Diversification of Angraecum (Orchidaceae, Vandeae) in Madagascar: Revised Phylogeny Reveals Species Accumulation through Time Rather than Rapid Radiation

Abstract

Angraecum is the largest genus of subtribe Angraecinae (Orchidaceae) with about 221 species. Madagascar is the center of the diversity for the genus with ca. 142 species, of which 90% are endemic. The great morphological diversity associated with species diversification in the genus on the island of Madagascar offers valuable insights for macroevolutionary studies. Phylogenies of the Angraecinae have been published but a lack of taxon and character sampling and their limited taxonomic resolution limit their uses for macroevolutionary studies. We present a new phylogeny of Angraecum based on chloroplast sequence data (matk, rps16, trnL), nuclear ribosomal (ITS2) and 39 morphological characters from 194 Angraecinae species of which 69 were newly sampled. Using this phylogeny, we evaluated the monophyly of the sections of Angraecum as defined by Garay and investigated the patterns of species diversification within the genus. We used maximum parsimony and bayesian analyses to generate phylogenetic trees and dated divergence times of the phylogeny. We analyzed diversification patterns within Angraecinae and Angraecum with an emphasis on four floral characters (flower color, flower size, labellum position, spur length) using macroevolutionary models to evaluate which characters or character states are associated with speciation rates, and inferred ancestral states of these characters. The phylogenetic analysis showed the polyphyly of Angraecum sensu lato and of all Angraecum sections except sect. Hadrangis, and that morphology can be consistent with the phylogeny. It appeared that the characters (flower color, flower size, spur length) formerly used by many authors to delineate Angraecum groups were insufficient to do so. However, the newly described character, position of the labellum (uppermost and lowermost), was the main character delimiting clades within a monophyletic Angraecum sensu stricto. This character also appeared to be associated with speciation rates in Angraecum. The macroevolutionary model-based phylogeny failed to detect shifts in diversification that could be associated directly with morphological diversification. Diversification in Angraecum resulted from gradual species accumulation through time rather than from rapid radiation, a diversification pattern often encountered in tropical rain forests.

Fig 1. Phylogenetic relationships within subtribe Angraecinae.

50% Bayesian majority-rule consensus tree from combined plastid data (matK, rps16 and trnL). Values above branches or at nodes represent posterior probability (PP) and bootstrap percentage (BP) support. Dashes represent branches that collapsed in MP strict consensus tree. Colors represent geographic distribution of taxa; in bold are Angraecum sensu Garay species. Abbreviations in brackets denote sections sensu Garay: Aca = Acaulia, Agd = Angraecoides, Ang = Angraecum, Arc = Arachnangraecum, Bor = Boryangraecum, Chl = Chlorangraecum, Fil = Filangis, Gom = Gomphocentrum, Had = Hadrangis, Hum = Humblotiangraecum, Lem = Lemurangis, Lep = Lepervenchea, Pct = Pectinaria, Per = Perrierangraecum, Psj = Pseudojumellea.

Fig 2

Posterior probability distributions for the speciation rates (in Ma) of morphological characters using equal rate speciation (μ0 ~ μ1) with the BiSSE model and equal rate extinction (λi ~ λj) with the MuSSE model: flower color (A), labellum position (B), flower size (C), and spur length (D). Abbreviation: v, very.

Fig 3. Configuration shifts from the 95% credible set sampled by BAMM from the Angraecinae phylogeny and evolutionary rates through time.

The intensity of colors on branches reflects the relative probability density of speciation rates (cool colors = slow, warm = fast). Black circles denote the position of the macroevolutionary regime shifts present in each sample. Blue curve indicates the mean speciation rate-through-time trajectory of Angraecinae in million years. Values above dendograms (f: configuration sampled frequency) indicate the marginal probability of rate shifts observed on branches across the posterior distribution of macroevolutionary rate shift configurations.

Fig 4. Best configuration shifts from the 95% credible set sampled by BAMM for the evolution of spur length across the phylogeny of Angraecinae.

Color intensity on branches reflects the relative probability density of the instantaneous rate of phenotypic evolution. Black circles denote the position of the macroevolutionary regime shifts present in each sample. Blue curve denotes the mean evolution rate-through-time trajectory.

Fig 5

Ancestral state reconstructions of floral characters in Angraecinae implemented in ‘diversitree’; colors represent character states of the labellum position (uppermost and lowermost) and pie charts represent the probability of ancestral states at nodes. C denotes taxa illustrated in the pictures to the right to represent the flower shape of each clade except for Jumellea which is represented by Jumellea comorensis (not sampled in the phylogeny). Photo: Andriananjamanantsoa.