Molecular Phylogenetics and Evolution 82 (2015) 111–117
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution
journal homepage: www.elsevier.com/locate/ympev
A phylogeny and biogeographic analysis for the Cape-Pondweed family Aponogetonaceae (Alismatales) ⇑ Ling-Yun Chen a, Guido W. Grimm b, Qing-Feng Wang a, Susanne S. Renner c, a Key Laboratory of Aquatic Botany and Watershed Ecology, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan 430074, Hubei, PR China b Swedish Museum of Natural History, Department of Palaeobiology, Svante Arrhenius Väg 7, 10405 Stockholm, Sweden c Systematic Botany and Mycology, University of Munich (LMU), Menzinger Strasse 67, 80638 Munich, Germany article info abstract
Article history: The monocot family Aponogetonaceae (Alismatales) consists only of Aponogeton, with 57 species occur- Received 2 August 2014 ring in Africa, Madagascar, India and Sri Lanka, Southeast Asia and Australia. Earlier studies inferred a Revised 21 September 2014 Madagascan or Australian origin for the genus. Aponogeton-like pollen is documented from the Late Cre- Accepted 9 October 2014 taceous of Wyoming, the early mid-Eocene of Canada, and the late mid-Eocene of Greenland. We Available online 22 October 2014 obtained nuclear and plastid DNA sequences for 42 species and generated a time-calibrated phylogeny, rooted on appropriate outgroups. Statistical biogeographic analyses were carried out with or without Keywords: the fossils incorporated in the phylogeny. The recent-most common ancestor of living Aponogetonaceae Alismatales appears to date to the mid-Eocene and to have lived in Madagascar or Africa (but not Australia). Three Aquatic plants Biogeography transoceanic dispersal events from Africa/Madagascar to Asia sometime during the Miocene could Integrating fossil geographic ranges explain the observed species relationships. As inferred in earlier studies, an ancient Australian species Molecular clock is sister to all other Aponogetonaceae, while the remaining Australian species stem from an Asian ances- tor that arrived about 5 million years ago. The family’s ancient Northern Hemisphere fossil record and deepest extant divergence between a single Australian species and an Africa/Madagascar clade are statis- tically well-supported and rank among the most unusual patters in the biogeography of flowering plants. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction of these families, all species-poor with together just 180 species (Les and Tippery, 2013; Iles et al., 2013). Aponogeton is a monocot genus of about 57 species, many of The disjunct geographic range of the Aponogetonaceae has been which are cultivated as aquarium plants. The genus occurs in Africa explained in mainly two ways. Thanikaimoni (1985) thought that (18 species), Madagascar (15 species), Sri Lanka, India, China, the family originated on Madagascar in the mid-Cretaceous and Southeast Asia, and the Malaysian region (together 11 species), dispersed from there to India, while the Indian plate was still close and Australia (13 species; van Bruggen, 1985; Hellquist and to Madagascar. He also followed Raven and Axelrod (1974: 600) in Jacobs, 1998; Jacobs et al., 2006; our Fig. 1). Aponogeton has long suggesting that the temperate Australian species A. hexatepalus been placed in its own family, the Aponogetonaceae, which might decent from a mid-Cretaceous dispersal event to Australia; together with 13 other families make up the Alismatales this is the only species of Aponogeton with two trimerous (Stevens, 2001 onwards), an order of mostly aquatic plants, includ- perianth whorls, while the remaining species retain only the inner ing sea grasses (Cymodoceaceae, Posidoniaceae, Ruppiaceae, whorl. Raven and Axelrod’s suggestion was based on a report of Zosteraceae), freshwater aquatics (Alismataceae, Aponogetona- Upper Cretaceous flowering and fruiting structures from ceae, Butomaceae, Hydrocharitaceae and Potamogetonaceae which Patagonia (Selling, 1947). Study of Selling’s material, however, also include a few sea grasses), plants of marshy habitats shows that it does not represent Aponogeton remains (Grímsson (Juncaginaceae, Maundiaceae, Scheuchzeriaceae, Tofieldiaceae), et al., 2014). and free-floating aquatics (most strikingly Lemnoideae and Pistia A different interpretation of the history of Aponogetonaceae in the Araceae). The Aponogetonaceae are the sister clade to eight was proposed by Les et al. (2005), who concluded that Aponogeton originated in Australia and expanded to its remaining distribution area from there. This was based on the position of A. hexatepalus as sister to the rest of the genus in DNA trees from nuclear and plastid ⇑ Corresponding author. sequences of ten species from Australia, six from Asia, three from E-mail address: [email protected] (S.S. Renner). http://dx.doi.org/10.1016/j.ympev.2014.10.007 1055-7903/Ó 2014 Elsevier Inc. All rights reserved. 