Changing perspectives on the biogeography of the tropical South Pacific: influences of dispersal, vicariance and extinction

Authors

  • Gunnar Keppel,

    Corresponding author
    1. Ecology Centre, School of Integrative Biology, University of Queensland, St. Lucia, Brisbane, QLD
      *Gunnar Keppel, Biology Division, Faculty of Science and Technology, University of the South Pacific, PO Box 1168, Suva, Fiji. E-mail: keppel_g@usp.ac.fj
    Search for more papers by this author
  • Andrew J. Lowe,

    1. Australian Centre for Evolutionary Biology and Biodiversity, School of Earth and Environmental Sciences, University of Adelaide, Adelaide, SA, Australia
    Search for more papers by this author
  • Hugh P. Possingham

    1. Ecology Centre, School of Integrative Biology, University of Queensland, St. Lucia, Brisbane, QLD
    Search for more papers by this author

*Gunnar Keppel, Biology Division, Faculty of Science and Technology, University of the South Pacific, PO Box 1168, Suva, Fiji. E-mail: keppel_g@usp.ac.fj

Abstract

Aim  The biogeographical patterns and drivers of diversity on oceanic islands in the tropical South Pacific (TSP) are synthesized. We use published studies to determine present patterns of diversity on TSP islands, the likely sources of the biota on these islands and how the islands were colonized. We also investigate the effect of extinctions.

Location  We focus on oceanic islands in the TSP.

Methods  We review available literature and published molecular studies.

Results  Examples of typical island features (e.g. gigantism, flightlessness, gender dimorphism) are common, as are adaptive radiations. Diversity decreases with increasing isolation from mainland sources and with decreasing size and age of archipelagos, corresponding well with island biogeographical expectations. Molecular studies support New Guinea/Malesia, New Caledonia and Australia as major source areas for the Pacific biota. Numerous studies support dispersal-based scenarios, either over several 100 km (long-distance dispersal) or over shorter distances by island-hopping (stepping stones) and transport by human means (hitch-hiking). Only one vicariance explanation, the eastward drift of continental fragments (shuttles) that may have contributed biota to Fiji from New Caledonia, is supported by some geological evidence, although there is no evidence for the transport of taxa on shuttle fragments. Another vicariance explanation, the existence of a major continental landmass in the Pacific within the last 100 Myr (Atlantis theory), receives little support and appears unlikely. Extinction of lineages in source areas and persistence in the TSP has probably occurred many times and has resulted in misinterpretation of biogeographical data.

Main conclusions  Malesia has long been considered the major source region for the biota of oceanic islands in the TSP because of shared taxa and high species diversity. However, recent molecular studies have produced compelling support for New Caledonia and Australia as alternative important source areas. They also show dispersal events, and not vicariance, to have been the major contributors to the current biota of the TSP. Past extinction events can obscure interpretations of diversity patterns.

Introduction

The biology of islands has long fascinated naturalists, biologists and biogeographers, and their study has led to many fundamental advances in ecology and evolution (Darwin, 1859; MacArthur & Wilson, 1967; Rolett & Diamond, 2004). Around the middle of the last century, islands of the tropical South Pacific (TSP) were the subject of much interest (van Steenis, 1952; Smith, 1955; Thorne, 1963; van Balgooy, 1971), although it is still the case that surprisingly little is known about their biogeography and ecology. The tropical South Pacific (TSP) is here considered to comprise the oceanic islands of Melanesia, Polynesia and Micronesia (Fig. 1), which are of different sizes, elevations and ages, and are widely distributed (Table 1), thus representing an ideal place to investigate biogeography and the comparative ecology of island systems. Recent advances in phylogenetic molecular techniques have generated renewed interest in South Pacific biogeography. However, most contemporary effort has focused on the predominant processes influencing New Zealand and New Caledonia (Pole, 1994, 2001; Winkworth et al., 2002, 2005; Heads, 2006a, 2008; Ladiges & Cantrill, 2007; Grandcolas et al., 2008). Much less attention has been devoted to the TSP.

Figure 1.

 The Pacific Ocean, showing the biogeographical areas of Malesia, Melanesia, Micronesia and Polynesia (after Mueller-Dombois & Fosberg, 1998). New Caledonia, Hawaii and New Zealand are not included in Polynesia or Melanesia because of their unique biotas. BA, Bismarck archipelago; CAL, New Caledonia; FIJ, Fiji; NG, New Guinea; SOL, Solomon Islands; VAN, Vanuatu.

Table 1.   Size, age and maximum elevation of archipelagos in the Southwest Pacific.
CountryTotal landmass (km2)Age (Myr)Elevation (m)Source
  1. Note that age refers to the time the archipelago is assumed to have become emergent.

  2. *Most of the present landmass only became emergent within the last 2 Myr.

New Guinea786,000> 1004884Bain (1973)
Bismarck Archipelago49,730c. 352150Wiebenga (1973)
Solomon Islands38,210c. 252331Hackman (1973); Petterson et al. (1999)
Vanuatu12,189c. 25*1878Mallick (1975);  Greene et al. (1994)
Fiji18,225c. 351324Gill (1970)
Samoa3040c. 51858Cribb & Whistler (1996)
Tonga691c. 41046Sager et al. (1994)
New Caledonia18,576> 1001629Avias (1973)

In this review of the biogeography of oceanic (non-continental) islands in the TSP, we synthesize relevant literature and evaluate recent molecular studies to investigate contemporary patterns of diversity, where the biota came from, modes of island colonization and the role of extinction. The latter process is often overlooked in biogeographical interpretations. We cover the range of evidence to support the major biogeographical theories for the region and finish by highlighting areas of likely fruitful future research. We focus in particular on whether data support the propositions that: (1) Malesia and Southeast Asia are indeed the major source areas of the biota on TSP oceanic islands; (2) dispersal is indeed the major process driving the colonization of oceanic islands in the TSP; and (3) there are species now persisting only in the TSP, which have become extinct in their source areas.

