possessing a number of species, namely, Tetramolopium with seven species, Lipochæta with eleven, Campylotheca with twelve, Dubautia with six, and Raillardia with twelve species. Of the four genera remaining all have only two species, and are restricted to two or three islands, Remya and Wilkesia being in both cases found in Kauai and Maui, whilst Argyroxiphium is confined to the adjacent islands of Maui and Hawaii, and Hesperomannia to those of Oahu, Lanai, and Maui. These four genera that are restricted to only two or three islands are the same before referred to as regarded by Hillebrand as the oldest, partly on account of their isolated generic position, and partly because in each case they only possess two species.

Although the early Hawaiian Compositæ were evidently originally transported to most of the islands of the group, it is noteworthy that their subsequent isolation from the rest of the world has in the later ages been repeated within the limits of the archipelago. Of the 56 species, all of which are now endemic, 28, or just half, as shown in the table on the following page, are confined to a single island. Of the remainder, almost all are restricted to two or three adjacent islands. Hillebrand gives only a solitary species, Lipochæta connata, as occurring in all the islands. This suspension, to a great extent, of the means of dispersal between the islands is also strikingly illustrated by the Lobeliaceæ.

We have only to mention the flora of Fiji and those of the adjacent groups of Samoa and Tonga to exclude them from any share in the early era of the Compositæ in the Pacific. The prevailing adventitious character of the Fijian Compositæ is indicated in the fact that the species of the majority of the genera are included by Seemann in his list of Fijian weeds. There are only one or two Fijian Compositæ, such as the mountain species of Lagenophora and the littoral species of Wedelia, that merit the special attention of the student of dispersal. So also with Samoa, Reinecke enumerates eight species, of which six are weeds either of aboriginal or of European introduction, the others being the littoral Wedelia above alluded to, and a species of Blumea found also in Fiji.

DISTRIBUTION OF THE ENDEMIC GENERA OF COMPOSITÆ IN THE HAWAIIAN ISLANDS.

+--------------+-----------------------------------------------------+-------+ | | Distribution of the Species. | | | +---------+----------+----------+----------+----------+ | | Genus. | One | Two | Three | Four | | | | | island. | islands. | islands. | islands. | General. | Total.| +--------------+---------+----------+----------+----------+----------+-------+ |Remya | 2 | — | — | — | — | 2 | |Tetramolopium | 1 | 4 | 2 | — | — | 7 | |Lipochæta | 3 | 4 | 3 | — | 1 | 11 | |Campylotheca | 5 | 4 | 3 | — | — | 12 | |Argyroxiphium | 1 | 1 | — | — | — | 2 | |Wilkesia | 2 | — | — | — | — | 2 | |Dubautia | 4 | — | 2 | — | — | 6 | |Raillardia | 9 | 1 | — | 2 | — | 12 | |Hesperomannia | 1 | 1 | — | — | — | 2 | +--------------+---------+----------+----------+----------+----------+-------+ | | 28 | 15 | 10 | 2 | 1 | 56 | +--------------+---------+----------+----------+----------+----------+-------+

We have now, I venture to think, gone far to establish the existence of an early “Composite” flora with mainly American affinities in the Pacific islands, an ancient flora of which only the remnants now occur in the uplands of Hawaii, Tahiti, and Rarotonga. That the achenes were originally transported in birds’ plumage is, as we have seen, probable; but we are still quite in the dark as to the causes of the subsequent suspension of the means of dispersal and of the resulting period of isolation, during which the original immigrant plants acquired their endemic characters. In our uncertainty, therefore, we will look to Fiji in the hope that in the absence of the early Compositæ from that group we may find a clue that will enable us to divest this problem of some of its difficulties.

It might be at first considered that since these peculiar genera of Compositæ occur in the higher levels of Hawaii and Tahiti their absence from Fiji might be connected with the relatively low altitude of those islands, a character that is concerned with the exclusion from the Fijian flora of many Hawaiian and Tahitian mountain plants (see Chapters XXIII. and XXIV.). But this view is at once negatived by the fact that Fitchia thrives in Rarotonga, an island which does not far exceed 2,000 feet in elevation. It is negatived also by the extensive development of shrubby and arborescent Compositæ in the Galapagos Islands, on the equator, in St. Helena in 16° South latitude, and in other tropical islands, which are less than, or do not exceed, the Fijian Islands in their altitude.

