Molecular phylogenetics and historical biogeography of Hawaiian Dryopteris (Dryopteridaceae)

Molecular Phylogenetics and Evolution 34 (2005) 392–407 www.elsevier.com/locate/ympev Molecular phylogenetics and historical biogeography of Hawaiian Dryopteris (Dryopteridaceae) J.M.O. Geiger¤,1, T.A. Ranker University Museum and Department of Ecology and Evolutionary Biology, University of Colorado, Boulder, CO 80309, USA Received 26 July 2004; revised 24 October 2004 Abstract The fern genus Dryopteris (Dryopteridaceae) is represented in the Hawaiian Islands by 18 endemic taxa and one non-endemic, native species. The goals of this study were to determine whether Dryopteris in Hawai’i is monophyletic and to infer the biogeographical origins of Hawaiian Dryopteris by determining the geographical distributions of their closest living relatives. We sequenced two chloroplast DNA fragments, rbcL and the trnL-F intergenic spacer (IGS), for 18 Hawaiian taxa, 45 non-Hawaiian taxa, and two outgroup species. For individual fragments, we estimated phylogenetic relationships using Bayesian inference and maximum parsimony. We performed a combined analysis of both cpDNA fragments employing Bayesian inference, maximum parsimony, and maximum likelihood. These analyses indicate that Hawaiian Dryopteris is not monophyletic, and that there were at least Wve separate colonizations of the Hawaiian Islands by diVerent species of dryopteroid ferns, with most of the Wve groups having closest relatives in SE Asia. The results suggest that one colonizing ancestor, perhaps from SE Asia, gave rise to eight endemic taxa (the glabra group). Another colonizing ancestor, also possibly from SE Asia, gave rise to a group of Wve endemic taxa (the exindusiate group). Dryopteris fusco-atra and its two varieties, which are endemic to Hawai’i, most likely diversiWed from a SE Asian ancestor. The Hawaiian endemic Nothoperanema rubiginosum has its closest relatives in SE Asia, and while the remaining two species, D. wallichiana and D. subbipinnata, are sister species, their biogeographical origins could not be determined from these analyses due to the widespread distributions of D. wallichiana and its closest non-Hawaiian relative.  2004 Elsevier Inc. All rights reserved. Keywords: Dryopteris; Hawaiian Islands; Molecular phylogenetics; trnL-F IGS; rbcL; Biogeography 1. Introduction The Hawaiian Island chain is approximately 80 million years old, however, most islands eroded to below sea level long ago. The oldest of the current high islands is Kaua’i, which is about 5.0 million years old and the youngest is the Big Island of Hawai’i, which is about 0.5 million years old (Carson and Clague, 1995). On the current high islands, most of the mid to high elevation * Corresponding author. Fax: +1 406 447 5476. E-mail addresses: [email protected] (J.M.O. Geiger), ranker@ colorado.edu (T.A. Ranker). 1 Present address: Carroll College, Department of Natural Sciences, 1601 North Benton Ave., Helena, Montana 59625, USA. 1055-7903/$ - see front matter  2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2004.11.001 Hawaiian Xora and fauna evolved in isolation for up to 5 million years (Clague, 1996; Price and Clague, 2002) before the arrival of the Polynesians about 1600 years ago (Kirch, 1982). The result of this isolation was the production of a unique and distinctive Xora. Among Hawaiian Xowering plant species, approximately 89% are endemic to the archipelago (Wagner et al., 1999b) and among Hawaiian pteridophytes, approximately 71% are endemic (Palmer, 2003). These are among the highest rates of endemism for any known Xora (Sohmer and Gustafson, 1987). The characteristically insular nature of the Xora is due to a limited number of original colonizers and the diversity of their origin (Fosberg, 1948; Wagner et al., 1999b). For angiosperms, it appears to have been a rare occurrence that more than one species of a J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 recognized genus reached the islands due to dispersal limitations (Carlquist, 1980) and it is commonly thought that dispersal events to the islands are not a large source of the observed diversity. It has been estimated that the ca. 1000 native angiosperm species were derived from only 272–282 natural introductions (Fosberg, 1948; Wagner et al., 1999b). Of the 161 native species of Hawaiian pteridophytes (Palmer, 2003), however, there were an estimated 115 colonizing ancestral species (Wagner, 1988). A likely explanation for this diVerence is that pteridophytes are generally more easily dispersed long distances due to their small, windblown spores, whereas the seeds or fruits of most angiosperms are much larger and less easily dispersed (Carlquist, 1980; Ranker et al., 1994, 2000; Smith, 1972; Tryon, 1970). Hence, within the archipelago (and elsewhere), speciation rates of pteridophytes could be depressed relative to those of angiosperms due to continued gene Xow between newly founded and source populations (Ranker et al., 2000). Biogeographical histories of endemic Hawaiian plants may be more complicated and obscure than those of plants of other volcanic oceanic archipelagos, mainly due to the large distances from source populations and the long existence of the chain of the Hawaiian Islands (Kim et al., 1998). Recent studies of speciWc groups of Hawaiian angiosperms have indicated that the diversity of Hawaiian species for many genera or groups of confamilial genera is usually the result of a single dispersal event to the Islands with subsequent phylogenetic radiations (e.g., the silversword alliance, Baldwin et al., 1991; Hesperomannia, Kim et al., 1998; Hawaiian geraniums, Pax et al., 1997). However, exceptions in which more than one dispersal event to Hawai’i of congeneric species has been discovered include Rubus (Howarth et al., 1997) and Scaevola (Howarth et al., 1999). In contrast to Xowering plants, of the two studies performed speciWcally on Hawaiian pteridophytes, one provides evidence of multiple dispersal events to the islands by a single species, Asplenium adiantum-nigrum (Ranker et al., 1994), while the other supports a single colonization hypothesis for the endemic genus Adenophorus plus its sister taxon Grammitis tenella (Ranker et al., 2003). Various groups of angiosperms in Hawai’i continue to be studied from a phylogenetic perspective (e.g., see Powell and Kron, 2002; Wagner and Funk, 1995), but to date there are only Wve published phylogenetic studies that include Hawaiian pteridophytes (HauXer and Ranker, 1995; Hennequin et al., 2003; Ranker et al., 2003, 2004; Schneider et al., 2004a) and few population genetic studies on Hawaiian fern groups (Ranker et al., 1994, 1996, 2000; Russell et al., 1999). Pteridophytes comprise about one-sixth of the native vascular plant species in Hawai’i and they physically dominate some communities, thus the paucity of 393 phylogenetic information about pteridophytes represents a gap in our understanding of the evolution of the Hawaiian Xora. 1.1. Background on Hawaiian Dryopteris Dryopteris Adans. (Dryopteridaceae) is a cosmopolitan genus comprising approximately 225 species, mostly occurring in temperate forests and montane areas of the tropics (Hoshizaki and Wilson, 1999). An estimated eight to 17 species occur in the Hawaiian Islands, all of which are endemic to the Islands except one, which is indigenous (D. wallichiana). Individual Hawaiian Dryopteris species generally occur on all or most of the islands, however, depending on the taxonomic classiWcation followed, there may be several single-island endemics. All Hawaiian taxa are found in mesic to very wet, montane forests. Dryopteris has been a diYcult group to understand taxonomically in Hawai’i (Wagner, 1995; Wagner et al., 1999a) and throughout the world (Hoshizaki and Wilson, 1999). There has been much debate and confusion regarding speciWc and subspeciWc classiWcations and in the understanding of the evolutionary relationships among the taxa, especially for the Hawaiian species (Fraser-Jenkins, 1986, 1994; Herat, 1979; Palmer, 2003; Wagner, 1993, 1995; Wagner et al., 1999a). There have also been conXicting views regarding the historical biogeography of these taxa and their origins in the Islands. Fosberg (1948) estimated that there were 25 Hawaiian Dryopteris species, each a result of a separate dispersal and colonization event to the Islands. Herat (1979) proposed that there were only eight species of Dryopteris in Hawai’i that resulted from Wve separate introductions. Fraser-Jenkins (1994) recognized nine species, with 10 varieties, and suggested at least three separate colonization events. In various publications, Wagner (1993, 1995) and Wagner et al. (1995b, 1999b) recognized as many as 17 species. Most recently, Palmer (2003) revised the classiWcation of