Molecular Phylogenetics and Evolution 55 (2010) 1008–1017
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution
journal homepage: www.elsevier.com/locate/ympev
Phylogeny and biogeography of the eastern Asian–North American disjunct wild-rice genus (Zizania L., Poaceae)
Xinwei Xu a, Christina Walters b, Michael F. Antolin c, Mara L. Alexander d, Sue Lutz e, Song Ge f, Jun Wen e,f,* a Freshwater Ecological Field Station of Liangzi Lake, Wuhan University, Wuhan 430072, China b USDA/ARS National Center for Genetic Resources Preservation, 1111 South Mason Street, Fort Collins, CO 80521, USA c Department of Biology, Colorado State University, Fort Collins, CO 80523-1878, USA d San Marcos National Fish Hatchery and Technology Center, 500 East McCarty Lane, San Marcos, TX 78666, USA e Department of Botany, National Museum of Natural History, MRC 166, Smithsonian Institution, Washington, DC 20013-7012, USA f State Key Laboratory of Systematic and Evolutionary Botany, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China article info abstract
Article history: The wild-rice genus Zizania includes four species disjunctly distributed in eastern Asia and North Received 3 September 2009 America, with three species (Z. aquatica, Z. palustris, and Z. texana) in North America and one (Z. latifolia) Revised 17 November 2009 in eastern Asia. The phylogeny of Zizania was constructed using sequences of seven DNA fragments Accepted 18 November 2009 (atpB-rbcL, matK, rps16, trnL-F, trnH-psbA, nad1, and Adh1a) from chloroplast, mitochondrial, and nuclear Available online 26 November 2009 genomes. Zizania is shown to be monophyletic with the North American species forming a clade and the eastern Asian Z. latifolia sister to the North American clade. The divergence between the eastern Asian Z. Keywords: latifolia and the North American clade was dated to be 3.74 (95% HPD: 1.04–7.23) million years ago (mya) Adh1a gene using the Bayesian dating method with the combined atpB-rbcL, matK, rps16, trnL-F, and nad1 data. Biogeography Intercontinental disjunction Biogeographic analyses using a likelihood method suggest the North American origin of Zizania and its Wild-rice migration into eastern Asia via the Bering land bridge. Among the three North American species, the Zizania organellar data and the haplotype network of the nuclear Adh1a gene show a close relationship between Z. palustris and the narrowly distributed endangered species Z. texana. Bayesian dating estimated the divergence of North American Zizania to be 0.71 (95% HPD: 0.12–1.54) mya in the Pleistocene. The non-monophyly of Z. palustris and Z. aquatica in the organellar and nuclear data is most likely caused by incomplete lineage sorting, yet low-frequency unidirectional introgression of Z. palustris into Z. aquatica is present in the nuclear data as well. Published by Elsevier Inc.
1. Introduction and evolution of the disjunct taxa in both continents have rarely been examined (Wen, 1999; Xiang et al., 2004; Wen et al., 2009). The intercontinental disjunction between eastern Asia and The wild-rice genus Zizania L. belongs to the rice tribe (Oryzeae, North America is a well-known biogeographic pattern in the north- Poaceae) and is an aquatic/wetland genus with four species dis- ern hemisphere and has attracted considerable attention from junctly distributed between eastern Asia and North America (Ter- plant biologists (Xiang et al., 1998, 2000; Wen, 1999, 2001; Manos rell et al., 1997). In North America, two annual species Z. palustris and Donoghue, 2001). The phylogenetic relationships of disjunct L. (with var. palustris and var. interior (Fassett) Dore) and Z. aquatica lineages, the timing of the disjunctions, and migration pathways L. (with var. aquatica and var. brevis Fassett), are widespread in the of many taxa have been investigated based on molecular data Great Lake region and along the Atlantic coastal plains, respec- (e.g., Xiang et al., 1998; Wen, 2000; Nie et al., 2005, 2006a,b; Peng tively, and have some overlapping ranges (Aiken et al., 1988; Ter- and Wang, 2008). Most of these studies have focused on woody rell et al., 1997). The perennial Z. texana Hitchcock is restricted to a plants or terrestrial herbs, but few studies have examined aqua- 2.4 km area of the upper San Marcos River in southcentral Texas. It tic/wetland plants. At present, investigators have mainly discussed is an endangered species geographically isolated from all other the questions on the processes of the formation of the interconti- Zizania taxa by at least 640 km (Terrell et al., 1978). Zizania latifolia nental disjunct pattern, such as estimating divergence times and (Griseb.) Turcz. ex Stapf is a perennial widely distributed in eastern inferring ancestral areas. Questions on subsequent diversification Asia (Wu et al., 2006). Of the four species, two are economically important as field crops. Zizania palustris has served as a traditional staple for native Americans for centuries (Johnson, 1969) and as a * Corresponding author. Address: Department of Botany, MRC 166, Smithsonian Institution, Washington, DC 20013-7012, USA. Fax: +1 202 786 2563. specialty commercial crop more recently (Hayes et al., 1989; Oelke, E-mail address: [email protected] (J. Wen). 1993). Zizania latifolia was once used as an important grain in
1055-7903/$ - see front matter Published by Elsevier Inc. doi:10.1016/j.ympev.2009.11.018 X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017 1009 ancient China and has been cultivated as an aquatic vegetable be- and 0.6 U GoTaq DNA polymerase (Promega) in a volume of cause the young shoots become swollen, soft and edible after being 25 lL under the following conditions: 3 min at 95 °C, followed by infected by the fungus Ustilago esculenta P. Henn. (Thrower and 35 cycles of 30 s at 94 °C, 30 s at 50–55 °C, and 90 s at 72 °C, and Chan, 1980; Zhai et al., 2001; Guo et al., 2007). then a final 5 min extension at 72 °C. Amplifications were carried Previous phylogenetic studies of the rice tribe have supported out in a PTC-225 Peltier Thermal Cycler. the placement of Zizania in Oryzeae (Zhang and Second, 1989; Duv- The amplification and sequencing primers of the organellar re- all et al., 1993; Ge et al., 2002; Guo and Ge, 2005; Tang et al., 2010). gions were from the following sources: (1) the atpB-rbcL spacer Zizania is most closely related to the South American genus (Manen et al., 1994); (2) the trnH-psbA spacer (Hamilton, 1999); Rhynchoryza Baill. At the infrageneric level, the Asian Z. latifolia is (3) the trnL-F spacer (primers ‘‘c” and ‘‘f” of Taberlet et al., 1991); well differentiated from the North American species by its chromo- (4) the rps16 intron (Oxelman et al., 1997); (5) the