During the age of the Compositæ it is reasonable to suppose that the dispersal was general over the Pacific. The absence of genera indicating this era from the islands of the Fijian region, that is, from Fiji, Tonga, and Samoa, would become intelligible if these groups were submerged during this age of the general dispersal of the order over this ocean. In my volume on the geology of Vanua Levu in Fiji, I have shown that these island-groups of the Western Pacific emerged from the sea towards the close of the Tertiary period, a conclusion that would enable us to assign the age of the general dispersal of the Compositæ over the tropical Pacific to an earlier portion of the same period.

In order, however, to make further progress in the discussion of this difficult problem we are obliged to approach it from the outside. We must in fact regard these genera from the standpoint of their position as members of the vast and ancient order of the Compositæ. It is now more than thirty years since Mr. Bentham completed his remarkable memoir on the classification, history, and geographical distribution of the Compositæ (_Journal Linnean Society, Botany_, London, Vol. 13, 1873). Like De Candolle, when dealing with the facts of distribution, he handled thousands of species, and as a result he drew certain inferences which are of prime importance to students of plant-dispersal. In his time the order included nearly 10,000 known species, and although this number has since no doubt been considerably increased, it is not likely that his main conclusions, in so far as they are free from purely hypothetical considerations, will be materially affected by the later discoveries.

Accepting the antiquity of the order, and regarding it as probably dating far back in geological time, he observes that the evidence points to a very wide dispersion of its original stock at an early period. Africa, West America, and possibly Australia, possessed the order at the earliest recognisable stage. There must have existed, he contends, at this early period some means of reciprocal interchange of races between these regions. Then followed a stoppage of communication, or a suspension of the means of dispersal, between the tropical regions of the Old and New Worlds; but long after communication was broken off in the warmer regions, it still existed, as he holds, between the alpine heights in those regions and also between the high northern latitudes of both hemispheres. Referring particularly to the Hawaiian Group, he considers that the large endemic element among the Compositæ indicates that the ancient connection, whether with America or with Australasia, has been so long severed as not to have left a single unmodified common form. Fitchia, the Tahitian genus, as we have already remarked, is regarded as the only remnant of an ancient Composite flora in the tropical islands of the South Pacific.

In the light of these reflections it will be interesting to glance at the general distribution of the shrubby and arborescent or woody Compositæ. Mr. Hemsley, having generally discussed the subject, arrived at the conclusion that, “although they form so large a proportion of the floras of St. Helena, Juan Fernandez, the Sandwich Islands, and some other islands, they are not specially insular.” There are scores of them, he goes on to say, in South America, Africa, Madagascar, India, Australia, and New Zealand from twenty to forty feet high, and more truly arboreous than the insular ones; whilst nearly every sub-order has its arboreous representatives. He was, however, unable to form any definite opinion of the method of distribution of the woody Compositæ. Taking those of St. Helena and Juan Fernandez, he observes that they are not more closely allied to the Compositæ of the nearest continents than they are to those of more distant regions. The occurrence of arboreous Compositæ, belonging in each case to different tribes, in so many remote oceanic islands, coupled with the distribution of the genera to which they bear the greatest affinity, seems, he observes, to indicate that they are the remains of very ancient types (_Introd. Bot. Chall. Exped._, pp. 19-24, 66, 68; also Parts ii. p. 61, and iii. p. 23).