The TSP constitutes an area of c. 40 million km2, of which only c. 150,000 km2 are currently above sea level. Its major archipelagos are the Bismarck Archipelago, Solomon Islands, Vanuatu, New Caledonia, Fiji, Samoa and Tonga, all of which, except New Caledonia, are entirely oceanic. New Caledonia is a continental fragment but, like New Zealand, much of its present landmass was submerged during the Oligocene (Yan & Kroenke, 1993; Petterson et al., 1999; Sutherland, 1999; Schellart et al., 2006; Trewick et al., 2007). The climate on most archipelagos is tropical, and tropical rain forest is the predominant native vegetation type (Mueller-Dombois & Fosberg, 1998). Several continental landmasses (Asia, New Guinea and Australia) border the TSP, but in the east there is mostly open ocean, with the Americas being some 10,000 km east of Tonga (Fig. 1).

Reconstructing the tectonic and geological history of the South Pacific has been problematic (Rodda & Kroenke, 1984; Yan & Kroenke, 1993; Kroenke, 1996; Hall, 2001; Neall & Trewick, 2008). The Melanesian archipelagos are believed to have originated within the last 40 Myr (Table 1) in a purely oceanic setting through volcanic activity caused by the collision of the Australian and Pacific plates. Although the exact mechanisms are still debated, continuing volcanic activity, tectonic uplift, submergence and plate rotations have interacted to produce complex patterns.

The biota of the Pacific islands displays characteristics typical of island inhabitants (Carlquist, 1966a, 1974), such as loss of flight in birds (megapodes and rails; Worthy, 2000, 2004) and insects (Zimmerman, 1942; Gillespie & Roderick, 2002), gigantism across a range of groups, loss of dispersibility and increased incidence of separate sexes in plants (Carlquist, 1966b,c; Whittaker & Fernández-Palacios, 2007). Few examples of gigantism persist in the present-day Pacific fauna, besides the giant Fijian longhorn beetle, Xixuthrus heros, which is one of the largest known insects (Yanega et al., 2004), and the giant rats of the Solomon Islands (Flannery, 1995). However, pre-human fossils document giant megapodes (Worthy, 2000), pigeons (Worthy, 2001), frogs and iguanas (Pregill & Worthy, 2003) from Fiji, and giant turtles from New Caledonia (Gaffney et al., 1984). These past giants were probably rapidly driven to extinction by early human colonizers (Steadman, 1995; Pregill & Worthy, 2003). Dioecy, a form of sexual dimorphism, is relatively high on Pacific islands, at both the species (Baker & Cox, 1984) and the genus (Webb et al., 1999) level, generally ranging from 10% to 25%. This compares with < 5% on the continental mainland in temperate regions (Baker & Cox, 1984), such as the British Isles (Clapham et al., 1962) and California (Freeman et al., 1980), but is similar to the levels of dioecy in tropical regions of the continental mainland (Gross, 2005).

Despite these island generalities, a biologist acquainted only with continental settings would probably be most struck by the absence of certain taxa and functional groups on the Pacific archipelagos. Good examples are the absence of native, flightless mammals east of the Solomon Islands (Flannery, 1996), and the absence of certain successional guilds of rain forest trees (Mueller-Dombois, 2008). Another phenomenon on islands is the presence of relictual taxa (Whittaker & Fernández-Palacios, 2007), such as the Degeneriaceae in Fiji (Bailey & Smith, 1942).

What are the patterns of diversity?

Species diversity

As predicted by the theory of island biogeography (MacArthur & Wilson, 1967), species diversity in the TSP tends to increase with island area, maximum elevation and proximity to the source area (van Balgooy, 1971; Diamond & Mayr, 1976; Diamond et al., 1976). This is supported by data from several taxonomic groups (Table S1 in Supporting Information, Fig. 2, van Balgooy, 1971; Whitmore, 1973; Stoddart, 1992). As one moves away from the inferred Malesian source area, islands in the TSP have lower species diversity. This pattern appears to be to the result of both isolation and area effects, specifically the differential dispersal abilities of taxa and a general decrease in the size of archipelagos moving east. New Guinea has an extraordinarily high diversity, which has been attributed to its being a melting pot of old and new Asian and Australian species (Diamond, 1984), and to the fact that islands from the Southwest Pacific accrete to the east coast (Pigram & Davies, 1987; Wells et al., 1999; Klootwijk et al., 2003). An exception to the general west–east diversity gradient is Vanuatu, which has a low diversity compared with Fiji and the Solomon Islands (Fig. 2; Schmid, 1975), a phenomenon that probably results from its younger geological age and smaller size (Table 1).

Figure 2.

 Graph illustrating how the number of seed plant genera (van Balgooy, 1971) varies with (a) increasing area and (b) increasing distance from New Guinea for the oceanic Bismarck, Solomon, Vanuatu, Fiji, Samoa and Tonga archipelagos and the continental island of New Caledonia.

Genetic diversity

Just as biogeographical theory predicts that species diversity will decrease with decreasing island size and increasing isolation, the island model of population genetics predicts a similar trend for genetic diversity (Wright, 1940). This is expected because of genetic bottlenecks during colonization and reduced population sizes on small islands (Vellend, 2003, 2005). In addition, it has been suggested that several founder events are required before reduction in genetic diversity occurs (Clegg et al., 2002). This hypothesis remains largely untested in the TSP literature, but some of the limited evidence supports the idea that oceanic island endemics have generally lower genetic diversity compared with continental taxa (Ito et al., 1998; Sarno et al., 2001). Reduced genetic diversity in islands compared with mainland populations was also found for Polynesian sandalwood (Santalum insulare Bertero ex A. DC.), although the total genetic diversity over all archipelagos studied was high (Butaud et al., 2005). In contrast, data for other taxonomic groups, for example predaceous diving beetles in Fiji, have not supported this theory (Monaghan et al., 2006). No attempts have yet been made to demonstrate a correlation between island size and genetic diversity. However, for five populations of a Pacific cycad, the population on the smallest island had the greatest genetic variation, which was attributed to its having the largest size (Keppel et al., 2002).

