Background

Parthenium argentatum Gray, commonly known as guayule, is a shrub in the Asteraceae that is native to the southwestern United States and northern Mexico. Parthenium argentatum produces high quality rubber in bark tissue, which is under development for biomedical uses. The U.S. Food and Drug Administration recently approved the first medical device made from P. argentatum natural rubber. Products made from P. argentatum latex are designed for people who have Type I latex allergies, induced by natural rubber proteins from Hevea brasiliensis. In addition to biomedical products, natural rubber is essential and irreplaceable in many industrial and consumer applications, and the price is rising under heavy demand, making natural rubber increasingly more precious. As an industrial crop that grows in temperate climates, P. argentatum represents a viable alternative source of high quality natural rubber.

One strategy for improving crops, such as the rubber-producing P. argentatum, is through chloroplast engineering [1–3]. Transformation of chloroplasts allows high-level production of foreign proteins because of the high number of chloroplasts per plant cell. As homologous recombination is the means by which foreign DNA is incorporated into the chloroplast genome, transformation is precise and predictable. Moreover, it has been shown that up to four genes can be inserted at once [4], enhancing the efficiency of metabolic engineering. From production of edible vaccines to bioplastics, transplastomic plants have been shown to provide a useful route to manipulate crops for industrial purposes [5].

Importantly from the point of view of minimizing environmental impact, expressing foreign proteins in the chloroplast results in transgene containment [6, 7]. It is thought that in the vast majority of plant species, chloroplasts are not transmitted by pollen, and so in these species, chloroplastidic transgenes would not be spread in that manner. Although, it is becoming clear that each case must be thoroughly verified [8, 9]. In the case of P. argentatum, transgene containment is important because it is currently cultivated as an industrial crop in its native region in the southwestern United States.

Construction of vectors for chloroplast transformation requires some knowledge of the chloroplast genome sequence to identify insertion sites. To date, just short of one hundred plastid genomes from angiosperms have been completely sequenced. The sequences are highly conserved [10]. Interestingly however, the order of genes in some groups, including the Asteraceae, Fabaceae and Poaceae, may be reversed by large inversions [11–13]. In the Asteraceae, the family of interest in this study, there is a second small inversion (~3 kb) nested within the larger inversion (~23 kb) [14]. The two inversions are always found together, implying that they occurred close in evolutionary time.

Chloroplast sequences are useful for identification of species, using a particular sequence as a DNA tag or barcode [15]. An ideal DNA barcode for general purposes would 1) have enough diversity to allow discrimination among species, but not so much that would prevent grouping of members of a species, 2) work in wide variety of taxa, and 3) provide the basis for reliable amplifications and sequences [16]. In plants, unlike in animals, the mitochondrial genome evolves too slowly to provide useful DNA barcode sequences. Although also possessing a relatively slow rate of evolution, several chloroplast sequences have been identified as fulfilling the criteria listed above [17–19]. Depending on the desired level of discrimination, the consensus conclusion appears to be that the low mutation rate in the chloroplast genome may require more than one barcode locus to be probed [18, 20, 21].

At present, classical breeding is being used to improve P. argentatum as a commercial source of natural rubber. Breeding efforts would be enhanced by informative chloroplast DNA barcodes. Because a very small amount of tissue is required for barcode analysis, purity of breeding lines can be determined at an early stage of seedling growth. In addition, barcodes would allow breeders and seed producers to discover seed lot contamination before advancing breeding lines for latex production. Having the ability to removing contaminating lines, especially when they represent lower rubber lines, would improve the efficacy of breeding efforts.

The focus of our research program is improvement of P. argentatum to enhance its commercial viability. We have chosen two approaches, biotechnology through chloroplast metabolic engineering and marker-assisted breeding. The P. argentatum chloroplast genome sequence that we report herein, supports our efforts in both approaches. In this article, we report the complete sequence of the chloroplast genome of P. argentatum and describe the development of DNA barcodes. The complete sequence of the P. argentatum chloroplast genome has enabled us to construct chloroplast transformation vectors based on the exact sequence of the large inverted regions, and to identify novel insertion sites in non-essential, non-coding regions. Barcode analysis with the matK gene and psbA-trnH spacer sequence allowed us to discriminate three of four Parthenium species from each other and from P. argentatum, and a subset of the P. argentatum lines from each other. These barcodes will be used in our breeding program.