matK gene some number and morphology (Duvall, 1987; Terrell et al., 1997), (Ge et al., 1999); and (6) the nad1 intron 2 (Demesure et al., whereas the relationships among three North American species are 1995; Guo and Ge, 2005). Two individuals from each population not fully resolved. Because of the overlap in distributional ranges were sequenced for the organellar regions and additional individ- and the existence of morphological intermediates, the two annual uals from sympatric populations and the Texas population were species, Z. palustris and Z. aquatica, were once treated as a single used. The nuclear Adh1a gene was amplified and sequenced with species (Fassett, 1924; Hitchcock and Chase, 1951). Later studies the Zizania-specific primers (Adh1aF1: 50-CTGACAGAGGTG- based on spikelet anatomy (Duvall and Biesboer, 1988a), artificial TAATGCTTA-30, Adh1aF2: 50-TCGGGACTTCGACCTTCAGT-30, and hybridization (Duvall and Biesboer, 1988b), and isozyme patterns Adh1aR, Xu et al., 2008) from six individuals of each population (Warwick and Aiken, 1986) supported the recognition of two an- of the annual species and 30 individuals of Zizania texana. The re- nual species. A close relationship between Z. texana and Z. aquatica gion of the outgroups was amplified and sequenced with primers is supported by the isoelectric focusing (IEF) profiles of seed pro- Adh1F6 and Adh1R7 (Zhang and Ge, 2007). tein (Duvall and Biesboer, 1989), whereas crossing behaviors sug- All PCR products were purified using the polyethylene glycol gest that Z. texana is more closely related to Z. palustris (Duvall, (PEG)/NaCl method of Kusukawa et al. (1990). Purified PCR prod- 1987). The latter relationship is supported by Horne and Kahn ucts were sequenced using the BigDye Terminator Cycle Sequenc- (1997) based on isozyme and nuclear ribosomal ITS sequence anal- ing Ready Reaction kit (Applied Biosystems). Sequencing reactions ysis. Nevertheless the hypothesis needs to be further tested be- were purified by gel filtration chromatography using Sephadex col- cause Horne and Kahn (1997) sampled only 3–4 individuals from umns (Amersham Pharmacia Biotech) and run on an ABI 3730xl a single population of each species and they used the distantly re- DNA analyzer (Applied Biosystems). The program Sequencher 4.5 lated Oryza sativa L. as the outgroup. The studies by Horne and (GeneCodes Corporation) was used to evaluate chromatograms Kahn (1997) and Xu et al. (2008) only examined the North Ameri- for base confirmation and to edit contiguous sequences. Individu- can or the eastern Asian species of Zizania, respectively. als in Zizania can be either homozygous or heterozygous at the The objectives of this study are to (1) construct the phylogeny Adh1a locus. For a heterozygote, sequences of two alleles can not to resolve the interspecific relationships in Zizania, (2) estimate be separated in the chromatogram when multi-point mutations the divergence times between intercontinental species and among or length differences caused by insertions or deletions exist be- intracontinental species, and (3) reconstruct the biogeographic his- tween the two alleles. In such cases, purified PCR products were tory of Zizania between eastern Asia and North America. cloned using the TOPO TA cloning kit (Invitrogen), and then the two alleles were determined separately by sequencing multiple clones. All sequences of different haplotypes were deposited into 2. Materials and methods the GenBank (Accession Nos. GU177208–GU177454).
2.1. Plant materials 2.3. Data analyses
Nineteen populations of North American Zizania were collected All sequences were aligned using the program Mafft 6.7 (Katoh in the USA and Canada (Table 1). All species and varieties were rep- et al., 2005) using the L-INS-i algorithm with ‘‘maxiterate” set to resented in our sampling: six populations of Z. palustris var. palus- 1000. We combined atpB-rbcL, matK, rps16, trnL-F, trnH-psbA, and tris (PWIA, PWIB, PWIC, PWID, PVTA, and PVTB), one Z. palustris nad1 data for phylogenetic analyses, collapsing all individuals with var. interior (PMN), eight populations of Z. aquatica var. aquatica the same haplotype to a single representative. Congruence be- (AWIA, AWIB, ASC, AMD, ADL, ANJ, AVTA, and AVTB), three Z. aqu- tween the chloroplast and the mitochondrial data was examined atica var. brevis (AQCA, AQCB, and AQCC), and one Z. texana (TTX) using the incongruence length difference (ILD) test (Farris et al., from the only known locality of the taxon. Young and healthy 1994) implemented in PAUP 4.0b10 (Swofford, 2003). This test leaves were collected in the field and dried with silica gel for sub- employed 100 replicates, each with 10 random sequence additions, sequent DNA extraction. Voucher specimens from each population and the resulting P value was used to determine whether the two were deposited in the United States National Herbarium (US). Pop- datasets had significant incongruence (0.05). ulations of Z. latifolia were described in Xu et al. (2008). Four clo- Maximum parsimony (MP) searches were performed with 1000 sely related species Rhynchoryza subulata (Nees) Baill., Zizaniopsis random taxon addition replicates followed by tree bisection-recon- miliacea (Michx.) Doell & Aschers, Luziola peruviana Juss. ex J.F. nection branch swapping in PAUP 4.0b10. Gaps were treated as Gmel. and Hygroryza aristata (Retz.) Nees were included as out- missing data. Parsimony bootstrap (PB) for the clades was exam- groups based on the recent phylogenetic study of the rice tribe ined with 1000 bootstrap replicates using the same options as (Tang et al., 2010). above. Maximum likelihood (ML) analysis was implemented in GARLI 0.951 (Zwickl, 2006) starting from random trees and using 2.2. DNA extractions, amplification, and sequencing 10,000,000 generations per search. The ML bootstrap (LB) support was estimated from 100 bootstrap replicates in GARLI. For the ML Total genomic DNA was extracted from silica-dried leaves using analysis the K81uf + I substitution model was identified under the DNeasy Plant Mini Kits (Qiagen). Polymerase chain reaction Akaike information criterion (AIC) implemented in Modeltest 3.7 (PCR) amplifications were performed using 10–30 ng of genomic (Posada and Crandall, 1998). Bayesian inference (BI) was imple-
DNA, 5 pmol of each primer, 0.2 mM of each dNTP, 2 mM MgCl2, mented in MrBayes 3.1.2 (Ronquist and Huelsenbeck, 2003). A 1010 X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017
Table 1 Voucher information of 19 Zizania populations from North America. Frequencies of the nuclear Adh1a haplotypes (1–63) are also indicated. All voucher specimens are deposited at the US National Herbarium (US) of the Smithsonian Institution.