Ecological/functional diversity

A possible consequence of the decreasing diversity and complexity of the TSP biota is an accompanying decrease in the diversity of ecological processes and functional groups, which would result in major changes in community structure and ecosystem functioning. No studies have been conducted to identify and quantify such changes. However, studies on the ecological dynamics of Pacific Islands show that some processes are very different from those in continental settings. For example, secondary succession is complex in continental tropical rain forests but occurs through a simple auto-succession involving the same species in some ecosystems of Hawaii (Stemmermann & Ihsle, 1993), New Zealand (Ogden, 1988) and the Galápagos Islands (Itow & Mueller-Dombois, 1988), supporting the notion of simpler ecological processes and functional groups. Auto-succession and the low diversity of native species associated with secondary succession on certain Pacific islands led Mueller-Dombois (2008) to propose that forests of remote islands are successionally impoverished and therefore more prone to invasive species that fill vacant successional niches. This theory will need to be thoroughly tested.

What are the sources of species?

Many source areas have been invoked for some portion of the biota of the TSP, but a Malesian source has long been believed to be by far the most important, mainly owing to floristic affinities between the regions (Fig. 3a; van Balgooy, 1971). For taxa that do reach islands, a diversity of unoccupied niches provides enhanced opportunity for colonization and speciation. Pulses of speciation, or adaptive radiations, have been observed across a range of island systems, with the islands of Hawaii providing many widely cited examples: the silversword (Asteraceae) alliance (Baldwin et al., 1991), Drosophila fruit flies (Carson & Kaneshiro, 1976), honeycreeper birds (Lovette et al., 2002) and other taxa (Simon, 1991). Within the TSP, radiations have been documented in the Solomon Islands, including ones for Sphenomorphus skinks, giant rats in the Uromys genus (Diamond, 1984), and Ficus (figs; Corner, 1969). Below, we systematically discuss the weight of evidence supporting the role of potential sources of the TSP biota.

Figure 3.

 Source areas of the biota of the tropical South Pacific based on (a) biogeographical similarities and (b) molecular studies. In (a) biogeographical similarity (beside the arrow) is estimated using the approximate percentage of genera shared between the insular Southwest Pacific and the source area (after van Balgooy, 1971; excluding genera with pantropical distributions). Hawaii’s contribution is given in dashed lines with a double-headed arrow because it is an oceanic island itself. In (b) the values beside the arrow indicate the percentage of molecular studies supporting the source area (from Table 3).

Malesia

Probably the most commonly invoked dispersal route into the TSP is that of a south-eastward pathway from Malesia (van Balgooy et al., 1996; Turner et al., 2001; Winkworth et al., 2002). Most bird and plant species in the Solomon Islands are believed to have originated in Malesia (Greenslade, 1968, 1969; van Balgooy, 1971; Diamond, 1984). A steadily decreasing species diversity eastwards of New Guinea (Table S1; Fig. 2), despite a confounding effect of archipelago size (= 0.92), on cursory inspection supports the notion that Malesia is the major source. However, the supposedly overwhelming contribution of a Malesian source area is not mirrored in molecular data to date (Table 2, Fig. 3).

Table 2.   Summary of molecular phylogenetic studies in the Southwest Pacific.
TaxonFamilyMolecular markersPart of Pacific sampledSource areasDispersal vs. vicarianceReference
  1. ETS, external transcribed spacer; ITS, internal transcribed spacer.

  2. AM, America; CAL, New Caledonia; FIJ, Fiji; HAW, Hawaii; MAL, Malesia; MEL, Melanesia; MIC, Micronesia; NHE, Northern Hemisphere; NG, New Guinea; NZ, New Zealand; PAC, entire Pacific range of the taxon; POL, Polynesia; SAM, Samoa; SEA, SE Asia; SOL, Solomon Islands; TAH, Tahiti; TOG, Tonga; VAN, Vanuatu.

  3. *Support for source area limited or equivocal. †Study suggests multiple colonization events.