these species and recognized 10 species and 11 varieties. He placed Wve species and nine varieties into two morphologically distinct groups. Palmer recognized an “exindusiate” group (as it will be referred to below), with 3- to 5-pinnate leaves, as likely being monophyletic. Included in this group are D. sandwicensis, D. tetrapinnata, and D. unidentata (and varieties). However, two species, D. mauiensis and D. crinalis (and varieties), which also lack indusia, were not recognized in this group. Palmer (2003) also recognized a second likely monophyletic group, which includes D. glabra plus varieties and D. hawaiiensis. However, Palmer suggested that D. hawaiiensis is probably unrelated to D. glabra. Species in this group, which will be referred to as the “glabra” group, have indusia and the leaves are characteristically 2-, 3-, to 5-pinnate. Palmer did not propose potential relationships for the remaining species (D. mauiensis, D. 394 J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 crinalis and varieties, D. wallichiana, D. subbipinnata, and D. fusco-atra and varieties) but suggested seven colonization events that gave rise to this species diversity of Dryopteris in Hawai’i. The goals of the present study were 3-fold. The Wrst goal was to elucidate the phylogenetic relationships among species of Hawaiian Dryopteris and their nonHawaiian relatives using chloroplast DNA (cpDNA) sequence variation. Our second goal was to test alternative hypotheses concerning the monophyly of Hawaiian Dryopteris. SpeciWcally we tested whether (a) Hawaiian Dryopteris is monophyletic or (b) Hawaiian Dryopteris is polyphyletic, consisting of monophyletic groups of species, each of which are descendants of distinct colonizing ancestors to Hawai’i. Our third goal was to infer the geographic origin(s) of Hawaiian Dryopteris. 2. Materials and methods 2.1. Taxon sampling and DNA extraction We used Arachniodes aristata as one outgroup taxon based on the family level rbcL phylogeny of Hasebe et al. (1995) where Arachniodes was consistently supported as the sister taxon to Dryopteris, when it was included in analyses. A second species, Nothoperanema rubiginosum, which is a Hawaiian endemic thought to be closely related to Dryopteris (Smith and Palmer, 1995), was also used as an outgroup taxon. Samples of all species of Hawaiian Dryopteris were collected in Hawai’i. Leaf material was stored in silica gel until DNA was extracted. Samples of non-Hawaiian taxa were supplied by colleagues, extracted from fragments of COLO or PTBG herbarium specimens, or spores were provided from the British Pteridological Society Spore Exchange and the American Fern Society Spore Exchange (Table 1). We sampled Dryopteris species broadly from around the PaciWc Rim and elsewhere. We also sampled common species and species that have been described as being morphologically similar to some of the Hawaiian species. When spores were provided, the spores were germinated and resulting gametophytes were cultured on a nutrient-enriched agar following the methods of Ranker et al. (1996). DNA was then extracted from the gametophytes (method cited below). Table 1 lists (when available) locality, collector, collection number, herbarium of deposit for voucher specimens, and GenBank accession numbers for all taxa and DNA sequences studied, including the outgroup species. We extracted total cellular DNA using the CTAB method of Doyle and Doyle (1987) modiWed by adding 3% PVP-40 and 5 mM ascorbic acid. Sample DNA concentrations were quantiWed using a miniXuorometer and standardized to 10 ng/
l. 2.2. PCR ampliWcation and sequencing We PCR-ampliWed and sequenced two segments of the cpDNA genome: a 1311 basepair (bp) fragment of the rbcL gene and the trnL (UAA) 3⬘ exon-trnF (GAA) intergenic spacer (IGS). PCR ampliWcation of rbcL was accomplished as in Ranker et al. (2003). Primer sequences are listed in Wolf et al. (1994) and Ranker et al. (2003) with the exception of D876R, which was designed speciWcally for this study. The sequence of D876R is: 5⬘-ATGAAGAAGCAGCCCYTTGTC-3⬘. PCR conditions were described in HauXer and Ranker (1995). AmpliWcation of the trnL-F IGS was achieved with primers “e” and “f” of Taberlet et al. (1991). PCR conditions were 94 °C (180 s), followed by 5 cycles of 94 °C (30 s), 45 °C (30 s), and 72 °C (30 s), followed by 37 cycles of 94 °C (30 s), 60 °C (30 s), and 72 °C (30 s), ending with 10 min at 72 °C after cycling was completed. The same primers were used individually for sequencing each strand of the spacer. PCR products were puriWed with the Promega Wizard PCR Preps PuriWcation System. Sequencing reactions were performed with the ABI Prism BigDye Terminator Cycle Sequencing Kit, employing 1/4 reactions. Sequencing products were puriWed with AutoSeq G-50 Sephadex columns from Amersham–Pharmacia Biotech. Sequences were detected on ABI automated sequencers at the Iowa State University DNA Sequencing and Synthesis Facility. 2.3. Phylogenetic analyses Sequence fragments were edited by visual inspection of electropherograms in Sequencher (Gene Codes) and aligned manually (rbcL) or with ClustalX (Thompson et al., 1997) and then manually adjusted to achieve more parsimonious alignments (trnL-F IGS). We conducted phylogenetic analyses on three diVerent data sets: (1) rbcL alone, (2) trnL-F alone, and (3) a combined data set of rbcL and trnL-F. For each data set we Wrst performed maximum parsimony (MP) analysis as implemented in PAUP* 4.0b10 (SwoVord, 2002). We conducted two MP analyses per data set, following two criteria: (1) with all characters unordered and equally weighted and (2) unordered and with a transition-totransversion bias. We estimated transition-to-transversion bias by comparing the length of the MP trees with transversions omitted to the length of MP trees with all variable sites included (see Martin and Naylor, 1997). The transversion bias was six for the rbcL data set and three for the trnL-F data set. For the combined data set, each data partition was deWned with its determined transversion bias using a step-matrix as described above. For the separate MP analyses we excluded all uninformative characters and employed the heuristic search algorithm with 1000 random addition sequence replicates with MulTrees activated, and with TBR branch 395 J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 Table 1 Species list, collection, and voucher information (when available), and GenBank accession numbers Species Collection locality Collector, number, and herbarium Native distribution GenBank accession number trnL-F rbcL Arachnoides aristata (Forst.) Tindale Dryopteris aemula (Aiton) Kuntze Dryopteris aYnis subsp. aYnis (Lowe) Fraser-Jenk. Dryopteris aYnis subsp. borreri (Newman) Fraser-Jenk. Dryopteris amurensis Christ Dryopteris aquilinoides (Desv.) C. Chr. American Samoa BPSSE Stowey Parish Lorence 9762; PTBG Unknown Crabbe 11824; COLO V. Vasak; COLO Marquesas W. Europe China, Japan AY268782 AY268816 AY268780 AY268851 AY268881 AY268849 AY268778 AY268847 AY268802 AY268803 AY268867 AY268868 AY268817 AY268882 AY268796 AY268801 AY268818 AY268777 AY268797 AY268808 AY268805 AY268774 AY268862 AY268866 AY268883 AY268846 AY268863 AY268873 AY268870 AY268835 AY268819 AY268884 AY268813 AY268820 AY268779 AY268878 AY268885 AY268848 AY268787 AY268775 AY268852 AY268844 AY268776 AY268793 AY268800 AY268760 AY268845 AY268857 AY268865 AY268837 Dryopteris ardechensis Fraser-Jenk. Dryopteris bissetiana (Baker) C. Chr. Dryopteris campyloptera (Kunze) Clarkson Dryopteris carthusiana (Villars) H. P. Fuchs Dryopteris carthusiana Dryopteris championii (Benth.) C. Chr. Dryopteris corleyi Fraser-Jenk. Dryopteris crassirhizoma Nakai Dryopteris crinalis (Hook. et Arn.) C. Chr. Dryopteris crispifolia Rasbach, Reichst. & Vida Dryopteris cystolepidota (Miq.) Makino Dryopteris dickinsii (Franch. & Sav.) C. Chr. Dryopteris dilatata (HoVm.) A. Gray Dryopteris erythrosora (D. Eaton) Kuntze Dryopteris expansa (C. Presl) Fraser-Jenk. & Jermy Dryopteris Wlix-mas (L.) Schott Dryopteris formosana (Christ) C. Chr. Dryopteris fragrans (L.) Schott Dryopteris fusco-atra var. fusco-atra (Hillebr.) W. J. Rob. Dryopteris fusco-atra var. lamoureuxii Fraser-Jenk. Dryopteris glabra (Brack.) Kuntze var. alboviridis (W. H. Wagner) D. D. Palmer Dryopteris glabra var. Xynnii D. D. Palmer Dryopteris glabra var. glabra Dryopteris glabra var. hobdyana (W. H. Wagner) D. D. Palmer Dryopteris glabra var. nuda (Underw.) Fraser-Jenk. Dryopteris glabra var. pusilla (Hillebr.) Fras-Jenk. Dryopteris glabra var. soripes (Hillebr.) Herat ex Fraser-Jenk. Dryopteris goeringiana (Kunze) Koidz. Dryopteris hawaiiensis (Hillebr.) W. J. Rob. Dryopteris hondoensis Koidz. Dryopteris intermedia (Muhlenb. ex Willd.) A. Gray subsp. maderensis Fraser-Jenk. Dryopteris juxtaposita Christ Dryopteris