112 L.-Y. Chen et al. / Molecular Phylogenetics and Evolution 82 (2015) 111–117
Madagascar, and one from Africa. Les et al. (2003) earlier inferred Australia. Lacking sufficient molecular data, Grímsson et al., of divergence times in Aponogeton, with 23.3 Ma for the split between course, were unable to study the modern family’s biogeography. the African A. distachyos and the Australian A. euryspermus, 15.3 Ma Here we provide a phylogeny and biogeographic analysis for the for that between A. distachyos and the Madagascan A. madagascar- Aponogetonaceae, based on sampling of African, Madagascan, iensis, 13.5 Ma for that of A. euryspermus and the Indian A. crispus, Indian and Chinese species (42 of the 57 species are included) and 15.8 Ma for that of the Indian A. rigidifolius and the Madaga- and formal statistical ancestral area reconstruction on a dated scan A. longiplumosa. These ages came from a strict clock model phylogeny, taking into account the pollen fossils. Ancestral area calibrated with a rate of 0.24% change per million years for trnK reconstructions (AAR) with fossil ranges included directly in the intron and 0.27% for the ITS region (such patristic distances need statistical runs may shed light on a clade’s earliest history, yet to be halved to calculate nucleotide change per site per million few such analyses have been carried out (but see Clayton et al., years). Les and Tippery (2013), after adding eight more species to 2009; Mao et al., 2012; Nauheimer et al., 2012). In such studies, the dataset of Les et al. (2005), upheld the earlier conclusion of fossils ranges are added into a Newick-format phylogenetic tree an Australian origin of the family. as sister to the clade to which the respective fossil has been Newly discovered fossil pollen from North America, Canada, assigned, using artificial branch lengths (since there is no DNA and Greenland, and rejection of Selling’s report of Upper Creta- sequence) that could reflect the fossil age. For the present study, ceous fossils from Argentina/Chile (Patagonia) caused Grímsson we explored this approach to test the hypothesis of Les et al. et al. (2014) to undertake a re-investigation of the family’s biogeo- (2005; Les and Tippery, 2013) of an Australian origin of graphic history using morphological data, compiled mainly from Aponogeton. literature, and molecular data available in GenBank. The three fos- sils are Aponogeton harryi from the Late Cretaceous (82–81 Ma) of Wyoming, USA, A. longispinosum from the early middle Eocene (ca 2. Materials and methods 46 Ma) of British Columbia, Canada, and A. hareoensis from the late middle Eocene (44–40 Ma) of West Greenland. Aponogeton and 2.1. Taxon sampling Butomus (Butomaceae) are the only Alismatales producing mono- sulcate pollen, and these two differ greatly in their exine sculptur- All sequences used in this study, with herbarium vouchers, geo- ing (Grímsson et al., 2014); Aponogeton pollen can therefore graphic origin, and GenBank accession numbers are listed in reliably be assigned as to family. These pollen finds imply that, Appendix S1 (see Supporting Information). The sequences come while the family today is confined to the tropics and subtropics from 45 individuals representing 11 of the 18 African species, 9 of the Old World, during the late Cretaceous and early Cenozoic, of the 11 Asian species, 10 of the 13 Australian species and 12 of it was widespread in the Northern Hemisphere from where it must the 15 Madagascan species. Based on Les and Tippery (2013) and have expanded its range into India, Africa, Madagascar, and Iles et al. (2013), Aponogetonaceae are the sister clade to the
(a)
