Toward a phylogenomic classification of magnoliids

Abstract

Premise: Magnoliids are a strongly supported clade of angiosperms. Previous phylogenetic studies based primarily on analyses of a limited number of mostly plastid markers have led to the current classification of magnoliids into four orders and 18 families. However, uncertainty remains regarding the placement of several families.

Methods: For the first comprehensive phylogenomic analysis of magnoliids as a whole, we sampled 235 species from 199 (74%) genera and representing all families and most subfamilies and tribes. We analyzed newly generated data from the Angiosperms353 probe set using both coalescent and concatenation analyses and testing the impact of multiple filtering and alignment strategies.

Results: While our results generally provide further support for previously established phylogenetic relationships in both magnoliids as a whole and large families including Annonaceae and Lauraceae, they also provide new evidence for previously ambiguous relationships. In particular, we found support for the position of Hydnoraceae as sister to the remainder of Piperales, though evidence was conflicting, and resolved the backbone of relationships among most genera of Myristicaceae. Different analytical strategies tended to have rather small effects on branch support and topology.

Conclusions: Although some of our results are limited by low gene recovery for a number of taxa and significant gene tree conflict for some relationships, this study represents a significant step toward reconstructing the evolutionary history of a major lineage of angiosperms. Based on these results, we present an updated phylogenetic classification for magnoliids, recognizing 21 families, summarizing previously established subfamilies and tribes, and describing new tribes for Myristicaceae.

Keywords: Angiosperms353; Canellales; Laurales; Magnoliales; Piperales; magnoliids; phylogenetic classification; phylogenomics.

Figure 1

Floral diversity in Canellales, Piperales, and Laurales. (A) Canellaceae: Canella winterana (L.) Gaertn.; (B) Winteraceae, Winteroideae: Drimys lanceolata (Poir.) Baill.; (C) Hydnoraceae, Hydnora visseri Bolin, E. Maass & Musselman; (D) Asaraceae: Asarum europaeum L.; (E) Aristolochiaceae: Aristolochia gigantea Mart. & Zucc.; (F) Saururaceae: Houttuynia cordata Thunb.; (G) Piperaceae, Piperoideae: Peperomia tetraphylla (G. Forst.) Hook. & Arn.; (H) Calycanthaceae, Calycanthoideae: Calycanthus occidentalis Hook. & Arn.; (I) Calycanthaceae, Idiospermoideae: Idiospermum australiense (Diels) S.T. Blake; (J) Siparunaceae: Glossocalyx longicuspis Benth.; (K) Gomortegaceae: Gomortega keule (Molina) Baill.; (L) Atherospermataceae: Doryphora sassafras Endl.; (M) Monimiaceae, Monimioideae: Peumus boldus Molina; (N) Lauraceae, Hypodaphnideae: Hypodaphnis zenkeri (Engl.) Stapf; (O) Lauraceae, Laureae: Persea americana L. Photo credits: (A) Kevin Nixon (http://www.diversityoflife.org/); (B, D, E, G–I, L–O): Hervé Sauquet; (C) Julia Naumann; (F) Laetitia Carrive; (J) Thomas L.P. Couvreur; (K) Ana Almeida.

Figure 2

Floral diversity in Magnoliales. (A) Myristicaceae, Mauloutchieae: Pycnanthus angolensis (Welw.) Warb.; (B) Myristicaceae, Scyphocephalieae: Scyphocephalium mannii (Benth.) Warb.; (C) Myristicaceae, Horsfieldieae: Horsfieldia sylvestris (Houtt.) Warb.; (D) Myristicaceae, Viroleae: Virola sebifera Aubl.; (E) Degeneriaceae: Degeneria vitiensis I.W. Bailey & A.C. Sm.; (F) Himantandraceae: Galbulimima baccata F.M. Bailey; (G) Magnoliaceae: Liriodendron tulipifera L.; (H) Magnoliaceae: Magnolia sororum Seibert; (I) Eupomatiaceae: Eupomatia laurina R. Br.; (J) Annonaceae, Anaxagoreoideae: Anaxagorea brevipes Benth.; (K) Annonaceae, Ambavioideae: Meiocarpidium oliverianum (Baill.) D.M. Johnson & N.A. Murray; (L) Annonaceae, Annonoideae: Cymbopetalum baillonii R.E. Fr.; (M) Annonaceae, Annonoideae: Xylopia aethiopica (Dunal) A. Rich.; (N) Annonaceae, Malmeoideae: Sirdavidia solannona Couvreur & Sauquet; (O) Annonaceae, Malmeoideae: Mitrephora diversifolia (Span.) Miq. Photo credits: (A, E, F, I): Hervé Sauquet; (B, D, G, J–N) Thomas L.P. Couvreur; (C, H, O) Zacky Ezedin.

Figure 3

Heatmap of exon recovery. Each cell corresponds to the number of exons retained with different filtering thresholds. The x‐axis is the minimum proportion of the exon length recovered (l), and the y‐axis is the minimum proportion of samples (i) recovered. For example, for 191 exons, at least 25% of the length of the exon was recovered in 75% of samples. The dashed white lines are contours to show the slope of the surface. Crosses indicate the thresholds used to generate the range of trees assessed in this study.

Figure 4

Summary of phylogenetic relationships in magnoliids obtained in the coalescent analysis of Angiosperms353 nuclear data. Triangles represent collapsed clades (for full detail, see Figure 5). Tip and clade labels correspond to the updated phylogenetic classification of magnoliids presented in this paper. Dashed branches received local posterior probabilities <0.95 in this analysis.

Figure 5

Phylogenetic relationships in magnoliids obtained in our main coalescent analysis (same tree as Fig. 4; also provided in high resolution in Appendix S6). This tree was reconstructed using the following methods: aligned with MAFFT, trimmed with gblocks option “–b2 = 0”, and filtering thresholds of r = 10%, l = 50% and i = 50%. This tree achieved the highest combined support score of those compared (Appendix S10). At each node, the larger grey circle indicates quartet support (QS), and the inner black circle represents local posterior probability (local PP) for the subtending branch. Note that branch lengths are in coalescent units.

Figure 5

Phylogenetic relationships in magnoliids obtained in our main coalescent analysis (same tree as Fig. 4; also provided in high resolution in Appendix S6). This tree was reconstructed using the following methods: aligned with MAFFT, trimmed with gblocks option “–b2 = 0”, and filtering thresholds of r = 10%, l = 50% and i = 50%. This tree achieved the highest combined support score of those compared (Appendix S10). At each node, the larger grey circle indicates quartet support (QS), and the inner black circle represents local posterior probability (local PP) for the subtending branch. Note that branch lengths are in coalescent units.

Figure 6

Topology tests to understand gene tree conflict for the placement of (A) Hydnoraceae and (B) Meiocarpidium. The left side is a summary tree with relevant branches colored and focal taxa bolded. Branch labels correspond to those used in the x‐axis labels in the bar plots on the right. The bar plots show the frequency of each quartet, labelled below the bar. For the tree in (B), we combined Magnoliaceae, Himantandraceae and Degeneriaceae into a single tip.