lacera (Thunb.) Kuntze Dryopteris lepidopoda Hayata Dryopteris mauiensis C. Chr. Dryopteris munchii A. Reid Smith Dryopteris odontoloma (Beddome) C. Chr. Dryopteris oreades Fomin. Caucasus NZ, Europe; SW Asia Label not legible Unknown; COLO E. Asia La Réunion Ranker 1536; Réunion; COLO Mauritius BPSSE Unknown W. Europe, France Cultivated, NY R. Moran; COLO China Partridge Island Cody 23484; COLO NE USA BPSSE Unknown NA/Euraisa Quebec, Canada Argus 9327; COLO NA/Eurasia Cultivated, NY R. Moran; COLO China AFSSE Unknown W. Europe AFSSE Unknown China, Japan Mau’i, Hawai’i Oppenheimer Hawai’i H50044; COLO BPSSE Unknown W. Europe, Portugal AFSSE Unknown Japan, Korea BPSSE Unknown China, Japan Siberia, Russia Krasnobovov 679; China, NZ COLO Cultivated, DBG Geiger 94; COLO China, Japan Unknown Nelson 7921; widespread COLO Boulder, Colorado Hogan 1421; COLO widespread Cultivated, NY R. Moran; COLO China, Japan Alaska, USA Kelso 83-221; COLO China, Japan Mau’i, Hawai’i Geiger 4; COLO Hawai’i Mau’i, Hawai’i Geiger 77; COLO Hawai’i AY268783 AY268841 Kaua’i, Hawai’i Geiger 25; COLO Hawai’i AY268768 AY268831 Kaua’i, Hawai’i Kaua’i, Hawai’i Mau’i, Hawai’i Geiger 23; COLO Geiger 24; COLO Palmer; COLO Hawai’i Hawai’i Hawai’i AY268764 AY268767 AY268773 AY268829 AY268830 AY268839 O’ahu, Hawai’i Geiger 90; COLO Kaua’i, Hawai’i Geiger 21; COLO Moloka’i, Hawai’i Geiger 41; COLO Hawai’i Hawai’i Hawai’i AY268785 AY268763 AY268786 AY268843 AY268828 AY268842 Cultivated, NY Mau’i, Hawai’i Cultivated, NY BPSSE R. Moran; COLO Geiger 74; COLO R. Moran; COLO Unknown China Hawai’i Japan NE USA AY268790 AY268784 AY268791 AY268821 AY268855 AY268840 AY268856 AY268886 AFSSE Cultivated, NY Cultivated, DBG Kaua’i, Hawai’i BPSSE Unknown R. Moran; COLO Geiger 96; COLO Geiger 29; COLO Unknown China China, Japan China Hawai’i Chiapas, Mesoamericana China Europe AY268810 AY268794 AY268789 AY268770 AY268822 AY268875 AY268860 AY268854 AY268833 AY268887 AY268807 AY268781 AY268872 AY268850 AFSSE Unknown Caucasus centralis V. Vasak; COLO (continued on next page) 396 J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 Table 1 (continued) Species Collection locality Collector, number, and herbarium Native distribution GenBank accession number Dryopteris paciWca (Nakai) Tag. Dryopteris pallida (Bory) C. Chr. ex Maire & Petitm. AFSSE AFSSE Unknown Unknown AY268814 AY268809 AY268879 AY268874 Dryopteris patula (Sw.) L. Underw. BPSSE Unknown AY268823 AY268888 Dryopteris polylepis (Franch. & Sav.) C. Chr. Dryopteris pulcherrima Ching Dryopteris pycnopteroides (Christ) C. Chr Dryopteris remota (A. Braun ex Doell) Druce, non Hayata Dryopteris sacrosancta Koidz. Dryopteris sandwicensis (Hook. & Arn.) C. Chr. Dryopteris sichotensis V. Komarov Dryopteris sieboldii (Van Houtte ex Mett.) Kuntze Dryopteris stenolepis (Baker) C. Chr. Dryopteris subbipinnata W. H. Wagner & Hobdy Cultivated, NY AFSSE Cultivated, NY Cultivated, NY AFSSE Kaua’i, Hawai’i Label not legible AFSSE BPSSE Mau’i, Hawai’i AY268798 AY268811 AY268799 AY268792 AY268812 AY268762 AY268804 AY268815 AY268824 AY268765 AY268864 AY268876 AY268859 AY268858 AY268877 AY268827 AY268869 AY268880 AY268889 AY268834 Dryopteris sublacera Christ Dryopteris tetrapinnata W. H. Wagner & Hobdy Dryopteris tokyoensis (Matsum. & Makino) C. Chr. Dryopteris unidentata var. paleacea (Hillebr.) Herat ex Fraser-Jenk Dryopteris unidentata var. unidentata (Hook. & Arn.) C. Chr. Dryopteris uniformis (Makino) Makino Dryopteris wallichiana (Spreng.) Hyl. Nothoperanema rubiginosum Smith & Palmer Cultivated, DBG Mau’i, Hawai’i Cultivated, NY Kaua’i, Hawai’i R. Moran; COLO Unknown R. Moran; COLO R. Moran; COLO Unknown Flynn 6675; PTBG Unknown; COLO Unknown Unknown Oppenheimer H50074; COLO Geiger 95; COLO Geiger 17; COLO R. Moran; COLO Geiger 28; COLO China, Japan W. Europe, Balearic Islands Guatemala, Chiapas, Mexico China, Japan China China Asia/Europe E. Asia Hawai’i NE Asia Japan China Hawai’i China Hawai’i Japan Hawai’i AY268788 AY268772 AY268795 AY268769 AY268853 AY268853 AY268861 AY268832 Kaua’i, Hawai’i Flynn 6666; PTBG Hawai’i AY268766 AY268825 AFSSE Kaua’i, Hawai’i Mau’i, Hawai’i Unknown Flynn 6671; COLO Geiger 2; COLO E. Asia Hawai’i Hawai’i AY268806 AY268761 AY268771 AY268871 AY268826 AY268836 Abbreviations: AFSSE, American Fern Society Spore Exchange; BPSSE, British Pteridological Society Spore Exchange; DBG, Denver Botanic Garden, Denver, CO, USA; NY, New York, USA. swapping. For the combined data set, the MP settings were the same as above, except 5000 random addition sequence replicates were performed. Also for the combined data set, we performed bootstrap (BS) analysis with 1303 repetitions and 10 random stepwise addition replicates each with the transversion bias excluded. We conducted a decay analysis of branch support (Bremer, 1988; Donoghue et al., 1992) on the combined data set with AutoDecay 4.0.1 (Eriksson, 1998). For each data set we used ModelTest (Posada and Crandall, 1998) to determine the model of evolution that best explained the data for use in maximum likelihood (ML) and Bayesian inference analyses. The TrN + I + G (Tamura and Nei, 1993) model best explained the rbcL data set, the HKY + G (Hasegawa et al., 1985) evolutionary model best Wt the trnL-F data set, and the TrNef + I + G (TrN equal base frequencies; Tamura and Nei, 1993) model best explained the combined data set. For the individual data sets, it was computationally diYcult to perform ML analyses due to the paucity of phylogenetically informative characters and the large number of taxa, even with the selected models enforced. We conducted a ML bootstrap analysis (ML BS), therefore, on the combined data set only, running 1000 repetitions and 10 random stepwise addition replicates each. We performed Bayesian analysis as implemented in MrBayes 2.01 (Huelsenbeck and Ronquist, 2001) on each data set, enforcing the models listed above. For each analysis, we ran 1,000,000 generations of which one tree was sampled per 100 trees generated. We obtained posterior probability values for each node in PAUP* by computing the majority-rule consensus tree of the last 8000 sampled trees, excluding the Wrst 2000 trees sampled during the “burn-in period.” We attempted to perform 1,000 random addition sequence replicates of the incongruence length diVerence test (ILD; Farris et al., 1994, 1995) with constant characters excluded as implemented in PAUP* to determine whether the rbcL and trnL-F data sets were combinable. However, after approximately 6 days of running time, the analysis had only reached replicate 25b, thus the results of this test were based on 25 replicates. As an additional means of assessing combinability, we compared the topologies of the strict consensus trees from the separate data sets with the Shimodaira–Hasegawa (SH) test as implemented in PAUP* (Goldman et al., 2000; Shimodaira and Hasegawa, 1999; Schneider et al., 2004b). The individual topologies did not diVer signiWcantly (P D 0.174). Because it is an apparently rare event that two or more congeneric species independently colonized the Hawaiian Islands, as our results suggest occurred with Dryopteris (see Section 3), we performed the topological tests discussed below to add conWdence to our results. J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 397 Fig. 1. Example of topologies used for testing alternative phylogenetic hypotheses. (A) Topology found by combined Bayesian analysis; (B) Hawaiian Clade VII and Hawaiian group of Clade V are monophyletic and diverge at the place in the topology where Clade VII diverged in the combined Bayesian analysis; (C) Hawaiian Clade VII and Hawaiian group of Clade V are monophyletic and diverge at the position in the topology where Clade V diverged in the combined Bayesian analysis. With the topology from the Bayesian analysis (which did not diVer signiWcantly topologically from either the ML tree or MP consensus tree) as our phylogenetic hypothesis, we performed likelihood-ratio tests employing the Shimodaira–Hasegawa (SH) test in PAUP* to compare diVerent hypothetical topologies (Shimodaira and Hasegawa, 1999). Using MacClade (Maddison and Maddi- son, 2000) we created topologies to test the support of the polyphyly of Hawaiian Dryopteris (see Section 3). We used roman numerals to label clades for explanatory simplicity (see Fig. 1). By constraining relationships, we tested the following hypotheses: (1) Hawaiian pairs of monophyletic groups (found in phylogenetic analyses; see Section 3) share common ancestry and thus evolved 398 J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 from a single colonizing ancestor (i.e., that the Hawaiian taxa of Clade V (glabra group) are sister to the Hawaiian taxa of Clade VII (D. subbipinnata + D. wallichiana) or that the Hawaiian Clade II (exindusiate group) is sister to the