44–40 Ma
46 Ma
82– 81 Ma
(b)
Fig. 1. Geographical distribution and photographs of Aponogetonaceae. Distribution was mainly compiled from literature: Lye (1989), Hellquist and Jacobs (1998), Jacobs et al. (2006), Zhou and Zhou (1992), and Sundararaghavan et al. (1982). (a) Aponogeton lakhonsis and its habitat (China; 28°380.00.2800N, 119°16016.5700E; 324 m alt.); (b) A. abyssinicus and its habitat (Kenya; 01°07041.8100S, 35°47.22.0800E; 1924 m alt.). Red stars indicate the location of Aponogetonaceae fossils (Grímsson et al., 2014). The straight line indicates the equator. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) L.-Y. Chen et al. / Molecular Phylogenetics and Evolution 82 (2015) 111–117 113
Cymodoceaceae, Juncaginaceae, Maundiaceae, Posidoniaceae, Pot- the major clades of Aponogeton. We used the BEAST package v. amogetonaceae, Ruppiaceae, Scheuchzeriaceae and Zosteraceae, 1.8.0 (Drummond et al., 2012), with Markov chain Monte Carlo and their divergence times were inferred by Iles et al. (2013). Given (MCMC) run for 10 million generations, sampling every 10,000 this existing framework, we included Tetroncium magellanicum,a generations. The first 10% of trees were discarded as burn-in, and species of Juncaginaceae, Scheuchzeria palustris, the single species the remaining trees were combined, using TreeAnnotator (part of of Scheuchzeriaceae, and as a more distant outgroup Butomus the BEAST package). As appropriate for our data, we applied a strict umbellatus, the single species of Butomaceae. clock model, calibrated with a matK substitution rate of 8.1 10 10 substitutions/site/year (s/s/y) from Lavin et al. (2005) and an ITS 2.2. Gene sequencing and alignment rate of 4.13 10 9 s/s/y from Kay et al. (2006). We did not use the 82–81 Ma, 46 Ma, or 44–40 Ma old pollen grains (Grímsson Total genomic DNA was extracted from herbarium specimens et al., 2014) as constraints because they cannot reliably be assigned or, in a few cases, silica-dried leaves using the DNeasy Plant Mini to any particular node in the extant species tree. Kit (QIAGEN GmbH, Hilden, Germany) or the NucleoSpin plant kit (Machery-Nagel, Düren, Germany). We sequenced the internal 2.5. Biogeographic analyses transcribed spacer regions (ITS1, ITS2) and 5.8S gene of the nuclear-encoded ribosomal DNA and the plastid trnK 5’ intron with For the AARs, we added the eight Alismatales families closest to an adjacent portion of the matK coding region. Primers used are Aponogetonaceae (see Taxon sampling) to the Newick tree listed in Appendix S2. PCR products were purified with the SanPrep obtained in the clock analysis, with their relationships constrained Column DNA Gel Extraction Kit (Sangon Biotech, Shanghai, China), to match the results of Janssen and Bremer (2004). The families’ then ligated to a pMD18-T vector (Takara Biotech Co., Dalian, geographic areas were coded as follows: Cymodoceaceae (16 China). Sequencing used either the primers M13F and M13R, spp., water-pollinated, largely in tropical and subtropical waters located in the vector, or the PCR primers and was carried out on of the Indo-West Pacific and Caribbean, but extending into Austra- an ABI 3730 automated sequencer at Tsingke Biotech Co. (Beijing, lia and Mediterranean; Kuo and McComb, 1998; Petersen et al., China). A total of 140 sequences were newly generated and 2014) coded as ‘South Asia – Southeast Asia – Australia’; Juncagin- submitted to GenBank. Alignment was done separately for the aceae ( 30 spp., sub-cosmopolitan, coastal and wetland herbs; von nuclear and plastid sequences using the MAFFT online version Mering and Kadereit, 2010) coded as ‘ambiguous’; Maundiaceae (1 (Katoh et al., 2009)(http://mafft.cbrc.jp/alignment/server/) with sp., East Australia; von Mering and Kadereit, 2010) coded as ‘South E-INS-i strategy and default setting. The aligned sequences were Asia – Southeast Asia – Australia’; Posidoniaceae (9 spp., Mediter- then inspected in MESQUITE (Maddison and Maddison, 2011) ranean, temperate Australia; Kuo and McComb, 1998) coded as and in a few cases manually adjusted. The total data set comprised ‘ambiguous’; Potamogetonaceae (85–90 spp., worldwide, but 114 ITS sequences and 72 trnK/matK sequences. Of these ten (eight mostly temperate; Haynes et al., 1998) coded as ‘ambiguous’; ITS