Population Species Collection voucher n Location Coordinate (N/W) Haplotype of Adh1a PWIA Zizania palustris var. palustris J. Wen 9918 6 Taylor Co., WI 45°1905100/ 44, 54(3), 55(4), 56(2), 57, 58 90°2604400 PWIB Zizania palustris var. palustris J. Wen 9936 6 Polk Co., WI 45°2304400/ 45, 49(2), 59(6), 60, 61, 62 92°1204900 PWIC Zizania palustris var. palustris J. Wen 9937 6 Polk Co., WI 45°2905000/ 59(12) 92°2902200 PWID Zizania palustris var. palustris J. Wen 9938 6 Douglas Co., WI 46°2604700/ 49(9), 59(3) 92°0905400 PVTA Zizania palustris var. palustris Xu et al. 106 6 Addison Co., VT 43°4804200/ 51(12) 73°1901000 PVTB Zizania palustris var. palustris Xu et al. 108A 6 Addison Co., VT 44°1303500/ 36(10), 52, 53 73°1603900 PMN Zizania palustris var. interior J. Wen 9963 6 Houston Co., MN 43°4901300/ 43(3), 45(3), 46, 47(2), 48, 49, 50 91°1604700 AWIA Zizania aquatica var. aquatica J. Wen 9912 6 Portage Co., WI 44°3102900/ 1(6), 7(3), 22, 41, 42 89°3505000 AWIB Zizania aquatica var. aquatica J. Wen 9914 and 9915 6 Wood Co., WI 44°2005100/89°5800500 7(12) ASC Zizania aquatica var. aquatica J. Wen 10018 6 Berkeley Co., SC 33°1104600/ 32(7), 33(2), 34(3) 79°5701200 AMD Zizania aquatica var. aquatica J. Wen 10401 6 Prince George Co., MD 38°4700500/ 1(2), 5, 6, 7(2), 8, 9(2), 10, 11, 12 76°4201200 ADL Zizania aquatica var. aquatica J. Wen 10408 6 Sussex Co., DL 38°3305000/75°4001600 1(6), 2(4), 3, 4 ANJ Zizania aquatica var. aquatica Xu et al. 101 6 Atlantic Co., NJ 39°3602700/ 1, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 74°3503000 AVTA Zizania aquatica var. aquatica Xu et al. 108B 6 Addison Co., VT 44°1303500/ 7(2), 27(2), 35, 36(2), 37, 38, 39, 40(2) 73°1603900 AVTB Zizania aquatica var. aquatica Xu et al. 109 6 Chittenden Co., VT 44°3703300/ 7(8), 40(4) 73°1402200 AQCA Zizania aquatica var. brevis J. Wen 10435 6 Montmagny Co., QC 46°5600900/70°4400300 24, 25(8), 26, 27(2) AQCB Zizania aquatica var. brevis J. Wen 10436 6 Montmagny Co., QC 47°0105200/70°2801100 25(2), 27(7), 28, 29, 30 AQCC Zizania aquatica var. brevis J. Wen 10437 6 L’Islet Co., QC 47°1004800/70°1801400 25(2), 27(7), 29(2), 31 TTX Zizania texana Texas WR 30 Hays Co., TX 63(60)
mixed model Bayesian analysis assigned model parameters for mented in NETWORK 4.5 (Bandelt et al., 1999). In this analysis, each gene partitions identified by AIC in Modeltest 3.7 (Table 2). indels were treated as single mutation events and coded as substi- Two independent runs of Metropolis-coupled Markov chain tutions (A or T). Genetic distances among species were calculated (MCMC) analysis were conducted simultaneously, with each run employing Tamura and Nei’s (1993) distance using MEGA 4 (Tam- having one cold chain and three incrementally heated chains and ura et al., 2007). all started randomly in the parameters space. Four million genera- We used the combined atpB-rbcL, matK, rps16, trnL-F, and nad1 tions were run and every 100 generations were sampled with the data to estimate the divergence time of Zizania between eastern first 25% of samples discarded as burn-in. Tracer 1.4 (Rambaut Asia and North America with 29 taxa sampled from the Oryzeae. and Drummond, 2004) was used to check whether the chains were Sequences of 21 species were obtained from GenBank. Some taxa converged. The remaining trees were sampled from the posterior were coded with missing data in the trnL-F region because only se- distribution to calculate the posterior probability (PP). quences of the trnL region were available. A likelihood ratio test For the nuclear Adh1a gene, we also performed MP, ML, and BI was carried out to determine if there was a significant difference analyses using non-redundant haplotypes only. The same options in evolutionary rates among lineages (Felsenstein, 1988). The as described above were used except that the best-fit model for hypothesis of a molecular clock was rejected (P < 0.05). Diver- Adh1a gene was TrN + I + G and 10 million generations were run gence-time estimation was performed using a Bayesian method in the BI analysis. Furthermore, a haplotype network of Adh1a se- as implemented in BEAST 1.4.7 (Drummond and Rambaut, 2007). quences was constructed using the Median-Joining model imple- We used the GTR model of nucleotide substitution with a gamma distribution with four rate categories, under an uncorrelated log- normal relaxed clock model (Drummond et al., 2006). Posterior Table 2 distributions of parameters were approximated using two inde- Characteristics of six chloroplast and mitochondrial regions and nuclear Adh1a gene pendent MCMC analyses of 20,000,000 generations with 10% (including outgroups). burn-in. Results were checked using Tracer 1.4 (Rambaut and Aligned Number of Number of Model Drummond, 2004) to ensure plots of two analyses were