Dispersal is fundamental to biogeography and the evolution of biodiversity on oceanic islands


*Robert H. Cowie, Center for Conservation Research and Training, University of Hawaii, 3050 Maile Way, Gilmore 408, Honolulu, Hawaii 96822, USA. E-mail: cowie@hawaii.edu


Vicariance biogeography emerged several decades ago from the fusion of cladistics and plate tectonics, and quickly came to dominate historical biogeography. The field has since been largely constrained by the notion that only processes of vicariance and not dispersal offer testable patterns and refutable hypotheses, dispersal being a random process essentially adding only noise to a vicariant system. A consequence of this thinking seems to have been a focus on the biogeography of continents and continental islands, considering the biogeography of oceanic islands less worthy of scientific attention because, being dependent on stochastic dispersal, it was uninteresting. However, the importance of dispersal is increasingly being recognized, and here we stress its fundamental role in the generation of biodiversity on oceanic islands that have been created in situ, never connected to larger land masses. Historical dispersal patterns resulting in modern distributions, once considered unknowable, are now being revealed in many plant and animal taxa, in large part through the analysis of polymorphic molecular markers. We emphasize the profound evolutionary insights that oceanic island biodiversity has provided, and the fact that, although small in area, oceanic islands harbour disproportionately high biodiversity and numbers of endemic taxa. We further stress the importance of continuing research on mechanisms generating oceanic island biodiversity, especially detection of general, non-random patterns of dispersal, and hence the need to acknowledge oceanic dispersal as significant and worthy of research.


Since the dispersal versus vicariance debate of the 1970s and early 1980s and the broad acceptance and melding of the theories of cladistics and plate tectonics, vicariance approaches to historical biogeography have dominated the last two or three decades. Only vicariant mechanisms were considered to offer testable patterns and refutable hypotheses, dispersal being a random process essentially adding only noise to a vicariant system (Morrone & Crisci, 1995; Humphries & Parenti, 1999). Dispersalist explanations were treated as simply narrative, appealing to individual explanations and hence ungeneralizable; and a priori assumption of dispersal was considered unnecessary (Rosen, 1976; Peake, 1981; Lynch, 1989; Morrone, 2005). Dispersal was therefore of little interest and hence given but ‘footnote acknowledgment’ (Lynch, 1989) by many vicariance biogeographers.

A consequence of this thinking seems to have been a focus on the biogeography of continents and continental islands. Acknowledgement that the distribution of the plants and animals of oceanic islands is strongly moulded by dispersal, while at the same time considering that dispersal cannot be rigorously modelled or tested, meant that the biogeography of oceanic islands became almost by definition random and hence uninteresting.

Long-distance dispersal, however defined, certainly occurs frequently and is often directional (Nathan, 2005; and other papers in Diversity and Distributions volume 11, number 2). Although some historical biogeographers have explicitly acknowledged the importance of dispersal and incorporated it into their analytical approaches (e.g. Brooks & McLennan, 1991, 2001; McLennan & Brooks, 2002; Halas et al., 2005), for the most part they have not been searching for general non-random patterns of dispersal that can explain biogeographical diversity. They have rather simply acknowledged that dispersal happens in particular scenarios and should be part of a combined vicariant-dispersal explanation of the distributions of the taxa in question.

Recently, a number of articles, both in this journal and others (Givnish & Renner, 2004; Renner, 2004; Cook & Crisp, 2005; Halas et al., 2005; McGlone, 2005; de Queiroz, 2005), have argued that dispersal must be seriously considered as an important process in evolution and speciation, and that its role has been underestimated in historical biogeography. We agree wholeheartedly. However, much of this commentary has been focused on dispersal in a continental or continental island context, and while trans-oceanic dispersal has been considered, this has been primarily in the context of explanations for the occurrence of related taxa on widely separated continents, notably in the Southern Hemisphere (Givnish & Renner, 2004; Sanmartín & Ronquist, 2004; Cook & Crisp, 2005). In this editorial we want to take these arguments further and to emphasize strongly that in oceanic island situations in particular not only is dispersal important but it is the critical initiating step in the generation of endemic biodiversity, without which vicariant evolution within these islands could not happen, and furthermore that general patterns of directional dispersal can be detected. The emphasis over the last few decades on almost exclusively vicariant scenarios as providing the only rigorously testable hypotheses of historical biogeography, with dispersal considered as just random noise, has stifled important research on the role of dispersal in oceanic biogeography.