and two trnK/matK, six new and four GenBank accessions) Ruppiaceae (1–10 spp., saline to fresh water, worldwide; Haynes showed markedly aberrant sequences (including evidence for et al., 1998) coded as ‘ambiguous’; Scheuchzeriaceae (1 sp., North pseudogeny), and were not considered in subsequent analyses. Temperate to Arctic; Haynes et al., 1998) coded as ‘Northern Hemisphere’; Zosteraceae (14 spp., worldwide, temperate to 2.3. Phylogenetic analyses subtropical; Kuo and McComb, 1998) coded as ‘ambiguous’. In some AARs, we added the ranges of the pollen fossils from Phylogenetic trees were obtained under maximum likelihood Wyoming, British Colombia, and Greenland in the Newick tree, (ML), using RAxML-HPC v. 7.4.2 (Stamatakis, 2006; Stamatakis using different placements and branch lengths for the fossils. Sce- et al., 2008) with separate partitions for the nuclear and plastid narios 1 and 2 assumed that all three fossils represent the ranges of data matrix. To select a model of nucleotide substitution, we ran extinct Aponogetonaceae sister lineages. Scenarios 3 to 8 assumed JModelTest v. 2.1.4 (https://code.google.com/p/jmodeltest2/; that the Eocene Canadian and Greenland fossils might represent Darriba et al., 2012), which selected the general time-reversible extinct sister groups to extant Aponogeton lineages. Fossils were (GTR) model with a gamma parameter (C) with four categories inserted into the tree according to their age, that is, the oldest fossil as the best fit for both data matrices. This model was therefore closest to the root and the two younger ones higher up in the tree. used in all analyses. Statistical support was assessed with the RAx- Each fossil was either given a short branch length (0.5 Ma), simu- ML fast bootstrap option; the number of necessary replicates was lating an extinct range, a branch as long as that of its sister clade, determined by the extended majority rule bootstop criterion simulating a range occupied for a long time, or an intermediate (Pattengale et al., 2010). Inferences were run on the original data branch length of 20 Ma. The geographic ranges of the fossils from matrices, which included several clones per individual, and strict Wyoming, Canadia, and Greenland and of the living species of Apo- individual consensus sequences. Where several homologous nogeton were categorized and coded as follows: ‘Africa – Madagas- sequences were available from different individuals of the same car’ or ‘South Asia – Southeast Asia – Australia’ or ’Northern species, we used the individual with best data coverage for the Hemisphere’. A few species of Aponogeton extend just north of final ML tree inference based on the concatenated ITS-trnK/matK the Tropic of Cancer (Fig. 1). The AAR was carried out in Mesquite data. Final trees were viewed in FigTree v. 1.4.0 (http:// v. 2.75 (Maddison and Maddison, 2011), using ML character state tree.bio.ed.ac.uk). reconstruction under the Markov k-state 1-parameter model (Mk1; Lewis, 2001). 2.4. Molecular clock dating
For molecular-clock dating, we used the combined matrix, but 3. Results took out 23 species (marked in Appendix S1) to reduce the number of very short or zero-length branches, which cause problems for 3.1. Evolutionary relationships and biogeography of Aponogetonaceae molecular-clock modeling while not adding information. Also excluded was the African species A. subconjugatus, as the signal The ITS matrix comprised 825 aligned nucleotides, the plastid from this species inflicts local topological ambiguity (Section 3). matrix 921 aligned nucleotides. In the absence of statistically sup- The final dating data matrix included 18 terminals representing ported topological incongruences (P87% ML bootstrap support) 114 L.