converging length variable sites informative selected by on the same area and then combined. Two calibration points were (bp) sites AIC used to estimate the divergence time within Zizania: 34.5 ± 6.8 atpB-rbcL 757 52 19 HKY + G million years ago (mya) for the stem node of Oryzeae, and 7 mya matK 1536 118 58 TVM + G as the minimum age for the stem node of Leersia. The first calibra- trnH-psbA 568 23 14 HKY + I tion point was obtained from Vicentini et al. (2008). They esti- rps16 820 53 15 HKY trnL-F 975 53 18 K81uf + I mated the divergence times of major grass lineages using six nad1 1510 31 15 TVM fossils as calibration points and several calibration schemes. There Combined 6166 330 139 K81uf + I are two reported macrofossil records for Oryzeae. One is the fertile organellar data lemmas and paleas (anthoecia) of Archaeoleersia nebraskensis Adh1a 1508 290 195 TrN + I + G Thomasson preserved as silicifications from the late Miocene in X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017 1011 the Ash Hollow Formation of the Ogallala Group in Nebraska. This The biogeographic range evolution of Zizania and its close rela- fossil compares most closely to the anthoecia of living Leersia ligu- tives was inferred using dispersal-vicariance analysis (DIVA; Ron- laris Trin. (Thomasson, 1980). The other fossil of the tribe was the quist, 1997) and a maximum likelihood based method, spikelets found in a Miocene excavation in Germany and was iden- LAGRANGE (Ree et al., 2005; Ree and Smith, 2008). The analyses tified as Oryza exasperate (A. Braun) Heer (Heer, 1855). Marshall were conducted on a fully resolved topology for Zizania and its et al. (1979) indicated that the top of the Ash Hollow Formation close relatives with nodal ages, obtained from Fig. 2 in Tang et al. in Nebraska is slightly younger than 7 mya. Tang et al. (2010) used (2010). Three areas of endemism were defined according to their 5 mya as minimum age constraint for the Oryza fossil. We thus distribution: (A) Asia, (B) North America, and (C) South America. constrained the split of Leersia and Oryza with the minimum age The maximum number of areas in ancestral ranges was set to be of 7 mya based on the Archaeoleersia record. two in both analyses. The A–C area was disallowed in the
Zizania palustris var. palustris_PVTB2 Z. palustris var. palustris_PWIA6; C1 Z. palustris var. palustris_PVTA1,4 Z. palustris var. palustris_PVTB1 Z. palustris var. palustris_PVTB6 Z. palustris var. palustris_PWIB1,6 64/70/0.89 Z. palustris var. palustris_PWID1,6 Z. texana_TTX2,13,23,27 Z. palustris var. palustris_PWIA1,3 Z. palustris var. interior_PMN3 Z. palustris var. interior_PMN1 93/96/1.00 Z. palustris var. palustris_PWIA2 Z. aquatica var. aquatica_AVTA1,3,5,6; B3,4 100/100/1.00 Z. aquatica var. aquatica_AVTB1 Z. aquatica var. aquatica_AWIA1,3; B1,4 & Z. aquatica var. brevis_AQCA1,2; B1,2; C2 --/58/1.00 Z. aquatica var. aquatica_AVTA2,B6 Z. aquatica var. aquatica_AVTA4 65/60/1.00 Z. aquatica var. brevis_AQCC6 61/73/1.00 Z. aquatica var. aquatica_ADL1,6; NJ5 100/100/1.00 Z. aquatica var. brevis_AQCC1 Z. aquatica var. aquatica_AMD5 81/84/1.00 Z. aquatica var. aquatica_AMD2 81/74/0.89 Z. aquatica var. aquatica_ASC3,5 Z. aquatica var. aquatica_ANJ1 Z. latifolia_YN 94/100/1.00 Z. latifolia_HLJ101,BJ10,SD210,ZJ01 Rhynchoryza subulata 100/100/1.00 Luziola peruviana Zizaniopsis miliacea Hygroryza aristata 5 changes
Fig. 1. The ML tree of Zizania inferred from combined atpB-rbcL, matK, trnL-F, trnH-psbA, rps16, and nad1 data. Numbers at nodes are bootstrap values obtained from MP and ML analysis and Bayesian posterior probabilities. The population codes are indicated in Table 1. 1012 X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017
54/54/0.95 Hap 36 Hap 52 Zizania aquatica Hap 43 Hap 56 Hap 54 Zizania palustris Hap 58 77/87/1.00 Hap 57 Zizania texana Hap 46 Hap 47 Hap 50 Hap 55 60/60/0.99 Hap 10 Hap 53 Hap 61 --/53/0.97 Hap 49 Hap 51 Hap 48 Hap 62 Hap 60 Hap 59 Hap 63 Hap 45 Hap 44 61/84/1.00 Hap 27 51/66/-- Hap 35 Hap 40 Hap 24 Hap 29 Hap 26 Hap 30 --/--/0.99 Hap 5 --/--/0.97 Hap 33 Hap 4 Hap 28 Hap 39 Hap 18 Hap 20 Hap 1 --/--/0.99 Hap 8 Hap 23 Hap 32 Hap 9 85/89/1.00 Hap 6 --/--/0.97 Hap 3 Hap 34 Hap 31 Hap 25 Hap 19 Hap 17 --/--/0.99 Hap 14 Hap 15 Hap 13 100/100/1.00 Hap 21 Hap 22 Hap 12 Hap 16 Hap 2 65/61/0.98 Hap 7 Hap 41 Hap 42 100/100/1.00 Hap 11 Hap 37 Hap 38 Hap B 100/100/1.00 Hap A Hap I 99/99/1.00 Hap D 86/92/0.99 Hap H Hap C Zizania latifolia Hap E 98/94/0.99 hap J Hap F 100/100/1.00 Hap G 100/100/1.00 Luziola peruviana Zizaniposis miliacea Rhynchoryza subulata Hygroryza aristata 5 changes
Fig. 2. The ML tree of Zizania inferred from the nuclear Adh1a data. Numbers at nodes are bootstrap values obtained from MP and ML analyses and Bayesian posterior probabilities.
