Evidence for overlooked mechanisms of long‐distance seed dispersal to and between oceanic islands

Ever since Darwin (1859), the dispersal of species from continents to oceanic islands, and between such islands, has been the subject of considerable speculation. These islands are those arising from the seafloor as a result of underwater geologic activity, typically volcanic or coralline, that have never been connected to continental land masses. In his seminal book Dispersal of Plants Throughout the World, Ridley (1930) collated considerable empirical evidence of long-distance dispersal (LDD) and proposed mechanisms responsible for the colonization of remote archipelagos. Later, van der Pijl (1982) summarized traits to define diaspore syndromes related to sea (hydrochory), wind (anemochory) and animal (zoochory) LDD. The latter includes endozoochory (dissemination of seeds in the disperser’s gut) and epizoochory (seeds externally attached to the disperser’s body). These three keystone publications aimed to build a framework to predict dispersal based on plant and vector features and empirical studies. Darwin (1859) performed sowing experiments with some seeds dispersed externally by waterbirds, finding that they often germinated, while Ridley (1930) found different aquatic species of unrelated genera (e.g. Juncus, Carex, Cyperus, Polygonum) along banks of isolated ponds, and associated this distribution to waterbirds. Although Ridley (1930) provided no direct evidence for waterfowl dispersal, he assumed that transport of seeds in mud on birds’ legs could be important for plants without special morphological adaptations for dispersal. Other botanists have since advanced this mechanism to explain the arrival to different oceanic islands of many small seeds lacking LDD adaptations. For instance, Carlquist (1967) estimated that c. 21% of plant species that have colonized Easter Island and 21% of those dispersed to the Juan Fernández archipelago did so in mud on the feet of birds because they lacked any special adaptation for LDD. Porter (1983) also argued that mud attached to animals was responsible for 42 out of 378 (11%) original dispersal events to the Galápagos. However, to our knowledge, there is limited evidence of LDD by terrestrial birds transporting seeds in their plumage or in mud on their legs and feet. Figuerola Proctor, 1968; Brochet et al., 2010). Given the inherent difficulties in studying seed dispersal to oceanic islands, it is often impossible to reconstruct how plants reached these islands (Thornton, 2007; Guillespie et al., 2012; Vargas et al., in press). Distances from the mainland to such islands can range from tens of km (e.g. Canaries, ≈ 100 km) to thousands of km (e.g. Galápagos, ≈ 1000 km, Hawaii, ≈ 4000 km), but it is expected that in all cases colonization will require mechanisms of dispersal distinct from those that enable short distance dispersal (e.g. within 1 km of a parent plant) and which have no immediate biogeographical consequences (Normand et al., 2011). Our main goal here is to examine some possibly overlooked LDD mechanisms involved in seed transportation to and between oceanic islands. In doing this, we must be clear about the use of the term mechanism in contrast to syndrome, and vector of dispersal. Here, dispersal syndromes are considered to be morphological adaptations of diaspores, which increase the likelihood of their dispersal. Dispersal mechanisms are defined as typical (i.e. predictable) pathways by which seeds attain LDD, regardless of the adaptations they might possess. Finally, dispersal vectors are the actual means by which seeds are dispersed, independently of dispersal syndrome or mechanism. Vegetative and reproductive parts of the plant are susceptible to be dispersed by different mechanisms, although fruits and seeds are primarily responsible for most cases of LDD colonization. When fruits and seeds are dispersed by a mechanism other than that to which they are particularly well adapted (see Ridley, 1930; van der Pijl, 1982), that is, a nonstandard dispersal mechanism (Higgins et al., 2003), such events have often been classified as stochastic (Clark et al., 2001). We consider as ‘stochastic’ those dispersal processes that are impossible to predict, for example those linked to exceptional conditions, such as storms or tsunamis. However, to dismiss such means of dispersal as attributable to chance alone might oversimplify the importance of deterministic, but poorly understood, processes. Considering the limited predictive power of morphological dispersal syndromes and the wide variety of nonstandard mechanisms (see Wilkinson, 1997; Vargas et al., in press), some authors have emphasized the likely importance of stochastic LDD events (Higgins Higgins et al., 2003; Nathan, 2006) and pointed out the need for better empirical data relevant to such processes. Next, we enumerate some potentially important but still poorly understood mechanisms of LDD, which we regard as having been overlooked in the past (summarized in Table 1): Some medium and large-sized birds, not typically frugivorous, frequently broaden their diet to include fruits/seeds of different species when these are abundant. The biogeographical importance of this in the migration of trees during the Holocene in Europe (often thought of as exclusively wind-dispersed) has been suggested by Wilkinson (1997), with birds acting as seed dispersers on large spatial and temporal scales. In an insular context, some bird species are widely distributed and have to cope with new ecological conditions for life on these islands, which frequently have a depauperate biota. Under such conditions, species have to obtain food from a wider range of sources, a phenomenon known as ‘niche expansion’ (Wright, 1980). This has been reported for several species of gulls on both islands (e.g. Gilham, 1952; Nogales et al., 2001; Calviño-Cancela Morton Cortés, 1994). In a recent review of frugivory and seed dispersal by gulls (Calviño-Cancela, 2011), a total of 31 species were found to disperse seeds of c. 100 plants worldwide, which illustrates the potential importance of this bird group as seed dispersers. Practically half of these plants produce dry fruits with most having no particular LDD