-Y. Chen et al. / Molecular Phylogenetics and Evolution 82 (2015) 111–117 the two data partitions were combined into a single matrix. With ble-checked). The nuclear and plastid matrix and the phylogenetic species of Juncaginaceae, Scheuchzeriaceae, and Butomaceae tree have been submitted to TreeBASE (accession number 15898). included, A. hexatepalus was resolved as sister to rest of the genus. The clock-dating matrix included 18 terminals representing the To avoid large gaps in the alignment, we excluded the three out- major clades of Aponogeton and was calibrated with average groups in some analyses, rooting instead on A. hexatepalus. A max- nuclear and plastid substitution rates (Section 2). The resulting imum likelihood phylogeny from the combined data shows six chronogram is shown in Fig. 3a and b, and Table 1 shows the geographically coherent clades within Aponogeton (Fig. 2). The 95% confidence intervals. The Aponogeton crown group is dated tropical Australian species are nested within a south/southeast to the mid-Eocene 39.8 (95% CI: 32–48) Ma, and the deepest diver- Asian clade (clade I, Fig. 2); sister to this clade is an African species, gence in the genus is between the Australian A. hexatepalus and the A. subconjugatus, but with low support. The Madagascan species remaining species. Among the interesting geographic disjunctions fell into two clades, one relatively undifferentiated clade (II) form- that between the South Asian A. satarensis and its Madagascan rel- ing part of the Asian + Australian + Madagascan crown group of atives is dated to 9 (7–13) Ma and that between the West African/ Aponogeton, and one (clade VI) with three genetically well differen- Southeast Asian pair A. azureus/A. robinsonii to 9–17 Ma. The clade tiated species that also include the southern Asian A. satarensis. of all tropical Australian species has a stem age of 5 (4–7) Ma. Except for A. subconjugatus, all African species form a grade with AARs that did not incorporate fossil geographic ranges sug- three highly supported subclades (III, IV, V), of which one gested that the most recent common ancestor (MRCA) of today’s comprised a surprising species pair: A. robinsonii from Laos and species of Aponogeton lived in Africa and Madagascar (Fig. 3a). Vietnam and A. azureus from Namibia (the sequences stem from The inclusion of fossil ranges in the AARs greatly affected at the wild-collected herbarium material (Appendix S1) and were dou- area inferred for the deepest nodes, with the minimum-branch
S.W. Australia BS >40 to <70 Aponogeton proliferus D Les 549 CONN Tropical Australasia BS >70 to <90 A. lancesmithii SWL Jacobs 8567 CONN S./S.E. Asia BS >90 A. queenslandicus D Les 550 CONN* Madagascar A. bullosus SWL Jacobs 8572 CONN Continental and E. Africa A. womersleyi H Bleher sn M PT 3 S.W. Africa (Namibia, Cape Prov.) A. queenslandicus D Les 552 CONN PT 3 A. vanbruggenii SWL Jacobs 8542 0.01 exp. subst./site A. elongatus D Les 551 CONN PT 3 A. euryspermus SWL Jacobs 8839 NSW A. kimberleyensis SWL Jacobs 8831 NSW A. lakhonensis Chen cly2013001 HIB Clade I A. echinatus H. Heine sn M (syn. of A. crispus) PT 4 A. crispus SWL Jacobs 8537 CONN A. undulatus SWL Jacobs 8539 CONN PT 4 A. stachyosporus D Les 564 CONN (syn. of A. undulatus) A. rigidifolius Berger sn M A. jacobsenii C. Kasselmann 24 MPT 5 A. bruggenii SR Yadav sn M A. natans J Bogner 1857 M PT 5 A. subconjugatus GS Perrottet 1009 M A. eggersii H Schoepfel sn voucher unclear A. boivinianus Collector unknown M A. longiplumulosus D Les 560 CONN A. gottlebei C Kasselmann 955c M PT 6 A. ulvaceus SWL Jacobs 8543 CONN PT 6 Clade II A. madagascariensis SWL Jacobs 8535 CONN PT 2 A. fenestralis C Kasselmann sn CONN* (syn. of A. euryspermus) A. tenuispicatus J Bogner 275 M A. masoalaensis J Bogner 2087 M PT 1 A. azureus H Merxmuller & Giess 30642 M Clade III A. robinsonii J Bogner 2314 M PT 2 A. fugax JC Manning et al 3105 NBG MO Clade IV A. angustifolius JJ Bos 270 M A. abyssinicus Wang 20128364 HIB A. stuhlmannii EA Robinson 5902 M M Chase 14232 K A. rehmannii H Merxmuller 2138 M Clade V A. distachyos D Les sn CONN PT 2 A. desertorum RDA Bayliss 6799 M PT 5 A. junceus C Viljoen sn CONN PT 5 A. decaryi J Bogner 2056 M PT 3 A. dioecus J Bogner 210 M Clade VI A. satarensis SR Yadav sn M A. capuronii HWE van Bruggen sn CONN A. hexatepalus G Sainty 434337 NSW PT 1
Fig. 2. Maximum likelihood phylogeny based on the combined nuclear and plastid data. PT = pollen type from Grímsson et al. (2014). The oldest fossil pollen from the Late Cretaceous (82–81 Ma) of Wyoming, North America, has pollen type PT3, the younger (Eocene) pollen grains from Canada and Greenland have pollen types PT1 and PT2. ⁄ = only trnK/matK sequences obtained. Species names in bold indicate species newly sequenced for this study. L.-Y. Chen et al. / Molecular Phylogenetics and Evolution 82 (2015) 111–117 115
