Afzelia bijuga.—The African home of the genus.—The double station of Afzelia bijuga, inland and at the coast.—The nature of the buoyancy of its seeds.—Summary relating to Afzelia bijuga.—Entada scandens.—Its station and distribution.—Darwin’s opinion of the plant.—The dispersal of its seeds by the currents.—Summary relating to the plant.—Cæsalpinia bonducella and C. bonduc.—Their station and distribution.—Their characters in various Pacific groups.—The parents of inland species.—Their dispersal by the currents.—The germination of their seeds.—A dream of vivipary.—The causes of the seed-buoyancy.—Summary of results.

In this chapter we have a study of Leguminous strand plants that are of great interest. It can be safely said that the student of plant-dispersal in the Pacific will be brought into contact with the problems here involved wherever he goes.

Afzelia bijuga (Gray).

This Old World tree, which belongs to the sub-family Cæsalpiniæ, is of great interest to the student of plant-dispersal. It is one of that large group of Indo-Malayan plants that extend into the Western Pacific, and give the prevailing character to the floras of such archipelagoes as that of Fiji. It is a large tree yielding a valuable timber used by the Fijians and Samoans for many purposes, such as for canoes, house-posts, clubs, kava bowls, &c., but it has not been recorded from the Tahitian region, and is unknown from Hawaii. In the fact of its being a littoral as well as an inland tree, it possesses a peculiar interest from the standpoint of plant-dispersal, and especially since this difference in station is associated with a difference in buoyancy, the seeds of the inland trees usually sinking, whilst those of the coast trees usually float, and often for a period of months.

A glance at the distribution of the genus will enable us to appreciate some of the points that will be touched upon in the following discussion; and it may be here remarked that the explanation of the distribution of these Leguminous trees will go far to make clear some of the most difficult points in plant-geography. Of the eleven species enumerated in the Index Kewensis, five belong to tropical Africa, occurring on both the east and west coasts as well as in the interior, three are confined to the mainland of tropical Asia, and two are peculiar to Malaya. In the last place we have the wide-ranging Afzelia bijuga, which, if it does not actually occur on the east coast of Africa, is found at all events in Madagascar and in the Seychelles, and is to be followed by the way of the Chagos Archipelago to the Malayan Islands and Queensland, and eastward to Fiji and Samoa.

The most suggestive feature in the distribution of the genus is to be seen in the frequent station of the species by rivers. We learn from Oliver’s Flora of Tropical Africa that these trees find a home along river-courses on both sides of the continent, as on the banks of the Congo, the Niger, the rivers of Senegambia, and the Zambesi, the Zambesi species being found also on the shores of Lake Nyassa. Since tropical Africa possesses about half of the species, it would seem highly probable that it is the home of the genus, and that from the rain-forests in the heart of the continent rivers flowing east and west have borne the buoyant seeds of the wandering species to the coasts of the Atlantic and Pacific Oceans. The operation that I witnessed on a miniature scale in the case of a species of Entada (E. scandens) in the Isthmus of Panama, as described in a later page of this chapter, has been in progress through the ages with the genus of Afzelia in the breadth of the African continent. According to the principle illustrated by Afzelia bijuga in the forests of Fiji, the seeds of the African forest-trees would, as a rule, possess no floating power; but now and then in the lapse of long periods of time buoyancy in some species would be developed, and such species would ultimately, through their buoyant seeds, find their station along the lower courses of the rivers.

To sustain this view it is not necessary that continuous rain-forests should now clothe the elevated regions in the interior of tropical Africa; but it is requisite that there should be sometimes a generic similarity between the plants of the East African and West African rain-forests; and it is evident that this is the case. Pechuel-Lösche, as quoted by Schimper (Plant-Geography, p. 299), describes the rain-forest on the Loango coast as covering the mountain ranges and as extending to the river-plains. In such a locality the operation would be rapid. In advancing this hypothesis I am referring to the possibility, however, of such an operation having effected the distribution of Afzelia in tropical Africa in the past rather than in the present. I would suggest that botanists in other habitats of the genus, as for instance in Queensland, might put it to the test of observation and experiment.

The interest that attaches itself to the story of the genus in its African home may be extended to the species that forms its outpost in the Pacific, and we shall see there a littoral species that doubtless had its home in the interior of a continent endeavouring, with a considerable measure of success, to become again an inland plant. Horne (p. 112), who was familiar with Afzelia bijuga at the two extremes of its range, namely, in the Mascarene Islands and in Fiji, speaks of it as characteristic of the shores of tropical regions; and Schimper, who includes it in the Indo-Malayan strand-flora, implies that it is more or less exclusively confined to the coast and its immediate vicinity (pages 121, 191-2). In the Seychelles, according to Mr. Button, this tree attains gigantic dimensions on the sandy flats. Still larger trees occur in the coral islands of the Chagos Archipelago; but in the atoll of Diego Garcia, as we learn from Mr. Bourne, it is almost extinct only some four or five trees existing there about twenty years ago, the increase of the tree being prevented through the destruction of the fallen seeds by the rats (Journ. Linn. Soc. Bot., vol. 22, 1887).

Afzelia bijuga may, therefore, be safely regarded as a littoral tree. We shall now see the importance of this conclusion when we come to consider its station in the Pacific islands, where it grows both inland and at the coast, and we have to decide to which station we must assign the priority. Speaking of its occurrence in Fiji, Dr. Seemann says it is “common in the forests all over Viti,” but makes no allusion to it as a littoral tree either in Fiji or elsewhere. On the other hand, Mr. Horne (p. 112) describes it as “generally growing on the shore or sandy beaches, and in rocky clefts, and by the sides of streams in the interior of Viti Levu and Vanua Levu.” It was on or near the coast in Fiji that the present writer was most familiar with this tree, sometimes bordering the sandy beach, at other times growing behind the mangrove-belt, or again thriving in the half sandy and half swampy soil of some low islet off the mouth of the Rewa. Especially is it to be found on those parts of the coast where the hill-slopes descend rapidly to the beach, or where some lofty spur from the mountains of the interior reaches the shore. It is also not uncommon on the banks of rivers both in their lower and upper courses. But it is as a forest-tree of the interior that it is most valued by both the white men and the natives on account of the superior quality of its timber in that station. There, far removed from stream or river, the Vesi, as the Fijians name Afzelia bijuga, takes its place amongst the lofty forest-trees, such as the Ndamanu (Calophyllum), the Ndakua (Dammara), and the Wathi-wathi (Sterculia). It is not often that one finds a tree in these islands that, like the Vesi, is able to make its home in almost any station, excepting, however, the “talasinga” or “sun-burnt” regions of the plains. Wherever tall trees grow gregariously in Vanua Levu, one will probably find Afzelia bijuga, whether beside a sandy beach, or bordering a swamp, or on a river’s bank, or on some rocky declivity, or on the great forest-clad mountain-slopes and plateaux of the interior. No doubt the same diversity of station is displayed in Samoa, where, according to Dr. Reinecke, the tree is most frequent in the “coast-bush.”