LAGRANGE analysis because it required prior extinction in their phylogenetic analyses were performed on the combined organellar intervening area. data. A single most parsimonious tree was generated with a length of 365 steps, a consistency index (CI) of 0.94, a CI excluding unin- 3. Results formative characters of 0.87, and a retention index (RI) of 0.95. The trees in ML and Bayesian analyses were the same in topology as the 3.1. Phylogenetic and haplotype network analysis single MP tree. The ML tree is presented in Fig. 1 along with boot- strap support values from MP and ML analyses and posterior prob- The statistics of the chloroplast atpB-rbcL, matK, trnL-F, trnH- abilities from Bayesian analysis. The monophyly of Zizania was psbA, and rps16 and mitochondrial nad1 regions are shown in Ta- strongly supported by all analyses (PB = 100%, LB = 100%, ble 2. A partition homogeneity test suggested that the chloroplast PP = 1.00). Among the multiple populations of Zizania, two distinct and the mitochondrial data sets were congruent (P = 0.354). Thus clades were resolved with robust support, corresponding to the X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017 1013
Asian clade of Z. latifolia (PB = 94%, LB = 100%, PP = 1.00) and the Table 3 North American clade of Z. aquatica, Z. palustris, and Z. texana Pairwise comparisons of genetic distances with the Tamura–Nei model among three North American Zizania species based on the nuclear Adh1a data. (PB = 100%, LB = 100%, PP = 1.00). In the North American clade, populations of the three species grouped into two separate clades. Z. aquatica Z. palustris Z. texana One clade included only populations of Z. aquatica with high sup- Z. aquatica 0.0116 ± 0.0023a port values (PB = 81%, LB = 84%, PP = 1.00); and the other clade in- Z. palustris 0.0116 ± 0.0023 0.0038 ± 0.0011b cluded all samples of Z. palustris and Z. texana, and seven samples Z. texana 0.0127 ± 0.0028 0.0084 ± 0.0025 from two populations of Z. aquatica with low support values a Distance between the two varieties of Z. aquatica (var. aquatica and var. brevis). (PB = 64%, LB = 70%, PP = 1.00). Populations of Z. aquatica appar- b Distance between the two varieties of Z. palustris (var. palustris and var. interior). ently did not form a monophyletic group (Fig. 1). Moreover, Zizania palustris and Z. texana were supported to be very closely related (Fig. 1). The Z. texana population only differed from all populations ana and Z. palustris (0.0084 ± 0.0025), and a higher infraspecific ge- of Z. palustris by no more than 7 mutations in the 5987 bp of netic distance was present in Z. aquatica (Table 3). aligned nucleotide positions of the combined atpB-rbcL, matK, trnL-F, trnH-psbA, rps16, and nad1 data, whereas it differed from 3.2. Biogeographic analysis two populations of Z. aquatica in the same clade by a minimum of eight mutations (indels treated as single mutation events). The chronogram and results of divergence-time estimation The ML tree from the nuclear Adh1a gene is shown in Fig. 2. The based on the combined data of atpB-rbcL, matK, rps16, trnL-F, and ML tree, the MP consensus tree and the Bayesian tree only had nad1 from the Bayesian approach are shown in Fig. 4. The disjunc- minor differences, and the support values of major clades in both tion of Zizania between eastern Asia and North America was esti- ML and MP analyses were congruent with posterior probabilities mated at 3.74 mya (with a 95% highest posterior density [HPD] from Bayesian analysis. The monophyly of Zizania and the two dis- interval of 1.04–7.23 mya), which yields time estimates in the late tinct clades corresponding to eastern Asia and North America with Tertiary. The age of the crown North American Zizania was esti- strong support were recovered, as in the combined organellar data mated at 0.71 mya (95% HPD: 0.12–1.54 mya). (Figs. 1 and 2). The North American clade was divided into two Both DIVA and LAGRANGE analyses suggested that the ancestral clades with weak support: one consisting of haplotypes exclusively area of the Zizania node was Asia and North America (Fig. 5). How- from Z. aquatica, and the other including all haplotypes from Z. ever, the DIVA results conflicted with those of LAGRANGE at some palustris and Z. texana and 14 haplotypes from Z. aquatica (Fig. 2). nodes. For the ancestral range of the Zizania-Rhynchoryza node, To clarify the relationship of North American wild-rice, a haplotype North America was supported by LAGRANGE, whereas both Asia network was constructed using the Median-Joining method. The and South America were inferred by DIVA (Fig. 5). 63 haplotypes of North American wild-rice were split into three groups (A, P, and T in Fig. 3) corresponding to the three species, 4. Discussion Z. aquatica, Z. palustris, and Z. texana (Fig. 3). Group A consisted of 40 haplotypes from Z. aquatica; group T included the single hap- 4.1. Phylogenetic relationships lotype (63) from Z. texana; and group P included 22 haplotypes, of which 20 were unique to Z. palustris, one (10) was unique to Z. aqu- Within Zizania, the eastern Asian Z. latifolia is sister to the clade atica, and one haplotype (36) was shared between Z. palustris and Z. of North American species consisting of Z. aquatica, Z. palustris, and aquatica (Fig. 3). Haplotype 45 from Z. palustris was closest to the Z. texana. This relationship is strongly supported by both organellar single haplotype 63 from Z. texana. Within group A, haplotypes and nuclear data (Figs. 1 and 2). The early divergence of Z. latifolia of the two varieties of Z. aquatica (var. aquatica and var. brevis) each within the genus was supported by a suite of morphological and did not form a monophyletic group, and one haplotype (27) was cytological characters. First, the Asian species has chromosome shared between them. Within group P, haplotypes of the two vari- number n = 17, whereas all the North American Zizania taxa have eties of Z. palustris (var. palustris and var. interior) were each not n =15(Brown, 1950; Dore, 1969; Chen et al., 1990). Second, the monophyletic, and two haplotypes (45 and 49) were shared be- panicle of Z. latifolia possesses mixed branches bearing both stami- tween them (Fig. 3). The genetic distance between Z. texana and nate and pistillate spikelets, whereas the North American species Z. aquatica (0.0127 ± 0.0028) was higher than that between Z. tex- have a panicle with the lower branches bearing male spikelets
Fig. 3. The Median-Joining haplotype network of the nuclear Adh1a fragment in North American Zizania. Circles and squares represent different haplotypes (1–63) with size proportional to their relative frequency. Black dots and crossed bars represent putative haplotypes. Each line connecting haplotypes represents one mutational step. Three groups are defined corresponding to the three species with A referring to Z. aquatica, P largely to Z. palustris (with exceptions of haplotypes 10 and 36), and T to Z. texana. 1014 X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017
Oryza meridionalis Oryza sativa Oryza glaberrima Oryza punctata Oryza officinalis Oryza rhizomatis Oryza australiensis Oryza coarctata Oryza brachyantha Oryza granulata Leersia perrieri Leersia oryzoides Leersia hexandra Leersia tisserantti Luziola leiocarpa Luziola peruviana Luziola fluitans Zizaniopsis villanensis Zizaniopsis miliacea Potamophila parviflora Hygroryza aristata Zizania texana 0.71mya (95%HPD:0.12-1.54) 2 Zizania palustris 3.74mya (95%HPD:1.04-7.23) 1 Zizania aquatica Zizania latifolia Rhynchoryza subulata Chikusichloa aquatica Ehrharta erecta Phyllostachys aurea 40 35 30 25 20 15 10 5 0 mya QUATER-
PLIO- NARY EOCENE OLIGOCENE MIOCENE CENE ML EL EMLEL
Fig. 4. Chronogram of Oryzeae inferred from combined atpB-rbcL, matK, trnL-F, rps16, and nad1 data using BEAST. Clade constraints are indicated with blank asterisks. Gray boxes indicate 95% highest posterior density intervals. Node indicated by number 1 is the disjunction between eastern Asian Zizania and North American Zizania. Node indicated by number 2 is the divergence among North American Zizania.