Plants
 MerytaAraliaceaeITS, ETSCAL, FIJ, NZNZ*Recent dispersalTronchet et al. (2005)
 Polyscias sect. TieghemopanaxAraliaceaeITS, trnLtrnF, 5S rDNAAUS, CAL, MEL, NG, NZCALRecent dispersalEibl et al. (2001)
 Schefflera (umbrella trees)AraliaceaeITS, trnLtrnFCAL, FIJ, NZCAL, NZ*Not tested/unclear†Plunkett et al. (2005)
 Agathis (kauri)AraucariaceaerbcLPACCAL*, NZ*Recent dispersalSetoguchi et al. (1998),  Knapp et al. (2007)
 HeterospatheArecaceaePRK, RPB2FIJMAL*Not tested/unclearNorup et al. (2006)
 Fitchia/OparanthusAsteraceaetrnLtrnF, matKPOLAMRecent dispersalMort et al. (2008)
 TetramolopiumAsteraceaeITSPACNG/HAWRecent dispersalLowrey et al. (2001)
 Mitthyridium (moss)CalymperaceaegpdFIJ, POL, SAM, VANMAL, AUSRecent dispersalWall (2005)
 CoriariaCoriariaceaerbcL, matKFIJ, NGAM*Recent dispersalYokoyama et al. (2000)
 CorynocarpusCorynocarpaceaerbcL, ITSAUS, CAL, NZ, SOL, VANMELRecent dispersalWagstaff & Dawson (2000)
 WeinmanniaCunnoniaceaeITS, trnLtrnF, trnLCAL, MEL, SAMCAL, maybe others*Not tested/unclear†Bradford (2002)
 Cycas (cycads)CycadaceaeallozymesPACMALRecent dispersalKeppel et al. (2008)
 CyrtandraGesneriaceaeITSFIJ, POL, MICSEA/HAW*Recent dispersalCronk et al. (2005)
 ScaevolaGoodeniaceaeITSPACAUSRecent dispersalHowarth et al. (2003)
 AglaiaMeliaceaeITS, rps16FIJ, SAM, SOLMALRecent dispersalMuellner et al. (2005, 2008)
 MetrosiderosMyrtaceaeITS, ETSPACCAL, NZRecent dispersalWright et al. (2000)
 SyzygiumMyrtaceaeITS, matK, ndhFPACMAL, NZ*Recent dispersal†E. Biffin (unpublished data)
 FuchsiaOnagraceaeITS, trnLtrnF, rpl16TAHNZRecent dispersalBerry et al. (2004)
 PittosporumPittosporaceaeITS, trnLtrnFFIJ, TOG, MICAUSRecent dispersalGemmill et al. (2002);  Chandler et al. 2007
 RetrophyllumPodocarpaceaetrnLtrnFCAL, FIJ, NZAM*Recent dispersal*Herbert et al. (2002);  Sinclair et al. (2002)
 CrossostylisRhizophoraceaechloroplast restrictionPACPacific/CAL*Recent dispersalSetoguchi & Ohba (1995)
 BikkiaRubiaceaeITS, trnL–FCALAMRecent dispersalMotley et al. (2005)
 MelicopeRutaceaeITS, trnL–F, trnET, trnHpsbAPOL, HAW, AUSAUS, MAL, HAW*Recent dispersal†Harbaugh et al. (2009)
 Santalum (sandalwood)SantalaceaeITS, ETS, trnKPACAUSRecent dispersal†Harbaugh & Baldwin (2007)
 PlanchonellaSapotaceaeITSCAL, FIJ, HAW, NG, NZ, TAH, VANCAL, MAL, AUSNot tested/unclear†Swenson et al. (2007)
Animals
 Acrocephalus (reed-warblers)Acrocephalidaecyt b, NADHE POL, AUS*Recent dispersalCibois et al. (2007)
 Rhantus (diving beetles)Dytiscidaecox1, cob, rrnL–tRNALeu– Nad1, 18S rRNA, H3CAL, FIJINHE*Not tested/unclearBalke et al. (2007)
 Iguanidae (iguanas)IguanidaeND4, tRNAFIJ, TOGAM*Not tested/unclearSites et al. (1996)
 Monarchidae (monarch fly catchers)MonarchidaeND2PACSEARecent dispersalFilardi & Moyle (2005)

Southeast Asia

Direct dispersal from the Southeast Asian continent has been suggested to be an important route for some taxa into the TSP (Sanmartín & Ronquist, 2004), but is difficult to distinguish from dispersal from Malesia unless the taxa in the often highly diverse regions of Southeast Asia and Malesia are extensively sampled. A recent molecular study suggests that monarch birds probably colonized the TSP directly from Asia without prior establishment in Malesia (Filardi & Moyle, 2005). The possible importance of Southeast Asia as a source area is underlined by the fact that it is the likely source area for the plant genus Lysmachia (Myrsinaceae) on the Hawaiian archipelago (Hao et al., 2004).

Australia

Although present-day biotic similarities with islands in the TSP are low (Fig. 2), Australia is the likely source of Pacific sandalwood (Santalum), Planchonella and Pittosporum species (Table 2), making it potentially a major source area for the biota of the TSP. Furthermore, dispersal events from Australia to New Zealand (Jordan, 2001; Winkworth et al., 2002, 2005; Perrie & Brownsey, 2007) and to New Caledonia (Bartish et al., 2005), well after their separation, have also been supported by molecular studies.

New Caledonia

Phytogeographic analysis suggests dispersal from New Caledonia into the TSP, via Vanuatu and Fiji (van Balgooy, 1971; Morat et al., 1986). Several molecular studies support New Caledonia as a source area, although it shares only about a quarter of the plant genera found on the various islands in the TSP (Fig. 3). New Caledonia is hence another potentially important source area for the biota of the TSP, as has been suggested for several plant genera (Polyscias sect. Tieghemopanax, Schefflera, Agathis, Weinmannia, Metrosideros, Planchonella; Table 2).

New Zealand

Dispersal from New Zealand into the TSP seems unlikely at first glance, owing to the long distances involved and the differences in climate. However, New Zealand is a likely source for the Pacific radiations of Metrosideros and Agathis (Table 2). A dispersal event from New Zealand has also been used to explain the presence of the genus Fuchsia on Tahiti (Berry et al., 2004).

America

Low floristic similarity suggests little impact of the American biota on that of the TSP (Fig. 3; van Balgooy, 1971). However, there is abundant evidence for a strong American contribution to the Hawaiian flora (Baldwin et al., 1991; DeJoode & Wendel, 1992; Howarth et al., 1997; Pax et al., 1997; Vargas et al., 1998; Costello & Motley, 2001; Lindqvist & Albert, 2002; Wanntorp & Wanntorp, 2003). In the TSP, molecular data point towards an American origin for Bikkia (Rubiaceae; Motley et al., 2005), the sister genera Oparanthus and Fitchia (Rubiaceae; Mort et al., 2008), and possibly Coriaria (Coriariaceae; Yokoyama et al., 2000) and Retrophyllum (Podocarpaceae; Herbert et al., 2002).

Other routes

An influence of the African biota on that of the TSP superficially appears extremely unlikely, but molecular data suggest that the Hawaiian sunflower genus Hesperomannia and Hawaian cotton (Kokia) were derived from an African source (Seelanan et al., 1997; Kim et al., 1998). Recent analyses also suggest that colymbetine diving beetles colonized the Pacific from the Northern Hemisphere, but wider sampling is required to confirm this (Balke et al., 2007). Similarly, woody violets and the genus Schieda (Caryophyllaceae) in Hawaii seem to have originated in subarctic regions (Ballard & Systma, 2000; Sakai et al., 2006). Considering that the Hawaiian archipelago is older than most other oceanic Pacific islands (although its oldest islands are now submerged), it is not surprising that taxa that initially colonized Hawaii spread to other Pacific islands, as has been suggested for sandalwoods (Harbaugh & Baldwin, 2007), Melicope (Harbaugh et al., 2009) and the daisy genus Tetramolopium (Lowrey et al., 2005).