From the variety in station it might be expected that corresponding variations in character would be found. There are differences, such as in the quality of the timber and in the size of the seeds between coast and inland trees; but the most important distinction in connection with the study of the dispersal of the species is to be found in the circumstance that whilst the seeds of the coast trees are, as a rule, buoyant, and often float for months, those of the inland trees usually sink, even after being kept for three or four years. I made a considerable number of experiments on the buoyancy of the seeds of this tree in Vanua Levu, and found that with the coast trees, as a rule, either all the seeds or the majority of them floated in sea-water, whilst with the inland trees either all of them or the majority of them sank. The buoyant seeds are able in most cases to float for a long time. Thus, in one experiment half were afloat after two months, and in another half were afloat after five months. It is probable that several of the exceptions, where inland seeds float, will prove to be connected with an inland station by a river. (I experimented on eight sets of seeds of coast trees from eight different localities, and found 70 to be the mean percentage of buoyant seeds. In the same way, four sets of seeds from four different inland localities gave 13 as the mean percentage of buoyant seeds.)

As in the case of Entada scandens, there is a rather fine adjustment between the mean specific weight of seeds and the density of water. If we place a number of the buoyant seeds in sea-water and begin to lower the density, some of the seeds will at once commence to float heavily and afterwards sink; and when the density has been lowered to approximately that of fresh water, usually about a third will be found at the bottom of the vessel. Out of 100 coast seeds, 70 will, as a rule, float in the sea and about 47 in the river; whilst of the same number of inland seeds, 13 on the average will float in sea-water and 8 or 9 in fresh water. The bearing of facts of this kind is especially discussed in Chapter X.

Coming to the causes of the floating-power of the seeds, we find that with the buoyant seeds the kernel floats, whilst with the non-buoyant seeds it sinks, the seed-tests in neither case possessing any floating-power. In this respect, therefore, the seeds of Afzelia bijuga belong, with the seeds of some other Leguminous littoral plants of the Pacific islands, such as Canavalia obtusifolia, Erythrina indica, and Sophora tomentosa, to the second section of the second non-adaptive group of buoyant seeds (page 107). But though we can in a measure explain the cause of the buoyancy, we are still ignorant of the manner in which the difference in the buoyant behaviour of coast and inland seeds has been brought about. It is possible that this may be connected with another difference between the coast and inland seeds, the latter being markedly smaller, and it is noteworthy that in my experiments the smaller seeds were generally those that sank. (Whilst the inland seeds averaged between ⁸⁄₁₀ and 1 inch, or 20 to 25 mm., in greatest diameter, 12 to 16 being required to make an ounce, the coast seeds measured 1 to 1²⁄₁₀ inch, or 25 to 30 mm., and only 10 or 11 were needed to weigh an ounce.)

There can be no question that the seeds are at times transported by the currents over wide tracts of sea, and this no doubt explains the occurrence of Afzelia bijuga in oceanic islands. They may be usually seen lying free in numbers on the ground beneath the tree or else still inclosed in the fallen dehiscing and decaying pods; and they might be swept sometimes into the sea or washed down into an adjacent stream. They thus came under my notice amongst the stranded beach drift at the mouths of estuaries in Fiji. But it is remarkable that the seeds have not apparently been recorded from the beach drift of other tropical regions. Penzig does not note them amongst the seeds stranded on the shores of Krakatoa. They did not occur amongst my collections from the beaches of Keeling Atoll or of the south coast of Java; nor does Schimper mention them amongst the drift of the Java Sea. In the Botany of the “Challenger” Expedition the species is not even referred to in any connection. Although, however, the capacity of these seeds for dispersal by currents is for the first time established by me, their fitness in this respect was surmised by Schimper (p. 191), when he placed the species in his list of tropical shore plants evidently distributed by the currents.

It will thus be gathered that we have yet much to learn in this matter; and I would recommend any resident in the tropics to take up this subject. When indeed we remember the fine adjustment existing between the specific weight of the seeds and the density of water, and recall the unknown factor determining the difference in buoyancy between the kernels of coast and inland seeds, we can understand how under particular conditions in certain portions of its range the seeds of Afzelia bijuga may perhaps never possess any floating power. It would seem, in fact, that the seeds are much more buoyant in the Western Pacific than they are in the Java Sea; or it may be that the tree is much less frequent; or that the stranded seeds are soon destroyed by crabs, such as is the fate of much seed-drift on the Keeling beaches; or lastly that, as in Diego Garcia, rats in destroying the fallen seeds are bringing about the extermination of the species.

Summary relating to Afzelia bijuga.

(1) Assuming that the genus has its home in the African continent, and that the species have frequently a riverside station, it is argued that the distribution of the genus on both sides of that continent can only be explained by its dispersal by rivers from a centre in the interior.

(2) Afzelia bijuga, a widely distributed shore tree of tropical Asia, occurs in Fiji both at the coast and in the inland forests.

(3) This double station is associated inter alia with a different buoyant behaviour of the seeds, those of the coast trees usually floating for long periods, whilst those from inland generally sink.

(4) There can be no doubt that this widely ranging littoral tree has been dispersed by the currents; but the specific weight of the coast seeds is on the average but slightly less than that of sea-water; and it is to this fine adjustment, always liable to be disturbed by variations in the environment, that the irregularities in the distribution of the species are to be attributed.

Entada scandens (Benth.).