Hawaiian taxa of Clade V (glabra group); see Fig. 1 for a graphical explanation of topological constraints); and (2) Hawaiian Dryopteris is monophyletic. For the pairwise hypotheses, each sister relationship was tested in two ways: (1) in the position on the tree where one of the two Hawaiian clades was resolved in the Bayesian phylogeny and (2) in the position on the tree where the second of the two clades was resolved in the Bayesian phylogeny. For example, when hypothesized that the glabra group (Clade V) is sister to D. wallichiana and D. subbipinnata (Clade VII), we Wrst constrained the glabra group of Clade V to be sister to the Hawaiian taxa of Clade VII to the exclusion of D. aYnis var. borreri, which is the sister taxon to the Hawaiian members of Clade VII (see Fig. 1). Second, we placed the Hawaiian taxa of Clade VII as sister to the Hawaiian taxa of Clade V to the exclusion of D. aemula + D. corleyi + Nothoperanema rubiginosum and Clade IV. To test whether Hawaiian Dryopteris is monophyletic, we constrained all Hawaiian species together and tested the likelihood of each of the four hypotheses against the likelihood of the Bayesian phylogeny. In four diVerent tests, we placed monophyletic Hawaiian Dryopteris in the position on the Bayesian tree where the Hawaiian Clades II, V, VII, and VIII diverged, respectively. We did not make any assumptions about the relationships among the Hawaiian Dryopteris species, thus the relationships within monophyletic Hawaiian Dryopteris were always unresolved in our tests. 3. Results resulted in 18,283 most parsimonious trees. Each tree had an L of 542 steps, a CI of 0.55, and a RI of 0.87. The strict consensus trees from both of these analyses (not shown) were not completely dichotomously resolved but did have the same topologies. The 50% majority-rule consensus tree obtained from 8000 equally likely trees from the Bayesian analysis was nearly completely dichotomously resolved (Fig. 2). Clade credibility ( D posterior probability, PP) values were often above 85% and were frequently above 95%. There were no major conXicts among the topologies resulting from the MP (not shown) and Bayesian analyses. From these analyses, Hawaiian Dryopteris was not supported as monophyletic (see Fig. 2). Dryopteris wallichiana and D. subbipinnata comprised a clade sister to non-Hawaiian D. aYnis var. borreri, but were in the same larger clade with the D. fusco-atra varieties. There was poor resolution among the groups in this clade, and the sister-taxon relationships of the D. fusco-atra varieties were not resolved. The exindusiate group was supported as monophyletic, however, with the inclusion of D. mauiensis and D. crinalis and this clade was sister to a group, which included the non-Hawaiian D. odontoloma, D. pallida, and D. tokyoensis. The glabra group was supported as paraphyletic, as this unresolved clade also included D. corleyi and D. aemula, which do not occur in Hawai’i. This group was sister to the Hawaiian species Nothoperanema rubiginosum, which had been included in the analysis as an outgroup. Additionally, D. hawaiiensis was in this unresolved clade. The glabra clade (including Nothoperanema rubiginosum) was strongly supported as sister to a clade that included the non-Hawaiian species D. erythrosora, D. cystolepidota, D. formosana, D. paciWca, D. championii, D. bissetiana, and D. sacrosancta. In neither of the analyses, MP or Bayesian, were any of the four Hawaiian groups supported as sister clades. 3.1. Phylogenetic analyses 3.3. TrnL-F only analyses Sequences of rbcL and trnL-F were obtained for all 63 samples of Dryopteris and the outgroups, Arachniodes aristata and Nothoperanema rubiginosum. Across the 63 Dryopteris rbcL sequences obtained, 1090 bp were invariant, 221 bp were variable, and 136 bp were parsimony informative. Across the 63 ingroup sequences of trnL-F obtained, 255 bp were invariant, 124 bp were variable, and 62 bp were parsimony informative. 3.2. rbcL only analyses The heuristic MP analysis of the rbcL data set with no transversion bias found 30,622 equally parsimonious trees. Each tree was characterized by a length (L) of 420 steps, a consistency index (CI) of 0.59, and a retention index (RI) of 0.84. The heuristic MP analysis of the rbcL data set run with a transversion-to-transition ratio of 6:1 The total aligned length of trnL-F IGS was 379 bp. The MP heuristic analysis of the trnL-F data set without a transversion bias found 16 most parsimonious trees. Each tree had L D 233, CI D 0.743, and RI D 0.88. When a transition-to-transversion bias of 3:1 was enforced in an MP heuristic search, 10 most parsimonious trees were found. These 10 trees each had a total L D 345, CI D 0.82, and RI D 0.90. The strict consensus trees (not shown) produced by both of these analyses did not diVer in topology and had well-resolved terminal relationships, although deep relationships were not resolved. Bayesian analysis resulted in an almost fully resolved 50% majority-rule consensus tree of the 8000 trees included (Fig. 3). Many of the clades had PP values above 90% and the topology was identical to the MP strict consensus trees. J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 399 Fig. 2. Topology from Bayesian analysis of the individual rbcL data set. Numbers above the branches represent posterior probabilities (PP). Abbreviations in parentheses following each taxon name represent the geographic distribution of that taxon. An abbreviation in an internal second set of parentheses indicates a restricted distribution of the taxon within the broader region. A slash between two regions with parentheses indicates it is distributed in both regions. A, Asia; Az, Azores; B, Boreal; CA, Central America; E, Europe; IO, Indian Ocean; IP, Indo-PaciWc; Ma, Madeira; W, Widespread; WA, Western Asia; and WE, Western Europe. 400 J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 Fig. 3. Topology from Bayesian analysis of the individual trnL-F IGS data set. Numbers above the branches represent posterior probabilities (PP). Abbreviations in parentheses following each taxon name represent the geographic distribution of that taxon. A distribution in a second set of parentheses indicates the restricted distribution of the taxon within a broader region. A slash between two regions with parentheses indicates it is distributed in both regions. A, Asia; Az, Azores; B, Boreal; CA, Central America; E, Europe; IO, Indian Ocean; IP, Indo-PaciWc; Ma, Madeira; W, Widespread; WA, Western Asia; and WE, Western Europe. J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 Analyses based on trnL-F did not support a monophyletic Hawaiian Dryopteris. Dryopteris wallichiana and D. subbipinnata were grouped with non-Hawaiian D. aYnis var. borreri. The D. fusco-atra varieties were grouped together and closely associated with the nonHawaiian D. stenolepis, D. sublacera, D. pycnopteroides, and D. uniformis. The exindusiate group was supported as monophyletic, including D. crinalis and D. mauiensis. This group’s relationship to any other clade was unresolved in this analysis. The glabra group was supported as monophyletic with its closest relatives being European D. corleyi and D. aemula. This clade was supported as sister to the same group of species indicated from the rbcL analysis and N. rubiginosum was sister to those two sister clades. This result is diVerent from the rbcL analysis in which N. rubiginosum was sister only to the glabra group + D. corleyi and D. aemula. Similar to the rbcL analysis, none of the four Hawaiian clades was supported as sister to another Hawaiian clade. 3.4. Combinability The ILD test was not signiWcant (P > 0.40) indicating that we could not reject the null hypothesis of data set homogeneity. Other than the placements of D. hondoensis, D. patula, D. pulcherrima, D. fragrans, and N. rubiginosum, there were not strong incongruencies in clade composition among single-sequence topologies (see Figs. 2 and 3). The SH test also indicated no signiWcant diVerence between the topologies (P D 0.174). Given the lack of topological incongruence, the result of the ILD test, the result of the SH test, and that the two sequenced regions are chloroplastic, we felt justiWed in combining the data sets for a more robust analysis. 