Speciation

Although discussion of the mode of speciation on island systems has traditionally favoured allopatric (on different islands or in different habitats on the same island) over sympatric (in the same habitat on the same island) processes, for example birds in the TSP (Diamond, 1977), Savolainen et al. (2006) offered support for sympatric speciation in two Lord Howe Island palms. Similarly, much of the Hawaiian Metrosideros diversification has occurred within islands (Percy et al., 2008). However, the concept of sympatry is scale-dependent. For example, populations occurring sympatrically in the same habitat on the same island may be allopatric at a finer scale, occupying different microhabitats. In fact, allopatric speciation and sympatric speciation may be viewed as opposing end points along a continuum of speciation with varying degrees of geographical segregation and under varying levels of gene flow (Whittaker & Fernández-Palacios, 2007; Fitzpatrick et al., 2008). Inferring the mode of speciation and the amount of gene flow during the speciation process is difficult because ongoing dispersal will have overwritten the historical patterns of genetic variation that diversification undoubtedly laid down during speciation (Arafeh & Kadereit, 2006).

Regardless of the mode of speciation, a varying proportion of species on an island, which is usually greatest on remote islands, is derived in situ through speciation (Gillespie et al., 2008). Although single colonizations, resulting in generic radiations, may be the case for some taxa (Baldwin et al., 1991; Soltis et al., 1996; Filardi & Moyle, 2005), many species groups have resulted from multiple colonizations, for example spiders (Tetragnatha), the plant genus Scaevola in Hawaii (Gillespie et al., 1994; Howarth et al., 2003), sandalwood (Harbaugh & Baldwin, 2007) and other groups in the TSP (Clegg et al., 2002; Bartish et al., 2005; Plunkett et al., 2005; Burridge et al., 2006). In some cases speciation has been very rapid, as evidenced by low levels of genetic variation amongst closely related taxa in the TSP (Setoguchi & Ohba, 1995; Filardi & Moyle, 2005; Balke et al., 2007; Harbaugh & Baldwin, 2007), Hawaii (Soltis et al., 1996; Baldwin & Sanderson, 1998) and New Zealand (Winkworth et al., 1999).

How have species colonized the TSP?

Over the last decades, two competing theories postulating the way organisms colonize and then speciate on islands have dominated the literature. One view holds that organisms are good dispersers and the other that organisms are generally poor dispersers, and that current distributions are almost entirely the result of tectonic and geological processes. The latter concept of vicariance biogeography (Nelson & Platnick, 1981) was strongly advocated after the discovery of plate tectonics, mostly because long-distance dispersal (LDD) was not considered feasible in the Pacific (Whitmore, 1969). Although there are various definitions of LDD (Nathan, 2005), we here consider it to comprise dispersal events over 200 km or more, facilitating dispersal between archipelagos in the TSP. Some still consider vicariance mechanisms to be a viable hypothesis for explaining the present-day distribution of the Pacific island biota (Ladiges et al., 2003; Heads, 2006b, 2008; Ladiges & Cantrill, 2007).

Early naturalists documented the tremendous potential for LDD (Darwin, 1859; Guppy, 1906; Ridley, 1930; Mayr, 1954), and staunch supporters of LDD events have insisted that almost the entire flora found on islands can be explained by dispersal events from continental regions (Carlquist, 1967). This theory assumes that plants are capable of crossing hundreds to thousands of kilometres of ocean. Pole (1994, 2001) brought attention to the issue in the South Pacific by asking the question: could the entire flora of New Zealand be derived through relatively recent long-distance dispersal? Rather surprisingly, many recent molecular data support the LDD scenario in New Zealand (Winkworth et al., 2002, 2005; Perrie & Brownsey, 2007). However, kauri (Stöckler et al., 2002; Knapp et al., 2007), moas (Cooper et al., 2001; Haddrath & Baker, 2001) and wrens (Ericson et al., 2002) are notable exceptions. Similarly, much of the Hawaiian biota, which is entirely derived through LDD, is believed to have resulted from relatively recent (< 5 Myr) divergence events (Price & Clague, 2002).

For many global systems, detailed biogeographical analyses agree that LDD is the only means by which many current patterns can be explained (Turner et al., 2001), although family-level disjunct distributions of related taxa occurring on east and west coasts of the Atlantic often pre-date the separation of Africa and South America and hence result from plate-tectonic processes (Renner, 2004). The relative importance of dispersal and vicariance is still controversial, both on a global scale (Ebach & Tangey, 2006) and in the Pacific (Keast & Miller, 1996). In the Pacific, vicariance has been considered the only possible explanation for the present-day distributions of certain taxonomic groups, such as conifers (de Laubenfels, 1996) and the Proteaceae (Weston & Crisp, 1996), whereas dispersal has been deemed more probable for other groups (Hill, 1996; Munroe, 1996; Wilson, 1996).

Two vicariance modes of colonization can be envisaged (Fig. 4). In one scenario, continental landmass(es) were previously present in the TSP, but have since disappeared under the sea in an Atlantis-like fashion. The other possibility is that pieces of continental or oceanic landmasses detached and drifted into the TSP, acting as shuttles transporting biota. Recent dispersal into the TSP could have occurred in three ways: dispersal via colonization of intermediate islands (stepping stones); direct dispersal from continental areas over long distances (LDD); and dispersal facilitated by humans (hitch-hiking). The latter implies very recent dispersal/colonization events (mostly younger than 3500 years). These different scenarios have received varying support from various sources (Table 3).

Figure 4.

