Island biogeography has traditionally predominantly focused on abiotic drivers of colonization, extinction, and speciation. Colonization of islands by plants, however, could be limited by the absence of symbionts. Most plants, for example, form symbioses with mycorrhizal fungi, whose limited dispersal to islands could act as a plant colonization filter. Our global-scale analyses of 200,000 plant species across ~2500 regions worldwide provide evidence for this hypothesis as plant species that associate with mycorrhizal fungi are disproportionately under-represented in native island floras compared to mainlands. In order to test patterns of mycorrhizal-plant distributions, we modeled regional plant species richness for mycorrhizal and non-mycorrhizal plants using generalized linear mixed effects models (GLMMs) and generalized linear models (GLMs). Plant species occurrence data, native status (native, naturalized), and all additional parameters associated with regional characteristics are from the Global Alien Naturalized Floras (GloNAF) and from the Global Inventory of Floras and Traits (GIFT gift.uni-goettingen.de) databases. We use four review papers to determine plant species mycorrhizal status. Ultimately, we identify mycorrhizal association as an overlooked driver of global plant biogeographic patterns with legacies on contemporary island biogeography.
Results/Conclusions
We find strong evidence that the colonization of islands by plants worldwide is driven by a mycorrhizal filter as mycorrhizal plant species are disproportionately under-represented in contemporary native floras of islands compared to that of mainlands (p < 0.0001). Moreover, on islands, the proportion of mycorrhizal plants increases with island age (p = 0.01) consistent with the expectation that mycorrhizal symbionts accumulate over time. The mycorrhizal filter on island colonization leaves legacies that affect the current risk of plant naturalizations. We find naturalized floras of islands to be disproportionately more mycorrhizal compared to mainland naturalized floras (p < 0.001) suggesting relaxation of the mycorrhizal filter by human means. In contrast, on mainlands, naturalized floras have a decreased proportion of mycorrhizal plant species, as expected from anthropogenic disturbance of symbionts, primarily though intensive agricultural soil modification. As further evidence of human-mediated impacts, the proportion of mycorrhizal plants in the naturalized floras of islands increases with human land use intensity (p < 0.001). Our results highlight that on islands, mycorrhizal species are the major invasion risk, while on mainlands, non-mycorrhizal species are the dominant risk. This has implications for management, supporting different prevention focuses on these two distinct land types.