The story of Entada scandens, a plant familiar to many of my readers under the name of the Queensland Bean, is a story of three continents, Africa, Asia, and America. From the point of view of its dispersal two features at once attract attention in the case of this giant-climber; in the first place its wide distribution over the tropics of the Old and New Worlds, and in the second place the great capacity of its large seeds, often two inches across, for dispersal by the currents. But before discussing these matters it will be necessary to glance at the distribution of the genus, since much light will thereby be thrown on some of the numerous difficult points affecting this extremely interesting tropical plant. Of the thirteen species enumerated in the Index Kewensis, seven are African, three are American, one is Burmese, one hails from Madagascar, and, lastly, there is the world-ranging Entada scandens, concerning whose home botanists are not agreed. Most of the species would seem to be inland plants, whilst Entada scandens thrives both inland and at the coast. Africa would thus appear to be, as with Afzelia, the principal home of the genus, but with America as a subsidiary centre.

In many points Entada scandens presents a parallel to Cæsalpinia bonducella, another Leguminous tropical plant which occurs also at the coast and inland. But since they both owe their wide distribution to their littoral station, it will be as coast plants that they will be most properly considered in this and the following chapter. Yet if the student were to regard the distribution of these two plants in a continental region as in India, where they extend inland to the Himalayas, he might fail to discern their true station. To accurately gauge the matter of their station, it is necessary for him to look at the plants as they occur in the islands of the Pacific. There he will first see the stranding of the seeds on a shore by the currents, then their germination and their development into giant-climbers over the littoral trees or into straggling bushes on the beach; and afterwards he will observe the plants of both species extending inland, and in these three stages he will learn their history in the Pacific; but a history, it may be observed, that in this region represents their efforts to return to an inland station, such as they once possessed in their birthplace in some distant region of the globe.

Dealing first with the station of Entada scandens, it may be remarked, as Dr. Seemann points out, that in Fiji it is most characteristic of the mangrove-formation. But it also occurs amongst the trees at the back of the mangrove swamp, on the beaches, on the banks of the estuaries, and at the edge of the inland forests where they border on the plains. Sometimes in the company of Derris uliginosa it grows not as a climber, but as a prostrate plant on the sandy beaches; and here, not being able to assume its normal habit of a climber, it does not seed. It is to be found at times far inland in open-wooded districts. Thus in Vanua Levu I found it growing in the Mbua district four miles inland, and 1,400 feet above the sea. Reinecke speaks of it in Samoa only in connection with the “urwald,” or primeval forest. Cheeseman describes it as most abundant in the interior of Rarotonga, covering the trees with a wide-spreading canopy of green. In the Malayan region Schimper refers to it as a plant of the beach-tree formation. In Ecuador and on the Panama Isthmus it grows not only at the coast, but also on the hill-slopes in the rear of the mangrove-belt.

With reference to the distribution of the plant, it may be remarked that, although it is found all round the tropics and possesses great capacity for dispersal by currents, there are certain difficulties in explaining its wide area and in accounting for its very peculiar distribution in the Pacific islands. It was doubtless in allusion to some of these difficulties that Mr. Darwin, in a letter to Sir Joseph Hooker, remarked: “Entada is a beast” (More Letters, &c., i, 93). There is at first the question of the identity of the species in the Old and New Worlds. It is here assumed that it is the same in both hemispheres; but it must not be forgotten that the identity is “not beyond doubt” (Bot. Chall. Exped. iv, 147).

Then there is the difficulty connected with its occurrence on both coasts of tropical America. In this respect it is at one with some other littoral plants, like Ipomœa pes capræ, as well as with the plants of the mangrove formation, as is pointed out in Chapter VIII. Whilst with the mangroves it is necessary to assume that they antedate the land connection between North and South America, this is not requisite in the case of Entada scandens, since it grows in the interior of the Panama Isthmus, and rivers on the north and south sides now carry its seeds seaward from the same “divide” to the Atlantic and Pacific Oceans, as described in Chapter XXXII.

But, as I have also shown in Chapter VIII, America forms with the West Coast of Africa a region characterised by the same tropical littoral flora. This region, on account of the arrangement of the currents, stands in a very peculiar relation with the Asiatic region, which comprises the rest of the tropics, and to a great extent possesses its own peculiar strand-flora. There are a few littoral plants, like Entada scandens, Canavalia obtusifolia, Sophora tomentosa, and Ipomœa pes capræ that occur in both areas; but the large majority are confined to one or other of them, either to the American region, including the African West Coast, or to the Old World region, which includes the African East Coast. The American region gives to the Old World, but it can receive nothing in return. For this reason, it is argued, we are compelled to regard most, if not all, of the cosmopolitan tropical shore plants that are dispersed by the currents, such as those above named, as having their home in the American region. Entada scandens would, therefore, from this standpoint have its home in America.

Then, again, there is the difficulty connected with the distribution of this plant on both sides of tropical Africa. Though Oliver in his Flora of Tropical Africa mentions this species only in connection with the West Coast, he says it is probably widely spread in that continent, and he refers to a pod in the Kew Museum indistinctly labelled “Lake Ngami.” I have not come upon any reference to its being a littoral plant on the East Coast, but since numerous littoral plants of tropical Asia are found on that coast its occurrence there or in the East African islands would be expected. However, as the genus has a centre in America, and as this species is regarded as of American birth, we are not called upon to employ the argument used in assigning to a non-American genus like Afzelia an African home. Since the African West Coast belongs to the American region of tropical shore plants dispersed by the currents, the presence of Entada scandens on that coast of Africa can be readily explained, whilst if it has reached the Malayan Archipelago from America by way of the Pacific, it would, by extending like many other Malayan coast-plants along the shores of the Indian Ocean, almost complete its circuit of the globe. It is in this fashion, I believe, that the other littoral plants,, like Cæsalpinia bonducella, Canavalia obtusifolia, and Ipomœa pes capræ, that are found all round the tropics, have performed the circuit of the globe with America as their home.

One may remark in passing that the double home of the genus in America and the Old World, though offering a serious difficulty in plant geography, has no immediate bearing on the present mode of distribution of Entada scandens. Questions relating to the distribution of tropical shore-plants that are dispersed by the currents at first resolve themselves into considerations of the arrangement of the currents. Entada is not alone amongst the genera containing littoral species in having a home both in the Old and in the New World. Carapa is another instance, and additional cases might be cited.

