Island floras are not necessarily more species poor than continental ones

Oceanic islands typically exhibit a lower number of species per unit area than continents owing to their geographical isolation (MacArthur & Wilson, 1967). This relationship demonstrates a remarkable consistency across spatial and taxonomic scales (MacArthur & Wilson, 1967) and has significant evolutionary consequences because island organisms are expected to respond to the empty niche space through a series of major transformations in their life-history traits (see Whittaker & Fern andez-Palacios, 2007). The biological, ecological and physiological changes are known as island syndromes, with the loss of dispersal power, changes in body size and the development of woodiness in herbaceous plant lineages being some of the most striking examples (Whittaker & Fern andezPalacios, 2007). In their seminal book, MacArthur & Wilson (1967) observed that the degree of impoverishment in bird faunas distributed across the archipelagos of Melanesia, Micronesia and Polynesia increased from islands near to New Guinea to intermediate and far islands (distance of > 3200 km from the mainland). This implies that the mismatch between species richness per unit area on continents and on islands increases with the degree of isolation of the latter, as species are progressively filtered out depending on their long-distance dispersal capacities (Whittaker & Fern andez-Palacios, 2007). However, there is mounting evidence for the existence of lineage-specific species–area relationships associated with long-distance dispersal capacities (Triantis et al., 2012; Pati~ no et al., 2014b) and this raises the question of whether communities of highly mobile organisms are necessarily less species rich on islands than on continents and, hence, whether they exhibit typical island syndromes. Here, data from a global survey of 421 land plant floras (Pati~ no et al., 2014b) were submitted to linear mixed-effects models to analyse variation in log(species richness) depending on log(area) while controlling for variation in the factors geology (continents, oceanic islands and continental islands), plant group (seed plants, ferns and fern allies, liverworts and mosses), biome and realm (the latter two as defined by Olson et al., 2001). The analyses demonstrate that whereas species richness patterns of seed plants conform to the expectations of the island impoverishment theory, spore-producing plants do not (Fig. 1). The species–area models obtained indicate that seed plants exhibit, on average, a continual decrease in species richness per unit area from continents to continental islands to oceanic islands (Fig. 1). Mosses only partially follow such a pattern, with oceanic islands showing the lowest species richness, but with continental regions and continental islands showing similar values for comparable units of area. In liverworts and ferns, the pattern is different. Whereas islands with a small area and presumably limited ecological complexity show impoverished floras, intermediate and large sized oceanic and continental islands display similar or even slightly higher levels of species richness than continents. Thus, oceanic island floras of liverworts and ferns with a log(area) ranging from three to four units (10– 10 km), and continental island floras with a log(area) from five to six (10– 10 km), exhibit higher species richness compared to continents, a pattern at odds with expectations (Fig. 1). Possible explanations for the patterns observed in spore-producing land plants are: (1) higher speciation rates on islands than on continents; (2) lower extinction rates on islands; (3) higher dispersal power than in seed plants; or (4) a higher carrying capacity on large islands per unit area than on continents. The generally very low rates of endemism in oceanic island ferns and bryophytes (Pati~ no et al., 2014a) suggest a lack of extensive island radiations in these groups, thus pointing to the other hypotheses. A global comparison of island and mainland environments revealed that wet climates typical of tropical and, especially, temperate rain forest biomes are over-represented on islands, whereas hot and dry desert climates are under-represented (Weigelt et al., 2013). These general features of islands are particularly suitable for organisms such as bryophytes and ferns that rely greatly on environmental humidity for water uptake (e.g. poikilohydry in filmy ferns and bryophytes) and exhibit lower temperature optima than flowering plants (Vanderpoorten & Goffinet, 2009; Mehltreter et al., 2010). Given the ecological requirements of these spore-producing plant groups with higher dispersal capabilities than in seed plants, such a combination of humidity and mild temperatures on islands could lead to a reduced extinction risk following colonization and a potentially higher carrying capacity than in continental regions of similar area. The fact that island spore-producing floras are not impoverished suggests that the shifts in the expression of reproductive traits observed in island bryophytes, such as the higher investment in asexual reproduction as well as the higher frequency of fertile shoots (Pati~ no et al., 2013), do not represent a response to empty niche space, as island biogeography theory predicts. Rather, such changes may reflect a response to the specific climate conditions that prevail on islands. Such environmental conditions may also be found on continents, but proportionally represent a smaller fraction of continental climates (Weigelt et al., 2013) and, hence, the emergence of a bias in life-history traits associated with such climates would not be linked to geographical isolation either. In seed plants, experimental support for the loss of dispersal capacity on islands is equivocal (contrast, e.g. Whittaker & Fern andez-Palacios, 2007 and Talavera et al., 2012) and in sporeproducing plants, a study on the dispersal capacity of the moss Rhynchostegium riparioides failed to demonstrate any significant difference in migration rates in insular and continental populations (Hutsem ekers et al., 2011). This raises the question of whether other typical island syndromes,