Time to abandon the loss of dispersal ability hypothesis in island plants: A comment on García‐Verdugo, Mairal, Monroy, Sajeva and Caujapé‐Castells (2017)

García‐Verdugo et al. (2017) recently tested the loss of dispersal ability hypothesis in a wind‐dispersed shrub from Southern Europe. Although the hypothesis has guided research for over 150 years, García‐Verdugo et al. (2017) results failed to substantiate its central prediction—the loss of seed dispersal potential in island populations. Here, I highlight several additional limitations of the hypothesis. First, García‐Verdugo et al. (2017) results are not unusual. Empirical support for the hypothesis is equivocal. Second, when reduced dispersal potential is documented, it may often evolve as a passive by‐product of selection for large seeds, for reasons that are wholly unrelated to their dispersal. Third, the hypothesis does not readily apply to all plant dispersal modes, particularly plants that produce fleshy fruits. These issues advocate a fresh approach to the study of how selection shapes the evolution of dispersal potential on islands.

port for the hypothesis is equivocal. Second, when reduced dispersal potential is documented, it may often evolve as a passive by-product of selection for large seeds, for reasons that are wholly unrelated to their dispersal. Third, the hypothesis does not readily apply to all plant dispersal modes, particularly plants that produce fleshy fruits.
These issues advocate a fresh approach to the study of how selection shapes the evolution of dispersal potential on islands.
When asked to explain the existence of flightless animals on isolated islands, Darwin (1859) responded with a famous analogy involving shipwrecked sailors (Lomolino, 2010). He argued there were two strategies for sailors to survive a shipwreck. The first was to be a strong enough swimmer to reach the shore. The second was to be a poor swimmer and cling to the wreck. This second strategy has played in the minds of island biogeographers ever since and ultimately led to a long-standing hypothesis of how dispersal evolves on oceanic islands. The loss of dispersal ability hypothesis predicts that islands are initially colonized by individuals with relatively good powers of dispersal, but through time, island populations evolve poorer disperser ability to avoid being swept out to sea (see Cody & Overton, 1996).
In a recent issue of the Journal of Biogeography, Garc ıa-Verdugo et al. (2017) provide an unusually thorough test of the loss of dispersal ability hypothesis in a wind-dispersed shrub, Periploca laevigata (Apocynaceae). They compared diaspore morphology and dispersal potential of P. laevigata populations in continental Europe and Africa to those on islands in both the Atlantic Ocean and Mediterranean Sea. They conducted molecular analyses to determine the age of island populations, as well as glasshouse experiments to establish whether diaspore morphology responds plastically to changes in environmental conditions. Garc ıa-Verdugo et al. (2017) results were inconsistent with the loss of dispersal ability hypothesis. Island populations of P. laevigata did not have reduced powers of dispersal relative to mainland populations. In fact, the opposite pattern was observed on the Cape Verde and western Canary Islands, which housed plants with comparatively high dispersal potential. The dispersal potential of island populations was also unrelated to when islands were colonized.
Results from their common garden experiments showed that while fruit morphology varied plastically with environmental conditions, geographical patterns in dispersal potential were largely under genetic control. Therefore, evolution appears to have shaped the evolution of fruit form in P. laevigata, but not in the way predicted by the loss of dispersal hypothesis.
The loss of dispersal ability is usually considered to be a cornerstone of the 'island syndrome', a suite of attributes that delineate island biotas from those on continents (Grant, 1998;Lomolino, 2010;Whittaker & Fern andez-Palacios, 2007). However, results from  2017), they found that two species actually exhibited greater dispersal potential on islands, in stark contrast to third species investigated by Fresnillo and Ehlers (2008) was different. Cirsium arvense (Asteraceae) showed a reduction in dispersal potential on islands, as predicted by the loss of dispersal ability hypothesis. However, a closer look at the morphology of its diaspores suggests that a loss of dispersal ability could have evolved for reasons other than dispersal potential.
Cirsium arvense produces wind-dispersed diaspores that are comprised of a seed (achene) attached to a feathery plume (pappus), which carries seeds aloft on air currents. Differences in seed dispersal distances can therefore arise not only from changes in the size of plumes, but also from changes in the size of seeds, the payload plumes transport. Seed size is known to have a pronounced effect on a wide range of important demographic processes, including early seedling survivorship, tolerance to environmental hazards and competitive ability (Geritz, van der Meijden, & Metz, 1999;Harms & Dalling, 1997;Leishman & Westoby, 1994;Leishman, Wright, Moles, & Westoby, 2000;L€ onnberg & Eriksson, 2013;Moles & Westoby, 2004;Rubio de Casas et al., 2017). This raises the possibility that differences in dispersal potential could evolve passively via selection for large seeds, which can be advantageous for a variety of reasons at later life history stages.
Sherwin Carlquist (1966a,b), an early pioneer in the study of seed dispersal dynamics on islands, clearly recognized this mechanistic pathway towards reduced dispersal ability: "Extreme cases of loss of dispersability. . . feature increase in fruit size without concomitant increase in appendages which serve in dissemination" (page 46, Carlquist, 1966a). Furthermore, he reasoned, "a poorly functioning dispersal apparatus may be retained because it does not have a strongly negative selective value" (page 47, Carlquist, 1966a). Fresnillo and Ehlers (2008) found that the loss of dispersal ability in C. arvense was not determined by decreased plume size. In fact, plume sizes tended to be bigger in island populations. Instead, declines in dispersal potential resulted from increases in seed size. These results are echoed by Garc ıa-Verdugo et al. (2017), who found that P. laevigata populations on islands in Macronesia tended to produce larger seeds than mainland populations. Recent comparative work on species inhabiting a handful of islands in the Southwest Pacific found consistent support for seed gigantism, regardless of dispersal mode or life-form, suggesting that selection may consistently favour increased seed size in island plants (Kavanagh & Burns, 2014). However, it has yet to be established whether increased seed sizes are linked evolutionarily to dispersal capacity or some other factor.