The next peculiarity in the geographical range of this species is concerned with its irregular distribution in the archipelagoes of the tropical Pacific. Notwithstanding its great capacity for dispersal by the currents, although it occurs in all the groups of the Western Pacific as well as in the Cook Islands, it has not been recorded from the Society Islands, the Paumotus, the Marquesas, and Hawaii. Since, however, its seeds have been gathered by Mr. Arundel on the beaches of Flint Island, lying about six degrees north of Tahiti (Bot. Chall. iv, 302), it is not unlikely that it will be found growing in other parts of Eastern Polynesia south of the equator. One might have looked for an explanation of its rarity in Eastern Polynesia to the absence of mangrove swamps, in which, as in Fiji, it is sometimes most at home; but this is negatived by its abundance in Rarotonga, where mangrove swamps do not exist.

The dispersal of Entada scandens by the currents.—This plant offers one of the most conspicuous examples of the transport of seeds across oceans through the agency of the currents. In the pages of many botanical works, from the close of the 17th century onward, reference is made to the transport of its beans (often in association with those of Mucuna urens and Cæsalpinia bonducella) by the Gulf Stream or other currents across the Atlantic to St. Helena, the Azores, the west coast of Ireland, the Hebrides, the Orkney Islands, the coasts of Scandinavia, and even as far north as Nova Zembla (see Hemsley’s Bot. Chall. Exped.; Sernander’s Skand. Veg. Spridningsbiologi, &c.). That the seeds of Entada scandens retain their germinating capacity after this ocean-transport has been demonstrated not only by the germination of stranded seeds on the shores of St. Helena, but also by the germination when sown at Kew of seeds drifted to the Azores, as well as by the results obtained by Lindman, who procured the germination of the seeds of this plant and of Mucuna urens that had been washed up on the Scandinavian beaches (see Sernander, pp. 7, 390).

One of the most interesting references to the conveyance by currents of these seeds to the coasts of Europe is to be found in Dr. Sernander’s recent work on the modes of dispersal of the Scandinavian flora, where he sums up the results of Lindman’s investigations respecting the Gulf Stream drift. The stranded seeds of Entada scandens, it appears, have been found all along the Norwegian coast, but occur most frequently north of the Söndmöre district. They have even been found in a sub-fossil condition in the peat-bogs of Tjörn on the Bohuslän coast in Sweden, having been originally stranded on a beach in that locality at some distant, but post-glacial, epoch. Few phenomena in plant-distribution are more suggestive than this ineffectual transport through the ages of these large tropical beans to coasts within the Arctic Circle. The seed, no longer under the care of the mother-plant, becomes a waif, exposed to the pitiless laws of the physical world which here prevail. It was not thus that the plant was reared, but it is in this haphazard fashion that its seeds are spread. The philosopher could unravel most of the tangled problems connected with present and past plant-dispersal, if he could follow the clue supplied by this stranded tropical seed on a Scandinavian beach.

It is a far jump from the North Cape to the coral islands of the Pacific and Indian Oceans; yet it is within the area covered by the drifting Entada bean. The stranded seeds occur commonly on the Fijian beaches and on other islands of the South Pacific; but I never found them in Hawaii. They were gathered by me on the shores of Keeling Atoll in the Indian Ocean, and on the south coast of Java. Penzig found on the Krakatoa beaches, in 1897, not only the stranded seed but the established plant. They came under my notice in numbers on the beaches of Ecuador and on the Pacific and Atlantic coasts of the Panama Isthmus; and, as I learned, they are equally common on the other parts of the coasts of Central America. Not uncommonly these stranded seeds in various parts of the world are to be found incrusted with polyzoa and tubicular annelids, which afford proof of prolonged flotation in the sea. These seeds are also to be frequently noticed floating in the drift of the tropical estuaries. Thus they came under my observation afloat in numbers in the Fijian estuaries, in the Guayaquil river, in the estuary of the Chagres at Colon, and in the mouth of a river on the Panama side of the isthmus.

The mode of liberation of the seeds is worthy of a passing remark. The huge pods, often several feet in length, ultimately break up into separate joints bearing the seeds. The joints may decay on the ground, and the seeds are thus freed; or not infrequently in a mangrove-swamp they fall at once into the water, and there they float, as may often be observed in Fijian rivers, until their decay sets free the seed.

The seeds of Entada scandens are often quoted, and justly so, as striking examples of the dispersal of seeds by currents. Yet in few plants could the nature or the structural cause of the buoyancy have so little claim to be considered as adaptive in its character. Quite half, and sometimes even the majority, of the seeds freshly liberated from the plant have no buoyancy at all. The mean specific weight of the seed is about that of sea-water, but markedly higher than that of fresh water; whilst the principal determining cause of the buoyancy is, as shown below, purely mechanical, and one that, whilst favouring the wide distribution of the species, could not be improved by or come within the scope of Natural Selection.

From experiments made in Fiji and Ecuador, it appears that at least 50 per cent., and often more than half, of the seeds when first liberated from the pod have no buoyancy in sea-water. Of those that float in sea-water, a proportion varying between one-third and one-half sink in fresh water, so that in the case of plants growing on the banks of a river only about one-fourth or one-third would be carried down to the sea. So fine is the adjustment of the specific weight of these seeds to the density of water, a subject discussed in its general bearings in Chapter X, that if one gathers a number of drift seeds on a beach, let it be in Fiji or in Ecuador, although, of course, all will float in the sea, only one-half or two-thirds will float in the neighbouring fresh-water stream. Those that float appear to be able to float almost indefinitely. This is sufficiently established by the transport of the seeds in a sound condition by the currents across the Atlantic, and by such evidence as the stranding of seeds incrusted with polyzoa and serpulæ on the beaches of Keeling Atoll. It has been also proved by the following experiment. Several years since, I placed a seed in a vessel of sea-water, where it still floated buoyantly in a perfectly sound condition twelve months afterwards.

With regard to the cause of the buoyancy, investigation shows that neither the seed tests nor the seed contents have any floating power, the buoyancy arising from a large central cavity produced by the shrinking and bending outward of the cotyledons during the drying and hardening of the maturing seed (see figure in Chapter XII). With the seeds that sink, this cavity is, as a rule, reduced to small dimensions, and may be represented only by a narrow slit. In some cases, however, where the cotyledons are unusually thick and heavy, even a large central cavity will not give floating power to the seed. There is an indication in my experiments that seeds from inland plants that have matured their pods in the forests sink in a much greater proportion than seeds of coast plants, or of those growing on the banks of estuaries. This we might expect, since in the shade of the forests the drying process that accompanies the setting and final maturation of the seed would be less complete and the intercotyledonary cavity smaller than with the seeds matured in more exposed situations. This is a point, however, that requires further investigation.