(a) 17 PG Aponogeton womersleyi 16 NEAu A. elongatus Africa - Madagascar 15 SEAs A. lakhonensis S and SE Asia - Australia 14 SAs A. crispus Northern Hemisphere 12 SAs A. natans Ambiguous 13 MG A. masoalaensis 10 MG A. madagascariensis 9 11 SEAs A. robinsonii NA A. azureus SA A. fugax 5 8 SAf A. rehmannii 7 EAf A. abyssinicus 6 Cape A. distachyos 2 SAf A. desertorum/A. junceus SAs A. satarensis 1 A 4 3 MG A. decaryi MG A. capuronii SWAu A. hexatepalus Zosteraceae Potamogetonaceae Ruppiaceae Cymodoceaceae Posidoniaceae Maundiaceae Juncaginaceae Scheuchzeriaceae
0D
(b) (c) A. womersleyi A. womersleyi A. elongatus A. elongatus A. lakhonensis A. lakhonensis A. crispus A. crispus A . natans A . natans A . masoal. A . masoal. A. madagasc. A. madagasc. A. robinsonii A. robinsonii A. azureus A. azureus A. fugax A. fugax A. rehmannii A. rehmannii A. abyssinicus A. abyssinicus A. distachyos A. distachyos A. des./A. junc. A. des./A. junc. A. satarensis A. hareoensis A. decaryi A. satarensis A. capuronii A. decaryi A. hexatepalus A. capuronii A. hareoensis A. longispinosum A. longispinosum A. hexatepalus A. harryi A. harryi Other families Other families
1000 90807060504030201a0 M 1000 90807060504030201a0 M
Fig. 3. Chronogram for the Aponogetonaceae obtained under a strict clock model calibrated with average nuclear and plastid substitution rates. The same chronogram was used to study the effects on ancestral area reconstruction (AAR) of including the fossil-derived former ranges of Aponogetonaceae in the Northern Hemisphere. The colored pie charts indicate the results of ML character state reconstruction, with the four states shown in the inset. (a) AAR without fossil geographic ranges included. (b) AAR including fossils as stem representatives (‘‘extinct grade scenario’’), with the fossils assigned a branch length of 20 Ma (Materials and Methods). (c) AAR including the fossils from Canada (A. longispinosum, 46 Ma) and Greenland (A. hareoensis, 44 Ma) as extinct sisters of modern Aponogeton species with similar pollen morphology (cf. Fig. 2); branch-lengths in the ingroup proportionally adjusted to accommodate the ages of the fossils. PG = Papua New Guinea, NEAu = Northeast Australia, SWAu = Southwest Australia, Cape = Cape Province, SEAs = Southeast Asia, SA = South Africa (country), SAs = Southern Asia, MG = Madagascar, NA = Namibia, SAf = Southern Africa (region), EAf = Eastern Africa. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) length model having the greatest effects, while the long-branch 4. Discussion model had the least effect (data not shown). With the short-branch length model, the MRCA of Aponogeton was inferred to have lived Aponogetonaceae have been the focus of three previous studies in the Northern Hemisphere (not shown). Even with the interme- (Les et al., 2003, 2005; Les and Tippery, 2013) that provided the diate branch lengths for the fossils of 20 Ma, we still inferred a basis for the present effort. Work on commonly cultivated plants, northern origin of modern Aponogeton (Fig. 3b). An Australian – such as the aquarium trade ‘‘favorite’’ Aponogeton, can be difficult Asian origin was never inferred with confidence regardless of because of hybrid forms that may have been created in cultivation where the fossils were placed or which branch length they and that may be propagated under man-made conditions. We were assigned. The AAR that included the fossils from Canada therefore focused on sequences from wild-collected material from (A. longispinosum, 46 Ma) and Greenland (A. hareoensis, 44 Ma) as Africa, Madagascar, India, and China. Even with this precaution extinct sisters of modern Aponogeton species with similar pollen (and insistence on vouchered material), we still obtained surpris- morphology (cf. Fig. 2) resulted in more ambiguous area ing species grouping, such as that between A. azureus from Nami- reconstructions for the deeper nodes (Fig. 3c) than did the ‘‘extinct bia and A. robinsonii from Laos and Vietnam, a relationship dated grade scenario’’ (Fig. 3b). to 9–17 Ma. Such a disjunction may be explained by transport of 116 L.-Y. Chen et al. / Molecular Phylogenetics and Evolution 82 (2015) 111–117
Table 1 Divergences time for the nodes within Aponogetonaceae shown in Fig. 3.