Results

Genome size and gene content, order and organization

The complete nucleotide sequence of the chloroplast genome of Parthenium argentatum is represented in a circular map (Figure 1; Genbank Accession GU120098). It is 152,803 bp in size and includes a duplicated region of inverted repeats (IR) of 24,424 bp. The IR are separated by small single copy (SSC) and large single copy (LSC) regions of 19,390 bp and 84,565 bp, respectively. The total G+C content of the whole chloroplast genome is 37.6%. The gene content and arrangement were observed to be similar to those in Lactuca sativa and Helianthus annuus [22], and Guitozia abyssinica (NC_010601), including one large (Inv1) and one small inversion (Inv2) in the LSC region. There are 85 genes coding for proteins (Additional file 1), including six that are duplicated in the IR regions. There are four rRNA genes that are also duplicated in the IR regions. In total there are 43 tRNA genes, seven of which are duplicated in the IR, one in the SSC, with the remaining 28 scattered in the LSC region.

Figure 1
figure 1

Representative map of the chloroplast genome of Parthenium argentatum (Genbank Accession GU120098). IR, inverted repeat; LSC, large single copy region; SSC, small single copy region; Inv1, inverted sequence 1; Inv2, inverted sequence 2. Gene names and positions are listed in Additional file 1.

Full size image

The size of the P. argentatum chloroplast sequence is larger than those of the three other Asteraceae chloroplast genomes (Table 1). It is close to the same size as the L. sativa genome, and 1.04 kb and 1.7 kb larger than the G. abyssinica and H. annuus genome, respectively, with the length differences primarily found in the LSC and SSC domains. The sequence differences between P. argentatum and each of the other three chloroplast genomes are concentrated in the noncoding regions of Inv2, and the SSC and LSC regions (Figure 2). The IR regions in P. argentatum are shorter than those of the three other species by 210-610 bp (Table 1, Figure 2).

Figure 2
figure 2

Chloroplast genomes of Parthenium argentatum , Helianthus annuus , Guizotia abyssinica and Lactuca sativa compared with mVISTA. A cut-off of 70% identity was used for the plot and the Y-scale represents the percent identity ranging from 50 to 100%. Blue represents exons, green-blue represents untranslated regions, and pink represents conserved non-coding sequences (CNS). Horizontal black lines indicate the position of Inv1, Inv2, IRa and IRb; SSC is flanked by IRa and IRb; grey arrows the direction of transcription.

Full size image
Table 1 Size comparison of Parthenium argentatum chloroplast genomic regions with those in other members of Asteraceae.
Full size table

Based on sequence comparison of the chloroplast genome of P. argentatum with H. annuus and L. sativa, two inversions of 22,890 bp and 3,364 bp were observed in P. argentatum, similar to those described by Kim et al. [14] and Timme et al. [22]. In P. argentatum, one end point of the 23 kb inversion was located between the trnS-GCU and trnG-UCC genes. The other end point is located between the trnE-UUC and trnT-GGU genes. The second 3.4 kb inversion was observed within the 23 kb inversion, which shares one end point just upstream of the trnE-UUC gene with the large inversion. The other end point of the 3.4 kb inversion is located between the trnC-GCA and rpoB genes (Figure 1).

Variation in chloroplast coding sequences of Asteraceae family members

Variation between coding sequences of P. argentatum and H. annuus, G. abyssinica or L. sativa was analyzed by comparing each individual gene (Additional file 1) as well as the overall sequences (Figure 2). In general, P. argentatum coding sequences are more similar to those in G. abyssinica (98.5% identical on average) and H. annuus (98.4%), than in L. sativa (97.2%). The greater average identity in G. abyssinica than in H. annuus is in large part due to deletions in the two copies of the ycf2 loci in H. annuus, otherwise, H. annuus is more similar overall than G. abyssinica. Fourteen genes in H. annuus and G. abyssinica were 100% identical to those in P. argentatum, compared to only four genes in L. sativa (Additional file 1). The most-divergent coding regions in the three genomes were ycf1, accD, clpP, rps16, and ndhA (Figure 2).