3.5. rbcL + trnL-F combined analyses The MP heuristic search, in which no transversion bias was enforced, found 80 most parsimonious trees, each tree with L D 662, CI D 0.64, and RI D 0.85. When each partition had a transversion bias applied (6:1 for rbcL and 3:1 for trnL-F), 48 most parsimonious trees were found with L D 1104, CI D 0.69, and RI D 0.87. The transversion-biased topology was better resolved, especially at deep nodes. The main diVerences between the two topologies were the relative resolutions at deeper nodes and the diVerential placement of D. fragrans. A SH test under the likelihood settings indicated no signiWcant diVerence between the topologies (P D 0.174). The ML bootstrap analysis, based on 1000 replicates of 10 random addition sequence replicates each, was nearly completely resolved except deep within the phylogeny. The resulting topology is not shown here, however, it was mostly congruent with the Bayesian and MP topologies. The only diVerences were the placements of D. fragrans, D. patula, D. pulcherrima, and D. sandwicen- 401 sis. The ¡log likelihood value of this topology was 6237.46. The 50% majority-rule consensus tree found from the Bayesian analysis is shown in Fig. 4 and resulted in nearly the same topology as the ML BS and the MP BS analyses (not shown), although based on a SH test it had a signiWcantly higher (P D 0.028) ¡log likelihood value (6205.28) than the ML topology. The major clades in Fig. 4 are identiWed with Roman numerals. Posterior probability values (PP) from Bayesian analysis, MP BS values, and decay values are indicated in Fig. 4, however, because the ML BS values were nearly identical to MP BS values and did not aVect our interpretation of the results, they are not reported here. Posterior probability values were, without exception, equal to or higher than MP BS and ML BS (not shown) values. It has been suggested that Bayesian analysis overestimates node support (Suzuki et al., 2002), however, others have suggested that the posterior probability estimation methods are more sensitive to phylogenetic signal in data sets than are bootstrapping methods (Alfaro et al., 2003). Consistent with the topologies of the individual data sets, the combined analyses strongly suggest that Hawaiian Dryopteris is not monophyletic. Again, D. wallichiana and D. subbipinnata were sisters to each other and non-Hawaiian D. aYnis var. borreri was strongly supported (PP 100, MP BS 95, d4) as sister to them. The D. fusco-atra varieties were sister to each other, falling within a clade that was supported with 99% PP support, a MP BS value of 70, and included the same species as listed in the trnL-F analysis. The glabra group was monophyletic with 100% PP support and a MP BS value of 72. European D. corleyi and D. aemula were strongly supported as sharing a common ancestor with the glabra group (PP 100, MP BS 100, d11), and N. rubiginosum was sister to the glabra-group + D. corleyi + D. aemula clade (PP 94, MP BS 63, d6). As in the rbcL analysis, the exindusiate group, including D. crinalis and D. mauiensis, was strongly supported as monophyletic (PP 100, MP BS 100, d13) and was weakly supported as sister to a clade including D. tokyoensis, D. odontoloma, and D. pallida. Similar to patterns shown in the single-sequence analyses, no Hawaiian clade was ever most closely related to another Hawaiian clade. When we tested the likelihood of alternative phylogenetic hypotheses using the topological incongruency SH tests, no hypothesis from the pairwise comparisons was as likely as the Bayesian topology (P values were not larger than 0.009 for any comparison and were <0.001 for eight of the 12 pairwise comparisons). Similarly, no hypothetical topology in which Hawaiian Dryopteris was monophyletic was as likely as the supported topology (P < 0.001 for each comparison). In every case, the constrained hypotheses were signiWcantly less likely than the Bayesian topology. Although not all plausible trees with each speciWc group 402 J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 Fig. 4. Topology from the combined Bayesian analysis. The numbers above branches, not in parentheses, are posterior probability values (PP) from Bayesian analysis, the numbers, in parentheses, above branches are decay values (d) from decay analysis, and the numbers below are bootstrap (BS) values from maximum parsimony (MP) analysis. For nodes not present in the MP BS analysis, no d or BS values are listed. J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 monophyletic were included in these analyses, due to the very low P values we believe the results would hold even if all plausible trees were tested. Additionally, forcing monophyly on the Hawaiian groups would require breaking up strongly supported clades in every case. 4. Discussion 4.1. Phylogenetic relationships Although our analyses were based only on cpDNA sequences, we will discuss putative relationships among Dryopteris taxa. As mentioned above, clades in Fig. 4 have been numbered with Roman numerals for simplicity in discussing the results. Each of the three data sets, rbcL, trnL-F, and combined, yielded similar results regarding relationships among Hawaiian taxa. Each type of analysis, MP, ML (combined only), and Bayesian, supports the hypothesis that Hawaiian Dryopteris is polyphyletic, and suggests that there are four or Wve independently derived groups (including N. rubiginosum) in Hawai’i, which do not share a unique common ancestor. Our results indicate that the Hawaiian clades have closer relatives in other parts of the world. We hypothesize then that there have been Wve successful colonizations of diVerent Dryopteris species in the Hawaiian Islands. This hypothesis assumes that mainland Dryopteris species did not arise from the Hawaiian taxa (but, see below). Dryopteris wallichiana and D. subbipinnata, part of Clade VII, consistently grouped together (Figs. 2–4; PP 84, MP BS 50, d1) and without exception were strongly supported as sister to D. aYnis var. borreri (PP 100, MP BS 95, d4). Dryopteris wallichiana is a pantropical species and is the only Hawaiian Dryopteris that is not endemic. It has been reported as a tetraploid in Hawai’i (n D 82; F.S. Wagner, pers. comm.; the base chromosome number in Dryopteris in x D 41), but Fraser-Jenkins (1994) rather suggests that it is an apomictic diploid that has been mistaken for a tetraploid. Plants of D. wallichiana from Asia have been counted three separate times as diploids (n D 41; Gibby, 1985; Khullar et al., 1988; Punetha, 1989) and Fraser-Jenkins (1994) suggests that it is normally an apomictic diploid. Dryopteris subbipinnata is morphologically similar to D. wallichiana, but diVers in that it has a longer petiole, is wider at the laminar base, and the pinnules are longer and more shallowly lobed (Fraser-Jenkins, 1994, and personal observations). It has been reported as a hexaploid (n D 123; Wagner (1993)), however, Fraser-Jenkins (1994) suggests it is likely an apomictic triploid rather than a hexaploid. Further cytological and experimental investigations are needed to clarify the ploidy levels and mating systems of D. wallichiana and D. subbipinnata. Nevertheless, it is not surprising that these analyses indicate that these 403 two species are closely related. It has been hypothesized that D. subbipinnata is of allopolyploid hybrid origin and that D. wallichiana is one of its parental species (Fraser-Jenkins, 1994). There is low molecular divergence between these two species (rbcL D 0.15% divergence; trnL-F D 0% divergence), suggesting that the separation of D. subbipinnata from D. wallichiana (and/ or its other parent) is a relatively recent event. A phylogeny based on a nuclear gene or other biparentally inherited markers might assist in elucidating whether D. subbipinnata is of alloploid hybrid origin or is an in situ autoploid derivative of D. wallichiana or another closely related unknown species. Clade VII was part of a larger clade (not numbered) that included two other Hawaiian taxa, D. fusco-atra var. fusco-atra and D. fusco-atra var. lamoureuxii, Clade VIII. Dryopteris fusco-atra var. fusco-atra has been reported to be a hexaploid with n D c. 123 (F.S. Wagner, pers. comm.). The two taxa were strongly supported as sister (Fig. 4; PP 100, MP BS 100, d13), and belonged to an unresolved clade including D. stenolepis, D. sublacera, D. pycnopteroides, and D. uniformis (PP 99; MP BS 70, d1). Additional data will need to be collected to identify the closest living relatives of D. fusco-atra. The two D. fusco-atr a varieties had identical sequences for trnL-F and for rbcL they were only 0.07% diVerent. Dryopteris fusco-atra var. lamoureuxii is only found on E. Maui and these results suggest that it may be only an ecotype of the more widespread taxon, D. fusco-atra var. fusco-atra. Additional study is needed to determine if D. fusco-atra is of allo- or autopolyploid origin and what other taxa may have contributed to its genome. Analyses based on trnL-F and the combined data set each supported a monophyletic glabra group as being closely related to D. corleyi, D. aemula, and N. rubiginosum (Clade V, Figs. 3 and 4, PP 100, MP BS 72, d2). In contrast, the rbcL analyses included D. corleyi and D. aemula within the glabra group (Fig. 2). The glabra group, as described by Palmer (2003), includes D. glabra, as