and the upper branches bearing female spikelets (Fassett, 1924; Zizania palustris B B Terrell et al., 1997; Wu et al., 2006). Third, the pistillate lemmas of Z. latifolia have numerous hairs and siliceous papillae and pits, AB B Zizania aquatica B whereas hairs and siliceous papillae and pits are absent or nearly A/B so in the pistillate lemmas of all North American Zizania species AC Zizania latifolia A B (Terrell and Wergin, 1981). Among the three North American wild-rice species, the geo- A Rhynchoryza C B graphically isolated Z. texana is supported to be most closely related AB AC Hygroryza A to Z. palustris based on our organellar and nuclear Adh1a data (Figs. 1 B and 3). The analysis by Horne and Kahn (1997) using ITS sequences Luziola BC also suggested a close relationship between the perennial Z. texana B C and the annual Z. palustris. This relationship is consistent to the fact B Zizaniopsis BC that Z. texana can hybridize only with Z. palustris var. interior under artificial conditions (Duvall, 1987). The interspecific genetic dis- Fig. 5. Ancestral area reconstruction for Zizania and its close relatives using DIVA tance with the Tamura–Nei model between these two species is and LAGRANGE. The tree was obtained from Tang et al. (2010) with branch lengths small, ranging from 0.006 to 0.01 (Table 3). Zizania texana has been proportionate to time. Three areas of endemism were defined according to their recognized as a highly distinct species based on its habitat in deep distribution: (A) Asia, (B) North America and (C) South America. The letters above running water, perennial life history, distinct morphology with branches and below branches are referred from DIVA and LAGRANGE, respectively. The ancestral areas inferred from LAGRANGE are the ones with the highest stoniferous rhizomes and geniculate culms, and geographically likelihood scores and the highest probabilities among the alternatives. isolated distribution (Terrell et al., 1978). X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017 1015
Zizania aquatica and Z. palustris were each not monophyletic in ka and northwestern Canada area after 7 mya associated with the both the organellar and the nuclear phylogenies (Figs. 1 and 2). global climatic cooling in the late Tertiary (White et al., 1997; Gra- This pattern of non-monophyly was probably due to incomplete ham, 1999). The Bering land bridge was suitable for exchanges of lineage sorting, not introgression, based on several lines of evi- temperate deciduous plants and remained available for floristic ex- dence. First, introgression (i.e., chloroplast capture) often results changes until about 3.5 mya (Wen, 1999). Thus dispersal via the in the discordance of nuclear and chloroplast phylogenies (Riese- Bering land bridge seems to be the most plausible explanation berg and Soltis, 1991; Soltis and Kuzoff, 1995), whereas our organ- for Zizania. Similar divergence times have also been reported in ellar and nuclear trees had similar topologies (see Figs. 1 and 2). other temperate aquatic and wetland intercontinental disjunct Both the organellar and the nuclear trees had two major clades, plants with an eastern Asian–North American distribution, such with one including samples from Z. aquatica only and the other as Acorus L. (A. calamus L. and A. americanus (Raf.) Raf., Tian et al., consisting of samples from both species (Figs. 1 and 2). Second, re- in press), Lysichiton Schott (L. camtschatcensis (L.) Schott and L. cently introgressions are expected to be presented in both species americanus Hulten & H. St. John, Nie et al., 2006b), and Symplocar- more commonly in the areas near contact zones or sympatric local- pus R.A. Salisb. ex Nutt. (S. foetidus (L.) Salisb. ex W. Barton and S. ities (Barbujani et al., 1994), whereas this pattern did not appear in renifolius Schott ex Tzvelev, Nie et al., 2006b). These results suggest our organellar and nuclear data (except haplotype 36). The haplo- that dispersal via the Bering land bridge in the late Tertiary may be types of Z. aquatica grouped into the mixed clades were not closely important for the migration of wetland plants between eastern related to haplotypes of Z. palustris from the sympatric localities. Asia and North America. Therefore, we prefer incomplete lineage sorting to explain the Our DIVA and LAGRANGE results suggested Asia and North non-monophyly of the two annual wild-rice species. America as the ancestral area of Zizania. A widespread Zizania Introgression, however, may have been supported in the sym- ancestor between Asia and North America is thus preferred by patric populations of Z. aquatica and Z. palustris in northwest Ver- the analyses. Using the phylogenetic framework of the rice tribe mont, USA. It is noted that Adh1a haplotype 36 was shared by Z. by Tang et al. (2010), the ancestral area of the Zizania–Rhynchoryza aquatica and Z. palustris and nested within group P (Fig. 3). More- clade was detected to be Asia and South America in DIVA, and over, it was only present in two sympatric populations (AVTA North America in LAGRANGE (Fig. 5). This conflict may be attrib- and PVTB from northwest Vermont, Table 1). This phylogeographic uted to the parsimony criterion not considering branch lengths pattern is seemingly in favor of introgression rather than incom- or time, and only vicariance being assigned for subdivision of wide- plete lineage sorting. Furthermore, only two haplotypes (10 and spread ancestors in DIVA (Clark et al., 2008). The LAGRANGE anal- 36) of Z. aquatica were nested within group P of Z. palustris, but yses incorporate geological information and allow prior none of the haplotypes of Z. palustris was nested within group A hypotheses for range and dispersal constraints, and infer the range of Z. aquatica (Fig. 3), suggesting possible unidirectional introgres- evolution in a likelihood framework (Ree and Smith, 2008). The sion from Z. palustris to Z. aquatica. This is consistent with the fact DIVA results of the ancestor of Zizania and Rhynchoryza in both Asia that a unilateral interspecific crossability barrier exists between Z. and South America seem to be unreasonable because no fossil spe- aquatica and Z. palustris. Artificial hybridization demonstrated that cies occurred in their intervening area North America, and long- interspecific hybrids were produced only when Z. aquatica was the distance dispersal between Asian and South America seems much ovulate parent (Duvall and Biesboer, 1988b). Unidirectional intro- less likely. We prefer the hypothesis that Zizania originated from gression was also suggested by the seed protein IEF data (Duvall its ancestor in North America and then dispersed from North and Biesboer, 1989). We plan to expand our study to sample across America to Asia based on the LAGRANGE results. A recent phylog- the distribution ranges of both annual species with special empha- eographic