 Graphical illustration of four theories for the mode of colonization of islands in the Southwest Pacific. Hitch-hiking on human transport is not illustrated. Arrows indicate the movement of plant diaspores. Dashed arrows show the hypothetical movement of landmasses. Current landmasses are traced by solid lines, and hypothesized past landmasses by dashed lines. Striped areas are continental landmasses; others are oceanic.

Table 3.   Current status of biogeographical, geological and molecular support for the various colonization-mode theories and the impact extinctions could have had on the current support.
TheoryDistribution of current biotaGeologicalMolecularImpact of extinction
AtlantisLimited (few ancient taxa with limited dispersal abilities, such as iguanas and Degeneriaceae in Fiji)No supportNo supportResilient
ShuttleDifficult to test because relatively recent shuttle dispersal would be similar to LDDLimited to Yan & Kroenke’s (1993) hotspot modellingDifficult to test because relatively recent shuttle dispersal would be similar to LDDResilient
Stepping stoneDifficult to test because stepping-stone dispersal would be similar to LDDSome support as several, now submerged, sea mounts would have been emergent during lower sea levelsNot yet specifically testedResilient
Long-distance dispersal (LDD)Excellent because presence of most taxa can be readily explained by LDDExcellent support, as all islands in the SW Pacific (except New Caledonia and New Zealand) are of oceanic origin.Excellent (Table 3)Sensitive/ overestimation
Express trainGood because many plants that are used by humans are found throughout the PacificNot applicableExcellent (Fisher, 1997; Matisoo-Smith et al., 1998; Austin, 1999; Zerega et al., 2004; Hinkle, 2007)Resilient

The Atlantis theory (vicariance)

Early vicariance biogeographers suggested that islands in the TSP were formerly part of a continental landmass. Taxa on remote islands belonging to lineages with a long fossil record, such as podocarp conifers (Podocarpaceae) and iguanas (Brachylophus spp.) in Fiji and Tonga, appeared to support this concept. This led Smith (1955) to propose that the Fiji region and its flora were remnants of a Melanesian continent that sank like the Atlantis of Greek mythology. Nur & Ben-Avraham (1977) envisaged a similar continent (called ‘Pacifica’), but proposed that it disintegrated and that parts of it accreted to other continents. Although not explicitly suggesting continental origins, Heads (2006b) proposed that Fiji’s flora could have survived on ephemeral islands and that the current islands received their flora from nearby islands that no longer exist. The presence of subsided seamounts (Neall & Trewick, 2008) supports this possibility, but geological studies have failed to find evidence of a definite origin of continental rocks on present TSP islands (Schellart et al., 2006).

Molecular studies have also so far failed to provide empirical evidence to support the Atlantis theory, but rather have supported recent LDD events. Most studies find low genetic differentiation between taxa, implying recent divergence. For example, the conifer genus Agathis, often cited in support of the vicariance origin of Pacific island biotas (Whitmore, 1969; de Laubenfels, 1996), was found to have recently speciated in the Pacific, with the likely source being New Caledonia (Setoguchi et al., 1998). Similarly, Pacific cycads (genus Cycas) appear to have recently expanded their distribution following the evolution of adaptations that facilitate oceanic dispersal (Hill, 1996; Keppel et al., 2008). In addition, LDD and recent divergence are now known to have played a major role in determining the current distribution of Nothofagus (Swenson & Hill, 2001; Cook & Crisp, 2005a; Knapp et al., 2005), long considered to support vicariance scenarios (Linder & Crisp, 1995; Swenson et al., 2001). The few molecular clock estimates attempted on TSP taxa (Berry et al., 2004; Filardi & Moyle, 2005; Wall, 2005; Cibois et al., 2007; Harbaugh & Baldwin, 2007) all suggest colonization and divergence events within the last 10 Myr, but were mostly calibrated based on evolutionary rates and divergence dates from other studies. More independently calibrated molecular data are needed to test vicariance scenarios in this region.

The shuttle theory (vicariance)

The theory of continental fragments drifting into the Pacific as shuttle transports for biota has, contrary to the Altantis scenario, received some geological support. It is highly likely that some Gondwanan elements rafted north-east on New Caledonia when it detached from east Gondwana. It has also been suggested, based on hotspot modelling, that a piece of land detached from New Caledonia and drifted rapidly north-east, colliding with the island of Viti Levu in Fiji some 6 Ma (Yan & Kroenke, 1993). This could explain the presence of the conifer genus Acmopyle (endemic to Fiji and New Caledonia with fossils in Australia and New Zealand) in Fiji. It is, however, unclear if the drifting island was emergent during the entire drifting process. Similarly, there is debate about what fraction of the biota survived the high Oligocene sea levels in New Zealand and New Caledonia (Pole, 1994; Cooper & Cooper, 1995; Hickson et al., 2000; Lee et al., 2001, 2007; McGlone et al., 2001; Landis et al., 2008).

Stepping stones (dispersal)

van Steenis (1979) believed dispersal to be limited to short rather than giant leaps, considering dispersal unlikely over > 200 km and improbable over > 500 km. As a result, dispersal to islands located far from the mainland was believed by him and some others to involve hopping across intermediate (stepping-stone or land-bridge) islands (van Steenis, 1962, 1969; Smith, 1963). Distances required for successful dispersal were shorter during the Last Glacial Maximum, some 18,000 years ago (Fig. 5), when sea levels were about 130 m lower, because additional islands existed and many present islands were larger and/or connected to neighbouring islands (Milliman & Emery, 1968; Gibbons & Clunie, 1986; Hope, 1996). However, these historical sea-level changes would mostly have facilitated connectivity between islands within archipelagos and would not have significantly reduced distances between archipelagos, except between Australia and New Caledonia (Steadman, 2006).

Figure 5.

 Land area during the Last Glacial Maximum (grey) and at present (black). Land area during the Last Glacial Maximum was estimated based on the 130-m depth contour (after Hope, 1996).