In a global analysis of island bryophytes, Patiño et al. (2013) highlight an additional confounding effect in the dispersal potential of island plants. They found that island bryophytes typically produce larger spores than their mainland counterparts. In bryophytes, spore size directly determines their dispersal potential. Larger spores are heavier, so they travel in air currents for shorter periods than smaller spores, thus leading to shorter dispersal distances. Reductions in dispersal potential in island bryophytes evolved via selection for bigger spores, potentially for a variety of reasons, which may, or may not, be directly related to their dispersal potential. In particular, Patiño et al. (2013) show that the dispersal capacity of spores is confounded by breeding system, as asexual reproduction in bryophytes tends to result in larger spores than sexual reproduction. Therefore, the loss of dispersal potential in island bryophytes may stem from selection for increased asexual reproduction on isolated islands (i.e. Baker's Law, see Pannell et al., 2015), rather than reduced dispersal potential. In a study of thousands of plant species, Grossenbacher et al. (2017) found that self-compatibility tends to be more prevalent on islands. A similarly extensive study on anemochorous plants on island across the globe could help establish whether the loss of dispersal ability is determined by changes in dispersal aides (e.g. the size of plumes), or changes in seed size.
A second problem with the loss of dispersal ability hypothesis is that it does not readily apply to some dispersal modes. Seed dispersal via ingestion by animals (endozoochory) is an exceedingly common dispersal mode in plants, and fleshy fruits have evolved independently on numerous occasions (Herrera, 2002). In fact, it is the predominant dispersal mode in woody plants, forested environments and in species that produce large seeds (Jordano, 2000). Any universal theory of how insularity shapes the dispersal potential of plants would obviously need to include this mode of seed dispersal.
In wind-dispersed plants species, dispersal potential is determined by the relative sizes of seeds and dispersal aides (e.g. plumes or wings). However, in plants that produce fleshy fruits, the relationship between seed dispersal distances and dispersal aides (the quantity or quality of fruit pulp) is more complex. Despite decades of research, consistent relationships between fruit pulp and dispersal distances have yet to be established (see Herrera, 2002). Rather than being determined by the investment made by parent plants into dispersal aides, once fleshy fruits are consumed, seed dispersal distances are determined largely by the behaviour of the animals that ate them.
The chemical composition of fleshy fruits can influence gut retention times and therefore dispersal distances (Murray et al., 1994).
Because frugivorous birds can only swallow fruits that are smaller than their gape size, restrictions in the size of frugivore assemblages associated with increased seed size might also curtail seed dispersal distances (see Burns, 2013). However, most fruit-eating animals inhabiting oceanic islands (e.g. birds, bats and reptiles) spend their entire lives in terrestrial environments. Seeds in the fruits they consume are therefore deposited in terrestrial environments. Although seabirds occasionally eat fleshy fruits, and they can sometimes be important dispersers of coastal plants (Calvino-Cancela, 2011), their relative contribution to terrestrial frugivore communities is typically negligible. As a result, seeds in fleshy fruits are unlikely to be deposited in the ocean. This cost of dispersal (see Bonte et al., 2012) forms the backbone of the loss of dispersal ability hypothesis. Its predictions hinge on the assumption that individuals with greater powers of dispersal are more likely to swept out to sea. On the other hand, individuals with poorer powers of dispersal are less likely to be dispersed at sea and are therefore at a selective advantage. Endozoochorous species appear to violate this assumption, and if so, the loss of dispersal ability hypothesis does not apply to them.
By providing testable predictions for how dispersal evolves on isolated islands, the loss of dispersal ability hypothesis has paved the way towards a better understanding of general trends in island evolution. However, Garc ıa-Verdugo et al. (2017) insightful study may be a turning point in our understanding of the dispersal capacity of island organisms. It now appears that the predictions of the loss of dispersal ability hypothesis are rarely upheld-island plants do not regularly evolve reduced dispersal potential. Furthermore, when the loss of dispersal ability is observed, it could have evolved passively, for reasons that are unrelated to dispersal potential. Lastly, its conceptual domain does not clearly encompass all modes of seed dispersal.
It is unclear whether the loss of dispersal ability hypothesis remains a valid hypothesis for island animals. Many flightless birds that are endemic to isolated islands evolved from volant ancestors (e.g. the dodo, Raphus cucullatus; kiwi, Apteryx spp.; moa-nalo, Thambetochenini). Selection could have acted directly on their capacity to fly to lower the probability they would be lost at sea, in accordance with the loss of dispersal ability hypothesis. Indeed, many island birds develop a behavioural reluctance to cross water bodies prior to the evolution of morphological changes to their wings (Diamond, 1981; c.f. Imbert and Ronce (2001)). Alternatively, selection could have favoured large body size without concomitant increases in wing size (see Mitchell et al., 2014;Yonezawa et al., 2017).
Critically evaluating the loss of dispersal ability hypothesis need not be a dead end to future research on the topic. Instead, widening the scope of research beyond its predictions may help to accelerate interest, and progress, in our understanding of the evolution of dispersal potential in island populations. Many interesting questions remain unanswered. For example, is seed gigantism a repeated pattern in island evolution? If so, why might large seeds be selectively advantageous? How often does the evolution of seed gigantism lead to an indirect loss of dispersal ability? Answers to these and other important questions await a fresh approach to the problem. Email: kevin.burns@vuw.ac.nz