well as D. hawaiiensis, although Palmer (2003) described D. hawaiiensis as “unrelated” to D. glabra. Our data suggest, however, that the hexaploid D. hawaiiensis (n D 123; F.S. Wagner, pers. comm.) is closely related to D. glabra, which has been counted as a diploid (n D 41; F.S. Wagner, pers. comm.). Sequences from rbcL showed only 0.3% base diVerences between D. hawaiiensis and six other taxa in the glabra group and it had identical trnL-F sequences as Wve other glabra group taxa. It is unknown whether D. hawaiiensis is an autopolyploid or an allopolyploid. Regardless of its polyploid origins, it appears either that D. hawaiiensis has inherited its chloroplast genome from D. glabra or that the two species share a very recent common ancestor. Circumscription of and relationships among the members of the glabra group remain unresolved. This is perhaps not surprising as this group has recently been 404 J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 the source of much taxonomic disagreement (FraserJenkins, 1994; Herat, 1979; Palmer, 2003; Wagner, 1993, 1988; Wagner et al., 1995a, 1999a). These authors have disagreed about whether the varieties of D. glabra are in fact morphologically distinct enough to be considered separate species. Our analyses indicate that there is little chloroplastic molecular divergence among the D. glabra varieties. Pairwise divergences varied from 0 to 0.32% for rbcL and from 0 to 0.15% for trnL-F. These molecular results alone do not provide evidence warranting the recognition of varieties as species or vice-versa. However, as more data are added, this may change. We have completed studies using more variable molecular markers and morphological data in an attempt to objectively resolve this issue, which will be reported elsewhere (Geiger and Ranker, unpublished). Analyses of the three data sets consistently placed two European species, D. corleyi and D. aemula, as closely related to, or within, the glabra group. Analyses based on rbcL placed both species within the glabra group, making the glabra group paraphyletic. However, analyses from both the trnL-F and combined data sets suggest that the two European species are basal to a monophyletic glabra group. Pairwise distances for rbcL between D. corleyi and members of the glabra group (range D 0.0030–0.0053) were greater than the pairwise distances between D. aemula and members of the glabra group (range D 0.0010– 0.0038). However, for trnL-F, D. corleyi and D. aemula shared identical sequences, thus there was no diVerence in the pairwise distance ranges between the two species when compared with the glabra group (range D 0.0030– 0.0064). That D. corleyi and D. aemula were supported as close relatives is not surprising as D. corleyi is thought to be an allopolyploid species that arose from chromosome doubling after a hybridization event between D. oreades and D. aemula (Fraser-Jenkins and Widen, 1993). From these analyses it would appear that D. corleyi inherited its chloroplast genome from D. aemula, as the two putative parental species, D. oreades and D. aemula, are resolved in very diVerent parts of the tree. Another interesting result regarding this group is that the Hawaiian endemic, Nothoperanema rubiginosum, was supported by both rbcL and the combined dataset as sister to the glabra + D. corleyi + D. aemula clade (Figs. 2, 4 PP 94, MP BS 63, d1), but the trnL-F analysis places N. rubiginosum as sister to the aforementioned clade and the non-Hawaiian clade (Fig. 3; Clade IV of Fig. 4). Nothoperanema rubiginosum has been variously treated in the genera Lastrea, Ctenitis, and Dryopteris (see Smith and Palmer, 1995). Smith and Palmer (1995) suggest that Nothoperanema is closely related to Dryopteris but is separated from it based on laminar scale diVerences. Our results suggest that N. rubiginosum should be placed in the genus Dryopteris to preserve the monophyly of the latter. As suggested by Fraser-Jenkins (1994) and by our results, further studies of the genus Nothoperanema in conjunction with the genus Dryopteris are needed to resolve whether Nothoperanema species should be treated in the genus Dryopteris. Our analyses suggest close relationships among members of the so-called exindusiate group, Clade II (Figs. 2– 4; PP 100, MP BS 100, d13). The leaves of these species range in length from 0.30–0.61 to 2.44 m (Palmer, 2003), and all species lack indusia. Our results support Palmer’s (2003) hypothesis that D. unidentata, D. sandwicensis, and D. tetrapinnata are closely related. Although not as closely related as the above mentioned three species, our data indicate that D. crinalis and D. mauiensis also belong in this clade, with all Wve species sharing a putative unique common ancestor. Pairwise distances among species in this group ranged from 0.0008 to 0.0076 for rbcL and from 0 to 0.0096 for trnL-F. The exindusiate group was supported as sister to the non-Hawaiian Clade I in the rbcL and combined analyses, although weakly, and its relationship to the rest of the ingroup was unresolved in the trnL-F analysis. None of the species in Clade I lack indusia. There are, however, non-Hawaiian exindusiate species of Dryopteris of various geographical origins, thus it is possible that we have not yet sampled the closest relatives of the exindusiate group. Another possible explanation is that we have sampled this group’s closest extant relatives, but the ancestor of the exindusiate Hawaiian species lost the indusium subsequent to colonization, and that the extant species inherited this trait. Sequence divergences among species within each of the monophyletic groups in Hawai’i were very low. This result is not unusual for closely related species in Hawai’i. Many studies have shown that although there is considerable morphological divergence between related Hawaiian species, there is often little molecular divergence (Helenurum and Ganders, 1985; Lowrey and Crawford, 1985; Lowrey, 1995; Witter and Carr, 1988; Witter, 1990). Our study provides another example of this phenomenon and may reXect recent diversiWcation of these clades. 4.2. Biogeographical implications In each analysis, the four Hawaiian clades of Dryopteris were more closely related to non-Hawaiian species than they were to each other (Figs. 2–4). The Hawaiian endemic, Nothoperanema rubiginosum, was strongly supported as closely related to, but not sister to, the Hawaiian glabra group. These results suggest that there were Wve separate introductions of dryopteroid ferns to the Hawaiian archipelago. This Wnding, although not entirely surprising when considering the dispersal ability of fern spores via wind, is contrary to what is believed to have occurred for several other endemic plant groups occurring in Hawai’i (Baldwin et al., 1991; Baldwin and Robichaux, 1995; Pax et al., 1997; Wagner et al., 1995b). In each of those groups, the evidence suggests J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 that there was a single colonizing ancestor and subsequent evolutionary radiation. Only three studies of Xowering plants have suggested multiple, independent colonizations of congeneric species to the Hawaiian Islands (Rubus, Howarth et al., 1997; Chamaesyce, Raz et al., 1998; and Scaevola, Howarth et al., 1999), and only one study has provided evidence for multiple colonizations of a single fern species (Ranker et al., 1994) in Hawai’i. However, Motley and Raz (2004) reported new evidence indicating that the endemic Hawaiian Chamaesyce taxa are monophyletic resulting from only one colonization event. Outside of Hawai’i, multiple colonizations of oceanic islands have been hypothesized for Lavatera (Fuertes-Aguilar et al., 2002) and Ilex (Cuenoud et al., 2000) in the Canary Islands, and Gossypium in the Galápagos (Wendel et al., 1995). Our results do not allow for a strongly supported hypothesis on the geographical origins of D. wallichiana and D. subbipinnata (Clade VII) in Hawai’i. These two species are clearly supported as having a diVerent phylogenetic history from the other Hawaiian species, but their geographical origin is unclear as D. wallichiana has a pantropical distribution. The close relationship with D. aYnis subsp. borreri does not help elucidate the matter, as it, too, is geographically widespread throughout Europe to W. Asia. Phylogeographic studies of populations of D. wallichiana may assist in clarifying the origins of the Hawaiian populations of this species. Dryopteris fusco-atra is related to Clades VI, VII, IX, and to D. stenolepis. Our data support the hypothesis that the presence of D. fusco-atra in Hawai’i is the result of an independent colonization event. The three species in Clade IX and D. stenolepis are all native to SE Asia. The two species in Clade VI are also native to SE and eastern Asia, speciWcally China