Long-distance dispersal (LDD)

Dispersal events over long distances are difficult to document (Nathan, 2006), and seed dispersal distances are usually inferred from biological observations. For example, based on gut retention times, seed dispersal distances up to a maximum of 100 and 300 km were inferred for the Pacific pigeon (Ducula pacifica) in Tonga (McConkey et al., 2004) and a bat (Cynopterus sphinx) in Indonesia (Shilton et al., 1999), respectively. LDD has therefore been contentious (van Steenis, 1979), despite long-standing evidence for efficient dispersal (Guppy, 1906; Ridley, 1930). However, support for LDD has been building globally, based on biogeographical (Sanmartín & Ronquist, 2004) and molecular (Systma et al., 2004; Alsos et al., 2007) data, and rare LDD events can have a big impact on biogeographical patterning. The efficiency of wind dispersal has been documented (Nathan et al., 2002), and extreme events such as tropical cyclones may greatly increase potential dispersal distances. Judged from present dispersal mechanisms for Pacific floras, bird-mediated LDD is even more efficient (Carlquist, 1967), and present dispersal agents and distances may be very different from pre-historic ones, as a result of the extinction of many large bird species (Olson & James, 1982; Steadman, 1995; Worthy, 2000, 2001; Meehan et al., 2002). Between Australia and New Zealand, herbaceous species with gender dimorphism, very small seeds and a preference for wet habitats have been found to disperse best across the Tasman Sea (c. 2000 km; Lord, 1999; Jordan, 2001; McGlone et al., 2001). Species that are small-seeded and animal-dispersed were also found to be good colonizers of the Krakatau Islands, following the effective sterilization of the islands in a volcanic eruption in 1883 (Whittaker et al., 1997).

The occurrence of LDD in other regions has been supported by molecular (Baum et al., 1998; Renner et al., 2001) and empirical (Close et al., 1978) studies. For example, around 20% of South American rain forest plants are believed to be derived from relatively recent (within the last 65 Myr and hence after the break-up of Gondwana), mostly African, immigrants through LDD (Pennington & Dick, 2004). In the Pacific, extreme LDD events are strongly supported by molecular studies. The spider genus Tetragnatha has been inferred to have colonized Hawaii, Tahiti and the Marquesas independently from continental sources (Gillespie, 2002). Similarly, molecular data for sandalwood suggests as many as five LDD events from Australia to islands as remote as Hawaii and the Juan Fernández Islands (Harbaugh & Baldwin, 2007), and the daisy Tetramolopium possibly crossed 6000 km of ocean to colonize Hawaii from New Guinea (Lowrey et al., 2001).

In the TSP, circumpolar westerly winds and ocean currents over the last 5–23 Myr, known as west-wind drift, are cited as the predominant cause of an eastward dispersal bias into the Pacific (Winkworth et al., 2002; Cook & Crisp, 2005b; Burridge et al., 2006). Such wind currents are expected to influence particularly spore-bearing and seed plants with wind-dispersal adaptations (Wright et al., 2000; Muñoz et al., 2004). Ocean currents provide an important means of dispersal for suitably adapted taxa, even for those intolerant of salt water (Guppy, 1906; Ridley, 1930; Censky et al., 1998; Calsbeek & Smith, 2003). The effectiveness of this mode of dispersal is well demonstrated by the strand flora of the TSP, in which many species are pantropical (Fosberg, 1984). Snakes and some geckos, which dispersed to the TSP before humans did (Fisher, 1997; Austin, 2000), are hypothesized to have rafted across on debris washed out during extreme weather events.

The hitch-hiking theory (dispersal)

Some taxa have colonized the Pacific east of the Solomon Islands only very recently (i.e. within the last 3000 years), probably through human-mediated transport. The lizard Lipinia noctua (Austin, 1999) and the gecko Gehrya mutilata (Fisher, 1997) probably hitch-hiked in the canoes of the first colonizers, whereas breadfruit (Zerega et al., 2004) and possibly Pacific rats (Matisoo-Smith et al., 1998) were deliberate early introductions. Early Pacific islanders were excellent navigators and have been implicated in the first introduction of chickens to South America (Storey et al., 2007). After this wave of pre-historic introductions, additional species, such as the mongoose (Herpestes javanicus) in Fiji (Morley, 2004) and the giant African snail (Achatina fulica) and fire ants (Wasmannia auropunctata) in Vanuatu and the Solomon Islands (Cowie, 2001; Wetterer & Porter, 2003), were accidentally, or intentionally, introduced. Many plants, ants and other taxa are also known to have been introduced by ancient (aboriginal) and recent (European) transport (Merrill, 1947; Wilson & Taylor, 1967).

Has extinction played a prominent role in achieving current distribution patterns?

In the TSP, the impact of extinction has been best documented for land birds. At least a fifth of all land bird species went extinct after the arrival of humans, but total extinction rates may amount to more than 50% of the pre-human bird fauna (Steadman, 2006). Accordingly, current diversity patterns may not reflect those of pre-human times for some taxa.

Source areas can also potentially be wrongly assigned because of extinctions. In the TSP, extinctions have been well documented for the Australian source area. Several rain forest taxa underwent local extinction as Australia became drier with the onset of the circum-Antarctic current some 35 Ma. Acmopyle, Ascarina, Nothofagus sect. Brassospora and Retrophyllum presently occur on Pacific Islands but no longer occur in Australia, where fossils of these taxa have been found (Coetzee & Muller, 1984; Hill & Carpenter, 1991; Hill & Brodribb, 1999; Hill, 2001).