syndrome, as is the case for many plants that arrive on and colonize oceanic islands (see Vargas et al., in press). Gulls can move hundreds and even thousands of kilometres from their home areas (Cramp Nogales et al., 2001). Thus, they may feasibly move seeds within their digestive tract from continents to islands and between islands, at least within the same biogeographic region. Besides gulls, the Corvidae (basically crows and ravens) are often reported to be seed dispersers. They are known to disperse seeds of > 30 species of gymnosperm and angiosperm species in North America (Wenny, 2001 and references therein), South America (Trucco see Wilkinson, 1997), Africa (Dean Corlett, 2000). The raven (Corvus corax) deserves special mention in the Canary Islands, where it legitimately disperses a minimum of 15 plant species (Nogales et al., 1999). Raptors may also be important as long-distance seed dispersers. In some continental areas, 13 of these bird species included in the families Cathartidae, Accipitridae and Falconidae directly consume fruits and disperse seeds of > 20 plant species (see review by Galetti strong flying-power; relatively long gut-passage time; and regular spatial and temporal movements. While most fleshy fruits are well adapted to promote seed dispersal by providing a nutritional reward to seed dispersers, seeds can also be important sources of energy (Norconk et al., 1998). Many frugivores have developed morphological attributes (e.g. particularly strong beaks) or techniques that allow them to digest the seed content, taking full advantage of ingested fruits. These frugivores have been called seed predators (Janzen, 1971), seed consumers (Hampe, 2001) or granivores (Bartuszevige Herrera Hulme, 2002). However, it is now widely accepted that many animals do not strictly fit into this dichotomy of seed predators vs legitimate seed dispersers. For example, rodents, ungulates and primates which usually act as seed predators can be successful seed dispersers (e.g. Price Clark et al., 2001). Although to date there is a scarcity of studies quantifying this effect in birds, two recent contributions reveal that at least avian seed predators can be very successful seed dispersers in oceanic islands, particularly of small seeds. Guerrero Hall, 1987; Dean Engel, 2000; Pearson Shanahan et al., 2001). In the Canary Islands, secondary dispersal via double endozoochory also involves frugivorous lizards and their common predators (shrikes and kestrels, Nogales et al., 1998, 2002, 2007), and it is feasible that this might be a common LDD mechanism in insular environments worldwide (Moore, 1999). Lizards are important frugivores in island ecosystems (see Olesen Pyle, 2002 for Hawaii; Wiedenfeld, 2006 for the Galápagos). Some legitimate dispersal vectors have been recorded and quantified on the recently formed volcanic islands, Surtsey (Iceland, North Atlantic Ocean) and Anak Krakatau (Indonesia, Indian Ocean) (see Thornton, 2007). However, there are marked differences between these two islands in the ways in which they have been colonized by plants and the dispersal mechanisms involved. Approx. 25 yr after the emergence of Surtsey from the seabed, 64% of the angiosperms now present on the island seem to have been brought to the island by birds, 27% by sea currents (including rafts) and 9% by wind (Fridriksson, 1992). By contrast, on Anak Krakatau it is estimated that 23% of plant species arrived through dispersal by animals (birds and bats), 55% by sea and 22% by wind (Thornton, 1992). The fact that birds seem to have played a more important role in plant colonization on Surtsey than on Anak Krakatau over an equivalent time period might be a result of the presence on Surtsey of seabird colonies (especially gulls) before it supported resident land birds, while this was not the case on Anak Krakatau (Thornton, 2002). Migratory birds (mainly frugivorous) were almost certainly involved in the colonization of Surtsey by angiosperms. Indeed, 60 out of 230 species of migrant birds observed in Iceland had been recorded on Surtsey by 1975 (Thornton, 2002), and following a population explosion in the gull colony after 1986, there was an increase in colonization of the island by angiosperms. Additionally, some new plants were recorded under the nests of omnivores (e.g. ravens, C. corax) and migratory birds, such as the snow bunting (Plectrophenax nivalis), a presumed ‘seed predator’. Indeed, a total of 87 seeds of various species were extracted from the gizzards of 32 of these birds, of which at least two plants (Polygonum persicaria and Carex nigra) germinated successfully (Fridriksson, 1975). The confirmation of these LDD events gives supporting evidence to the mechanisms we propose here. We advocate that future research should focus on nonstandard LDD mechanisms such as those presented here, and others not yet reported, in studies of the colonization of oceanic islands by plants. The three mechanisms herein discussed require explicit testing in different oceanic archipelagos. Such research should take advantage of molecular tools that are likely to be highly effective in the study of LDD to oceanic islands (see Queiroz, 2005; Vargas, 2007 and references therein), relating genetic fingerprints of plants to source populations and migration routes. While we have focused here on LDD events related to animal movements, other specific but still poorly known LDD mechanisms, such as anemochory (by wind) or hydrochory (by water), should also be evaluated in the context of dispersal to oceanic islands (see Guillespie et al., 2012). In this way, a greater understanding will be gained of mechanisms of dispersal to oceanic islands, which in the past have been placed in the stochastic, ‘black-box’ drawer. We would like to thank D. J. Crawford, H. Adsersen and E. W. Schupp for their valuable comments on an earlier version of the manuscript. Furthermore, three anonymous referees and the editor, Richard Abbott, had an important input into this contribution. This study is framed within a biodiversity project in the Galápagos Islands, financed by BBVA Foundation (Spain). We also thank the Charles Darwin Foundation and the Parque Nacional de Galápagos for giving us logistic support while working there.