Gene Trees and Species Trees: Lessons from morning glories

1. Gene Trees and Species Trees:Lessons from morning glories Lauren A. Eserman & Richard E. Miller Department of Biological Sciences Southeastern Louisiana University

2. Introduction • DNA sequences are an important source of data for phylogenetic reconstruction • Single-gene trees were considered exciting and sufficient at one time • Chase and 41 other authors, 1993 • Phylogeny of angiosperms using rbcL

3. Introduction • Next sequenced additional gene regions • Philosophical argument for “total evidence” –more data will strengthen the ability to determine species relationships • Used concatenated datasets to implement this idea • Still dominates the way species trees are estimated

4. Introduction • Population genetics and coalescent theory emphasize that genes have unique histories • Gene trees do not always reflect the true species history

5. Introduction • Gene tree heterogeneity can come about by: • Gene duplication events • Horizontal gene transfer • Incomplete lineage sorting (deep coalescence) • Branch length heterogeneity Edwards, 2009 • This provides evidence against concatenation

6. Introduction • Paradigm shift in systematics? (Edwards, 2009) • Moving away from notion that gene trees show true species relationships • Promotes synergism of phylogenetic systematics with population genetics and coalescent theory

7. Introduction • Application of the paradigm shift: • Use collective information from multiple gene trees to estimate a species tree • Consider conflicting results, valid alternative hypotheses for species relationships

8. Research Objectives • Explore how gene trees with different phylogenetic signal influence the estimated species tree • Using 28 gene trees • Effects of concatenation on estimated species tree • Alternative objective is to obtain an understanding of species relationships for the organisms of interest (not discussed here)

9. Study Organisms • Morning glories are generally species of the genus Ipomoea (not monophyletic) • Focus on tribe Ipomoeeae • Ipomoea + 9 other genera • c. 900 species • Distributed throughout the subtropics and tropics of the world Ipomoea nil

10. Methods1. Bayesian phylogenetic analysisof 28 gene trees • Obtained 28 gene regions for species of Ipomoeeae based on our research and additional genes from GenBank • Number of taxa ranged from 6 to 129 • Alignments using MAFFT and manually adjusted • Models of nucleotide substitution chosen using jModelTest • Gene trees constructed using MrBayes v3.1.2 • 4 runs, 4 chains sampling every 100-200 generations • Runs were continued until stationary distribution was estimated • Burn-in determined as asymptote in plots of total tree length by generations • Convergence criteria: • Same topology among 4 runs • PP of clade support ±3% among 4 runs • Majority rule consensus tree constructed from a combination of post-burnin trees from all 4 runs

11. Methods1. Synthesis of 28 gene trees • ITS tree used as working hypothesis • Densest taxon sampling (129 species) • Good intragenericresolution • NOTE: Not assuming this is the species tree – rather, a working hypothesis to compare to other gene trees • Topology and clade support of 27 other gene trees compared to ITS gene tree

12. DFRB-2 myb1 Results1. Same relationships between ITS and other genes PHAR PHAR MINA MINA

13. DFRB-2 Results2. Individual species with unique positions not shown in any other gene tree bHLH3 PHAR PHAR MINA

14. Results3. Major alternative topology in CHSE CHSE PHAR MINA

15. bHLH2 waxy 1 Results4. Identify new unnamed clades ‘OBSC’ PHAR MINA PHAR TRIC MINA CALO BATA ‘AMNI’ ‘VIOL’ BATA

16. Methods2. Concatenated dataset UF3GT DFRB-1 • To address the issue of concatenation, constructed concatenated dataset using 10 genes • All gene trees showed similar topologies CHI

17. Results10-gene concatenated dataset • Maintains topologies of individual gene trees

18. Concatenated dataset • What happens when one more gene is added? • Add CHSE to 10-gene concatenated dataset • Alternative topology • All coding region • No indels

19. Results11-gene concatenated dataset • Exhibits topology of CHSE – new gene overwhelms this analysis

20. 10-gene concatenated dataset 11-gene concatenated dataset

21. BEST AnalysisBayesian Estimation of Species Trees(Liu, 2008) • Incorporates a multispecies coalescent model to estimate species tree from many gene trees • Methods: • 11-gene concatenated dataset • 2 runs, 4 chains • 8 million generations (did not reach convergence on topology)

22. BEST Analysis • Results: • Clade present in CHSE appears in BEST tree • Overall topology differs • Species pairs supported throughout

23. Discussion • Analysis of 28 gene trees • Provides an estimate of species tree • Alternative hypotheses for species relationships have emerged • Overall congruence among gene trees

24. Discussion – Concatenated datasets • Total evidence philosophically justified but misleads results because of gene tree heterogeneity • Shown clearly in 11-gene concatenated dataset • Left with idea that we have two alternative hypotheses of species relationships • Two estimates of the species tree

25. Discussion – Concatenated datasets • Can now appreciate how a single gene can overwhelm results of a concatenated dataset • Topology of CHSE dominated

26. Acknowledgements Research Assistants: A. McDaniel, K. Robichaux, W. Terry, S. Major, H. Echlin, F. St. Cyr Seed Donations: M. Clegg, M. Rausher, J. A. McDonald, J. Miller P. Tiffin, B. Zufall, S.M. Chang Ipomoea purpurea