Acmopyle comprises one species in New Caledonia and another in Fiji, and, although colonization of Fiji from New Caledonia is the most logical explanation, fossil evidence points towards independent colonization events for the two islands from an Australian source area (Hill & Carpenter, 1991). An alternative scenario could be a colonization event from New Zealand, where fossils are known (Pole, 1997). Another possible example of misinterpretation of source areas is that of iguanas on Fiji and Tonga, which have their only extant relatives in America (Sites et al., 1996). This has been attributed to an exceptional LDD event involving iguanas rafting from America to Fiji (Gibbons, 1981). However, the discovery of Cretaceous iguana fossils in China (Gao & Hou, 1995) shows that iguanas once had a wider distribution. This therefore opens the possibility that they dispersed via the Malesia/Southeast Asia routes but subsequently went extinct in the putative Asian source area and possibly from other Pacific islands.

Similarly, extinction may affect the interpretation of modes of colonization. For example, if a taxon colonized Fiji from New Guinea through stepping-stone dispersal during the Last Glacial Maximum and subsequently went extinct in the ‘stepping stones’, the present-day distribution and molecular patterns would be indistinguishable from those of a single LDD event. Therefore, some evidence for LDD needs to be more cautiously interpreted because extinctions can inflate the estimated number of inferred LDD events (Table 3).

Pacific biogeography: present and future

Although the central role of dispersal in explaining the current distribution of the Pacific biota is now hardly disputed (but see Heads, 2006b, 2008; Nelson, 2006; Ladiges & Cantrill, 2007), questions that still need to be addressed on oceanic islands in the TSP relate to whether vicariance events have occurred at all and what their relative importance is. Current evidence strongly supports Carlquist’s (1967) assertion of the overriding importance of LDD in producing the current Pacific biota (Table 3). Even when extinctions are considered, LDD plays a central role in the biogeography of oceanic islands (Cowie & Holland, 2006) and explains many of the current patterns of the biota on oceanic islands in the TSP, as has also been identified for Hawaii (Simon, 1991; Wagner, 1991; Wagner & Funk, 1995).

Whether, and how far, taxa have spread into the Pacific depends on their dispersal abilities and on the ecology/habitats of the recipient area (Carlquist, 1967). Several plant groups are particularly good dispersers, and plants and animals have very different dispersal patterns (Sanmartín & Ronquist, 2004). In plants, coastal taxa are often well dispersed by floating or rafting propagules. These super-dispersers generally have very wide ranges, and this dispersal ability has resulted in many strand species becoming naturally pantropical (Fosberg, 1984). Such well-dispersed taxa have been called ‘super-tramps’, a term first coined for extremely dispersive Melanesian birds, such as fruit doves, white eyes and rails (Diamond, 1974), but recently applied to the plant genus Cyrtandra (Cronk et al., 2005), which has colonized and speciated on many TSP islands.

The high dispersal ability of many taxa in the TSP suggests that taxa of the TSP may have colonized surrounding continents. There is good evidence that some bird species have colonized Australia from the oceanic Pacific (Filardi & Moyle, 2005; Cibois et al., 2007). The plant genus Corynocarpus may have colonized New Guinea, Australia, New Caledonia and New Zealand from Melanesia (Wagstaff & Dawson, 2000). However, extinctions in the ‘true’ source areas could be responsible for this interpretation.

More molecular studies are needed to determine whether the Malesia–TSP route is indeed the most important colonization pathway. Such studies could also shed more light on the roles of stepping-stone dispersal and hitch-hiking dispersal models. Published studies (Table 2) confirm Malesia as an important source area but highlight Australia, New Caledonia and, to a lesser extent, New Zealand as having played a much greater source role than previously considered to be likely. The importance of this south-western source area needs to be further investigated, because its importance may have been overestimated as a result of an access bias to territories under American, Australian, French or New Zealand administration, which could result in comparatively more supportive data for these areas. Taxa from Melanesian islands and New Guinea need to be sampled more thoroughly if we are to quantify how frequently, along which routes, and over what distances dispersal occurs. Another plausible source of bias is extinction, which could be related to changes in the climate of Australia and New Caledonia. For example, the rain forest biome, which once covered most of Australia, has retracted to a few remnant pockets in Northeast Australia, with the extinction of several taxa (Martin, 2006).

Islands in the TSP are centres of speciation (Keast, 1996), as well as recipients of colonists and potential museums of relictual taxa. Despite its central role in Pacific biogeography, the TSP has often been ignored in biogeographical, molecular and taxonomic studies. For example, the Solomon Islands, Vanuatu and Samoa lack published floras, and, even in a botanically well-explored country such as Fiji, new plant species are still being discovered (Miller, 1988; Fuller et al., 1997; Gardner, 1997). Therefore there is an urgent need to increase the research effort in the understudied regions of the TSP, so that we can gain a better understanding of the composition, biogeography, ecology and evolution of the ecosystems and biota of the TSP.

Acknowledgements

We thank M. Pole, L. Cook, Y. Buckley and E. Biffin for kindly reviewing this paper. We are grateful to the University of Queensland for providing an International Postgraduate Research Scholarship (IPRS) and an International Postgraduate Living Allowance Scholarship (ILAS) to G.K., to the Australia and Pacific Science Foundation (APSF) for providing funding for G.K. to visit A.J.L. to work on this paper, to the Ecology Centre at the School of Integrative Biology of the University of Queensland for supporting G.K., and to the Australian Research Committee (ARC – DP 0665859) for support to A.J.L.

Biosketches

Gunnar Keppel is a lecturer in plant biology at the University of the South Pacific in Fiji and a PhD student at the University of Queensland. With a keen interest in the ecology, biogeography and taxonomy of plants in the insular Southwest Pacific, his current focus is on the diversity and origin of lowland tropical rain forests in this region.

Andrew Lowe is a professor of plant conservation biology at the University of Adelaide and the Head of Science at the State Herbarium in Adelaide. His research focus is on plant ecological and evolutionary genetics. Two of his recent projects look at the phylogeography of Pacific plant taxa.

Hugh Possingham is an ARC Research Fellow and the director of the Ecology Center at the University of Queensland. His research focus is on the environmental applications of decision theory, but his scientific interests are many, including birds and biogeography.

Editor: Robert McDowall

Ancillary