Node no. Description of node Mean node age and 95% HPD (Ma) 1 Crown age of Aponogeton 39.8 (32.2–48.0) 2 Stem of A. satarensis, A. decaryi and A. capuronii 29.3 (25.7–33.2) 3 Crown of A. satarensis, A. decaryi and A. capuronii 22.2 (17.7–26.7) 4 Crown of A. satarensis and A. decaryi 9.4 (6.5–12.6) 5 Stem of A. abyssinicus, A. rehmannii, A. distachyos and A. desertorum 28.8 (25.3–32.6) 6 Crown of A. abyssinicus, A. rehmannii, A. distachyos and A. desertorum 20.2 (16.8–23.9) 7 Crown of A. abyssinicus, A. rehmannii and A. distachyus 19.3 (15.8–22.8) 8 Stem of A. abyssinicus and A. rehmannii 10.0 (7.4–12.6) 9 Stem of A. fugax 28.1 (22.4–31.8) 10 Stem of A. robinsonii and A. azureus 22.2 (18.8–22.5) 11 Crown of A. robinsonii and A. azureus 12.8 (8.9–16.8) 12 Stem of A. madagascariensis and A. masoalaensis 13.8 (11.5–16.5) 13 Crown of A. madagascariensis and A. masoalaensis 8.1 (5.9–10.8) 14 Stem of A. natans 10.7 (8.6–12.8) 15 Stem of A. crispus 9.8 (7.8–12.0) 16 Crown of A. elongatus, A. womersleyi and A. lakhonensis 5.2 (3.6–7.0) 17 Crown of A. elongates and A. womersleyi 2.7 (1.6–4.1)
seeds in mud on the feet of birds that migrate annually along the In conclusion, the biogeography of the aquatic family Aponog- East Asia/East Africa flyway, a well-established route used by over etonaceae matches previous generalizations and molecular clock- 330 species of passerines and other birds (Feduccia, 2003; BirdLife based inferences about the extreme mobility of aquatic plants, first International, 2014). These species breed in Asia and have their stressed by Les et al. (2003) and since supported in many studies, non-breeding sites in eastern and southern Africa. Transport on for example, for Hydrocharitaceae (Chen et al., 2012a) Alismata- birds also seems the most plausible explanation for the other ceae (Chen et al., 2012b), Alismatales (Chen et al., 2013), Nymphae- trans-oceanic disjunctions inferred here and earlier by Les et al. ales (Löhne et al., 2008), and Hydatellaceae (Iles et al., 2014). It (2003, 2005). These bird migration patterns go back to the Miocene would be interesting to repeat Darwin’s (1863: Letter to Hooker; or earlier (Feduccia, 2003) and may have helped to shape more Veak, 2003) famous experiment of germinating seeds from the ancient disjunctions within Aponogeton like A. azureus–A. robinsonii. mud stuck on a single partridge’s leg. Darwin raised 32 seedlings To the extent that species sampling overlaps, the relationships belonging to several families from the earth ball stuck to the bird’s found here are congruent with those found by Les et al. (2005) leg. Similar data on the possible presence and viability of Aponog- and Les and Tippery (2013). Also in agreement with these authors’ eton seeds on aquatic birds would help link the indirect inferences work, we found that the temperate Australian species A. hexatepa- from molecular clock-dated phylogenies (of Aponogeton and lus is a distinct sister species of all other Aponogeton. However, they migratory birds, as their putative main vectors) to the ecology of inferred a crown age of Aponogeton of 23.3 Ma (Les et al., 2003), this understudied family. Especially puzzling is the morphological, while we inferred 39.8 Ma, probably mostly because of denser spe- genetic, and ecological diversity of Aponogeton on Madagascar (van cies sampling (in the absence of reliably assignable fossils both Bruggen, 1985), a relatively small region. However, similar to other studies used strict clock models, calibrated with average substitu- aquatic groups, Aponogetonaceae are rarely collected and tion rates). Regardless which crown age may be closer to the truth, extremely