and Japan. Although the closest relative of D. fusco-atra has yet to be identiWed, this study provides support for this species being of East Asian origin. The exindusiate group, Clade II, is supported as a third monophyletic Dryopteris clade in Hawai’i. Dryopteris tokyoensis and D. odontoloma of Clade I, which is the sister group to Clade II in the rbcL and combined analyses (Figs. 2 and 4), are geographically distributed throughout E. Asia. Dryopteris pallida of Clade I is a W. European taxon. It is possible that the Hawaiian Clade II arose from a SE Asian species that dispersed to Hawai’i and a related species also dispersed to W. Europe. However, the relationship between Clades I and II is weakly supported by the combined analysis (Fig. 4; PP 71, MP BS 55, d3). It is probable that we have not yet sampled the closest relatives of the exindusiate group, and thus any speciWc biogeographical hypotheses are likely to be misguided at this time. Our data indicate that the Hawaiian Dryopteris taxa of Clade V, the glabra group, are closely related to the W. European species D. aemula and D. corleyi (Fig. 4, PP 405 100, MP BS 100, d11). The relationship between D. aemula and D. corleyi is discussed above. Hillebrand (1888) recognized the close relationship between D. aemula and Hawaiian D. glabra as did Fraser-Jenkins (1994). FraserJenkins (1994) comments that “it is unlikely to have been mere chance long-distance dispersal with such a restricted species [D. aemula].” Indeed, it is diYcult to explain this type of biogeographical disjunction. However, from our data, and from previous morphological observations by Hillebrand (1888) and Fraser-Jenkins (1994), it would seem that there must have been a long distance dispersal event of D. aemula to the Hawaiian Islands, and the lack of gene Xow between the Hawaiian populations and W. European populations allowed the genetic separation of D. glabra. There are, however, at least two alternative hypotheses. The sister clade to the Hawaiian + W. European Clade V + Hawaiian N. rubiginosum, Clade IV, is comprised of seven species, all of which occur in SE Asia. It is possible that D. aemula and D. glabra share an unsampled common ancestor (or an extinct common ancestor) that was from SE Asia. This ancestor could have recently dispersed to both W. Europe and Hawai’i separately. Alternatively, D. aemula could have arisen in the Hawaiian Islands and dispersed to W. Europe, since in our analyses it is embedded in a clade of endemic Hawaiian species. We cannot reject any of these scenarios with our phylogenetic hypotheses. Nothoperanema rubiginosum may represent a Wfth dispersal event to the Hawaiian Islands. This species was sister to the Hawaiian glabra group + D. corleyi + D. aemula. Given the close relationship of the entire clade to the SE Asian Clade IV, a SE Asian origin for this species is the best supported hypothesis. However, there are Wve other described species of Asiatic Nothoperanema that should be sampled and included in subsequent phylogenetic analyses to best infer the geographical origins of this Hawaiian endemic. Our analyses suggest that three, perhaps four, of the Wve Hawaiian clades have Asian geographical associations. We may have expected to Wnd this pattern for several reasons. First, Fraser-Jenkins (1994) commented that the Hawaiian fern Xora appears to have a considerable resemblance to the Sino-Japanese and Sino-Himalayan Xoras and Fosberg (1948) hypothesized that 14 of the 25 Dryopteris species he recognized were of IndoPaciWc origin. Second, the jetstream occurs as a latitudinally undulating band of fast-moving air, »5500–17,000 meters in altitude, Xowing from west to east. The jetstream accelerates as it moves eastward from SE Asia (up to 195 kph) and decelerates as it moves over the Hawaiian Islands (down to »115 kph; Carlquist, 1980). Spores could be moved up into the jetstream by storms and transported from SE Asia/Malaysia to Hawai’i in 2– 4 days. Third, the taxa included in our analyses are mostly native to Asia, thus our analyses could be biased. Until we have sampled all Dryopteris and Nothopera- 406 J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 nema species and obtained sequence data from both chloroplast and nuclear markers, we will not have a complete picture of the origins of each Hawaiian clade, at least from a molecular perspective. Acknowledgments This study was funded by grants from the National Science Foundation (DEB-0104962), the Department of Ecology and Evolutionary Biology at the University of Colorado, the University of Colorado Graduate School, the University of Colorado Museum, and the Botanical Society of America. The authors thank Dan Palmer for sharing his time and knowledge of Dryopteris in Hawai’i with us. We also thank Florence Wagner for her contributions. For Weld and collection assistance we thank everyone at the National Tropical Botanic Garden, especially David Lorence, Ken Wood, Tim Flynn, and Steve Perlman; the Maui Pineapple Company, in particular Hank Oppenheimer; and Patti Welton and the National Park Service. References Alfaro, M.E., Zoller, S., Lutzoni, F., 2003. Bayes or bootstrap? A simulation study comparing the performance of Bayesian Markov Chain Monte Carlo sampling and bootstrapping in assessing phylogenetic conWdence. Mol. Biol. Evol. 20, 255–266. Baldwin, B.G., Robichaux, R.H., 1995. Historical biogeography and ecology of the Hawaiian Silversword alliance (Asteraceae). In: Wagner, W.L., Funk, V.A. (Eds.), Hawaiian Biogeography: Evolution on a Hot Spot Archipelago. Smithsonian Institution Press, Washington, DC, pp. 259–285. Baldwin, B.G., Kyhos, D.W., Dvorák, J., Carr, G.D., 1991. Chloroplast DNA evidence for a North American origin of the Hawaiian Silversword Alliance (Asteraceae). Proc. Natl. Acad. Sci. USA 88, 1840–1843. Bremer, K., 1988. The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42, 795–803. Carlquist, S., 1980. Hawai’i: A Natural History, 2nd ed. PaciWc Tropical Botanical Garden, Lawai. Carson, H.L., Clague, D.A., 1995. Geology and biogeography of the Hawaiian Islands. In: Wagner, W.L., Funk, V.A. (Eds.), Hawaiian Biogeography: Evolution on a Hot Spot Archipelago. Smithsonian Institution Press, Washington, DC, pp. 14–29. Clague, D.A., 1996. The growth and subsidence of the HawaiianEmperor volcanic chain. In: Miller, S.E. (Ed.), The Origin and Evolution of PaciWc Island Biotas, New Guinea to Eastern Polynesia: Patterns and Processes. SPB Academic Publishing, Amsterdam, pp. 35–50. Cuenoud, P., Martinez, M.A.D., Loizeau, P.A., Spichiger, R., Andrews, S., Manen, J.F., 2000. Molecular phylogeny and biogeography of the genus Ilex L. (Aquifoliaceae). Ann. Bot. 85, 111–122. Donoghue, M.J., Olmstead, R.G., Smith, J.F., Palmer, J.D., 1992. Phylogenetic relationships of Dipsacales based on rbcL sequences. Ann. Missouri Bot. Gard. 79, 333–345. Doyle, J.J., Doyle, J.L., 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. 19, 11–15. Eriksson, T., 1998. AutoDecay ver. 4.0.1 (program distributed by the author). Department of Botany, Stockholm University, Stockholm. Farris, J.S., Källersjö, M., Kluge, A.G., Bult, C., 1994. Testing signiWcance of incongruence. Cladistics 10, 315–319. Farris, J.S., Källersjö, J., Kluge, A.G., Bult, C., 1995. Constructing a signiWcance test for incongruence. Syst. Biol. 44, 570–572. Fosberg, F.R., 1948. Derivation of the Xora of the Hawaiian Islands. In: Zimmerman, E.C. (Ed.), Insects of Hawai’i. University Press, Honolulu, pp. 107–119. Fraser-Jenkins, C.R., 1986. A classiWcation of the genus Dryopteris (Pteridophyta: Dryopteridaceae). Bull. Brit. Mus. Nat. Hist. 14, 183–218. Fraser-Jenkins, C.R., 1994. Dryopteris (Pteridophyta) of Hawai’i—a monographic study. Thaiszia-J. Bot., Kosice 4, 15–47. Fraser-Jenkins, C.R., Widen, C.J., 1993. Phloroglucinol derivatives in Dryopteris ardechensis and D. corleyi (Pteridophyta, Dryopteridaceae) and their putative ancestors. Annales Botanici Fennici 30, 43– 51. Fuertes-Aguilar, J., Ray, M.E., Francisco-Ortega, J., Santos-Guerra, A., Jansen, R.K., 2002. Molecular evidence from chloroplast and nuclear markers for multiple colonizations of Lavatera (Malvaceae) in the Canary Islands. Syst. Bot. 27, 74–83. Gibby, M., 1985. Cytological observations on Indian subcontinent and Chinese Dryopteris and Polystichum (Pteridophyta: Dryopteridaceae). Bull. Brit. Mus. Nat. Hist. (Bot.) 14, 1–42. Goldman, N., Anderson, J.P., Rodrigo, A.G., 2000. Likelihood-based tests of topologies in phylogenetics. Syst. Biol. 49, 652–670. Hasebe, M., Wolf, P.G., Pryer, K.M., Ueda, K., Ito, M., Sano, R., Gastony, G.J., Yokayama, J., Manhart, J.R., Murakami, N., Crane, E.H., HauXer, C.H., Hauk, W.D., 1995. Fern phylogeny based on rbcL nucleotide sequences. Am. Fern. J. 85, 134–181. Hasegawa, M., Kishino, K., Yano, T., 1985. Dating