Vicariance and dispersal in oceanic island systems

The majority of the myriad islands of the Pacific have been formed in situ and not by the fragmentation of large land masses. As the Pacific tectonic plate moves steadily north-westwards, island arcs, such as the Marianas and some of the Fijian islands, are created at its western edge as it meets the tectonic plates to the west (Polhemus, 1996). However, most of the islands of the central Pacific have been formed as the plate moves over stationary hot spots in the underlying mantle, which from time to time send magma up through the plate, forming long chains of volcanoes, each volcano sequentially younger than the one that preceded it, and which will have moved north-westwards away from the hot spot (Price & Clague, 2002). This is how the Hawaiian Islands, the islands of French Polynesia (Tahiti, etc.), the Samoan Islands, and many others have been formed. Only few Pacific islands are of continental origin, for example New Caledonia and New Zealand. In the Atlantic, similar geological forces have created volcanic island chains such as the Canary Islands and the Madeiran archipelago.

Thus, the biogeography and endemic biodiversity of these oceanic islands, both arc and hot-spot islands, which have never had a connection to a continental land mass, are fundamentally a product of oceanic dispersal. That vicariance has played a role in the radiation of lineages is not to be denied, as for instance among the Hawaiian islands of Molokai, Maui, Lanai and Kahoolawe, the so-called ‘Maui-nui’ island group. Because of a combination of fluctuations in sea level and the dynamic processes of building, subsidence and erosion of these sequentially produced islands, the islands of Maui-nui have at various times been emergent volcanic peaks separated by sea-water channels, then combined into larger land masses, and finally broken again into a number of smaller islands (Price & Elliott-Fisk, 2004). Within certain lineages, intra-island speciation may also have been driven by various selective forces, such as sexual selection (Kaneshiro & Boake, 1987) and reproductive isolation brought about by fragmentation as deep valleys formed through erosion of previously continuous habitat (Holland & Hadfield, 2002). However, in the most fundamental sense, oceanic dispersal is the key to evolutionary radiation on these islands.

General patterns of dispersal

Strict vicariance biogeography has essentially considered dispersal as noise in the system, stochastic in nature, and therefore exhibiting no general patterns and permitting no refutable hypotheses, hence being ultimately uninteresting (Humphries & Parenti, 1999). But there are indeed general dispersal patterns in oceanic systems, related to ocean currents, predominant wind patterns (trade winds, hurricane tracks), the geographical arrangement of islands (facilitating use as stepping stones) and bird migration routes (Peake, 1981; Ballard & Sytsma, 2000; Hoskin, 2000; Givnish & Renner, 2004; Renner, 2004). Asymmetry of dispersal – the predominance of dispersal in one direction rather than another – has recently received particular attention as a potentially confounding factor for biogeographical reconstructions that ignore it (Cook & Crisp, 2005; McGlone, 2005). In addition, there are well-documented patterns in which taxa with inherently worse dispersal abilities are unable to colonize more geographically isolated islands (Peake, 1981; Whittaker, 1998), and disharmonic biotic distributions arise as a result.