poorly represented in collections, meaning that our both reject Thanikaimoni’s (1985) suggestion of a mid-Cretaceous understanding of their geographic ranges is still highly incomplete. age of A. hexatepalus, an idea almost certainly influenced by Unfortunately, this is also true of the plants’ basic morphology. Selling’s (1947) erroneous report of an Upper Cretaceous Grímmson et al. (2014, their Figs 2 and 3) reconstructed species Aponogeton inflorescence from Patagonia. relationships in Aponogeton from a morphological matrix The precise range extension(s) by which Aponogetonaceae compiled mostly from van Bruggen (1985) and original reached the Southern Hemisphere (Fig. 1) remains unclear, species descriptions, but none of the deeper morphological clusters although the current data probably reject the hypothesis of Les they found correlates with the clades found here with molecular et al. (2005; Les and Tippery, 2013) of an Australian origin of the data. family. One possibility is that, following extinction in the Northern Hemisphere at the end of the Eocene climate optimum, the family survived in Africa and/or Madagascar and then dispersed to Aus- Author contributions tralia almost immediately, explaining the temperate Australian species A. hexapetalus. Noting the absence of Aponogeton in the LYC generate the data and carried out analyses; GG and LCY well-studied record of Europe, Grímsson et al. (2014) instead sug- generated alignments and biogeographic analyses; LCY, GG, and gested expansion across the Bering Land Bridge and then to Asia, SSR wrote the paper; SSR devised the project; QFW provided finan- Australia, and Africa, but this does not match well with the tree cial support and material. topology and divergence ages found here. All Asian species are dee- ply nested within Aponogeton. A potential vector for a (late) Paleo- gene direct North America – Africa dispersal are again migratory birds (above), whose East Atlantic flyway includes Canada, Green- Acknowledgments land, Europe, and all of West Africa south to the Cape (BirdLife International, 2014). A fuller understanding may come from the We thank Josef Bogner, Munich, and John Manning, discovery of more fossils, although our experiments with the inclu- Kirstenbosch, for plant samples, and M. Silber, Munich, for sion of no longer occupied geographic ranges (in this case in Wyo- assistance in the laboratory. Financial supports came from the ming, Canada, Greenland) highlight the difficulty of ancestral area Chinese Academy of Sciences (Research Grant XDAO5090305 to reconstruction in deep time and over almost the entire globe. QFW). L.-Y. Chen et al. / Molecular Phylogenetics and Evolution 82 (2015) 111–117 117
Appendix A. Supplementary material Les, D.H., Crawford, D.J., Kimball, R.T., Moody, M.L., Landolt, E., 2003. Biogeography of discontinuously distributed hydrophytes: a molecular appraisal of intercontinental disjunctions. Int. J. Plant Sci. 164, 917–932. Supplementary data associated with this article can be found, in Les, D.H., Moody, M.L., Jacobs, S.W.L., 2005. Phylogeny and systematics of the online version, at http://dx.doi.org/10.1016/j.ympev.2014.10. Aponogeton (Aponogetonaceae): the Australian species. Syst. Bot. 30, 503–519. 007. Lewis, P.O., 2001. A likelihood approach to estimating phylogeny from discrete morphological character data. Syst. Biol. 50, 913–925. Löhne, C., Yoo, M.J., Borsch, T., Wiersema, J., Wilde, V., Bell, C.D., Barthlott, W., Soltis, References D.E., Soltis, P.S., 2008. Biogeography of Nymphaeales: extant patterns and historical events. Taxon 57, 1123–1146. Birdlife International, 2014. East Asia/East Africa Flyway.