Genetic characterization of Zostera asiatica on the Pacific Coast of North America

Genetic characterization of Zostera asiatica on the Pacific Coastof North AmericaS.L.Talbot a,*,S.Wyllie-Echeverria b,D.H.Ward a,J.R.Rearick a,G.K.Sage a,B.Chesney c,R.C.Phillips d,äa U.S.Geological Survey,Alaska Science Center,1011E.Tudor Rd.,MS701,Anchorage,AK99503,United Statesb UW Botanic Gardens,College of Forest Resources,P.O.Box354115,University of Washington,Seattle,WA98195,United Statesc National Oceanic and Atmospheric Administration,National Marine Fisheries Service,501W.Ocean Blvd.,Suite4200,Long Beach,CA90802,United Statesd Institute of Biology of the Southern Seas,2Nakhimov Ave.,Serastopol,Crimea99011,UkraineReceived10March2005;received in revised form8March2006;accepted14March2006AbstractWe gathered sequence information from the nuclear5.8S rDNA gene and associated internal transcribed spacers,ITS-1and ITS-2(5.8S rDNA/ ITS),and the chloroplast maturase K(matK)gene,from Zostera samples collected from subtidal habitats in Monterey and Santa Barbara(Isla Vista)bays,California,to test the hypothesis that these plants are conspecific with tica Miki of Asia.Sequences from approximately520 base pairs of the nuclear5.8S rDNA/ITS obtained from the subtidal Monterey and Isla Vista Zostera samples were identical to homologous sequences obtained from Z.marina collected from intertidal habitats in Japan,Alaska,Oregon and California.Similarly,sequences from the matK gene from the subtidal Zostera samples were identical to matK sequences obtained from Z.marina collected from intertidal habitats in Japan, Alaska,Oregon and California,but differed from tica sequences accessioned into GenBank.This suggests the subtidal plants are conspecific with Z.marina,not tica.However,we found that herbarium samples accessioned into the Kyoto University Herbarium,determined to be Z. asiatica,yielded5.8S rDNA/ITS sequences consistent with either Z.japonica,in two cases,or Z.marina,in one case.Similar results were observed for the chloroplast matK gene;we found haplotypes that were inconsistent with published matK sequences from tica collected from Japan. These results underscore the need for closer examination of the relationship between Z.marina along the Pacific Coast of North America,and Z. asiatica of Asia,for the retention and verification of specimens examined in scientific studies,and for assessment of the usefulness of morphological characters in the determination of taxonomic relationships within Zosteraceae.Published by Elsevier B.V.Keywords:Zostera marina;Z.japonica;tica;ITS;matK1.IntroductionDespite taxonomic controversies associated with the relationships within the seagrass family Zosteraceae(Posluszny and Tomlinson,2001;Les et al.,2002;summarized in Kuo and den Hartog,2001),all treatments consider that the genus Zostera(subgenus Zostera according to Les et al.,2002) comprises at least four species.These include tica Miki, Z.caespitosa Miki,Z.caulescens Miki and Z.marina L.,which all inhabit the North Pacific(Short et al.,2001).Z.marina occurs throughout the Northern Hemisphere,while tica, Z.caespitosa and Z.caulescens are found only in the North Pacific region(Short et al.,2001).In addition,Z.japonica Asch and Graebn.,in the genus Nanozostera(Posluszny and Tomlinson,2001)or subgenera Zosterella(Kuo and den Hartog,2001;Les et al.,2002),occurs in the North Pacific (Short et al.,2001;Aioi and Nakaoka,2003;Lee and Lee,2003; Wyllie-Echeverria and Ackerman,2003).Three of these species,tica,Z.marina and Z.japonica,inhabit the Pacific Coast of North America(Wyllie-Echeverria and Ackerman,2003).Z.marina populations are thought to have been present in North Pacific waters during the Pliocene(Domning,1976),and are now found along the coast and in bays and estuaries from Alaska to Mexico.Morphological variation and differences in/locate/aquabot *Corresponding author.Tel.:+19077863582;fax:+19077863636.E-mail address:sandy_talbot@(S.L.Talbot).äDeceased.0304-3770/$–see front matter.Published by Elsevier B.V.doi:10.1016/j.aquabot.2006.03.011biotic processes,primarily influenced by water temperature (Setchell,1927,1929;Phillips and Lewis,1983;Phillips et al., 1983)led to the identification offive varieties along the Pacific coast(Backman,1991).Varietal status is linked to variation in somatic characteristics such as leaf width,seed morphology and frequency offlowering(Setchell,1927,1929;Backman, 1991;Wyllie-Echeverria et al.,2003).Tall,wide-leaved(6–20mm)plants with heavier seeds live primarily in subtidal regions whereas short,thin-leaved(1.5–4mm)plants with lighter seeds grow primarily in intertidal areas(Backman, 1991;Wyllie-Echeverria et al.,2003).Z.japonica,probably introduced early in the20th century through oceanic trade with Japan(Harrison and Bigley,1982), is a short,thin-leaved(0.75–1.5mm)plant morphologically distinguished from intertidal Z.marina by the presence of an open leaf sheath and retinacula(Den Hartog,1970;Phillips and Menez,1988;Kuo and den Hartog,2001).This species commonly co-occurs with intertidal Z.marina in the Pacific Northwest from southern British Columbia to southern Oregon (Harrison and Bigley,1982;Wyllie-Echeverria and Ackerman, 2003).Recently the species was found in Humboldt Bay, California(Susan Schlosser,University of California Sea Grant Extension Program,Eureka,CA,USA,personal communica-tion)but no populations are reported elsewhere in California, or in Mexico.Phillips and Wyllie-Echeverria(1990)provided evidence that the third Zostera species,tica,occurs in subtidal waters along the California coast of the Pacific.Diagnostic characteristics used to describe this taxon included the plant’s restriction to deep water(À5toÀ17m mean lower low water), a wide leaf(12–18mm),the notched characteristic of the vegetative leaf tips,a smooth seed coat,and theflowering and seed phenology,all characteristics of tica(Miki,1932). Leaf tip morphology and‘‘appearance of the seed surface’’are also suggested by Kuo and den Hartog(2001)as appropriate characteristics to differentiate‘‘large Zostera species’’such as tica and subtidal Z.marina.While the Jepson Manual(Thorne,1993)accepts the occurrence of a third Zostera species in California,theflora gives precedence to the name Z.pacifica S.Watson.Never-theless,because Zostera in general,and Z.marina in particular, demonstrates wide plasticity in a number of morphological and life history characteristics(Backman,1991),it is possible this taxon represents a deep-water adapted Z.marina instead. Because genetic markers routinely used in phylogenetic reconstructions can also be used to clarify species status,we used nuclear and chloroplast DNA markers to determine the status of this wide-leaved Zostera found in California waters.The nuclear ribosomal DNA-ITS markers[i.e.,internal transcribed spacers1and2(ITS-1and ITS-2)and associated nuclear DNA(nrDNA)sequences;Jansen et al.,1998],have been nominated as a target for DNA barcoding for plants (Stoeckle,2003;Kress et al.,2005).Although nrDNA-ITS genes are repeated throughout the plant nuclear genome,rapid concerted evolution often promotes intra-genomic uniformity of repeat units(Baldwin et al.,1995;Dubouzet and Shinoda, 1999;but see Gaut et al.,2000;Hughes et al.,2002)and as a result of relaxed constraint on mutation,the ITS regions are characterized by high levels of polymorphisms at lower taxonomic levels(Baldwin et al.,1995).Furthermore,ITS regions are small(<700base pairs(bp)in most angiosperms), facilitating their use as taxon-specific markers particularly when working with substandard DNA sources,such as herbarium specimens.DNA from such sources is typically degraded,precluding polymerase chain reaction(PCR) amplification of larger fragments of less polymorphic loci.Here we use the nrDNA ITS-1and ITS-2regions’potential as a molecular diagnostic marker to determine the specific status of an unknown marine angiosperm.The present study, involving ITS sequencing,and independent corroborative evidence from the chloroplast DNA(cpDNA)maturase K (matK)gene,was undertaken to test the hypothesis that wide-bladed,subtidal Zostera occurring off the coast of Monterey Bay and the Santa Barbara Channel,California,is tica.2.Materials and methods2.1.Voucher specimens examined and certification of speciesAll sequences were compared against sequence information accessioned in GenBank(/)by Les et al.(2002)and Olsen et al.(2004)for the nrDNA5.8S rDNA/ITS-1and ITS-2gene region(5.8S rDNA/ITS)of Z. marina and Z.japonica.Similarly,cpDNA matK sequences were compared with those accessioned in GenBank by Tanaka et al.(2003)and Kato et al.(2003)for Z.marina,Z.japonica, and tica.No data are available in public databases for the 5.8S rDNA/ITS family for tica.We therefore include in our analyses homologous sequence information for both genes from herbarium specimens(Kyoto University Museum,KYO) determined to be tica(KYO547-2,547-3and547-4).An accessioned specimen representing North Pacific Z.marina was used to obtain homologous information from both loci(UAM Herb43460,collected from Wide Bay,Alaska Peninsula,AK, USA).Similarly,a specimen representing the individuals used by Phillips and Wyllie-Echeverria(1990)to describe tica from Tomales Bay,California,was used to represent the original described specimen(WEH94937-1).2.2.Plant materialTwelve samples from extant Z.marina populations were obtained from the intertidal regions of Alaska[Izembek(n=2) and Kinzarof(n=2)lagoons on the Alaska Peninsula;Crab Bay(n=2)in the Alexander Archipelago in southeastern Alaska],Yaquina Bay,Oregon(n=2)and Keil Cove(n=4), San Francisco Bay,California.Four samples were obtained from Hokkaido,Japan(Notsuke Bay,n=2;Lake Furen,n=2). Samples of Z.japonica were obtained from Boundary Bay, British Columbia(n=2),Puget Sound,Washington(n=1), and Humboldt Bay,California(n=2).Specimens representing the North American Pacific coast tica were collected along the California coast,off the mouth of Corral Creek nearS.L.Talbot et al./Aquatic Botany85(2006)169–176 170Isla Vista on16October2001,and near Wharf2of Del Monte State Park in Monterey Bay on9December2003.These collections were made at water depth of5–10m,and very close to the same locations where Phillips and Wyllie-Echeverria (1990)collected specimens in southeastern Monterey Bay and Santa Barbara Channel,California.Samples from the Isla Vista site are vouchered(S.S.Talbot,U.S.Fish and Wildlife Service, Anchorage,AK,USA).boratory analysesGenomic DNA was extracted from approximately20–40mg dry weight of leaf tissue using protocols outlined in Mun˜iz-Salazar et al.(2005).Subsequent to extraction,the DNA was suspended in60m L of TE,quantified usingfluorometry,diluted to50ng m LÀ1using nanopure water,and stored atÀ208C until further analysis.To check for errors due to contamination during the extraction process,a negative control was included in all extraction procedures.Since contamination is a possible confounding factor in analyzing herbarium specimens,all herbarium samples were extracted independently of other samples.In addition,DNA was extracted twice from herbarium samples,each in separate extraction sessions,under sterile conditions including ultraviolet sterilization andflame-ster-ilization,where appropriate,of all extraction materials and solutions;the second extraction was subjected to subsequent independent laboratory analyses(quantification,PCR ampli-fication,sequencing;see below).Sequences from the 5.8S rDNA/ITS-1and ITS-2gene complex were obtained from product generated via PCR amplification,using the primers ZomaITSF(50-AAATTC-CACCCAYCATGC)and ZomaITSR(50-CRATGCCTTTGGT-TAGACCA).Sequences from the chloroplast matK gene were obtained using the primers matK851F(50-CAAGGAAAATC-CATTATGGC)and matK1205R(50-TCTACATATACGAC-CAAACC).Primers were synthesized with added universal sequences(M13F:50-CACGACGTTGTAAAACGAC or SP6: 50-GATTTAGGTGACACTATAG;M13R:50-GGATAA-CAATTTCACACAGG)on the50end of the light and heavy strand primers,respectively.PCR products were subsequently purified using Quantum Prep1PCR Kleen Spin Columns(BIO-RAD,Hercules,CA,USA)and subjected to tailed primer (Oetting et al.,1995)bi-directional cycle sequencing reactions (LI-COR,1999)using SequiTherm EXCEL II TM DNA Sequencing Kits-LC(Epicentre,Madison,WI,USA),under conditions recommended by the manufacturer and employing IRD700and800labeled M13F(À29)or SP6,and M13R primers,respectively.Sequenced products were electrophoresed on a3.7or5.5%polyacrylamide(KB-Plus TM,LI-COR,Lincoln, Nebraska)66or41cm gel,using a LI-COR LongReadIR TM 4200two-dye automated sequencing system.Sequences were analyzed using LI-COR eSeq TM imaging software and aligned using AlignIR2.0TM(LI-COR,Lincoln,Nebraska).3.ResultsLocation,source,and archival data associated with all samples are listed in Table1.We obtained data from approximately520bp of the 5.8S rDNA/ITS-1and ITS-2 (5.8S rDNA/ITS)and348bp of the cpDNA matK genes.Two haplotypes,differing at98nucleotide sites,were observed at the5.8SrDNA/ITS locus among all samples sequenced(Fig.1). Differences included47insertions/deletions(indels),23transi-tions and28transversions.Four haplotypes(haplotypes matK2-4)were observed at the matK locus among all samples sequenced(Figs.2and3).Data from tica(haplotypeS.L.Talbot et al./Aquatic Botany85(2006)169–176171 Table1Location of samples analyzed(including latitude/longitude),sample source,and archival informationSpecies Geographic location N Source Archive ID atica Corral Creek near Isla Vista,CA,USA;34.408N119.98W3Field collection USFWS-SSTtica Del Monte State Park,Monterey Bay,CA,USA;36.608N121.898W3Field collection ASC MONT1-3tica Shikotan Island,Kurile Islands,Japan,(latitude/longitude not listed on specimen)1Kyoto University Museum,J.Ohwi718?8August1931KYO547-2tica Yezo,Kuril Islands,Japan,(latitude/longitude notlisted on specimen)1Kyoto University Museum,U.Faurie,2938KYO547-3tica Kunashiri Island,Kuril Islands,Japan,(latitude/longitude not listed on specimen)1Kyoto University Museum,Anon.KYO547-4tica Hog Island,Tomales Bay,CA,USA;38.128N122.558W1Wyllie-Echeverria WEH94937-1Z.marina Wide Bay,Alaska Peninsula,AK,USA;57.488N159.098W1University of Alaska Museum UAM Herb43460 Z.marina Izembek Lagoon,AK,USA;55.468N163.208W2USGS-Alaska Science Center ASC IZE03,04Z.marina Kinzarof Lagoon,AK,USA;55.298N162.678W2USGS-Alaska Science Center ASC KIL02-02,21 Z.marina Crab Bay,Alexander Archipelago,AK,USA;57.828N135.398W2USGS-Alaska Science Center ASC CRAB01,06 Z.marina Yaquina Bay,OR,USA;44.628N124.038W2USGS-Alaska Science Center ASC YAB01,02Z.marina Keil Cove,San Francisco Bay,CA,USA;37.528N122.268W4USGS-Alaska Science Center ASC KCCE01,02 Z.marina Notuke Bay,Hokkaido,Japan;43.348N145.218E2USGS-Alaska Science Center ASC NB03,04Z.marina Lake Furen,Hokkaido,Japan;43.168N145.278E2USGS-Alaska Science Center ASC LF02,03Z.japonica Boundary Bay,BC,Canada;49.078N123.018W2USGS-Alaska Science Center ASC BOUN02-25,28 Z.japonica Puget Sound,WA,USA;48.338N122.568W1USGS-Alaska Science Center ASC W1-3Z.japonica Humboldt Bay,CA,USA;46.778N124.238W2USGS-Alaska Science Center ASC HUM02-17,18 Archive ID indicates herbarium or DNA voucher identification numbers.a USFWS-SST:Stephen S.Talbot,U.S.Fish and Wildlife Service,Region7collection;KYO:Kyoto University Museum Herbarium;WEH:S.Wyllie-Echeverria, personal collection;UAM Herb.:University of Alaska Museum Herbarium;ASC:U.S.Geological Survey,Alaska Science Center genomic DNA archive.matK1),obtained from GenBank (Tanaka et al.,2003:Genbank Accession no.AB096161,vouchered at TNS:Tanaka 00212)are included in Figs.2and 3for comparison.Twenty-eight sites (19transitions,8transversions and 1multistate site)differed among all five matK haplotypes shown in Fig.2.3.1.Voucher specimens examined and certification of speciesNinety-nine percent sequence homology was observed at the 5.8S rDNA/ITS locus between accessioned voucher specimen UAM Herb 43460,representing Z.marina from the Pacific coast of North America,and sequences from Z.marina reported by Les et al.(2002:AY077986,vouchered at CONN:Yarish s.n.).A single site substitution (G $A transition)was plete sequence homology was observed at the matK locus between voucher specimen Z.marina UAM V0135799,and sequences from Z.marina reported by Tanaka et al.(2003:AB096164;accessioned at TNS:Tanaka 99190).We thus use UAM Herb 43460to represent North Pacific Z.marina for subsequent comparisons.Complete sequence homology was observed at the 5.8S rDNA/ITS locus between the five samples of Z.japonica collected from Humboldt Bay,Boundary Bay,and PugetS.L.Talbot et al./Aquatic Botany 85(2006)169–176172Fig.1.Variable nucleotide sites in sequences from the 18S rDNA (positions 14–72),the internal transcribed spacer 1(ITS-1,positions 86–224),5.8S ribosomal RNA gene (5.8S rDNA,positions 236–390)and the internal transcribed spacer 2(ITS-2,positions 396–502)of Zostera japonica (Les et al.,2002:AY077991),Z.marina (UAM Herb 43460),samples collected from Isla Vista and Monterey Bay and herbarium specimens,all ascribed to tica Miki (Phillips and Wyllie-Echeverria,1990;KYO 547-2,-3and -4,respectively).Nucleotide position (read vertically)refers to the location of each variable site in the sequences.Dots (Á)indicate similarity with the first haplotype (Z.japonica );dashes (-)indicate deletions;question marks (?)signify missingdata.Fig.2.Nucleotide sequence data from the chloroplast matK gene,describing four haplotypes observed among Zostera sampled from the Pacific coast of North America,and Japan.Nucleotide position (read vertically)refers to the location of each variable site in the sequences.Dots (Á)indicate similarity with the first haplotype (tica ).Sound,with sequences reported for Z.japonica by Les et al. (2002:AY077991;accessioned at CONN:Iida s.n.).Similarly, complete sequence homology was observed at the matK locus among the same5samples and sequences reported for Z. japonica by Tanaka et al.(2003:AB096166;accessioned at TNS:Tanaka00111)and by Kato et al.(2003:AB125361, apparently not vouchered).Sequence homology at the 5.8S rDNA/ITS locus was observed for the70bp segment obtained for herbarium sample KYO547-4,determined to be tica from Japan(Table1), and the same segment from Z.marina voucher UAM Herb 43460(see above).Complete sequence homology was also observed at the5.8S rDNA/ITS locus between both herbarium samples KYO547-2and KYO547-3,determined to be Z. asiatica from Japan(Table1),and samples of Z.japonica from Puget Sound,Humboldt and Boundary Bay.Similarly,we observed complete sequence homology between sequences obtained at the matK gene from herbarium samples KYO547-2 and KYO547-3,with sequences obtained from Z.japonica collected from Puget Sound and Humboldt and Boundary bays (Figs.2and3).Despite multiple attempts,we were unable to obtain sequence information from herbarium sample KYO547-4for the matK gene.Also despite multiple attempts,we were unable to obtain sequence information from the Tomales Bay herbarium specimen(WEH94937-1,Table1),which we used to represent the originally-described Californian tica population,for either locus.3.2.Field samples examinedAllfivefield samples collected to represent Z.japonica were characterized by the same haplotype(Fig.1)at the5.8S rDNA/ ITS locus.All16field samples collected to represent Z.marina possessed another haplotype,shared with Z.marina voucher UAM Herb43460(Fig.1).One haplotype(matK3)was observed at the matK locus among thefivefield samples representing Z.japonica(Figs.2 and3).Two haplotypes(matK2and matK4)were observed among the16field samples representing Z.marina(Figs.2and 3).Samples possessing haplotype matK2were from Japan, Alaska and Oregon;this haplotype was also observed in Z.marina voucher UAM Herb43460(see above).Samples possessing haplotype matK4were sampled from Keil Cove, San Francisco Bay,or Hokkaido,Japan(Fig.2).Comparison of these haplotypes with sequence data from tica accessioned into GenBank(Tanaka et al.,2003:AB096161; Kato et al.,2003:AB125360)demonstrates that Z.marina sampled from the Pacific coast of North America differs from Z. asiatica at three to four sites at this portion of the matK gene (Figs.2and3).3.3.Analysis of subtidal samples from Monterey and Isla VistaThe single haplotype of the5.8S rDNA/ITS locus observed among all plants occurring in the subtidal in Monterey Bay and Isla Vista was completely homologous to sequence from all samples of Z.marina from Japan and the Pacific coast of North America and with the Z.marina voucher UAM Herb43460 (Fig.1).Five of the six subtidal California samples shared complete sequence homology at385bp of the cpDNA matK locus with Z.marina samples collected from Keil Cove,San Francisco Bay(haplotype matK4)and two specimens of Z. marina collected in Lake Furen,Hokkaido,Japan(Figs.2and 3).One of the subtidal Zostera samples from Monterey possessed a single site substitution at position22(haplotype matK5)relative to haplotype matK4(Figs.2and3).Basic local alignment search tool analysis(BLAST;Altschul et al.,1997) of sequences archived into GenBank verified that haplotype matK5differed at a single site from Z.marina in Japan reported by Kato et al.(2003:AB125357),and at two sites from Z. marina reported by Tanaka et al.(2003:AB096164).However, haplotype matK5differed at four sites from sequences reported from tica by Kato et al.(2003:AB125360)and Tanaka et al.(2003:AB096161).Complete sequence data from novel haplotype matK5are accessioned in GenBank(DQ485983).4.DiscussionSequence data from the nuclear5.8S rDNA/ITS locus from the wide-bladed plants collected in subtidal habitats in Monterey Bay and Isla Vista,California,hypothesized to be conspecific with tica(Phillips and Wyllie-Echeverria,S.L.Talbot et al./Aquatic Botany85(2006)169–176173Fig.3.Number and location of samples containing matK haplotypes described in Fig.2.JP:Japan(Hokkaido);AK:Alaska;OR:Yaquina Bay;CA:San Francisco Bay;BC:Boundary Bay,British Columbia;WA:Puget Sound,Washington;CA:Humboldt Bay,California.H:successful amplification of product in herbarium specimens;x:no amplification.Location information is given in Table1.1990),are identical to sequences collected from specimens of Z.marina from elsewhere along the Pacific Coast of North America,and Japan.Within-species variation at this locus is low(Olsen et al.,2004;Talbot et al.,unpublished data.),but between-species variation is high.BLAST comparison of our data with homologous data collected by other researchers(Les et al.,2002;Olsen et al.,2004;Ficca et al.,2000)demonstrates 97to100%homology within Z.marina at this locus(data not shown),whereas homology among Zostera species ranges from 81%(Z.marina and Z.japonica,this study)to89%(Z.marina and Z.novazelanidica,Les et al.,2002).Unfortunately,no sequence data generated by another laboratory are publicly available for this gene locus for tica,and sequence data we obtained from herbarium specimens determined to be Z. asiatica were,surprisingly,consistent with either Z.marina,or Z.japonica(see Section3).Nevertheless,if the subtidal plants collected from Monterey and Isla Vista are conspecific with a species other than Z.marina,we would not expect to see complete sequence homology with Z.marina at this locus.Sequence data from the subtidal plants at the cpDNA matK gene are identical to sequence from Z.marina in San Francisco Bay and elsewhere in the species’worldwide range,and differ from sequence tica collected in Japan(Tanaka et al.,2003). The novel haplotype(matK5)observed in one of the subtidal Zostera samples collected from Monterey shows closer affinity to Z.marina than tica.Collectively,these data suggest that the Zostera occupying subtidal habitats in Monterey and Santa Barbara Bay(Isla Vista)are conspecific with Z.marina.Field-based identification of Z.marina depends upon the assessment of morphological and ecological(habitat)compar-isons.Diagnostic characteristics of Z.marina include leaves 1.5–20mm wide,generally obtuse tips,seed surfaces with16–25distinct longitudinal ridges,with a shallow water habit(0–10m below mean low tide:Den Hartog,1970;Phillips and Menez,1988;Backman,1991;Kuo and den Hartog,2001). However,meristic characters and morphology are all subject to phenotypic variation within Z.marina.Although this pheno-typic variability is well-documented throughout the native range of Z.marina(Phillips et al.,1983;Backman,1991),few studies have compared meristic and other morphological characteristics of Z.marina from North America with Z. asiatica Miki from Asian populations.Phenotypic variation and ambiguity of qualitative characters,and the potential difficulty in applying these measurements to the highly plastic Z.marina on the west coast of North America,underscores the value of using genetic data to corroborate species determinations.Nevertheless,the wide-bladed plants originally described by Phillips and Wyllie-Echeverria(1990)from Santa Barbara were certainly outside the range of normal phenotypic variation observed within Z.marina,and within the range of diagnostic characters for tica Miki.The researchers describing the species have significantfield experience with Z.marina on the west coast of North America,and their determination of this plant as conspecific with tica Miki is not unreasonable. Although Hayes(2003)describes only two species of Zostera in North America,other researchers have also accepted the occurrence of three species of Zostera in California:Z.japonica Aschers and Graeben,Z.marina L.,and Z.pacifica S.Watson. The latter nomenclature is ascribed to subtidal Zostera in Monterey and Santa Barbara bays,presumably the same taxon described as tica Miki by Phillips and Wyllie-Echeverria (1990).The surprising result of sequence homology of the KYO herbarium specimens of tica from Asia with either Z. japonica or Z.marina may be due to cross-sample contamina-tion of the herbarium extracts in the laboratory.We maintained strict sterile conditions in the laboratory,however,and identical results from isolated extraction and amplification experiments for two genes suggest contamination in the laboratory is not a likely explanation for these results.Nevertheless,we cannot rule out contamination resulting from exposure to samples of congeners in the herbarium.Tanaka et al.(2003)have investigated the relationship between tica and Z.marina, using matK sequence information;the two species differ at,on average,four sites for the span of nucleotide sequence data compared at that locus.If the affinities of the herbarium specimens we observed in this study reflect the true genotypes of these specimens,it is apparent that the morphological, phenological and ecological characters currently used to differentiate these Zostera species are ambiguous.Even at higher phylogenetic levels,results of ongoing taxonomic investigations involving Zosteraceae are not con-cordant.Posluszny and Tomlinson(2001)divide the family into four genera(Heterozostera,Nanozostera,Phyllospadix and Zostera)based on morphological and developmental character-istics,whereas studies combining molecular and morphological analyses(Les et al.,2002)collapse the number of genera in this family to two(Phyllospadix and Zostera).Both classification schemes accept the genus Zostera within the family,but disagree as to the number of subgenera and species within this genus.Les et al.(2002)split the genus into three subgenera and nine species, while Posluszny and Tomlinson(2001)recognize four species and no subgenera.These newfindings are not concordant with earlier divisions,based primarily on vegetative characteristics, which placed three genera under the Zosteraceae and two subgenera and eleven species in the genus Zostera(summarized in Kuo and den Hartog,2001).The discordance within Zosteraceae between morphologically-and molecular-based taxonomies may in part fall into the category of‘‘imperfect taxonomy’’(Funk and Omland,2003),in which species are oversplit when intraspecific variation is misidentified as species-level variation.Because traditional morphology-based taxonomic systems are limited significantly by(1)difficulty in detecting cryptic species,(2)phenotypic plasticity and genetic variability and(3) dichotomous keys that depend upon characters available during a single stage of the life history of an organism,and that require high levels of expertise,the appeal of‘‘molecular taxonomy’’is great(Hebert et al.,2003).However,like taxonomies that rely on morphological characters,molecular taxonomies require a thorough investigation of the geographical pattern of variation within and between putative species(Funk and Omland,2003). Preferably,neutral genetic markers should be used concurrently with phenotypic characters to delineate Zostera species,S.L.Talbot et al./Aquatic Botany85(2006)169–176 174。