DNA barcode analysis of Parthenium

To differentiate Parthenium taxa, a molecular approach was used in which we analyzed four different chloroplast DNA regions, which were shown to be useful DNA barcodes in past studies [16, 18, 23, 24]. These regions were the trnL-UAA intron, rpoC, matK and the non-coding spacer region between the two genes psbA-trnH. Tests were conducted on DNA of three Parthenium species (P. incanum, P. tomentosum, and P. schottii) and three cultivated lines of P. argentatum (AZ2, AZ3 and Cal6) (data not shown). The best differentiation of Parthenium species and lines within P. argentatum was obtained with the psbA-trnH spacer region barcode. There were 5 indel sites in 400 bp of DNA in the six lines tested. When 1000 bp of the matK DNA barcode were analyzed, a total of 12 indel sites were found. In 600 bp from the trnL-UAA intron region, only one indel site was observed. Obtaining good sequence from the rpoC spacer region was difficult, but in 500 bp, four indel sites were identified. Therefore, due to the higher number of informative sites, the matK and psbA-trnH DNA barcodes were used for further studies of Parthenium taxa.

The matK DNA barcode

After re-evaluation of the 1000 bp sequence of matK, an efficient barcode for Parthenium species was defined. Using the Parth-matK-F and Parth-matK-R primers, matK DNA sequences were examined in Parthenium species, lines of P. argentatum and AZ101, a hybrid of P. argentatum cv. 11591 × P. tomentosum. We sampled 601 nucleotides in the matK gene, which yielded fourteen potentially informative, variable positions (2.3%), with eight nucleotide substitutions (1.3%) and six length mutations (indels) (1.0%). Although the psbA-trnH spacer region in P. integrifolium DNA did amplify with the psbA-trnH barcode primers, the matK locus did not amplify with the matK-barcode primers. This matK barcode was effective at differentiating P. schottii, P. hysterophorus, and P. tomentosum from each other and from a group that included P. incanum, P. argentatum lines and one hybrid (Figure 3). This barcode did not differentiate P. incanum from the seven P. argentatum lines and the hybrid (Table 2).

Figure 3
figure 3

Differentiation by mat K barcode (Genbank Accession 1230803) in Parthenium species. UPGMA in Jukes-Cantor mode, with gamma correction, was used to construct the tree, with statistical support for tree branches evaluated by bootstrap analysis (1000 replicates), indicated above the node. Helianthus annuus is included as an outgroup.

Full size image
Table 2 Population information for analyses of Parthenium species using DNA barcode sequences.
Full size table

The psbA-trnH DNA barcode

The non-coding spacer region between psbA and trnH was used to differentiate several Parthenium species, lines of P. argentatum and a hybrid of two Parthenium species (Table 2). A 469 bp region was amplified via PCR using the psbA-F and trnH-R primers. This region produced the best differentiation (Figure 4). We sampled 456 nucleotides in the psbA and trnH spacer, which yielded fourteen potentially informative, variable positions (3.1%), with eleven nucleotide substitutions (2.4%) and three length mutations (0.7%). First of all, we found that there was 100% consensus in the barcode sequence among samples tested of line AZ1 (n = 21), AZ4 (n = 15), Cal6 (n = 17), AZ101 (n = 3), P. incanum (n = 6) and P. tomentosum (n = 5). On the other hand, there was a second barcode sequence within line AZ2 (minority barcode in 6.5% of total, n = 31), AZ3 (minority barcode 6.7%, n = 15), AZ5 (minority barcode 20%, n = 15), AZ6 (minority barcode 15%, n = 20) and 11591 (50% alternative barcode, n = 20). The minority or alternative barcodes differed from the corresponding common barcode by one to three bases.