It will thus be seen that in respect of buoyancy the seeds of Entada scandens are to be referred to the mechanical or non-adaptive group of buoyant seeds, described in Chapter XII, which comprises several other Leguminous strand-plants, including Cæsalpinia bonducella. I especially studied the various stages in the development of the buoyancy of seeds in this mechanical group in the case of the species of Cæsalpinia just named, and the description of the process as given under that plant will apply to all.

Summary relating to Entada scandens

(1) This plant, which has been distributed by the currents over the tropics of the globe, has its station in the mangrove swamp, on the beach, by the estuary, and in the inland forest.

(2) It is regarded as an American plant that has reached the shores of the Indian Ocean by crossing the Pacific, and the coast of West Africa by crossing the Atlantic.

(3) Its occurrence on both coasts of America is attributed to its having a focus of dispersal in the forests of Central America, from which its seeds have been transported by the rivers to the shores of the Atlantic and Pacific Oceans.

(4) Its irregular distribution in the Pacific islands, to wit, its absence from Hawaii and its rarity in the Tahitian region, is not to be easily explained, but it is more than likely that it will be subsequently recorded from other localities in Eastern Polynesia.

(5) Although the seeds offer a striking example of dispersal by currents, since they are to be found stranded on beaches over much of the globe, from within the Arctic Circle to the Coral Sea, in few plants could the character of the buoyancy and the structure connected with it have so little claim to be considered as adaptive in their nature. At least 50 per cent. of the seeds sink in sea-water, and the cause of the buoyancy of the other seeds is only to be connected with the large size of a cavity produced by the shrinking of the embryo within the seed tests during maturation.

Cæsalpinia

This genus is represented in the tropics of both the Old and the New World by some eighty species of trees, shrubs, and climbers, some of which are noted for their dye-woods, and others for the beauty of their flowers. In the Pacific islands the botanist is only concerned with three widely distributed species, all more or less littoral in their station, and in great part dispersed by the currents, namely, Cæsalpinia nuga (Ait.), C. bonducella (Flem.), and C. bonduc (Roxb.).

With Cæsalpinia nuga we have little to do, since, although widely distributed in tropical Asia and the Malayan region, and reaching to both New Guinea and North Australia, it has not apparently penetrated into the Pacific further east than the Solomon and New Hebrides groups. I found it growing on the coasts of the larger islands of the Solomon group, but no observations were made on its mode of dispersal. However, as its seeds were identified at Kew (Bot. Chall. Exped. iv, 311) amongst my collections of stranded drift from those islands, it would appear to be to some degree dispersed by the currents, though since it does not extend far into the Pacific, its capacity for dispersal by this agency would seem to be limited. Schimper includes it among the strand-plants of the Indo-Malayan region.

It is with the other two species, Cæsalpinia bonducella and C. bonduc that we are especially interested. Their extremely hard, marble-like seeds at once attract attention, and when pale in colour they look not unlike quartz pebbles as they lie stranded on a beach. The prickly pods and the recurved prickles of the leaf-branches often make these plants provokingly evident to a stranger. Though usually to be characterised when growing on a beach as straggling shrubs, they will often climb trees when opportunities occur, and they then display themselves as stout-stemmed climbers. I have seen one or other of them in the mangrove swamps of Fiji ascending the Bruguiera trees to a height of 30 feet and more, the stem quite bare below, but leafing and flowering in the tree-branches above.

From the standpoint of dispersal there are few more interesting plants in the Pacific islands; but their discussion raises several difficult questions, and it will be, therefore, requisite to treat them somewhat in detail. With regard first to the diagnostic characters between the species, it may be observed that, as a rule, they are sufficiently evident, such, for instance, as the number, size, and form of the leaflets, the presence or absence of foliaceous stipules, and the colour of the seeds, though, as shown below, the seed-colour in the case of Fijian plants does not always present a constant distinction. Yet as I found in Fiji the difference between the two species is not in all cases well pronounced, and intermediate forms occur, about which it is sometimes difficult to decide to which of the two species they should be assigned.

Mr. Hemsley remarks (Bot. Chall. Exped. iii, 114, 145, 300) that the two species have been often confused. I venture to think that this has been in some cases due to the occurrence of these intermediate forms. One has only to look at the different “distributions” given by botanists for C. bonduc, as indicated below, in order to suspect that the cause of confusion has been at times with the plants themselves. When in Fiji I paid a good deal of attention to this subject, and the results of the comparison of the foliage and seeds of the plants obtained from fourteen different localities in Vanua Levu are given below.

It will be seen in this table that I distinguish in Fiji three littoral forms and one inland or mountain variety, which may perhaps be a distinct species. Those of the strand include Cæsalpinia bonducella, C. bonduc, and an intermediate form. C. bonduc is typically distinguished by its large leaflets, by the absence of foliaceous stipules, and by its pale yellow seeds; whilst C. bonducella is similarly characterised by its small leaflets, its foliaceous stipules, and its lead-coloured or darkish grey seeds. But in the first species the colour of the seeds may often be yellow mixed with pale-grey, or almost white; whilst in the second species the seeds may be stained with brownish-yellow patches.

It seemed to me when examining fresh specimens in Hawaii and Fiji that the ultimate colour of the seed is a good deal determined by the degree of alteration of the original olive-green colour of the immature seed. All gradations may be noticed from the olive-green of immaturity to the yellow, pale grey, and dirty white hues of the mature seeds of Cæsalpinia bonduc and to the lead or slate-colour of those of C. bonducella. It almost appeared as if the changes might be compared to the bleaching which a dark volcanic rock undergoes in the weathering process through the hydration and removal of the iron oxides.

Note.—The characters of the Fijian plants are from my own observations. Drake del Castillo is quoted for Tahiti, and Hillebrand for Hawaii. Reinecke observes that the pods of C. bonducella in the inland forests have no prickles.

