Ants sow the seeds of global diversification in flowering plants

Abstract

Background: The extraordinary diversification of angiosperm plants in the Cretaceous and Tertiary periods has produced an estimated 250,000-300,000 living angiosperm species and has fundamentally altered terrestrial ecosystems. Interactions with animals as pollinators or seed dispersers have long been suspected as drivers of angiosperm diversification, yet empirical examples remain sparse or inconclusive. Seed dispersal by ants (myrmecochory) may drive diversification as it can reduce extinction by providing selective advantages to plants and can increase speciation by enhancing geographical isolation by extremely limited dispersal distances.

Methodology/principal findings: Using the most comprehensive sister-group comparison to date, we tested the hypothesis that myrmecochory leads to higher diversification rates in angiosperm plants. As predicted, diversification rates were substantially higher in ant-dispersed plants than in their non-myrmecochorous relatives. Data from 101 angiosperm lineages in 241 genera from all continents except Antarctica revealed that ant-dispersed lineages contained on average more than twice as many species as did their non-myrmecochorous sister groups. Contrasts in species diversity between sister groups demonstrated that diversification rates did not depend on seed dispersal mode in the sister group and were higher in myrmecochorous lineages in most biogeographic regions.

Conclusions/significance: Myrmecochory, which has evolved independently at least 100 times in angiosperms and is estimated to be present in at least 77 families and 11 000 species, is a key evolutionary innovation and a globally important driver of plant diversity. Myrmecochory provides the best example to date for a consistent effect of any mutualism on large-scale diversification.

Figure 1. Convergent evolution of elaiosomes as an adaptation for seed dispersal by ants (myrmecochory) in angiosperm plants.

(A) Rhytidoponera metallica ant holding a seed of Acacia neurophylla by the elaiosome during seed transport [Photograph by Benoit Guenard]. (B) Myrmecochore diversity hotspots (in black) and number of myrmecochorous plant lineages in major biogeographic regions (in shades of grey). Lineages distributed in more than one region (not shown) are Holarctic (n = 14), Old World (n = 5), pan-tropical (n = 2), or worldwide (n = 10).

Figure 2. Diversification in sister lineages of myrmecochorous and non-myrmecochorous plants by seed dispersal mode in the sister group (A–B) and by biogeographic region (C–D).

Top panels show the number of contrasts in which the mymecochorous lineage is more diverse (closed bars), or when the non-myrmecochorous lineage is more diverse (open bars). Bottom panels show species diversity contrasts (difference in log-transformed species numbers; mean±1 S.E.) to illustrate the magnitude of differences between sister lineages, with positive values indicating more species in the myrmecochorous lineage than in its sister group. The number of contrasts is shown above the X axis in bottom panels. General linear mixed models with sister-group dispersal mode (fixed effect) and distribution type (random effect) showed that dispersal mode did not influence either the direction (Model 1: logistic regression, F _4,85 = 0.751, P = 0.560) or the magnitude of contrasts (Model 2: F _4,85 = 0.756, P = 0.557). The evaluation of the random effect of biogeographic region by comparing intercept S.D. to residual S.D. showed that variation by biogeographic distribution type also was not influential (Model 1: intercept S.D.: 0.024<residual S.D.: 0.999; Model 2: intercept S.D.: 0.249<residual S.D.: 0.798).