Figure 4
figure 4

Differentiation by psb A- trn H spacer region barcode (Genbank Accession 1230807). This barcode was analyzed in Parthenium species, P. incanum, P. tomentosum, P. schottii, P. integrifolium, hybrid AZ101 (P. argentatum × P. tomentosum) and P. argentatum lines AZ1, AZ2, AZ3, AZ4, AZ5, AZ6, Cal6, C156, C86 and cv. 11591. UPGMA in Jukes-Cantor mode was used to construct the tree, with statistical support for tree branches evaluated by bootstrap analysis (1000 replicates), indicated above the node. Minority barcodes are indicated by #'s after the name of the line. Helianthus annuus is included as an outgroup.

Full size image

The psbA-trnH spacer barcode differentiated P. hysterophorus, P. integrifolium and P. schottii from each other and from all the other species and lines. The psbA-trnH spacer barcode of P. argentatum cultivar 11591 and the two breeding lines C156 and C86 was different from those of the remaining P. argentatum lines, P. tomentosum and P. incanum. The barcode of AZ101, which is a hybrid between P. argentatum cultivar (cv.) 11591 and P. tomentosum, is similar to or identical to that of P. tomentosum. Parthenium incanum's barcode clustered with two AZ2 variants and a plant of unknown parentage, indicating their close relationship. Analysis with both the psbA-trnH spacer and matK barcodes provided further differentiation (Figure 5). The combined barcodes of AZ101 and P. tomentosum are more similar to each other than to all those of the P. argentatum lines together with P. incanum. Drilling deeper, the barcodes of cv. 11591/C156/C86 are different from those of P. incanum and all the remaining P. argentatum lines.

Figure 5
figure 5

Barcode differentiation using the combined mat K sequence and the spacer region of psb A- trn H. Combined barcodes were analyzed in Parthenium species, P. incanum, P. tomentosum, P. schottii, hybrid AZ101 (P. argentatum × P. tomentosum) and P. argentatum lines AZ1, AZ2, AZ3, AZ4, AZ5, AZ6, Cal6, C156, C86 and cv. 11591. UPGMA in Jukes-Cantor mode was used to construct the tree, with statistical support for tree branches evaluated by bootstrap analysis (1000 replicates), indicated above the node. Helianthus annuus was used as an outgroup.

Full size image

Discussion

Comparative genome organization and structure

Asteraceae is one of the largest families of flowering plants with approximately 1,500 genera and 23,000 species. Production of secondary metabolites is a key feature of this diverse family. For example, several genera within the Asteraceae produce high molecular weight rubber in the cytosol, including Lactuca sativa [25] and Taraxacum kok-saghyz [26], and the species of interest to our studies, Parthenium argentatum. To support efforts to improve the levels of rubber production in this industrial crop, the sequence of the chloroplast genome of P. argentatum was determined. This information is useful for our efforts in chloroplast engineering. The barcodes we present will be used in breeding of commercially important lines in the genus Parthenium.

Within the Asteraceae, the P. argentatum chloroplast sequence represents the fourth complete sequence. This sequence reveals that the chloroplast genomes of P. argentatum, H. annuus, G. abyssinica and L. sativa are identical in gene order and content (Figure 1; Figure 2). The four genomes differ slightly in length, with the chloroplast genome in P. argentatum somewhat longer than those in L. sativa, G. abyssinica and H. annuus, respectively (Table 1). Two inversions in the chloroplast genome are shared by two of the three subfamilies of the Asteraceae [14, 22] and are present in P. argentatum (Figure 1). In H. annuus, the IR-located gene ycf2 has an internal deletion of 455 bp that is not found in the three other genomes. The large chloroplast gene ycf2 specifies an expressed protein [27], whose function has not yet been determined, although ycf2's homology to ATPases was noted by Wolfe [28]. Our protein domain analysis [29] suggests similarity with conserved domains of the ATPase AAA family that perform chaperone-like functions involved in assembly or disassembly of protein complexes. In some chloroplast genomes, particularly in grasses, ycf2 is entirely absent [30]. Despite that fact, knockout studies in Nicotiana tabacum demonstrated that ycf2 is essential for survival [31]. There must be sufficient coding sequence remaining in H. annuus to provide any essential ycf2 function. Interestingly, ycf2 is one of the eight fastest evolving genes in the chloroplast genome (Additional file 1; [32]). Notably, this rapid evolution has taken place in the framework of the more slowly evolving IR region as a whole (Figure 2; [33]). Another notable size difference in coding regions is found in the SSC region. The SSC region of the chloroplast genome of P. argentatum is 791 to 1162 bp longer than that in the other species (Table 1). Within the SSC region, the ycf1 gene has a 3'-deletion in H. annuus, G. abyssinica and L. sativa (Figure 2). Similar to ycf2, ycf1 encodes a protein of unknown function that is also essential [31]. It appears to be a multi-pass transmembrane protein, with no clear association to known functional domains.

