by Fabian Acker

October 1996

Summary  

Evolutionary paths on islands differ from those on continents, these differences becoming more pronounced the further the islands are from mainland sources. The size and defined geographical boundaries of islands have engendered the term "evolutionary laboratories", and in this context, useful insights into theories of evolution have been gained.

However, the size, geography and climate vary enormously from island to island and within islands themselves and explanations appropriate to one island or group, may not be applicable to another. So it is important, although not always straightforward, to separate those mechanisms which distinguish islands from mainlands, and those mechanisms which are unique to islands.

While it may be possible to refine and test evolutionary theories in general with the study of islands, many specific questions have also been raised about island development, the answers to which are not always clear. For instance, apparently similar factors such as a limited diversity of pollinators may cause plants to evolve different reproductive strategies and may often exhibit rapid divergence.

Isolated territories such as savannahs surrounded by forest, or constrained habitats such as lakes, also have characteristics comparable with islands in their limited genetic diversity and clear boundaries. The range of fauna and flora discussed included snails, lizards, butterflies, plants, deer, mice, sheep and fish.

In opening the meeting, Professor Spencer Barrett of the University of Toronto, Canada, noted the paucity of pollinators on islands compared with mainland areas. As an example, he pointed out that in the Hawaiian Islands only 15% of known families of insects were represented, with only six native species of hawkmoths, two species of butterflies and no bumble bees.

A similar lack of pollinators is common to most islands, and flies, beetles, wasps and other insects, which can migrate more easily than larger insects, carry out much of pollination, rarely discriminating between types of plant. As a result, the opportunities for specialised plant/pollinator relationships to evolve are very limited.

As evidence of this situation, Barrett points to the work of several investigators who have shown that island flowers are predominantly white or dull in colour and have simple access routes to their pollen. Most New Zealand flowers, for instance, are small, white and simply shaped, whereas many of those in Australia are coloured brightly and are elaborately shaped.

Further evidence in favour of the influence of the pollinator on the evolution of flower shape and size is the direct relationship observed between the size of the flower and that of the pollinator.

The effect of wind on pollination may also affect adaptive strategies, he says, and several researchers argue that it can be highly effective because it is independent of insects and animals. Also where it is strong and pervasive, it promotes pollen dispersal and cross-fertilisation. But he says that more research is needed to determine its role more fully.

Reproduction strategies in plants are also affected by the same factors and one major question that still needs to be answered is how self-fertilisation as against cross-fertilisation is favoured. Although there are still not enough data to draw firm conclusions, evidence so far substantiates the theory which says that self-fertilising plants will be more successful than cross-fertilising ones in island conditions.

He gave a number of examples to support this view. In Greece and Turkey, for instance, there are 14 species of Nigella ("love-in-a-mist") which depends mainly on cross-fertilisation, whereas in the Aegean Islands, there are only two, and they are self-fertilising. A similar pattern exists in the Kikladhian Islands, where the largest islands have a range of self-fertilising plants and the smaller ones do not.

Nevertheless, the genetic advantages of cross-fertilisation and the evolution of different forms can, in certain islands such as Jamaica, allow for the existence of both kinds of fertilisation. However, the mechanisms which have given rise to these situations, and in particular the characteristics of the founder plants, are not yet clear.

Although two island archipelagos, the Caribbean and the Hawaiian, began their formation at about the same time, they are widely different in the way in which they were formed, and in their proximity to the mainlands. As a result, evolutionary paths are also different, and this difference, according to Professor Hope Hollocher, of Princeton University, USA, provides a useful way of studying the mechanisms of speciation (that is the divergence of a common ancestral group into different species). Her work has concentrated on the fruit fly Drosophila.

The ages of the Hawaiian Islands range from that of Kauai (5-6 million years) to Hawaii itself (0.5 million years). Yet of the 500 species of the Hawaiian Drosophilidae family already named, and a further 350 to be named, evidence points to just two genera as the ancestors of the present fauna, which began diverging 24 million years ago.

This suggests that the process began on islands that have already submerged, and research in various fields including chromosome analysis supports the idea that speciation progressed from older to younger islands, and in some cases, from older parts to younger parts within islands following or in conjunction with fauna colonisation.