the human–ape splitting by a molecular clock of mitochondrial DNA. J. Mol. Evol. 22, 160–174. HauXer, C.H., Ranker, T.A., 1995. RbcL sequences provide phylogenetic insights among sister species of the fern genus Polypodium. Am. Fern. J. 85, 361–374. Helenurum, K., Ganders, F.R., 1985. Adaptive radiation and genetic diVerentiation in Hawaiian Bidens. Evolution 39, 753–765. Hennequin, S., Ebihara, A., Ito, M., Iwatsuki, K., Dubuisson, J-Y., 2003. Molecular systematics of the fern genus Hymenophyllum s. l. (Hymenophyllaceae) based on chloroplastic coding and noncoding regions. Mol. Phylogenet. Evol. 23, 283–301. Herat, T.R., 1979. A Systematic Study of the Genus Dryopteris in Hawai’i. University of Hawai’i, Honolulu. Hillebrand, W., 1888. Flora of the Hawaiian Islands: A Description of their Phanerogams and Vascular Cryptogams. Lubrecht and Cramer, Monticello, NY. Hoshizaki, B.J., Wilson, K.A., 1999. The cultivated species of the fern genus Dryopteris in the United States. Am. Fern J. 89, 1–98. Howarth, D.G., Gardner, D.E., Morden, C.W., 1997. Phylogeny of Rubus subgenus Idaeobatus (Rosaceae) and its implications toward colonization of the Hawaiian Islands. Syst. Bot. 22, 433–441. Howarth, D.G., Gustafsson, M.H.G., Motley, T.J., 1999. Phylogenetics and biogeography of Scaevola in the PaciWc basin. XVI International Botanical Congress Abstracts, 4539, St. Louis, Missouri, USA. Huelsenbeck, J.P., Ronquist, F., 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17, 754–755. Kim, H., Keeley, S.C., Vroom, P.S., Jansen, R.K., 1998. Molecular evidence for an African origin of the Hawaiian endemic Hesperomannia (Asteraceae). Proc. Natl. Acad. Sci. USA 95, 15440–15445. Kirch, P., 1982. The impact of the prehistoric Polynesians on the Hawaiian ecosystem. PaciWc Sci. 36, 1–14. Khullar, S.P., Sharma, S.S., Verma, S.C., 1988. SOCGI plant chromosome number reports VI. J. Cytol. Genet. 23, 38–52. Lowrey, T.K., 1995. Phylogeny, adaptive radiation, and biogeography of Hawaiian Tetramolopium (Asteraceae, Astereae). In: Wagner, W.L., Funk, V.A. (Eds.), Hawaiian Biogeography: Evolution on a J.M.O. Geiger, T.A. Ranker / Molecular Phylogenetics and Evolution 34 (2005) 392–407 Hot Spot Archipelago. Smithsonian Institution Press, Washington, DC, pp. 195–220. Lowrey, T.K., Crawford, D.J., 1985. Allozyme divergence and evolution in Tetramolopium (Compositae: Astereae) on the Hawaiian Islands. Syst. Bot. 10, 64–72. Maddison, D.R., Maddison, W.P., 2000. MacClade 4. Sinauer Associates, Massachusetts. Martin, A.P., Naylor, G.J.P., 1997. Independent origins of Wlter-feeding in megamouth and basking sharks (Order Lamniformes) inferred from phylogenetic analysis of cytochrome b gene sequences. In: Yano, K., Morrissey, J.F., Yabumoto, Y., Nakaya, K. (Eds.), Biology of Megamouth Shark. Tokai University Press, Tokyo, pp. 39– 50. Motley, T.J., Raz, L., 2004. Origins, evolution, and systematics of Hawaiian and PaciWc Chamaesyce (Euphorbiaceae). Botany 2004 Abstracts, 522, Snowbird, Utah, USA. Palmer, D.D., 2003. Hawai’i’s Ferns and Fern Allies. University of Hawai’i Press, Honolulu. Pax, D.L., Price, R.A., Michaels, H.J., 1997. Phylogenetic position of the Hawaiian geraniums based on rbcL sequences. Am. J. Bot. 84, 72–78. Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14, 817–818. Powell, E.A., Kron, K.A., 2002. Hawaiian blueberries and their relatives—a phylogenetic analysis of Vaccinium sections Macropelma, Myrtillus, and Hemimyrtillus (Ericaceae). Syst. Bot. 27, 768–779. Price, J.P., Clague, D.A., 2002. How old is the Hawaiian biota. Geology and phylogeny suggest recent divergence. Proc. R. Soc. Lond. B 269, 2429–2435. Punetha, N., 1989. Cytological observations on ferns of Kumoan (N.W. Himalaya). Aspects Plant Sci. 11, 459–465. Ranker, T.A., Floyd, S.K., Trapp, P.G., 1994. Multiple colonizations of Asplenium adiantum-nigrum onto the Hawaiian archipelago. Evolution 48, 1364–1370. Ranker, T.A., Gemmill, C.E.C., Trapp, P.G., 2000. Microevolutionary patterns and processes of the native Hawaiian colonizing fern Odontosoria chinensis (Lindsaeaceae). Evolution 54, 828–839. Ranker, T.A., Gemmill, C.E.C., Trapp, P.G., Hambleton, A., Ha, K., 1996. Population genetics and reproductive biology of lava-Xow colonising species of Hawaiian Sadleria (Blechnaceae). In: Johns, R.J. (Ed.), Pteridology in Perspective. Royal Botanical Gardens, Kew, pp. 581–598. Ranker, T.A., Geiger, J.M.O., Kennedy, S.C., Smith, A.R., HauXer, C.H., Parris, B.S., 2003. Molecular phylogenetics and evolution of the endemic Hawaiian genus Adenophorus. Mol. Phylogenet. Evol. 26, 337–347. Ranker, T.A., Smith, A.R., Parris, B.S., Geiger, J.M.O., HauXer, C.H., Straub, S.C.K., Schneider, H., 2004. Phylogeny and evolution of grammitid ferns (Grammitidaceae): a case of rampant morphological homoplasy. Taxon 53, 415–428. Raz, L., Albert, V.A., Motley, T.J., 1998. Molecular systematics of Hawaiian Chamaesyce (Euphorbiaceae). Am. J. Bot. 85 (supplement), 153–154. Russell, A.E., Ranker, T.A., Gemmill, C.E.C., Farrar, D.R., 1999. Patterns of clonal diversity in Dicranopteris linearis on Mauna Loa, Hawai’i. Biotropica 31, 449–459. Schneider, H., Russell, S.J., Cox, C.J., Bakker, F., Henderson, S., Rumsey, F., Barrett, J., Gibby, M., Vogel, J.C., 2004a. Chloroplast phylogeny of Asplenioid ferns based on rbcL and trnL-F spacer sequences (Polypodiidae, Aspleniaceae) and its implications for biogeography. Syst. Bot. 29, 260–274. Schneider, H., Smith, A.R., CranWll, R., Hildebrand, T., HauXer, C.H., Ranker, T.A., 2004b. Unraveling the phylogeny of polygrammoid ferns (Polypodiaceae and Grammitidaceae): exploring aspects of 407 the diversiWcation of epiphytic plants. Mol. Phylogenet. Evol. 31, 1041–1063. Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116. Smith, A.R., 1972. Comparison of fern and Xowering plant distributions with some evolutionary interpretations for ferns. Biotropica 4, 4–9. Smith, A.R., Palmer, D.D., 1995. Ctenitis rubiginosa, from Hawai’i, belongs in Nothoperanema. Am. Fern J. 85, 63. Sohmer, S.H., Gustafson, R., 1987. Plants and Flowers of Hawai’i. University of Hawai’i Press, Honolulu. Suzuki, Y., Glazko, G.V., Nei, M., 2002. Overcredibility of molecular phylogenies obtained by Bayesian phylogenetics. Proc. Natl. Acad. Sci. USA 99, 16138–16143. SwoVord, D.L., 2002. PAUP*. Phylogenetic analysis using parsimony (* and other methods). Sinauer Associates, Sunderland, MA. Taberlet, P., Gielly, L., Pautou, G., Bouvet, J., 1991. Universal primers for ampliWcation of three non-coding regions of chloroplast DNA. Plant Mol. Biol. 17, 1105–1109. Tamura, K., Nei, M., 1993. Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Mol. Biol. Evol. 16, 1292–1299. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The ClustalX windows interface: Xexible strategies for multiple sequences alignment aided by quality analysis tools. Nucleic Acid Res. 24, 4876–4882. Tryon, R., 1970. Development and evolution of fern Xoras of oceanic islands. Biotropica 2, 76–84. Wagner Jr., W.H., 1988. Status of the Hawaiian fern Xora. Fiddlehead Forum 15, 11–14. Wagner Jr., W.H., 1993. New species of Hawaiian pteridophytes. Contr. Univ. Michigan Herb. 19, 63–82. Wagner Jr., W.H., 1995. Evolution of Hawaiian ferns and fern allies in relation to their conservation status. PaciWc Sci. 49, 31–41. Wagner Jr., W.H., Wagner, F.S., Flynn, T., 1995a. Taxonomic notes on the pteridophytes of Hawai’i. Contr. Univ. Michigan Herb. 20, 241– 260. Wagner Jr., W.H., Wagner, F.S., Palmer, D.D., Hobdy, R.W., 1999a. Taxonomic notes on the pteridophytes of Hawai’i-II. Contr. Univ. Michigan Herb. 22, 135–187. Wagner, W.L., Funk, V.A., 1995. Hawaiian Biogeography: Evolution on a Hot Spot Archipelago. Smithsonian Institution Press, Washington, DC. Wagner, W.L., Weller, S.G., Sakai, A.K., 1995b. Phylogeny and biogeography in Schiedea and Alsinidendron (Caryophyllaceae). In: Wagner, W.L., Funk, V.A. (Eds.), Hawaiian Biogeography: Evolution on a Hot Spot Archipelago. Smithsonian Institution Press, Washington, DC, pp. 221–258. Wagner, W.L., Herbst, D.R., Sohmer, S.H., 1999b. Manual of the Flowering Plants of Hawai’i. University of Hawai’i Press/Bishop Museum Press, Honolulu. Wendel, J.F., Schnabel, A., Seelanan, T., 1995. An Unusual Ribosomal DNA-Sequence from Gossypium Gossypioides Reveals Ancient, Cryptic, Intergenomic Introgression. Mol. Phylogenet. Evol. 4, 298– 313. Witter, M.S., 1990. Evolution in the Madiinae: evidence from enzyme electrophoresis. Ann. Missouri Bot. Gard. 77, 110–117. Witter, M.S., Carr, G.D., 1988. Adaptive radiation and genetic diVerentiation in the Hawaiian Silversword alliance (Compositae: Madiinae). Evolution 42, 1278–1287. Wolf, P.G., Soltis, P.S., Soltis, D.E., 1994. Phylogenetic relationships of Dennstaedtioid ferns: evidence from rbcL sequences. Mol. Phylogenet. Evol. 3, 383–392.