Furthermore, to reinforce the notion that dispersal within hot-spot archipelagos is far from a stochastic process, we can point to many examples of studies in numerous plant and animal groups that have detected common patterns of dispersal that generally conform to the so-called ‘progression rule’ (Fig. 1) that dispersal and colonization, frequently accompanied by lineage bifurcation, proceeds from older to younger islands (Wagner & Funk, 1995; Roderick & Gillespie, 1998; Juan et al., 2000; Hormiga et al., 2003; Nepokroeff et al., 2003; Holland & Hadfield, 2004). As an island first appears above the surface of the ocean it is available for colonization, and its most likely colonizers come from the nearest land mass, which is the next youngest island in the chain. As the plate continues to move, new islands form and the process is repeated. Sometimes colonizing species pass over one or more islands in colonizing the newest island, sometimes there are back-colonizations from younger to older islands; although stochastic dispersal patterns have been observed (Wagner & Funk, 1995), particularly for species with a high dispersal ability such as flying insects, birds and plants with wind-dispersed seeds (Holland & Hadfield, 2004), the progression rule pattern holds broadly for many taxa.

Figure 1.

An example of a non-stochastic dispersal pattern observed for many plant and animal lineages, the progression rule pattern of island colonization, here depicted for a hypothetical Hawaiian island lineage. The volcanoes of the six main Hawaiian islands arose over a hot spot in the south-east. As the Pacific plate moves northwestwards it carries with it each sequentially formed island (island ages shown in parentheses, in millions of years). According to the progression rule, the initial colonization event occurs on the oldest island, Kauai, accompanied by subsequent lineage splitting as individuals disperse down the island chain from volcano to volcano (black circles). More complex patterns, involving radiations within islands, back-colonizations and dispersal that passes over an intermediate island, are often superimposed on the basic progression rule pattern.

Future directions in oceanic island biogeography

Nonetheless, we still know rather little about the dispersal processes and mechanisms that bring about the initial colonizations of these isolated archipelagos. We know, for instance, that there have been multiple colonizations of the Hawaiian Islands from more than one region of the Pacific rim (Gillespie et al., 1994; Rundell et al., 2004). In the marine realm, coastal organisms, especially those lacking a dispersal phase, behave rather like terrestrial organisms; but most marine dispersal in the Pacific has been outward from the centres of diversity in the East Indies (Briggs, 1999). While there have been numerous studies of within-archipelago phylogenetic diversification and biogeography, especially in Hawaii and the Canary Islands (see references above), there remain few comprehensive ocean-wide reconstructions of the phylogeny of any major widespread group of plants or animals, and, at least in the Pacific, little research addressing both geographical origins and routes of colonization ocean-wide. Some of the few examples include morphological research on a subgenus of Pacific blackflies (Simuliidae) (e.g. Craig et al., 2001; Spironello & Brooks, 2003) and molecular work on the Pacific tree subgenus Metrosideros (Wright et al., 2000).

The downplaying of the role of dispersal may have arisen in the past in part because of difficulty in resolving the underlying evolutionary histories within and among island species. Recent studies that have revealed non-stochastic dispersal and colonization pathways have done so largely as a result of the development of new laboratory techniques and multivariate statistical methods, and in particular the availability of polymorphic molecular markers. DNA-based approaches are revolutionizing our ability to tease apart and reconstruct otherwise cryptic biogeographical pathways.

Although the oceans cover nearly three-quarters of the earth's surface, the combined land area of all oceanic islands represents a miniscule fraction of the earth's total land area. For instance, the islands of the Pacific (excluding continental New Guinea and New Zealand) have a total area of about 106,000 km2, approximately the area of Guatemala and less than 0.1% of the earth's total land area. Yet the terrestrial biotas of these islands are immensely diverse: for example the diversity of land snails in the Hawaiian Islands (Cowie, 1996a) is comparable to that of the whole of North America north of Mexico (Pilsbry, 1938–1948) and exhibits unrivalled endemism, exceeding 95%. This oceanic island biodiversity has provided biologists since Darwin with fundamental insight into the processes and patterns of evolution (Emerson, 2002). Evolution of the biotas of oceanic islands is therefore of great interest and significance, and it is important that the major force that has shaped these biotas, namely dispersal, should not be dismissed but acknowledged and researched all the more intensely. Phylogenies are the prerequisite for ascertaining the detailed pattern of dispersal and colonization that has resulted in the huge diversity among many widespread groups of organisms on oceanic islands. Developing these comprehensive phylogenies is a wide-open opportunity for research that will lead to major advances in our understanding of the biogeography of an important component of the earth's biodiversity. Also, by re-emphasizing the role of dispersal in historical biogeography, we see an opportunity, although not an easy one (McGlone, 2005), to test hypotheses of its non-randomness, thereby refining our biogeographical interpretations with a more holistic approach that perhaps will better reflect biological reality.