The further discussion of this subject would lead us into a wide field of inquiry, quite beyond the scope of this work. There is, however, an inference that I think we may legitimately draw from geological evidence in this region. With respect to the antiquity of the woody Compositæ of the Pacific as illustrated by the endemic genera, both Mr. Bentham and Mr. Hemsley view them as belonging to ancient types. Mr. Wallace, in his _Island Life_, a book that becomes more and more indispensable for the student of dispersal as years progress, dwells on the importance of these ancient Compositæ in the floral history of the Pacific islands. We may look upon the Hawaiian Compositæ, he remarks, as representing the most ancient portion of the existing flora, carrying us back to a very remote period when the facilities for communication with America were greater than they are now. The date of this period of oceanic dispersal of the Compositæ we can now approximately determine, since these plants are absent from the Fijian region, an area of submergence during the Tertiary era. Before the island-groups of the Fijian region had emerged towards the close of the Tertiary period the achenes of the early Compositæ had been dispersed far and wide over the tropical Pacific.

But this is not all that we can infer from the convergence of these independent lines of botanical and geological investigation. Mr. Bentham observes that the tribes of the Compositæ had acquired the essential characters now employed in classification before the dispersion of the order over the Pacific. Since this general dispersion took place, as we hold, during the Tertiary submergence of the island-groups of West Polynesia (Fiji, Tonga, Samoa), it follows that the birth of the tribes of the Compositæ antedates that period. If this interesting order could supply us with a “datum-mark” in the history of the Pacific floras, it would be stated in terms of the development of specific and generic characters, but not of those of a tribe.

_Summary of Chapter._

(1) The Hawaiian Islands present the same contrast with the Fijian and Tahitian groups as regards the development of new species in the case of the flowering plants that they offer in the case of the vascular cryptogams (ferns and lycopods). But the contrast is intensified, and it is further emphasised as respecting the flowering plants by the evolution of a large number of endemic genera.

(2) This great preponderance of peculiar species and genera in Hawaii is not to be connected with the relative antiquity of the group but with its degree of isolation.

(3) The earliest stage of the flowering plants of the islands of Hawaii and of Eastern Polynesia (the Tahitian region) is indicated by the endemic genera, particularly those of the Compositæ and Lobeliaceæ. Such genera are numerous in Hawaii, and occur also in the Tahitian region, as in Tahiti and Rarotonga; but do not exist in the groups of the Fijian region (Fiji, Tonga, and Samoa).

(4) The endemic genera of the Hawaiian Compositæ are mainly American in their affinities. The relationship of the solitary Tahitian genus (Fitchia) is still a subject of discussion.

(5) In the Hawaiian Islands, as well as in Tahiti and Rarotonga, the plants of the endemic genera of Compositæ are, as a rule, arborescent or shrubby; and in the first two localities they are mainly restricted to the higher levels.

(6) In discussing the mode of dispersal of the achenes of the original genera we have also to explain why the process of dispersal has been in the main suspended.

(7) It is shown that the achenes of these early Compositæ were in all probability suited for dispersal in birds’ plumage.

(8) Yet the isolating influence that cut off these genera from the outside world has, in later ages, been active within the limits of the Hawaiian archipelago, with the result that half the species are not found in more than a single island. Inter-island dispersal has, therefore, been also largely suspended.

(9) The absence of endemic genera of Compositæ from Fiji, Tonga, and Samoa cannot be attributed to unsuitable climatic conditions connected with the relatively low elevation of those islands as contrasted with those of Hawaii, since a species of Fitchia abounds in Rarotonga, which is not far over 2,000 feet in elevation. Shrubby and arborescent Compositæ of peculiar types also occur in the Galapagos and other tropical islands not more elevated than the Fijis.

(10) These endemic genera are the remains of an ancient Composite flora in the islands of the tropical Pacific, and ages have elapsed since the severance of their connections with regions outside.

(11) According to Mr. Bentham the Compositæ were distributed over Africa, West America, and possibly Australia, at an early period, but subsequent to the differentiation of the tribes of the order. Some means of reciprocal interchange of races between these regions then existed. Then followed a suspension of the means of dispersal between the tropical regions of the Old and New Worlds except between the alpine heights of those latitudes.

(12) It is inferred by the author of this volume that the general dispersion of the early Compositæ over the Pacific took place during the Tertiary submergence of the island-groups of the Fijian region (Fiji, Tonga, and Samoa), and that their absence from that region may be thus explained. At the time of this general dispersion, as above pointed out, the tribes of the Compositæ had been already differentiated.