In a comparative study of individual genes of P. argentatum, H. annuus, G. abyssinica and L. sativa, we identified several sequences with high levels of differences along their length, the most divergent including the already mentioned ycf1, and clpP, rps16, accD, and ndhA (Additional file 1). Interestingly, three of these genes, ycf1, accD and clpP, are essential plastid genes in some taxa, but not others [31, 34–37]. The presence of non-coding intronic sequences in both ndhA and rps16 contributes to the divergence in those two loci [38, 39]. These divergent sequences among the four Asteraceae chloroplast genomes identify the fastest evolving regions containing coding sequences.

Metabolic engineering of plants by inserting transgenes in the chloroplast would potentially be made more efficient with knowledge of chloroplast sequences, based on the conclusions of one group that chloroplast transformation efficiency was significantly enhanced when vectors were constructed with 100% homologous sequences [40]. Other groups have shown that precise homology may not be essential, as tobacco sequences [41] were sufficient to allow recombination in tomato [42], potato [43], and petunia [44]. The chloroplast genome sequence of P. argentatum was used to design a 100% specific chloroplast transformation vector (unpublished data), to maximize the possibility of successful recombination. Improving crop plants via chloroplast transformation is a viable strategy [1, 5] that will be pursued in this industrial crop.

DNA barcodes

Chloroplast genomic sequences were used to develop DNA barcodes to discriminate at the species level and below. The matK barcode contained sufficient information to differentiate three Parthenium species (tomentosum, hysterophorus and schottii) from each other and from P. argentatum and P. incanum. However, the matK-barcode did not differentiate P. incanum from P. argentatum or P. agentatum lines from each other (Figure 3). The psbA-trnH spacer barcode provided additional differentiation at the species level and below (Figure 4, 5). Interestingly, when the matK gene and the psbA-trnH spacer barcode information was combined, P. tomentosum and cv. 11591 were differentiated from the remaining P. argentatum lines and P. incanum. Using the combined barcodes, we observed that they were more similar in P. argentatum AZ1 to AZ6 and Cal6 lines overall than they were in the P. argentatum cv. 11591, breeding lines C-156 and C86, and hybrid line AZ101 (Figure 5). To understand the pattern of differentiation, it would be useful to have precise information about the pedigrees of all the lines. Unfortunately, in most cases that is either lacking or incomplete. We know that AZ4 and AZ5 were selected from the same seed lot [45] and their combined barcodes are very similar (Figure 5). We cannot trace the ancestors of AZ4, AZ5 and AZ6 to understand the history of their relatedness to AZ1, AZ2, AZ3 and Cal6. The barcodes of the two P. argentatum lines AZ2 and AZ3 were not different, which is not surprising as AZ2 and AZ3 were selections from the same 11591 seed lot [45], however, it would be expected that their majority barcodes would be more similar to 11591 than they are. The psbA-trnH DNA barcode analysis demonstrated that two plants of AZ2, #8 grown in a field at Higby and #16 grown in a field at the Maricopa Agriculture Center (MAC) have a different psbA-trnH barcode than the common DNA barcode sequence of AZ2 (Figure 4). These do not appear to be pure AZ2 derivatives and may represent seed contaminants. Several of the P. argentatum lines were homogeneous according to the psbA-trnH spacer sequence, including AZ1, AZ4, and Cal 6. Other lines were less homogeneous, including AZ2, AZ3, AZ5, and AZ6, with a minority sequence present in 6 to 20% of the individuals tested. From our own observations in the field, P. argentatum accessions are highly heterogeneous in growth habit, suggesting that seed lots are composed of highly mixed genetic populations. This would not be unexpected for open-pollinated, self-incompatible, field-grown lines. Our barcode data support the heterogeneity and provides information that will be used immediately to differentiate breeding populations.