Our current research on succineid land snails aims to do this, by developing and testing phylogenetic hypotheses for the entire family in the Pacific, permitting us to infer the pathways via which these snails have colonized the islands from continental regions, by large trans-oceanic jumps or in a stepping-stone manner from island group to island group, gradually making their way across the ocean. The patterns probably do not follow oceanic currents, as most land snails would be unlikely to survive such journeys faced with long-term exposure to salt water. But they may follow bird migration routes – succineids have been recorded attached to birds travelling long distances (Anonymous, 1936; Rees, 1965; Boag, 1986). And they may follow generalized hurricane pathways – snails can fly (Kirchner et al., 1997), perhaps especially if attached to a leaf. Such dispersal mechanisms have often been suggested for snails (e.g. Vagvolgyi, 1975; Peake, 1981; Hausdorf, 2000).

An estimated 29 successful colonizations are necessary to explain the diversity of Hawaiian land snails (Ziegler, 2002), which translates into roughly one every million years, if the oldest colonization was to the island of Kure, or one every 175,000 years if the oldest colonization was to Kauai (Price & Clague, 2002). On average, hurricanes with maximum wind speeds of 64 knots (c. 119 km h−1) (minimum hurricane intensity) and maximum wind speeds of 125 knots (c. 232 km h−1) come within 250 nautical miles (463 km) of the Hawaiian Islands every 7 and 137 years, respectively, with hurricanes of intermediate wind speed coming at intermediate frequencies (Chu & Wang, 1998). That snails, with some exceptions that have evolved in situ (Cowie, 1996a), tend to be smaller the more isolated the island offers support to the idea that birds and the wind are the more important mechanisms (Peake, 1981; Cowie, 1996b).


Thus, the immense diversity of land snails and other organisms on oceanic islands (e.g. Ziegler, 2002) is fundamentally dependent on dispersal, and that dispersal probably exhibits patterns both on a small intra-archipelago scale and on a larger ocean-wide scale. These patterns remain poorly understood but deserve considerable research attention.

We hope, therefore, that the trend identified by de Queiroz (2005)– the resurrection of oceanic dispersal as important in historical biogeography – is real and that the straightjacket of strict vicariance biogeography is being loosened to include once again the plurality of mechanisms and processes that make evolutionary biology the exasperating but ever fascinating discipline that it is. At least in the Pacific, there are excellent opportunities with numerous plant and animal groups not only to address the origins and diversification of this important component of the earth's biodiversity but also to contribute profoundly to the advancement of the discipline of biogeography.


Our research on Pacific island biogeography is currently supported by the US National Science Foundation, grant DEB-0316308. We thank Richard Field for helpful comments on a draft of this article.


Robert H. Cowie is an Associate Researcher in the Center for Conservation Research and Training at the University of Hawaii. His research addresses issues related to the origins and determinants of biodiversity and its conservation, as well as the origins and impacts of introduced species. His primary focus is on non-marine snails in the islands of the Pacific and freshwater snails in South and Central America.

Brenden S. Holland is a Junior Researcher in the Center for Conservation Research and Training at the University of Hawaii. His research uses molecular techniques to address phylogenetic and phylogeographical questions in Pacific island biodiversity, focusing primarily on molluscs, both marine and non-marine.

Editors: David Richardson and Robert Whittaker