analysis of the eastern Asian Z. latifolia showed that the sis on the zones of sympatry to test the introgression hypothesis all three clades of Adh1a haplotypes occurred in northeastern Chi- between Z. aquatica and Z. palustris. na while two clades and one clade occurred in eastern and central/ At the infraspecific level, the two varieties of Z. aquatica (var. southern China, respectively (Xu et al., 2008). The haplotype occur- aquatica and var. brevis) are geographically separated by a distance rence in Z. latifolia suggested that this species colonized in the of ca. 130 km, and are morphologically easily distinguishable by southward direction from the northeast part of eastern Asia and their plant height and the length of their lemma awns (see Darby- achieved its wide distribution across the entire eastern Asia. The shire and Aiken, 1986; Aiken et al., 1988). The two varieties of Z. phylogeographic structure of Z. latifolia thus supports the hypoth- palustris, on the other hand, overlap in their distribution and some- esis that Zizania migrated from the New World to the Old World times it is difficult to separate var. palustris from var. interior mor- via the Bering land bridge and then migrated/dispersed south- phologically (Terrell et al., 1997). The genetic distance between wardly in eastern Asia. var. aquatica and var. brevis is three times higher than that be- Bayesian dating estimated the divergence of Zizania aquatica tween var. palustris and var. interior based on the nuclear Adh1a from other North American Zizania to be 0.71 (95% HPD: 0.12– data (Table 3). Only one of 42 haplotypes in Z. aquatica was shared 1.54) mya, suggesting that the North American Zizania taxa began between var. aquatica and var. brevis, whereas two of 21 haplo- to diverge in the middle Pleistocene. The intracontinental diversi- types in Z. palustris were shared between var. palustris and var. fication in North America was thus relatively recent, consistent interior (Fig. 3). The shared haplotypes and non-monophyly of each with the electrophoretic seed protein profiles data (Duvall and variety in both species (Figs. 1 and 3) indicated recent divergence Biesboer, 1989). These workers detected no taxon-specific banding and/or recurrent gene flow at the infraspecific level in the North patterns in North American Zizania. Furthermore, in a survey for 11 American wild-rice, supporting their recognition at the variety le- enzyme systems in 33 populations of the two annual species in vel taxonomically. North America, only four enzymes are species specific (Warwick and Aiken, 1986), showing a low level of genetic divergence be- 4.2. Biogeographic diversification of Zizania tween the two species. Future phylogeographic study with a broader sampling scheme should be useful and are planned to fur- The Bayesian dating using the combined atpB-rbcL, matK, rps16, ther analyze the population history and diversification of North trnL-F, and nad1 sequence data suggests the divergence time of American Zizania. Zizania between eastern Asia and North America to be 3.74 (95% The perennial Zizania texana differs from the two annual North HPD: 1.04–7.23) mya in the late Tertiary. Palynological evidence American species in several morphological characters including its indicated that temperate taxa became important elements in Alas- prostrate, submerged habit and geniculate culms that root at the 1016 X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017 basal nodes (Duvall, 1987; Terrell et al., 1997). Yet our results sug- Clark, J.R., Ree, R.H., Alfaro, M.E., King, M.G., Wagner, W.L., Roalson, E.H., 2008. A gested that it diverged recently from the widespread North Amer- comparative study in ancestral range reconstruction methods: retracing the uncertain histories of insular lineages. Syst. Biol. 57, 693–707. ican Z. palustris (Fig. 4). The unusual morphology was noted to be Darbyshire, S.J., Aiken, S.G., 1986. Zizania aquatica var. brevis (Poaceae): a 1983 not maintained in the common garden (Duvall, 1987, p. 77), and distribution survey and a scanning electron microscope study of epidermal may be predominantly determined by the unique habitat in the features. Naturaliste Can. (Rev. Ecol. Syst.) 113, 355–360. Demesure, B., Sodzi, N., Petit, R.J., 1995. A set of universal primers for amplification San Marcos River of Texas where the only population of Z. texana of polymorphic noncoding regions of mitochondrial and chloroplast DNA in inhabits. It is also notable that no polymorphism was detected in plants. Mol. Ecol. 4, 129–131. Z. texana in the nuclear Adh1a gene (Table 1 and Fig. 3), even Dore, W.G., 1969. Wild rice. Can. Dept. Agric. Publ. no. 1393, Ottawa, Ontario, Canada. though our samples of the Adh1a haplotype network covered all Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogenetics the stands of Z. texana in the San Marcos River (from river sections and dating with confidence. PLoS Biol. 4, 699–710. A to K, see figure in Richards et al., 2007). The lack of Adh1a gene Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7, 214. diversity may be due to the small population size within a narrow Duvall, M.R., 1987. A systematic evaluation of the genus Zizania (Poaceae). Ph.D. geographic range of Z. texana. This species was presumed to be pre- dissertation, University of Minnesota, St. Paul. dominantly asexual because monitoring of the stands in the river Duvall, M.R., Biesboer, D.D., 1988a. Anatomical distinctions between the pistillate since the 1960s revealed little or no flowering (Emery, 1977; spikelets of the species of wild-rice (Zizania, Poaceae). Am. J. Bot. 75, 157–159. Duvall, M.R., Biesboer, D.D., 1988b. Nonreciprocal hybridization failure in crosses Power, 1996). However, high heterozygosity and few duplicate between annual wild-rice species (Zizania palustris Zizania aquatica: Poaceae). genotypes were detected in Z. texana using microsatellite makers, Syst. Bot. 13, 229–234. suggesting that sexual reproduction occurs more often (Richards Duvall, M.R., Biesboer, D.D., 1989. Comparisons of electrophoretic seed protein profiles among North-American populations of Zizania. Biochem. Syst. Ecol. 17, et al., 2007). The difference in the level of genetic diversity revealed 39–43. by the Adh1a gene in our study and microsatellite markers (Rich- Duvall, M.R., Peterson, P.M., Terrell, E.E., Christensen, A.H., 1993. Phylogeny of North ards et al., 2007) is most likely due to the much higher mutation American Oryzoid grasses as construed from maps of plastid DNA restriction sites. Am. J. Bot. 80, 83–88. rates in microsatellite DNA. Emery, W.H.P., 1977. Current status of Texas wild rice. Southwest. Nat. 22, 393–394. Farris, J.S., Källersjö, M., Kluge, A.G., Bult, C., 1994. Testing significance of incongruence. Cladistics 10, 315–319. 