In Fiji all three coast forms may be found on the same beach, or they may exist apart. The large-leaved species (C. bonduc) appears to be much the most frequent in Vanua Levu; and the intermediate form is common enough to disturb the serenity of the observer’s mind when he is anxious to diagnose rather than to collect cumbersome specimens. The mountain form, which came under my notice as a climber in the forest at an elevation of 1,700 feet on the slopes of Koro-mbasanga in Vanua Levu, acquires from the lanceolate shape of its leaflets quite a character of its own, though it comes nearest to Cæsalpinia bonducella. Mountain forms also occur, as indicated in a later page, in the forests of Samoa and in Tahiti; but in the first-named group they are referred by Reinecke to C. bonducella, and in Tahiti by Drake del Castillo to C. bonduc. In the Samoan forests the inland plants possess pods deprived of the prickles that are so characteristic of the beach plants. Before one can pronounce definitely on the relation between the coast and inland forms in any of the groups, a thorough investigation of the connections between the two shore-species is needed. I am inclined to think that they will prove to belong to a single dimorphic (or perhaps polymorphic) species.

The distribution of Cæsalpinia bonducella and C. bonduc.—Botanists agree in giving C. bonducella a distribution around the tropics of the globe; but they are not at all unanimous with respect to the other species. According to Mr. Hemsley this species is by no means so universally dispersed as C. bonducella. It is unknown from Africa and Australia; but it is generally characteristic of tropical Asia and the Malay Archipelago. The same authority alludes to specimens in the Kew Herbarium from Florida and the West Indies (Bot. Chall. iv, 300). Drake del Castillo gives both species a range through the tropics, whilst Schimper seems in doubt about the occurrence of C. bonduc in the New World, and Mr. Burkill makes no allusion to its American habitat in his paper on the Tongan flora. The cause of this confusion is doubtless to be mainly attributed to the variation in characters of the plants, and to the occurrence of intermediate forms.

We should be scarcely consistent if we assumed that of two kindred shore-species dispersed by the currents one had its home in America and the other in the Old World. The same home must belong to both. According to the principle laid down in Chapter VIII, and referred to under Entada scandens, it is held that a strand-plant, with its home in Asia, on account of the arrangement of the currents could never reach the American continent, and that American shore-plants are for the most part native-born except those hailing from the African West Coast, which, however, lies within the American province of tropical strand-plants. From this standpoint Cæsalpinia bonducella would be regarded as now having its home in the New World, and since it is found on both the Pacific and Atlantic coasts of that continent (as well as on both coasts of Africa), it is assumed, as with Entada scandens, that it has reached the African West Coast by crossing the Atlantic, and the African East Coast by way of the Pacific and Indian Oceans. The genus, I may remark, is distributed over the tropics of the eastern and western hemispheres.

As regards the general distribution of the two species in the Pacific islands, it would appear from the writings of Seemann, Hillebrand, Hemsley, Drake del Castillo, Reinecke, Cheeseman, and Burkill that with the exception of Hawaii and Samoa, where Cæsalpinia bonducella alone occurs, and of Rarotonga where C. bonduc alone is found, they are generally associated in the larger groups, as in Fiji, Tonga, Tahiti, and the Marquesas.

The station of Cæsalpinia bonducella and C. bonduc.—Both the species are to be regarded as littoral plants likely to stray inland. The first-named is described in the Botany of the “Challenger” Expedition as essentially a sea-side plant, though flourishing equally well inland, and in India extending to the Himalayas as far as Kumaon, and up to elevations of 2,500 feet. Schimper speaks of both species as characteristic of the Indo-Malayan strand-flora, and he quotes Kurz when referring to C. bonduc as a constituent of the beach-jungle of Pegu.

In the Pacific islands they are typically littoral in their station; but they may extend inland, and in one or two groups they are only known in their inland station. Dr. Seemann speaks of both species only in connection with the beaches in Fiji, and alludes to Cæsalpinia bonducella (p. 72) as sometimes climbing over the mangroves. In Vanua Levu both came under my notice on the beaches, and in their immediate vicinity, usually as straggling bushes, whilst at times they were to be observed climbing the mangroves at the borders of the adjacent swamp. In this island of the Fijis they do not, as a rule, stray far from the beach, and strange to say are not to be ranked amongst those seashore plants that frequent the “talasinga” regions or inland plains. Judging from the mountain form found in the forests of Koro-mbasanga, if they extend inland in Fiji they prefer the forests and become differentiated in character. In Tahiti, as we learn from Nadeaud and Drake del Castillo, C. bonducella occurs on the beach and extends inland to the mouths of the valleys; whilst C. bonduc is only recorded from the mountains at elevations of 600 to 700 metres (2,000 to 2,300 feet). Jouan is quoted by Mr. Hemsley as remarking that C. bonduc is as common in the Marquesas as brambles are in Europe (Bot. Chall. Exped. iii, 145). In Rarotonga, according to Cheeseman, C. bonduc is restricted to the interior. In Samoa, as we are informed by Reinecke, C. bonducella is frequent both in the coast districts and in the mountain-forests. In the Samoan mountains the pods lose their prickles, and from this circumstance, as well as from the extremely widespread distribution of the species over the islands, the German botanist concludes that the plant has been for ages established in the group.

In Hawaii, Cæsalpinia bonducella, which alone occurs, rarely figures as a beach plant; but it is found, as Hillebrand observes, in the lower plains of all the islands. In the large island of Hawaii I found it not on the scanty beaches of the coast, but on the partly vegetated surface of the old lava-flows at distances varying usually between a hundred yards and a mile from the sea, but extending at times a few miles inland, and in one locality reaching an elevation of 2,000 feet above the sea. It was mostly observed by me on the dry side of the island, where, associated with Erythrina monosperma, the Cactus, and the Castor-Oil plant, it thrives in very arid localities, where the rainfall is only a few inches in the year. Farther inland, where the old lava-surfaces were more vegetated, it was associated with such shrubs as Osteomeles anthyllidifolia and Cyathodes tameiameiæ. Dr. Hillebrand, writing of a generation and more ago, says that in his time the plant was less common than formerly.

