The monocot family Bromeliaceae comprises approximately 3,140 species distributed in tropical and subtropical regions of the New World from the southern United States to southern Argentina. The family is subdivided into eight subfamilies; the most species-rich are Tillandsioideae and Bromelioideae. Taxonomic concepts within Bromeliaceae are highly problematic, since discriminating morphological characters have been shown to be homoplastic or plesiomorphic. The present study aims to provide a robust phylogenetic framework for Bromeliaceae, especially for the most diverse and complicated subfamilies Bromelioideae and Tillandsioideae. Resulting phylogenies provide a basis to estimate the usefulness of morphological characters and to propose or strengthen hypotheses concerning evolutionary traits, biogeography, age and origins of bromeliads.
The main questions raised are:
(1) Do additional sequence data from the plastid genome and a wider sampling within Bromeliaceae provide a better resolved, robust phylogenetic framework? What are the reasons for the low DNA sequence divergence observed up to now?
(2) Can nuclear DNA sequences be successfully implemented for phylogenetic reconstruction? What are the challenges to optimize nuclear markers and do they perform better than plastid loci?
(3) Can the resulting phylogeny based on plastid and nuclear DNA sequences together with the re-evaluated morphological characters provide a reasonable, stable classification?
To provide a more robust phylogenetic hypothesis for the classification of Bromeliaceae, eight rapidly evolving plastid DNA markers (atpB-rbcL, matK, ndhF, psbA-trnH, rpl32-trnL, rps16, trnL intron, and trnL-trnF) and 90 bromeliad species were included in the current study and analysed using maximum-parsimony, maximum-likelihood, and Bayesian approaches. Results support the formerly proposed eight-subfamily classification based on the single plastid gene ndhF. Support values for five of the subfamilies increase, but that for Lindmanioideae, Puyoideae, and Bromelioideae decrease as a result of expanded taxon sampling, including several more divergent species. The initially proposed monophyletic origin of Puyoideae cannot be unambiguously confirmed. Calibration of the resulting phylogeny against time and biogeographic analysis reveals that Bromeliaceae originated in the Guayana Shield about 100 million years ago (Ma) and spread radially into adjacent areas ca. 16—13 Ma. Extant lineages arose between 20 and 5 Ma. Andean uplift facilitated diversification of core Tillandsioideae about 14.2 Ma and Bromelioideae 10.1 Ma, the latter having their greatest diversity in the Brazilian Shield due to dispersal from the Andes. The most species-rich genera did not appear before 8.7 My with a high diversification between 5 and 4 Ma, which is most likely the reason for the comparatively low sequence divergence.
To test the usefulness of nuclear regions for phylogenetic reconstruction in Bromelioideae, DNA sequences of part of the low-copy nuclear gene phosphoribulokinase (PRK) and five plastid loci (matK, 3'trnK intron, trnL intron, trnL-trnF, and atpB-rbcL) were investigated. Phylogenetic trees obtained from analyses of the PRK sequences do not contradict trees obtained from plastid markers. The PRK matrix shows a significantly higher number of potentially PICs (phylogenetically informative characters) than the plastid dataset (16.9% vs. 3.1%), which improves resolution and support in the resulting trees. Although PRK is not able to resolve relationships completely, the combined analysis with plastid markers yields good support for several uncertain relationships observed previously. The early diverging lineages can be identified (“basal bromelioids”) and the remainder of the subfamily clustered into a highly supported clade (“eu-bromelioids”). Results indicate that taxonomic circumscriptions within “core bromeliads” are still insufficient, and relationships complex and difficult to solve. Several genera appear polyphyletic, and Aechmea as well as Quesnelia remain the most complicated genera of the subfamily. Most-parsimonious character state reconstructions for two evolutionary traits (tank habit, sepal symmetry) indicate that both characters have undergone few transitions within the subfamily and thus are not as homoplasious as previously assumed. The comparative study of nuclear DNA sequences within Tillandsioideae shows that some nuclear markers are able to provide more information and a higher degree of resolution in phylogenetic trees than plastid markers. However, their utility does not depend only on sequence variability, but also on methodological challenges in using traditional Sanger-sequencing. The internal transcribed spacer of nuclear ribosomal DNA (ITS nrDNA) is not recommended as a suitable marker for phylogenetic investigations of Bromeliaceae due to the presence of strong secondary structures which create problems in amplification and sequencing as well as its low number of PICs for resolving deeper nodes. Amplified fragments of the genes malate synthase (MS) and RNA polymerase II, beta subunit (RPB2) are not helpful due to their small size and limited number of PICs. Glucose-6-phosphate isomerase (PGIC), nitrate reductase (NIA), and xanthine dehydrogenase (XDH) need to be further investigated. Phosphoribulokinase (PRK) and phytochrome C (PHYC) are useful nuclear markers and able to provide considerable resolution in phylogenetic trees, but some relationships are poorly supported.
The combined analysis of nuclear DNA sequence data (PRK, PHYC) with the already existing plastid DNA sequence data (atpB-rbcL, matK, rbcL, partial rbcL-accD, rps16 intron, partial trnK intron, trnL intron and trnL-trnF) shows a significant increase of resolution within phylogenetic trees of Tillandsioideae. Nine accepted genera can be re-circumscribed and three new genera are described taxonomically based on morphology. For morphologically distinct species groups within Racinaea and Tillandsia, two new subgenera are erected. Viridantha has been downgraded to subgeneric rank. Poor sampling within the Cipuropsis-Mezobromelia clade and missing support for clades within Tillandsia prevent the recognition of further taxonomic groups.
Figures - uploaded by
Michael H J BarfussAuthor contentAll figure content in this area was uploaded by Michael H J Barfuss
Content may be subject to copyright.