Classical breeding efforts will be enhanced by using the informative chloroplast DNA barcode we describe herein. We assessed the genetic purity of a small population of P. argentatum using the psbA-trnH barcode and were able to show, as described above, which lines had undergone homogenization and which had not (Figure 5). Knowledge of the purity of lines and the presence of contaminating seeds, will further our breeding efforts of lines that are being advanced for latex production.

Our barcode study was useful in providing support for the maternal parent of the hybrid plant, AZ101. AZ101 is a vigorous interspecific hybrid, low in rubber concentration, but high in biomass production [46]. The line is the result of an open-pollinated cross between P. argentatum cv. 11591 and P. tomentosum cv. stramonium [45]. AZ101 most likely inherited its chloroplast genome from P. tomentosum, as AZ101 and P. tomentosum are not differentiated by the combined barcode system (Figure 5). Although we do no know the reason for the difference, our results are not the same as those from the non-DNA analyses by Ray and co-workers [47]. More extensive analysis of differences at the DNA level is necessary.

According to the literature, there are about a dozen species of Parthenium growing on the North American continent. However, P. argentatum is the only species with commercially viable amounts of rubber. Other species such as P. incanum and P. tomentosum produce primarily resinous materials [48]. The substrate for rubber biosynthesis is isopentenyl pyrophosphate (IPP) [49, 50]. Chloroplasts have been shown to contribute to the pool of IPP in plant cells [e.g., [51]; unpublished data, Kumar and Whalen]. If the levels of chloroplastic IPP production vary from line to line, it may be possible to breed for enhancements in substrate production by controlling the maternal parent. This suggests that hybrids could be developed using a maternal parent that produces more rubber like AZ2 combined with a higher biomass from a line like AZ101, to produce a superior plant. More experiments are necessary to understand the role of the maternal parent in rubber biosynthesis.

Our preliminary results on lack of PCR amplification from mature pollen DNA of targets within the IR regions (data not shown), suggest that chloroplasts are not present in the mature pollen and thereby are likely to be maternally inherited in P. argentatum. Use of plastid specific barcodes derived from the genome sequence, will allow us to definitively track any paternal inheritance in future experiments. With the recent finding of paternal inheritance in a weedy Helianthus species [52], as well as in species previously considered to lack paternal inheritance in pollen, such as Arabidopsis thaliana [8, 9], it is crucial that extensive studies are performed, especially if a strategy for transgene containment depends on not transferring transgenes in pollen.

Conclusion

The genome sequence of the P. argentatum chloroplast will enrich the sequence resources of plastid genomes in commercial crops. The availability of the complete plastid genome sequence may facilitate improved transformation efficiency by using the precise endogenous flanking sequences and regulatory elements in chloroplast transformation vectors. The DNA barcoding study forms the foundation for genetic identification of commercially important lines of P. argentatum that are producing natural rubber latex for biomedical applications.

Methods

Isolation of chloroplasts and DNA amplification, and sequencing

A mature, greenhouse-grown Parthenium argentatum line AZ2 plant was placed in the dark for 2-days before harvesting young leaves. Chloroplasts were isolated from leaves using a 30-52% sucrose-gradient according to both Palmer [53] and Jansen et al. [54]. Genomic DNA from chloroplasts was isolated using the GeneElute Plant Genomic Miniprep kit (Sigma-Aldrich Co.). The resulting DNA was amplified using the REPLI-g whole genome amplification kit (Qiagen, Inc.). Amplified DNA was digested with EcoRI and BstBI and examined by agarose gel electrophoresis to confirm the clear banding pattern, which indicated that the amplification product was chloroplast and not nuclear DNA.