The Methods of Dispersal of Cæsalpinia bonducella and C. bonduc.—We come now to the modes of dispersal of these plants; and in so doing we have to choose between the agencies of birds and of currents. The seeds of C. bonducella are on the average ⁷⁄₁₀ of an inch (18 mm.) in diameter, whilst those of C. bonduc are rather smaller (⁶⁄₁₀ of an inch or 15 mm.). As far as their size and character go, it would seem scarcely likely that birds could transport these seeds across an ocean; but our knowledge of the agency of birds is of a very imperfect nature. Yet their occasional dispersal by birds is not improbable. When I was in the Keeling Islands the residents informed me that the seeds of C. bonducella are sometimes found in the stomachs of sea-birds, such as frigate-birds and boobies. (See Note 59.)

However, it has long been known that the seeds of one or both of these species are carried great distances by the currents; but it is to be gathered that the older botanists, in alluding to this fact, more usually referred under the synonym of Guilandina bonduc to Cæsalpinia bonducella. De Candolle, loth to attach much importance to the effective transport of seeds by currents, was compelled to admit this species in his scanty list of current-dispersed plants (see Note 33). For more than two centuries it has been known that the seeds of C. bonducella are carried in the Gulf Stream drift to the coast of Europe from the American side of the Atlantic; and ever since they were recorded by Sloane in 1696 as stranded in a fresh condition on the beaches of the Orkney Islands, they have been found washed up on other localities, as on the coasts of Ireland and of Scandinavia and on the shores of the islands of the Western Atlantic. According to Robert Brown, a plant was raised from a seed cast up on the west coast of Ireland; and with respect to Scandinavia, Dr. Sernander informs us that the seeds of Cæsalpinia bonducella, like those of Entada scandens and of Mucuna urens, are of frequent occurrence amongst the “Gulf Stream products” stranded on the Norwegian coasts. The seeds of this species are commonly washed ashore at St. Helena, and there are specimens in the Kew Museum that were stranded on Tristan da Cunha. (Those interested in the subject will find it discussed by Mr. Hemsley in the Botany of the “Challenger” Expedition, and also by Dr. Sernander in his recent work on Scandinavia.)

The seeds of Cæsalpinia bonducella have been also found stranded on beaches in other parts of the world. Thus Prof. Schimper found them in the beach-drift of the south coast of Java. Prof. Penzig noticed them amongst the stranded seeds of the Krakatoa beaches; but it does not appear that the plant had established itself up to the date of his visit in 1897, or fourteen years after the great eruption. They have been picked up on the other side of the Indian Ocean on the east shores of Africa (Bot. Chall. Exped. iv, 300). They came frequently under my notice stranded on the beaches of Keeling Atoll in the same ocean; and seedlings sprouting from the seeds were sometimes to be seen growing amongst the drift just above the high-tide level. The seeds of both C. bonducella and C. bonduc have been found also on the shores of Jamaica. Those of both species are not uncommon amongst the stranded drift of the Fijian beaches; but notwithstanding a careful search I found only a solitary seed of C. bonducella in the Hawaiian beach-drift, a circumstance explained below as arising from the usual non-buoyancy of Hawaiian seeds.

That the seeds of Cæsalpinia bonducella stranded on the coasts of an oceanic island are able to germinate and reproduce the plant is, of course, established by the distribution of the species; and we have just observed that the process was noticed by the author on Keeling Atoll where the plant has found a home. It is to be noted that the plant collected by Darwin in this atoll was identified by Prof. Henslow as C. bonduc; but the plant observed by me was more like C. bonducella, and the stranded seeds collected by me were referred at Kew to this species. Some curious considerations arise from the fact that although, just as in the Keeling Islands, the plants of C. bonducella have evidently established themselves from drift seed in one locality in the Bermudas, they do not seem to have done so either on the shores of Krakatoa, or of St. Helena, where, although they are frequently washed ashore, Mr. Melliss never met with an instance of germination (see Bot. Chall. Exped. iv, 300, and Penzig). This is doubtless in part the result of the destructive efforts of the crabs, which, as I have shown in my paper on Keeling Atoll, nibble off the shoots of many germinating seeds in beach drift.

The readiness or non-readiness of seeds to germinate on a beach, and the nature of the conditions essential for the process, are matters that are directly concerned with their effective dispersal by currents. On account of the stony character of the seeds of these two species, it might be expected that germination would only take place under exceptional conditions. It should, however, be observed that the fine transverse striæ on their outer surface represent original fissures or cracks in the epidermis of the soft immature seed; and as such may be regarded as lines of weakness in the seed-tests. If a pod is opened before the seeds are mature, we find the seeds about twice the size of maturity, and so soft that they can be indented by the nail. The transverse striæ that mark the mature seed are displayed as indistinct cracks in the epidermis; and if the immature seed is exposed to the sun, in a few hours these cracks gape widely, and the seed has the grooved appearance of a top. If a pod opens prematurely on a plant, as sometimes happens, the immature seeds will be noticed with the epidermis scaling off. It is evident that the “setting” or the induration of the seed-coats and the final great contraction of the seed take place in the pod before dehiscence. From these remarks it would seem probable that seeds lying exposed to the fierce rays of the sun on a tropical beach would be liable to develop cracks along the old fissures, and that such cracks by permitting the entrance of moisture would favour germination.

My experiments show that high temperature under moist conditions will not of itself induce germination or in any way affect the seed. Thus in two sets of experiments, in 1890 and 1902, I failed to induce the germination of seeds which, after floating a year in sea-water, were kept in moist soil at a high temperature. In one case a temperature varying from 80° to 110° F. was sustained for several weeks, and in the other experiment a temperature of 80° to 90° was kept up for five months. When, however, an incision was made into the epidermis, or the seed-coats were partially penetrated with a file, the seeds swelled up in a day or two, and in a few days began to germinate.

The rapid transformation of the stone-like seed into a softened, swollen, germinating mass ranks amongst the numerous little wonders of the plant world. The seed, in fact, assumes again the appearance of immaturity, and in so doing it suggests to us that the rest-stage exemplified in the hard, pebble-like seed is but an adaptation to general climatic conditions, and that in a region of great heat and humidity, where there are no seasons, and where the sun’s rays are for ever screened off by mist and cloud, it could be dispensed with altogether. One of my Hawaiian dreams was to establish vivipary in Cæsalpinia bonducella by subjecting the maturing pod on the plant to very warm and humid conditions, my expectation being that the soft, swollen seed would at once proceed to germinate in the pod, and that the final process of setting, as indicated by the induration and contraction of the coats, or in other words the rest-stage, would be done away with. The dream, however, bore some fruits in enlarging my standpoint in the matter of vivipary, and I have referred to the subject in Chapter XXXI.

The seed-shell, about 1·5 mm. in thickness, consists of three coats: the outer skin very tough and waterproof; the inner skin seemingly permeable; and the intermediate layer of hard prismatic tissue, the “prismenschicht” of Schimper (p. 164). This middle layer absorbs water rapidly and in large quantity, so that if a fragment of the shell is placed in water it will be found after a day’s soaking to be three times as thick as it was in the dry state. If one files a seed, or makes a small incision, so as to expose the middle layer without piercing the inner coat, and then places it in water, it will be noticed that the middle layer at once begins to absorb water; and within a couple of days the whole seed will swell and attain the size it possessed in the so-called immature condition. During the process the outer skin stretches, usually without rupturing; and all three coats, previously so hard that a heavy blow with a hammer is required to break the seed, become in a day or two soft enough to be easily cut with a knife. The seeds thus treated swell in two days to three times their original size and increase their weight fourfold. Water finds its way to the nucleus or embryo partly through the dilated inner opening of the micropylar passage and partly through the inner skin. The nucleus then swells up into a fleshy mass, filling the seed-cavity, and in two or three days more germination begins.

I pass now to the discussion of the buoyancy of the seeds. Considering that both species occur in oceanic islands, and that the currents are active agents in transporting the seeds, their behaviour under experiment appears at first sight to be full of anomalies. Thus, it was ascertained at Kew (Bot. Chall. Exped. iv., 301), both with comparatively fresh and with older seeds, that those of Cæsalpinia bonducella floated in salt water, whilst those of C. bonduc sank; but in the record given of the experiment no mention is made of the original station of the parent plants; and it will be shown later on that the station of the plant, whether at the coast or inland, has an important determining influence on the buoyancy.

In Fiji I found that almost without exception the seeds of littoral plants of Cæsalpinia bonducella floated both in sea-water and in fresh water. On the other hand, in Hawaii the seeds of this species, obtained from three typical localities removed inland from the beach, sank without exception, even after drying for several months; and the only buoyant seed noted in these islands was a solitary seed collected from the beach drift. In Hawaii, however, as before remarked, the species is not strictly a littoral plant, occurring as it does in the lower levels, but not necessarily in the vicinity of the coast. In the case of seeds of littoral plants of C. bonduc in Fiji, I found that sometimes all floated in sea-water and sometimes only a portion of them, whilst their specific weight was on the whole rather greater than that of the seeds of the other species. Thus, in one experiment half the seeds floated in sea-water and a quarter in fresh water, whilst with seeds from another locality 90 per cent. of the seeds floated in sea-water and 80 per cent. in fresh-water; and in a third set of seeds all floated in both waters.

The above experiments on Fijian seeds all relate to littoral plants. In the instance, however, of the inland species from the mountains of Vanua Levu, all the seeds sank in sea-water, even after being kept for five years. If we follow the indications of these several experiments we shall find that Cæsalpinia presents another illustration of the general principle established in Chapter II that the seeds of inland plants sink and those of coast plants float.

My data, therefore, show that with the seeds of Cæsalpinia buoyancy goes with station and not necessarily with species. It is probable, therefore, that with the two widespread species, C. bonducella and C. bonduc, varying results will be obtained with seeds from different localities, whether insular or continental, according to the original station. The typically buoyant seeds of the former species may, as we have seen in Hawaii, lose their floating powers when they grow inland; and the seeds of an inland species from the mountains of Fiji sink at once. It is essential in interpreting the results of experiments on the seeds of these plants to be acquainted with the stations; and in this respect those of the Tahitian plants may be regarded as probable test cases. We have seen that in Tahiti, C. bonduc is an inland plant, and C. bonducella usually a beach plant; and I have no doubt that experiments in that island on the seeds of these two species from the particular stations just referred to will give results in agreement with the principle here laid down.

With reference to the duration of the floating powers of these seeds it may be observed that a seed of Cæsalpinia bonducella, originally found stranded on the beaches of Keeling Atoll, floated after a year in sea-water as buoyantly as at the commencement of the experiment. Seeds of Fijian littoral plants of both C. bonducella and C. bonduc floated in my experiments after two and a half years’ immersion in sea-water, showing no change whatever. Some of the seeds removed at the end of the first year were filed and placed in soil, when they germinated healthily. In Chapter IX it is pointed out that some buoyant seeds of other Leguminous plants, such as Mucuna urens, would be apt to germinate abortively and to sink in crossing the more heated areas of tropical seas. The seeds of Cæsalpinia, judging from my experiments and observations noted on page 84, seem to be quite proof against such risks. This was well brought out in an experiment where seeds of the two species of Cæsalpinia were kept afloat for two and a half years in a vessel of sea-water together with seeds of Mucuna and Strongylodon. None of the Cæsalpinia seeds attempted to germinate in the sea-water; but with the other genera some of the seeds began to germinate, and sank in the course of the first warm season, when the water-temperature ranged from 75 to 90° Fahr.

The seeds develop their buoyancy during the great contraction that, as before described, marks the final setting of the seed-coats and the ultimate maturation, as it is termed, of the seed. During this shrinking process the kernel also shrinks within the seed-tests, and cavities are thus produced within the seed-shell, on the relative size of which depends the buoyancy of the seed, neither the seed-shell nor the kernel possessing independent floating-power. These cavities, as illustrated in the figures given in Chapter XII, are of two kinds. That usually produced, being the one that mainly determines the buoyancy, is a large central hollow caused by the arching outwards of the cotyledons during the shrinking process, such as is found also in the seeds of Entada scandens, Mucuna urens, and some other Leguminous littoral plants. With such seeds the kernel never rattles when the seed is shaken, since the cotyledons lie in close contact with the seed-shell. The other kind of cavity is produced between the seed-shell and the kernel by the general or partial shrinking of the kernel away from the shell, the cotyledons remaining in apposition, as shown in the figures. When the shrinking away from the shell is general, the kernel lies loose within the shell, and the seed rattles when shaken. When the shrinking is partial the cavity is on one side of the seed and the kernel is fixed.