One established theory, originally proposed by the zoologist Ernst Walter Mayr, suggested that so-called founder events were engines for speciation. The basis of the theory is that the original colonisers, being few in number, would carry only a small proportion of the variability of the whole population. In addition to having a limited gene bank, there would be a high level of inbreeding. In such a case, some traits might be lost, while others might be favoured. This could lead to rapid changes in the descendent population, as it would be influenced by quite different selection pressures compared with its antecedents.

None of the Caribbean islands have the same level of isolation from the mainland as the Hawaiian group, and the Caribbean Drosophilidae are closely related to those on the Central American mainland.

Their formation was chaotic compared with the Hawaiian group’s, which has been likened to a conveyor belt system, with new islands being pushed up by subsea volcanic activity, and moved by plate tectonics, while older ones were submerging. The Greater Antilles were formed about 80 million years ago and moved to their present site about 50 million years ago. Islands were pushed together, and later forced apart during that time. Jamaica was completely under water about 30 million years ago.

The Lesser Antilles appear to have developed where they are today, and water levels between the islands are high, and there is no evidence of land bridges between the group and South America. It is assumed, therefore, that colonisation of the Lesser Antilles must have been over water.

Dr Josephine Pemberton and others from the University of Edinburgh described a study of red deer on the Isle of Rhum in the Inner Hebrides, and Soay sheep on the island of Hirta of the St Kilda group in the Outer Hebrides, to examine the assumption that small isolated populations tend to lose genetic diversity particularly when the population fluctuates.

The reasoning behind this view is that the surviving individuals (of a declining population) do not carry the wide range of genes that is present in the full population, and hence their offspring will be similarly limited.

But she points out that this assumption does not necessarily take into account any selection processes during a population reduction, and this could retard genetic loss. In addition, it is often assumed that mating frequency remains constant irrespective of population fluctuations, implying that a declining population does not stimulate an increased mating frequency and vice versa, although there is not much evidence for or against this view.

In fact, the study indicated that both these assumptions are flawed. It was shown that selective processes are present in fluctuating populations, thereby slowing the loss of variation, and in one of the populations, mating patterns also vary systematically in such a way as also to retard genetic loss.

Small mammal differentiation in islands, the subject of a paper by Professor R J Berry of University College London, assessed the merits of the founder principle in explaining the different ways in which small mammals, such as mice, rats and voles, evolved on a number of islands, in particular those around the United Kingdom. He said that there was no evidence to suggest this principle accounted for their differentiation.

Berry maintained that many "myths" based on dubious arguments have been generated concerning some of the small mammals living in the islands near Britain, giving as one of a number of examples, the view that moles on Guernsey and the Orkneys, had a common ancestor, which, through isolation, differentiated to their present forms, when the "land bridge" between Orkney and the mainland was broken.

He pointed out that there was no evidence of a "land bridge" between Orkney and Scotland after the retreat of the Pleistocene ice, and that the Orkney voles almost certainly colonised the islands much later when the first humans arrived. Therefore, theories based on the assumption that voles spread through south-eastern England from Guernsey and into the Orkneys were based on a false premise. He quoted evidence to show that the Guernsey voles have more in common with those of northern Europe, and those in the Orkneys are closer to those in southern Europe, particularly in the Iberian Peninsula.

He gave other examples of mistaken assumptions about small mammals in islands near the United Kingdom, all of which led to what he considered to be the erroneous hypothesis that isolation automatically leads to differentiation. He contended that the significance of founding populations was that they provided a mechanism for rapid or sudden changes in gene frequency (because they had been introduced suddenly into a new environment) and this provided a basis for differentiation.

He argued that, in general, many island populations arose from colonisation rather than the presence of relicts (that is populations remaining from when islands were still attached to the mainland).

The speciation of birds was discussed by Professor P R Grant and B R Grant from Princeton University, USA. They took Darwin’s finches on the Galapagos Islands as the basis for their study, because these birds, unlike many of those on other islands, have not apparently had any of the species that developed from one parent group driven to extinction.

In one proposed model for the speciation process there are three stages: in the first, a population is established by immigrants from the mainland. In the second, some of the birds fly to another island and evolve characteristics more favourable to the new conditions. This may be repeated several times. The third stage is when contact occurs between members of two groups.