5. Conclusions Fassett, N.C., 1924. A study of the genus Zizania. Rhodora 26, 153–160. Felsenstein, J., 1988. Phylogenies from molecular sequences: inference and Our molecular data support the monophyly of the wild-rice reliability. Ann. Rev. Genet. 22, 521–565. Ge, S., Li, A., Lu, B.R., Zhang, S.Z., Hong, D.Y., 2002. A phylogeny of the rice tribe genus Zizania. Biogeographic analyses suggest that the genus most Oryzeae (Poaceae) based on matK sequence data. Am. J. Bot. 89, 1967–1972. likely originated in North America and then dispersed into eastern Ge, S., Sang, T., Lu, B.R., Hong, D.Y., 1999. Phylogeny of rice genomes with emphasis Asia via the Bering land bridge. The North American Zizania species on origins of allotetraploid species. Proc. Natl. Acad. Sci. USA 96, 14400–14405. Graham, A., 1999. Late Cretaceous and Cenozoic history of North American formed a clade, which is sister to the eastern Asian Z. latifolia. The Vegetation North of Mexico. Oxford University Press, New York, New York, USA. divergence of the intercontinental disjunction in Zizania is esti- Guo, H.B., Li, S.M., Peng, J., Ke, W.D., 2007. Zizania latifolia Turcz. cultivated in China. mated to be in the late Tertiary. The divergence of modern North Genet. Resour. Crop Evol. 54, 1211–1217. Guo, Y.L., Ge, S., 2005. Molecular phylogeny of Oryzeae (Poaceae) based on DNA American Zizania species occurred recently in the Pleistocene. sequences from chloroplast, mitochondrial, and nuclear genomes. Am. J. Bot. 92, Incomplete lineage sorting is responsible for the non-monophyly 1548–1558. of the two widespread species Z. palustris and Z. aquatica, whereas Hamilton, M.B., 1999. Four primer pairs for the amplification of chloroplast intergenic regions with intraspecific variation. Mol. Ecol. 8, 521–523. low-frequency unidirectional introgression of Z. palustris into Z. Hayes, P.M., Stucker, R.E., Wandrey, G.G., 1989. The domestication of American aquatica has occurred as well. A close relationship between the wild-rice (Zizania palustris, Poaceae). Econ. Bot. 43, 203–214. narrowly distributed endangered species Z. texana and Z. palustris Heer, D., 1855. Flora Tertiaria Helvetiae. Winterthur, Paris, France. is supported in this study. Hitchcock, A.S., Chase, A., 1951. Manual of the Grasses of the United States, second ed. U.S. Dept. Agric. Misc. Publ., Washington, DC. Horne, F., Kahn, A., 1997. Phylogeny of North American wild rice, a theory. Acknowledgments Southwest. Nat. 42, 423–434. Johnson, E., 1969. Archeological evidence for utilization of wild rice. Science 163, 276–277. This study was supported by grants from the John D. and Cath- Katoh, K., Kuma, K., Toh, H., Miyata, T., 2005. MAFFT version 5: improvement in erine T. MacArthur Foundation (to J. Wen), the Chinese Academy of accuracy of multiple sequence alignment. Nucleic Acids Res. 33, 511–518. Kusukawa, N., Uemori, T., Asada, K., Kato, I., 1990. Rapid and reliable protocol for Sciences (to J. Wen and S. Ge) and the National Natural Science direct sequencing of material amplified by the polymerase chain reaction. Foundation of China (to J. Wen and T.-S. Yi), and by the Laboratory Biotechniques 9, 66–72. of Analytical Biology of the Smithsonian Institution. We thank Pat- Manen, J.F., Natali, A., Ehrendorfer, F., 1994. Phylogeny of Rubiaceae–Rubieae inferred from the sequence of a cpDNA intergene region. Plant Syst. Evol. 190, rick Reeves, two anonymous reviewers, and Associate Editor Lena 195–211. Hileman for constructive comments, and Guy Jolicoeur, Greg Manos, P.S., Donoghue, M.J., 2001. Progress in northern hemisphere Kearns, Ken and Bess Weston, Dan Redondo, Dail Laughinghouse phytogeography: An introduction. Int. J. Plant Sci. 162, S1–S2. Marshall, L.G., Butler, R.F., Drake, R.E., Curtis, G.H., Tedford, R.H., 1979. Calibration of and Jinmei Lu for advice and field assistance. The views presented the great American interchange. Science 240, 272–279. here do not necessarily reflect those of the U.S. Fish and Wildlife Nie, Z.L., Sun, H., Beardsley, P.M., Olmstead, R.G., Wen, J., 2006a. Evolution of Service. biogeographic disjunction between eastern Asia and eastern North America in Phryma (Phrymaceae). Am. J. Bot. 93, 1343–1356. Nie, Z.L., Sun, H., Li, H., Wen, J., 2006b. Intercontinental biogeography of subfamily References Orontioideae (Symplocarpus, Lysichiton, and Orontium) of Araceae in eastern Asia and North America. Mol. Phylogenet. Evol. 40, 155–165. Nie, Z.L., Wen, J., Sun, H., Bartholomew, B., 2005. Monophyly of Kelloggia Torrey ex Aiken, S.G., Lee, P.F., Punter, D., Stewart, J.M., 1988. Wild Rice in Canada. New Benth. (Rubiaceae) and evolution of its intercontinental disjunction between Canada Press Limited, Toronto. western North America and eastern Asia. Am. J. Bot. 92, 642–652. Bandelt, H., Forster, P., Rohl, A., 1999. Median-joining networks for inferring Oelke, E.A., 1993. Wild rice: domestication of a native North American genus. In: intraspecific phylogenies. Mol. Biol. Evol. 16, 37–48. Janick, J., Simon, J.E. (Eds.), New Crops. Wiley, New York, pp. 235–243. Barbujani, G., Pilastro, A., Dedomenico, S., Renfrew, C., 1994. Genetic variation in Oxelman, B., Liden, M., Berglund, D., 1997. Chloroplast rps16 intron phylogeny of North Africa and Eurasia: Neolithic demic diffusion vs. Paleolithic colonization. the tribe Sileneae (Caryophyllaceae). Plant Syst. Evol. 206, 393–410. Am. J. Phys. Anthropol. 95, 137–154. Peng, D., Wang, X.Q., 2008. Reticulate evolution in Thuja inferred from multiple Brown, W.V., 1950. A cytological study of some Texas Gramineae. Bull. Torrey Bot. gene sequences: Implications for the study of biogeographical disjunction Club 77, 63–76. between eastern Asia and North America. Mol. Phylogenet. Evol. 47, 1190–1202. Chen, S.L., Qin, H.Z., Jin, Y.X., Shu, P., 1990. Preliminary studies on systematic Posada, D., Crandall, K.A., 1998. Modeltest, testing the model of DNA substitution. position and evolutionof Zizania L. (Gramineae). In: Proc. Intern. Symp. Bot. Bioinformatics 14, 817–818. Gard. Nanjing, pp. 593–605. X. Xu et al. / Molecular Phylogenetics and Evolution 55 (2010) 1008–1017 1017
Power, P., 1996. Reintroduction of Texas wildrice (Zizania texana) in Spring Lake: Tian, H.L., Zhou, S.L., Nie, Z.L., Wen, J., in press. Phylogeny and biogeographic some important environmental and biotic consideration. In: Proceedings of the diversity of Acorus (Acoraceae). Taxon. Second Conference on Southwestern Rare and Endangered Plants on 1995 Vicentini, A., Barber, J.C., Aliscioni, S.S., Giussani, L.M., Kellogg, E.A., 2008. The age of
September 11–14 in Flagstaff, AZ. General Technical Report RM-GTR-283. USDA the grasses and clusters of origins of C4 photosynthesis. Global Change Biol. 14, Forest Service Rocky Mountain Forest and Range Experiment Station, Fort 2963–2977. Collins, CO. Warwick, S.I., Aiken, S.G., 1986. Electrophoretic evidence for the recognition of two Rambaut, A., Drummond, A.J., 2004. Tracer v1.4. Available from: