Solomon Benor

Department of Biotechnology,College of Biological and Chemical Engineering,Addis Ababa Science and Technology University,Addis Ababa,Ethiopia

Keywords:Corchorus Jute Malvaceae s.l.Phylogeny Pseudocorchorus

A B S T R A C T A molecular phylogenetic analysis of the genus Corchorus(Grewioideae,Malvaceae s.l.)is presented,based on sequences of the nuclear ribosomal DNA internal transcribed spacer(ITS)region for 144 accessions representing 47 species.Several other genera from the subfamily Grewioideae,namely Pseudocorchorus,Triumfetta,Sparrmannia,Entelea,and Grewia,were included as outgroups.The monophyly of the genus was well supported by all phylogenetic analyses(maximum likelihood,Bayesian approaches,and parsimony),and Corchorus was divided into four major clades.The majority of African species formed a statistically highly supported and distinct clade separated from the other pantropically distributed species.Several endemic species from Australia,New Caledonia,and tropical America were nested within this distinct clade,indicating dispersal from Africa to the rest of the pantropics.Based on the taxa included in this study,the two cultivated species(C.olitorius and C.capsularis)shared a common ancestry with wild species of C.africanus,C.brevicornatus,C.pseudocapsularis,C.pseudo-olitorius,C.urticifolius,C.pilosus,C.orinocensis,and C.cunninghamii.Pseudocorchorus,previously considered an accepted genus,was nested within the genus Corchorus and shared a common ancestry especially with C.depressus and C.siliquosus.Based on morphological and anatomical similarity as well as the results of the present molecular findings,inclusion of the six Pseudocorchorus species into Corchorus is proposed,with Pseudocorchorus as a synonym of Corchorus.Of the included outgroup taxa,Triumfetta is the closest sister to Corchorus,while the common ancestor of Corchorus/Pseudocorchorus,Triumfetta,Sparrmannia,and Entelea is Grewia.A further phylogenetic study with more taxa mainly from Australia,together with additional molecular markers and morphological investigation,would help to test the hypothesis on the biogeography and growth form evolution of the genus Corchorus.

1.Introduction

The genus Corchorus consists of some 50–60 shrub,subshrub,or annual and perennial herb species distributed in the tropics, subtropics, and warm temperate regions of the world[1,2].The majority of species are found in Africa and Australia[2–7]with minor numbers in tropical America and southern Asia [8]. Only two species, C. olitorius and C.capsularis,are cultivated and young leaves are consumed as leafy vegetables,while the stem bark is used as a cheap source of natural fiber,jute[2–5].According to Benor et al.[6],the two cultivated species originated in Africa and later dispersed to Asia. The genus Corchorus was previously included in Tiliaceae. However, as Tiliaceae were found to be paraphyletic,several of the taxa including Corchorus were transferred to the subfamily Grewioideae,forming a monophyletic group within the family Malvaceae s.l.[9,10].

Most of the Corchorus species are morphologically similar and identification in the vegetative phase is extremely difficult.Stipules and setae are shared by several species,and their presence or absence and color are among morphological characters whose careful examination may help in distinguishing species in the vegetative stage.Identification of Corchorus species is possible mainly during the reproductive stage,especially using capsule and seed morphologies.The first and preliminary subgeneric classification of the genus Corchorus was made by Candolle[11].According to Candolle[11], five sections comprising 19 species were described.However,several other species were left undescribed and C.depressus was treated in a former genus Antichorus as A.depressus L.Oceanopapaver neocaledonicus Guillaumin was also previously considered a single species in the former genus Oceanopapaver[12].However,based on morphological as well as molecular findings, it is now placed under the genus Corchorus as C.neocaledonicus[13,14].The genus Corchorus is morphologically and anatomically similar to Pseudocorchorus[1,15],the latter described under the tribe Pseudocorchoreae[16].It consists of six species:P.alatus,P.cornutus,P.rostratus,P.greveanus,P.mamillatus,and P.pusillus,all of which are endemic to Madagascar.Although morphological and anatomical similarity is obvious, phylogenetic relationships between these two genera have not been conducted at the molecular level prior to this study.

Several modern classification problems in plant taxonomy are resolved through molecular phylogenetic approaches.One among several of the molecular markers used in plant systematics is the internal transcribed spacer (ITS) gene marker of ribosomal DNA. It is employed in fungus and plant taxonomy for its high copy number,aiding in amplification,and also for its high degree of variation between species[17].Literature on the molecular phylogeny of the genus Corchorus is,however,scarce.Despite the agricultural and industrial importance of Corchorus, the evolutionary relationships between wild and cultivated species are unknown.Thus,only scarce information on the phylogeny of Corchorus exists and its monophyly has not previously been tested. Moreover, little is known about the biogeographic hypothesis of the genus.Accordingly,the objectives of this study were 1)to infer the phylogeny and test the monophyly of the genus Corchorus, 2) to examine the evolutionary relationship between wild and cultivated Corchorus species,3)to assess the phylogenetic relationship between Corchorus and Pseudocorchorus,and 4)to develop a first hypothesis on the biogeography and evolution of growth forms in the genus Corchorus.

2.Materials and methods

2.1.Taxon sampling

Corchorus accessions included in this study represent species from all life cycles and growth forms.Forty-seven Corchorus species,accounting for ＞80%of the accepted taxa of the genus,were included in this study(Table 1).The species included in this study represent the entire pantropical distribution of the genus.Most of the species from Ethiopia were collected under field conditions[4],while species from other geographical origins were represented by herbarium voucher specimens and seeds that were obtained from the Royal Botanic Gardens,Kew;Munich Botanical Garden;the National Herbarium of Addis Ababa University;the National Museum of Natural History in Paris;the Herbarium of the Botanical Garden and Botanical Museum in Berlin; the National Herbarium of the Netherlands; the International Jute Study Group Bangladesh;the Botanical Museum of the University of Vienna;the World Vegetable Center(Taiwan,China;Tanzania);the U.S.Department of Agriculture Agricultural Research Service, the Jardin Botanique Exotique de Menton, the Innsbruck University Botanical Garden, the Missouri Botanical Garden,and the National Plant Genetic Resource Center,Namibia.Seven Corchorus species,namely C.africanus,C.gillettii,C.junodii,C.longipedunculatus,C.parvifolius,C.pinnatiparititus,and C.pseudo-olitorius,were represented by DNA aliquots obtained from the Jodrell Laboratory of the Royal Botanic Gardens,Kew.Additional ITS sequences were downloaded from the GenBank nucleotide database.Each ingroup taxon was represented by at least three accessions.The outgroup taxa included in the study were from the following genera:Pseudocorchorus(five species,all endemic to Madagascar),Triumfetta(five species from different regions),Sparrmannia (two species, both restricted to Africa), the monotypic genus Entelea,endemic to New Zealand,and one species from the genus Grewia.Selection of outgroup taxa was based on results of previous findings on relationships in Malvaceae s.l.[9,14,18,19].A list of the outgroup taxa appears in Table 1. Herbarium vouchers were deposited at the National Herbarium,Addis Ababa University,Ethiopia;and Leibniz Institute of Plant Genetics and Crop Research,Germany.

2.2.Molecular methods:DNA extraction,PCR amplification,purification,and sequencing of studied taxa

Fresh and silica-dried leaves,herbarium voucher specimens,and seed materials were used for genomic DNA extraction.Starting material for DNA extraction varied from 10 to 50 mg in fresh/silica-dried leaves and herbarium voucher specimensto 30–80 mg in seeds.The genus Corchorus is known to have high amounts of mucilaginous substances[5,20,21],which interfere with genomic DNA and make extraction processes difficult.Thus,standardization of extraction and subsequent amplification procedures were performed using several methods and protocols including modified CTAB[22]and SDS-KAc protocols, as well as Qiagen DNeasy (Hilden,Germany)and Invisorb(Invitek,Berlin,Germany)commercial kits.

Table 1–List of the genera Corchorus and Pseudocorchorus and outgroup taxa used in the phylogenetic analyses.

Table 1(continued)

Table 1(continued)

PCR amplification of the nuclear ribosomal DNA internal transcribed spacer (ITS) region was performed using the universal primers ITS-A(5′-GGAAGGAGAAGTCGTAACAAGG-3′) and ITS-B (5′-CTTTTCCTCCGCT TATTGATATG-3′) [13].With them the entire ITS region(ITS1,5.8S rDNA,and ITS2)could be amplified in case of DNA from fresh and silica-dried leaves as well as from seeds.However,most of the Corchorus species investigated in this study were present as herbarium specimens.Thus,the ITS region was amplified in two parts,using primer ITS-A together with internal primer ITS-C(5′-GCAATTCACACCAAGTATCGC-3′) and ITS-B together with ITS-D (5′-CTCTCGGCAACGGATATCTCG-3′) according to Blattner [23]. This amplification procedure results in sequences with overlapping regions in the 5.8S rDNA gene,allowing combining the partial sequences into contigs of the entire ITS region.Two to three microliters of template DNA was used for a final PCR master mix reaction of 30 μL consisting of 3.0 μL 10×PCR-buffer, 1.5 μL of 25 mmol L?1MgCl2,1.2 μL of 2.5 mmol L?1dNTPs,50 pmol of each primer,1.5 U Taq DNA polymerase(Qiagen)and 6.0 μL Q-solution(Qiagen).The volume of PCR master mix reaction for each of the universal and internal primer combinations (primer combination A/C, or B/D) in old herbarium samples was increased to 50 μL.PCR thermal cycler steps were as follows:initial melting of the reaction mixtures at 95°C for 3 min,followed by 38 cycles at 95°C for 30 s denaturation,45 s at 53°C for primer annealing, 1 min at 68°C for primer extension,and a final elongation step of 8 min at 70°C.Amplified PCR products were purified using a Qiagen PCR purification kit and sequenced with the ABI dye-terminator technology on an ABI 3730xl automatic DNA sequencer(Applied Biosystems).As sequencing primers the PCR primers were used,or in some cases the nested primers ITS-SF and ITS-SR instead of ITS-A and ITS-B were also used[24].

2.3.Phylogenetic analyses

Forward and reverse sequences were examined,where necessary manually edited with Chromas 1.45[25],and combined into single consensus sequences of the ITS region for each individual using Bioedit 7.0.9.0[26]and Geneious 5.3.6[27].A final multiple alignment of the sequences was constructed using CLUSTAL X Windows interface software[28].

Before phylogenetic analyses,the model of evolution for DNA sequences was inferred using Modeltest 3.7[29]based on the Akaike information criterion(AIC).The GTR+I+G model was found to be the most appropriate model. Maximum likelihood and Bayesian inferences were employed as follows.Bayesian inference was performed with MrBayes 3.1.2[30]with two runs of four chains for 10,000,000 generations and sampling a tree every 1000 generations.Convergence of the analyses was checked afterwards and the first 25%of trees discarded as burn-in. Maximum likelihood analysis was carried out using RAxML[31]with GTRGAMMA model.A maximum parsimony(MP)analysis was performed in PAUP*[32]with a heuristic search,saving all shortest trees with TBR branch swapping. Bayesian posterior probabilities [33] in Bayesian inference and 1000 replicates of bootstrap analysis[34] for maximum likelihood were used to assess the statistical strength of clades.The outputs of each phylogenetic analysis were displayed using FigTree 1.3.1[35]and the trees were annotated in Adobe Illustrator CS5 software(Adobe Systems Inc.,San Jose,CA).

3.Results

Successful extraction of genomic DNA from fresh greenhouse leaves and silica-dried materials was easily achieved using all extraction methods(Qiagen,Invisorb,modified CTAB and SDS-KAc).With the Qiagen method,interference by mucilaginous substances was substantially reduced by increasing the volume of buffer AP1 from 400 to 500–600 μL,depending on the weight of starting material. However, for nearly all historical herbarium specimens, best DNA quality was achieved by use of the modified CTAB protocol[22]and the Invisorb commercial kit.In some cases,in order to improve the final yield,the extracted DNA was further precipitated with NH4Ac or purified using QIAquick PCR purification kit(Qiagen).Extraction and subsequent PCR amplification for a few historical herbarium specimens,namely C.parviflorus(Benth.) Domin, C. rostrisepalus Domin, and C. torresianus Gaudich.,were not successful.

The ITS length for Corchorus species ranged from 658 bp to 677 bp,with the shortest sequence detected in C.urticifolius(Table 2).The ITS sequences of ingroup and outgroup taxa were deposited in the EMBL nucleotide database(sequence accession numbers FR874941–FR875090).The two analyses(maximum likelihood and Bayesian inference)showed similar phylogenetic trees.Comparison of the model-based phylogenetic approaches with results from maximum parsimony analysis(data not shown)revealed similar tree topologies.

In both Bayesian inference and maximum likelihood analyses,Corchorus consisted of four major clades(Figs.1 and 2).Clade I(green),which is statistically highly supported(bootstrap support:97,posterior probability:1.0),is formed by the majority of African (endemic) species plus several endemic species from other parts of the pantropics,which were nested in the African taxa. These endemic species include C. elachocarpus, C. leptocarpus, C. walcottii, and C.sidoides from Australia;C.hirsutus from tropical America;and C.neocaledonicus from New Caledonia.Within clade I,the endemic species mentioned above formed a sister relationship with C. junodii, C. velutinus, C. kirkii and C.longipedunculatus,which are endemic to southern African.

The remaining species,pantropically distributed Corchorus species including other species from Africa,were grouped in clades II(orange),III(blue)and IV(purple–gray).Unlike clade I,these clades were statistically not highly supported,although high support was found for several subclades of the large clades.The evolutionary relationships of cultivated(C.olitorius and C.capsularis)and wild species are displayed in clade II.Based on the taxa included in this study,the two cultivated species shared a common ancestor with C. africanus, C.brevicornatus,C.pseudocapsularis,C.pilosus,C.orinocensis,C.pseudo-olitorius,C.urticifolius,and C.cunninghamii(bootstrap support:53,posterior probability:0.73).Within this subclade,C.olitorius is most closely related to C.orinocensis and C.pilosus(bootstrap support:54,posterior probability:0.9).The relationship of the second cultivated species,C.capsularis,to wild taxa is not statistically supported,and different groupings were obtained by maximum likelihood and Bayesian inference. By maximum likelihood it was grouped with C.pseudocapsularis,C.brevicornatus and C.africanus but without statistical support ≤50%.This relationship was not retrieved by Bayesian inference and the subclade was characterized by several polytomies. The relationship among three wild relatives,namely C.africanus and C.brevicornatus,which are endemic to East Africa,and C.pseudocapsularis,distributedfrom eastern to southern Africa, are statistically highly supported(bootstrap support:98,posterior probability:1).In the same clade,C.tridens,C.fascicularis,C.fruticosus,and C.hirtus shared a statistically highly supported common ancestor with the endemic southern Africa species,C. saxatilis(bootstrap support:100,posterior probability:1).Individuals of the pantropically distributed species C.aestuans formed a single group(clade III with blue color in RAxML,and clade IV with blue color in MrBayes trees).

Table 2–Summary statistics of ITS sequences generated from the genus Corchorus included in the study.

In all analyses(Bayesian inference,maximum likelihood,and maximum parsimony),Pseudocorchorus was nested within Corchorus, contradicting its generic classification in the current angiosperm phylogeny group classification.In this phylogenetic study,the two taxa(Corchorus+Pseudocorchorus,displayed in the clade colored in purple-gray) formed a statistically highly supported monophyletic group(bootstrap support: 100, posterior probability: 1). As shown in both analyses,Pseudocorchorus shared a common ancestor especially with two of the Corchorus species C.depressus and C.siliquosus(bootstrap support:54,posterior probability:0.74).

Of the included outgroup genera,Triumfetta was the closest sister to Corchorus(bootstrap support:100,posterior probability: 1). The two Sparrmannia species (S. ricinocarpa and S.palmata)shared a clade with the monotypic genus Entelea(bootstrap support:100,posterior probability:1).Based on this study conducted using an ITS marker,the genus Grewia is the common ancestor for Corchorus/Pseudocorchorus, Triumfetta,Sparrmannia,and Entelea.

4.Discussion

4.1.Phylogeny of the genus Corchorus

The present study confirmed that Pseudocorchorus is nested within Corchorus,so that it may be necessary to include it under the genus Corchorus.The majority of African species formed a statistically highly supported clade,and several endemic species from other pantropical regions, namely tropical America,New Caledonia,and Australia,were nested within the African clade,indicating dispersals from Africa to other pantropical regions.The present study supports earlier findings.The present study supports earlier findings[2,6],reconfirming monophyly of the genus Corchorus,and wild taxa from Africa were found to be maternal ancestors for the two cultivated species.Corchorus neocaledonicus is nested within Corchorus,supporting the annotation of Tirel et al.[13]and molecular findings of Whitlock et al.[19].In this study,C.neocaledonicus is nested within with those species restricted to eastern and southern African regions.

Most of the taxonomic classification of Corchorus species is supported by the present molecular phylogenetic study.However, relatively similar sequences were obtained and differentiations between some species were not clearly recognized. For example, accessions of C. fascicularis, C.fruticosus, and C. hirtus were nested with few sequence differences among them.Together with the few morphological differences among these species, this finding may indicate the need for further study of the taxonomic status of these species.

Fig.1–Maximum likelihood tree showing evolutionary relationships in Corchorus species and the phylogenetic position of Pseudocorchorus.Bootstrap values are indicated above the branches.Numbers next to species name indicate accession numbers,and OG represents outgroups.

Edmonds[36],in his review of African Corchorus species,pointed out the highest similarity between C.trilocularis and C.confusus,and suggested a further study of the taxonomic classification in these species.In addition,the difficulty of differentiating between these species using herbarium specimens was reported by Moeaha[37].Although the taxonomic treatments of the two species were questioned,a new species known as C. argillicola M.J. Moeaha & P.J.D. Winter was described by Moeaha[37]from a specimen formerly identified as C.confusus[38].Corchorus argillicola is described by[37]as a new species owing to its capsule morphology,mainly in the distribution of trichomes and lack of longitudinal ridges.The life history of C.argillicola is reported to be different from C.trilocularis owing to its perennial life cycle,although the latter species has been reported as annual,biennial,or perennial[39–41].In the present molecular phylogenetic study of the genus Corchorus,accessions of C.trilocularis and C.confusus were grouped together,supporting the remarks of[36]in taxonomic treatment revision of these species. The ITS sequences of C. trilocularis and C. confusus and a newly submitted sequence from[37]were relatively similar,forming a single group with high statistical support(bootstrap support:92;posterior probability:1).Thus,although C.argillicola is described as a new species,the present study showed no clear differentiation between these species.In addition,C.sulcatus is morphologically similar to C.asplenifolius in its procumbent habit and with capsules borne on recurved pedicels, but taxonomic recognition for the former species was made due to differences in leaf shape,leaf pubescence,and short and prostrate branches[42,43].In the present study,sequences of C. psammophilus, C. sulcatus, C. pinnatiparititus, and C.asplenifolius were relatively similar and formed a single group in the phylogenetic tree (bootstrap support: 99; posterior probability:1).

Fig.2–Bayesian inference tree showing the evolutionary relationships in Corchorus species and the phylogenetic position of Pseudocorchorus.Posterior probability values are indicated above the branches.

Although the five sections described by Candolle [11]comprise very few of the currently described species, an attempt was made to determine whether these sectional classifications were consistent with the present phylogenetic groupings.The present study did not support the previous classifications and several of the species of a given section were randomly grouped in the four major clades.For example,C.trilocularis and C.olitorius,which belong to Section Coretoides,were distantly clustered in the ITS phylogeny as belonging to two different major clades. Moreover, C. hirsutus and C.fascicularis were grouped under Section Guazumoides,but this relationship was not supported by the present molecular phylogenetic analyses.

The present phylogenetic study revealed preliminary but not strongly supported evolutionary relationships between cultivated and wild species.Based on taxa included in this study, C. olitorius, which is one of the most widely pantropically distributed species of the genus Corchorus,shared a common ancestry with two endemic species from tropical Americas,namely C.orinocensis and C.pilosus.The two diploid cultivated species formed a subgroup with diploid(C.pseudocapsularis),tetraploid(C.orinocensis and C.cunninghamii),and other wild species with unknown ploidy level(C.africanus,C.brevicornatus,C.pilosus,C.pseudo-olitorius,and C.urticifolius).The finding that diploid cultivated species were grouped together with tetraploids might indicate a polyploidization event,which might further clarify the evolutionary relationship.However,sequences from polyploid species were not cloned(but rather were directly sequenced),and the ploidy level of the wild species is not known,owing to the use of leaf specimens obtained from herbarium vouchers.In addition,some other endemic species from Australia were not included.Thus,the evolutionary relationship between wild and cultivated species is based on the taxa included in this study,and hence a broader sampling of other wild species especially from Australia might shed more light on the evolutionary relationship between wild and cultivated taxa.

4.2.Pseudocorchorus Capuron

Formerly, Pseudocorchorus greveanus and P. rostratus were included in the genus Corchorus as C.greveanus Baill[44]and C. rostratus Danguy [45], respectively. Later, Capuron [16]reclassified and these two species and transferred them to a new genus name,Pseudocorchorus,together with four other species.According to his description,Pseudocorchorus differs from Corchorus by its zygomorphic flowers, unilaterally inserted stamens,and subbasifixed,poricidal anthers unique to the former taxa. However, the description and reclassification of C.rostratus Danguy into P.danguyanus has been criticized for several errors in annotation of specimens[15].The present phylogenetic study indicates for the first time that Pseudocorchorus is nested within Corchorus,questioning the taxonomic assignment of the former to the genus level.Pseudocorchorus species were found to be members of Corchorus and shared a common ancestry especially with C.siliquosus,a species endemic to tropical America,and C. depressus, which is common to Africa and Asia.Thus,results of the present study support the highly morphological and anatomical similarities between Corchorus and Pseudocorchorus.Therefore,based on the present molecular findings and previous reports of overall morphological similarities,the inclusion of all Pseudocorchorus taxa under the genus Corchorus is proposed with the taxonomic treatments of Pseudocorchorus as a synonym of Corchorus.

Although Pseudocorchorus is nested within Corchorus,its relationship with the other clades of Corchorus was not identical in the two analyses,making a conclusive biogeographic hypothesis difficult. By maximum likelihood, the clade that includes Pseudocorchorus was sister to the other two clades comprising species with pantropical distributions.However,this relationship was not supported by Bayesian inference,and all the clades including the Pseudocorchorus clade were grouped in a polytomy.By maximum parsimony a sister relationship between clade I and the clade formed by Pseudocorchorus,C.depressus,and C.siliquosus was obtained.

The genus Triumfetta was found to be the closest sister to the monophyletic group Corchorus.The ITS phylogeny revealed little differentiation between Sparrmannia and the monotypic genus Entelea(bootstrap support:100,posterior probability:1),supporting the previous molecular findings[14].Entelea,described as a monotypic genus,is a shrub or small tree endemic to New Zealand.Morphologically and anatomically it is identical to Sparrmannia [46], and is distinguished taxonomically from the South African Sparrmannia only by the stamens being all fertile[47].Based on the present and previous molecular phylogenetic findings,the taxonomic placement of Entelea as a monotypic genus might be questioned.Future phylogenetic study including all Sparrmannia species, together with Entelea, might help to better understand the phylogenetic relationships between these two genera.

4.3. Implication of the phylogenetic tree on life cycle and growth form relationships

All outgroup taxa(Grewia,Sparrmannia,Entelea,and Triumfetta)included species that are either shrubs or trees.With the exception of two species(C.longipedunculatus and C.schimperi),the major clade formed by endemic African species consists of perennial species,whereas the remaining three clades are composed of both annual and perennial species.In terms of growth forms,all of the pantropical species(clade II and C.aestuans),except C.cunninghamii,are herbs,whereas clade I and the other clade that comprises Pseudocorchorus consist of both herbs and shrubs.Thus,in comparison to the outgroup taxa a change of perennial to annual life cycle from clade I to the rest of the clades is seen.The finding that the shrub-type growth form of the outgroup taxa is reflected mainly in clade I may indicate an evolution of growth form from endemic African shrubs to pantropically distributed herbs.

5.Conclusions

In this phylogenetic study, the two taxa Corchorus and Pseudocorchorus formed a statistically highly supported monophyletic group, and hence taxonomic treatments of Pseudocorchorus as a separate genus are not supported.The inclusion of all Pseudocorchorus taxa under the genus Corchorus is therefore proposed, with taxonomic treatments of Pseudocorchorus as a synonym of Corchorus. Although the taxonomic classification of Corchorus species is supported by the present study, treatments of several species remain unclear.Several of the species included in the present study were represented by herbarium voucher specimens.Thus,a revision of the genus and further taxonomic work using living materials, together with additional molecular approaches,might help to resolve problems in taxonomic treatments of Corchorus species.The finding that several endemic species from Australia,tropical America and New Caledonia were nested within endemic species of the southern to eastern African regions might support the hypothesis that the genus evolved in Africa and later dispersed to the other pantropical regions.A further comprehensive study with additional taxa from Australia might help to test hypotheses about historical biogeography and growth form evolution in the genus Corchorus.

Acknowledgments

The work was supported by the German Academic Exchange Service(DAAD)and the Leibniz Institute of Plant Genetics and Crop Research(IPK),Germany.The author thanks the following institutions,which kindly provided herbarium specimens,seeds and DNA aliquots:The Royal Botanical Gardens,Kew,UK;Munich Botanical Garden,Germany;The National Herbarium,Addis Ababa University,Ethiopia;The Museum of Natural History in Paris, France; The Herbarium of the Botanical Garden and Botanical Museum Berlin,Germany;National Herbarium of the Netherlands;International Jute Study Group,Bangladesh;Botanical Museum of the University of Vienna; The World Vegetable Center (Taiwan, China;Tanzania); The Agricultural Research Service, USDA; The Jardin Botanique Exotique de Menton, France; Innsbruck University Botanical Garden,Austria;International Livestock Research Institute,Ethiopia;National Plant Genetic Resource Center,Namibia;and Missouri Botanical Garden,USA.The author thanks the research staffs of Experimental Taxonomy and Herbarium,IPK for providing technical support during laboratory work,as well as for facilitating the acquisition of herbarium specimens from various global institutes.I thank Prof James Nelson,Kansas State University,for his valuable comments and suggestions that greatly improved the quality of the manuscript.Sequence analyses,data interpretation,and manuscript preparation were performed at the Department of Biotechnology,Addis Ababa Science and Technology University,Addis Ababa,Ethiopia.