Genome sequencing, assembly and annotation

Parthenium argentatum chloroplast genome sequencing was carried out using 454 Sequence Technology (Agencourt Biosciences, Corp). Random sequences were assembled into a draft genome sequence using Newbler as described by Chaisson et al. [55]. The whole genome was annotated using DOGMA (Dual Organellar GenoMe Annotator; [56]) to identify coding sequence, rRNAs, and tRNAs using the plastid/bacterial genetic code. To analyze the similarity of the chloroplast genes in P. argentatum and the other members of the Asteraceae, H. annuus (NC_007977), L. sativa (NC_007578), and G. abyssnica (NC_010601), the percent identity of nucleotide sequences within the open reading frame was calculated based on alignments made with ClustalW [57] and BLAST 2 SEQUENCES [58]. Inversions in the chloroplast genome of P. argentatum were identified by comparing the sequence in the inversion region [11] with that in L. sativa, H. annuus and Nicotiana tabacum (NC_001879). The end points of the inversion were determined as described by Timme et al. [22]. The mVISTA program in Shuffle-LAGAN mode [59] was used to compare the DNA sequences of the chloroplast genomes of the four species of Asteraceae, using the sequence annotation information of P. argentatum (Figure 2).

Identification of Parthenium species and lines

To differentiate various Parthenium species and lines, a chloroplast DNA barcode system was developed. Four regions of the Parthenium chloroplast genome were explored, including the intron in trnL-UAA, the rpoC and matK genes, and the non-coding spacer between psbA-trnH. Plant genomic DNA was isolated from young plants (3-4 weeks old) of available Parthenium species, cultivars, and lines using DNeasy Plant Mini Kit (Qiagen, Inc.). PCR was carried out with Phusion DNA Polymerase according to manufacturer's instructions (New England Biolabs, Inc.). The primers, TrnL-F, 5'-CGAGTTGGGGATAGAGGGACTTGAAC-3' and TrnL-R, 5'-GATATGGCGAAATAGGTAGACGCTACGGAC-3' were used to amplify trnL-UAA; for rpoC, rpoC1-F, 5'-CATAGGAGTTGCTAAGAGTCAAATTCGG-3' and rpoC2-R, 5'-CCTTTTCTAGATCTTGATTCACGTAGAAATTCCGC-3'; for matK, matK-F, 5'-GAATTTCAAATGGAGAATTCCAAAGC-3' and matK-end-R, 5'-CGAGCTAAAGTTCTAGCACAAGAAAGTCG-3'; and for psbA-trnH, psbA-F, 5'-GGAAGTTATGCATGAACGTAATGCTC-3' and trnH-R, 5'-CGCGCATGGTGGATTCACAATC-3'. PCR products were sequenced in both directions. Sequences were compared and any sequences with differences from the majority sequence were re-sequenced in both directions. Barcode differentiations were visualized using the UPMGA best tree method in Jukes-Cantor mode and then bootstrapped with 1000 replicates according to manufacturer's instructions in MacVector (MacVector, Inc.). Helianthus annuus was included as an outgroup.

Based on preliminary analysis of selected taxa of Parthenium, the central region of the matK gene was the best for finding divergence in Parthenium species. DNA from P. schottii, P. tomentosum, P. incanum, a cultivar of P. argentatum cv. 11591, nine lines of P. argentatum (AZ1, AZ2, AZ3, AZ4, AZ5, AZ6, C156, C58 and Cal6) and AZ101 (a hybrid of P. argentatum cv 11591 × P. tomentosum) was amplified via PCR with a 60°C annealing temp, using primers Parth-matK-F, 5'-CAAGCTCATCTGGAAATCTTGGTTCAGGCTC-3' and Parth-matK-R, 5'-GCCAACGATCCAACCAGAGGCATAATTGG-3'. The PCR products were sequenced in both directions using the same primers. In addition, the non-coding spacer region between the two genes psbA-trnH (500 bp) was used to further differentiate the Parthenium taxa. DNA was amplified with the PCR using primers psbA-F and trnH-R at an annealing temperature of 58°C. PCR products were sequenced in both directions with the following primers, psbAF1-seq, 5'-GCTGCTATTGAAGCTCCATC-3' and Rev1-seq-trnh Gua, 5'-CCTTGATCCACTTGGCTACATCCG-3'.