If interbreeding is successful, then speciation has not yet occurred. Alternatively, interbreeding may reduce fitness, and the offspring do not survive. Eventually selection may reinforce the differences between the two groups, and speciation will have taken place.

But they argue that hybridisation does not always reduce fitness and the offspring may have characteristics such as longer beaks, which are advantageous in a particular environment. This process could also contribute to speciation.

Two problems arise in this proposed model. One is that the potential to hybridise for birds can remain for, on average, 22 million years after speciation, and the other is that almost one in 10 bird species have hybridised at least once.

Another factor which has only recently been considered, is the evolutionary timescale of Darwin’s finches and their descendants. The starting point of speciation had been assumed to be half a million years ago, but has recently been revised to 2.8 million years ago.

This means that the geological changes in the Galapagos chain must now be taken into account in the evolution of these birds, because during that time, some islands have been submerged and others have been formed (possibly by splitting). Until recently, it had been assumed that the geography of the various islands had been constant during the finches’ evolution.

The paper by Professor Brian Clarke of the University of Nottingham and others, Speciation of Land Snails on Islands, warned against the possibility of gaining misleading inferences from molecular data in attempting to elucidate times or the mechanics of divergence. The basis of the paper was the evolutionary history of a Pacific snail (the Partula genus), and in particular its development on one of the small islands in the Société group, of which Tahiti is the most well known.

Two species of the Partula genus exhibit unusually different characteristics, one being right-handed in that the aperture of the shell and genitals are on the right-hand side of the snail, whereas the other form is a mirror image of the first, with aperture and genitalia on the left. Hybridisation is not impossible, but difficult, and therefore the two forms tend to remain distinct.

Each one tends to occupy a particular ecological niche on the island, one living on shrub leaves, the other on tree trunks.

The prevailing climate in the north of the island is dry, and the snails of one group tend to take a squat rounded form while in the south, where it is more humid, the form is long and slender. It was assumed that the dry and wet micro-climates favoured the different shapes, the squatter ones being better able to conserve water.

However, another group of snails of the same genus exhibits characteristics which are exactly opposite (that is squat and rounded in the humid climate, and long and slender in the dry climate) which suggests that climate may not have been the major influence.

A number of tentative suggestions were put forward to account for the different ways in which the two groups evolved, but what emerged clearly from the study, is that where the snails were similar to each other in shape and size, that is in the areas intermediate between the two micro-climates, their enzymes were substantially different; where the physical differences were quite distinct, the enzyme differences were small.

The evolutionary history of cichlid fish species in the East African lake of Tanganyika provided the basis of a paper by Professor A Meyer et al of the State University of New York, USA. These fish have evolved into many species with a wide range of behavioural and physical traits, more than any other vertebrates according to Meyer, and therefore offer a rich field for research into speciation.

Lake Tanganyika is between 9 and 12 million years old, which implies its cichlid species are older and more genetically distinct than the other two great African lakes, Victoria and Malawi. Geological evidence suggests that in Pleistocene times (about 200 000 years ago), the lake was split into three isolated ones, and Meyer attempts in his paper to distinguish between the effects of the geographic isolation (either because of the supposed split, or within the lake itself), and the biological development within the lake on the evolution of species.

For example, the varying water level, which is a feature of the lake, leads to pools isolated by beaches or rocks, which accounts for cichlids which have successfully adapted to this shallow habitat, specifically in the size of their swim bladders and tooth shapes. If the conditions last long enough, another species could develop.

These conditions are not tied to one locality, and in one instance it was observed that at two similar locations 20 km apart, the cichlids in both shared identical characteristics.

Trying to differentiate between the relative influences of geological changes, and selection on the evolution of species, was the subject of a paper by Professor Roger Thorpe and A Malhotra of the University College of North Wales. They pointed out the difficulties, for instance, of determining whether differences such as colour pattern or body size in the Canary Island lizards, could be attributed to founder effects and drift, or adaptation to the different ecological conditions on the various islands.

DNA sampling and molecular analysis can give a fuller explanation of speciation than inferences from geological events alone; if they are used in conjunction, many evolutionary processes can be determined with a high degree of certainty.

As an example of the process, he referred to the lizard population on the western Canaries, which had not been joined to one another or to the mainland. Therefore changes to the lizards (or any other flora and fauna for that matter) must have occurred following colonisation of each island. DNA testing showed that the divergences which arose corresponded in sequence to the geological origin of each island, displaced by an appropriate time interval, validating the value of such an approach.

Professor D Schluter, University of British Columbia, Canada, considered the role of reproductive isolation and contrasting selection pressure in leading to speciation. He said that these factors, commonly accepted as engines for speciation, had rarely been tested, which encouraged him to see what evidence could be found to support this view.

He directed his investigation to the evolution of fish in some North American lakes that had been formed after the last Ice Age, that is, the lakes were less than 15 000 years old. The geological history of the area meant that these lakes had the characteristics of islands in that the fauna were limited but varied and the environment was self-contained and independent.

He found that fish in the lakes shared four characteristics: rapid evolution into non-mating forms, retention of their characteristics despite constant gene flow, a high degree of differentiation (sticklebacks, for instance, evolved into plankton-eating forms and plant-eating forms), and a high fertility of hybrids.

Because of the similar evolutionary paths taken by species in lakes, that were widely separated but similar in ecological constraints, he concluded that the argument for the accepted view was strong, although more field studies were needed to explore the role of ecological environments in speciation.

Sir Ghillean Prance discussed the island characteristics of areas within the Amazonian rain forest, where savannahs isolated by forest, and high granite outcrops (known as inselbergs), were home to distinct species of flora and fauna differing from their surroundings and in many cases from each other. In both, the dense rain-forest acts as an impediment to migration, and the isolated areas develop their specific ecologies.

Most inselbergs range in height from 300-800 m, and the upper reaches become dry in arid periods. As a result, many plants have evolved to deal with drought, and it has been estimated that 55% of plant species on inselbergs are specific to those habitats, and not to the surrounding rain-forest.

The amount of speciation between these areas is similar in character to that between islands in the same chain, in that the older the areas, the greater the degree of differentiation. Some savannahs, for instance, appear to have been formed from larger ones, which split as the climate changed from dry to wet, and therefore have a common heritage. On the other hand, one particular category has generally retained its character as it is based on white sand with little nutrient value. As a result, even when climatic changes favour the expansion of the rain-forest, these sand-based savannahs still tend to remain isolated.

Professor J R G Turner of the University of Leeds and L B Mallet of University College London discussed the evolution of poisonous or harmful butterflies displaying "warning" colours. "Palatable" butterflies develop other strategies; unpalatable ones simply need to maintain a consistent pattern which is remembered by predators. Five or six distinct patterns can exist simultaneously.

The paper too used examples from "islands" in the South American rain-forests, although not the ones considered by Prance, but others isolated from the forest and self-contained by specific and localised ecology.

The authors suggest that evolution can take place either as a result of a random change in the genetic structure of the plant or animal under study, or as an adaptive response to random changes in the environment, or both.

The value of "ring capture" is explored also as explanation. This is the name given to the mimicking process by which butterfly coloration may converge to a common pattern. Where any two groups are similar to each other in coloration, one group may eventually "capture" the other, if for instance, the capturing group is more distasteful or more numerous than the captured one. But where the patterns are widely different, there will be no "capture", as the mutations of any one group would have to go through too many random changes to mutate to the other, and butterflies of distinctly different pattern will co-exist successfully.

If, by random changes, individuals move out of a successful ring, with a slight modification to the parent pattern, even though they are equally harmful, they are immediately at risk to predators, and are unlikely to evolve further.

Changes will occur more extensively when the so-called islands are isolated for long periods. Random changes in the flora can lead to extinction of what was once a successful butterfly, and this in turn leads to more pronounced changes due to the subsequent loss of predators and host plants and so on.

The area may be recolonized by another population, or, when it again coalesces with other similar areas following major climatic change, but the earlier species becomes extinct.

In his paper on the Anolis lizard on the Caribbean islands, Dr Jonathan Losos, argued for the value of studying a narrowly defined group (of plant or animals) as against a broader study with a wider range, to gain a better understanding of evolutionary forces.

So far, 138 species of the lizard have been identified, with as many as 54 on one island. They are also present on islands 250 miles away from Cuba, one of the Caribbean islands, although 85% of them are endemic to a single island or island bank.

There is a clear relationship between island size and number of species; as islands expand or shrink because of geological forces, habitats will also expand or be lost (or disappear), and species dependent on a particular niche will become extinct. For instance, some small islands in the Bahamas have little vegetation and only one species of lizard. Larger islands, with more and a greater variety of vegetation, have more species.

Colonisation, when conditions are appropriate, will probably restore the "lost" species, but on very large islands the process of speciation is different; the study postulated that there might be a critical size below which speciation would not occur, suggesting that the minimum area would be less than Jamaica (4 450 miles²), but larger than Great Abaco (776 miles²).

The study considers a number of ways in which speciation and habitat are connected in the larger islands. One hypothesis is that there exists a core of six types of lizard on most or all of the islands, each of which specialises in a particular habitat. These diversify into some which develop in their own particular areas, leading to reproductive isolation, and some which develop in the same area, occupying different habitat niches.

The reproductively isolated groups provide a pool which provides enhanced opportunities for further diversification.

Professor Spencer Barrett

Department of Botany

University of Toronto

25 Willcocks Street

Toronto

Ontario M5S 3B2 tel: 00 1 416 978 2011

CANADA fax: 00 1 416 978 5702

Professor Hope Hollocher

Department of Ecology and Evolutionary Biology

Princeton University

Princeton

NJ 08544-1003 tel: 00 1 609 258 6742

UNITED STATES OF AMERICA fax: 00 1 609 258 1712

Dr Josephine Pemberton

Institute of Cell, Animal and Population Biology

University of Edinburgh

West Mains Road tel: 0131 650 5505

Edinburgh EH9 3JT fax: 0131 667 3210

Professor R.J. Berry

Department of Biology

University College London

Gower Street tel: 0171 380 7170

London WC1E 6BT fax: 0171 380 7096

Professor Peter Grant F.R.S.

Department of Ecology and Evolutionary Biology

Princeton University

Princeton

NJ 08544-1003 tel: 00 1 609 258 5156

UNITED STATES OF AMERICA fax: 00 1 609 258 1334

Professor Brian Clarke F.R.S.

Department of Genetics

School of Biological Sciences

Queen’s Medical Centre

Clifton Boulevard tel: 0115 970 9397

Nottingham NG7 2UH fax: 0115 970 9906

Professor A. Meyer

Department of Ecology and Evolution

State University of New York

Stony Brook

NY 11794-5245 tel: 00 1 516 632 7434

UNITED STATES OF AMERICA fax: 00 1 516 632 7626

Professor Roger Thorpe

School of Biological Sciences

University of Wales

Bangor tel: 01248 382312

Gwynedd LL57 2UW fax: 01248 371644

Professor D. Schluter

Department of Zoology and Centre for Biodiversity

University of British Columbia

Vancouver V6T 1Z4 tel: 00 1 604 822 2211

CANADA fax: 00 1 604 822 5785

Professor Sir Ghillean Prance F.R.S.

Royal Botanic Gardens

Kew

Richmond tel: 0181 332 5111

Surrey TW9 3AB fax: 0181 948 4237

Professor J.R.G. Turner

Department of Genetics

University of Leeds tel: 0113 243 1751

Leeds LS2 9JT fax: 0113 244 1175

Professor Jonathan Losos

Department of Biology

Campus Box 1137

Washington University

St Louis

MO 63130-4899 tel: 00 1 314 935 6706

UNITED STATES OF AMERICA fax: 00 1 314 935 4432

Enquiries to: Ref: PR 44 (96)

Miss Anna Link

Science Promotion Section

The Royal Society

6 Carlton House Terrace

London

SW1Y 5AG

Direct line: 0171 451 2581 4 October 1996

On 6 and 7 December 1995, the Royal Society held a scientific meeting on Evolution on islands. The enclosed document was prepared afterwards to summarise key issues raised by the speakers and to provide a list of helpful contacts for future reference.

The document does not necessarily constitute the views of the Royal Society and views expressed in it should not be attributed to the Society. The document is free of copyright and may be used without reference to source.