Adaptive radiations should not be simplified: The case of the danthonioid grasses

Abstract

Although much of extant diversity is probably the product of evolutionary radiations, the special case of adaptive radiations has not yet been thoroughly explored. Adaptive radiations are postulated to occur when a lineage is exposed to new ecological opportunities, where it can diversify ecologically. We argue that adaptive radiations have two characteristics. Firstly, the diversification rate accelerates initially, and is then followed by a density-dependent slow-down. Secondly, traits relevant to the new ecological opportunity should evolve at or just before the radiation. We also argue that a correct identification of adaptive radiations is dependent on the phylogenies underlying the diversification dynamics being sampled adequately (i.e. comprehensive species sampling), and that the traits should be treated continuously if they exhibit a biological continuum and not be over-simplified into binary traits. Here, we test the hypothesis that much of the extant diversity of the south-temperate grass subfamily Danthonioideae is the result of two geographically separated but contemporaneous adaptive radiations, in response to Late-Miocene—Pliocene aridification and increasingly seasonal climates. We show that both Pentameris (83 African species) and Rytidosperma (73 Australasian-South American species) exhibit accelerations in diversification rates followed by linear density-dependent declines. We also show that two selected traits show differential evolutionary regimes with different evolutionary optima, and that these are linked to changes in the diversification rate. These results are consistent with these being adaptive, and putatively parallel, radiations. However, by mapping traits over the whole danthonioid phylogeny, it is evident that no identified trait or trait combination is sufficient and necessary for adaptive radiations. Furthermore, we show that simplifying the traits to binary gives a strong but potentially erroneous link between trait shift and diversification rate shift.