De novo transcriptome assembly of Pueraria montana var. lobata and Neustanthus phaseoloides for the development of eSSR and SNP markers: narrowing the US origin(s) of the invasive kudzu

De novo transcriptome assembly of Pueraria montana var. lobata and Neustanthus phaseoloides for... Background: Kudzu, Pueraria montana var. lobata, is a woody vine native to Southeast Asia that has been introduced globally for cattle forage and erosion control. The vine is highly invasive in its introduced areas, including the southeastern US. Modern molecular marker resources are limited for the species, despite its importance. Transcriptomes for P. montana var. lobata and a second phaseoloid legume taxon previously ascribed to genus Pueraria, Neustanthus phaseoloides, were generated and mined for microsatellites and single nucleotide polymorphisms. Results: Roche 454 sequencing of P. montana var. lobata and N. phaseoloides transcriptomes produced read numbers ranging from ~ 280,000 to ~ 420,000. Trinity assemblies produced an average of 17,491 contigs with mean lengths ranging from 639 bp to 994 bp. Transcriptome completeness, according to BUSCO, ranged between 64 and 77%. After vetting for primer design, there were 1646 expressed simple sequence repeats (eSSRs) identified in P. montana var. lobata and 1459 in N. phaseoloides. From these eSSRs, 17 identical primer pairs, representing inter-generic phaseoloid eSSRs, were created. Additionally, 13 primer pairs specific to P. montana var. lobata were also created. From these 30 primer pairs, a final set of seven primer pairs were used on 68 individuals of P. montana var. lobata for characterization across the US, China, and Japan. The populations exhibited from 20 to 43 alleles across the seven loci. We also conducted pairwise tests for high-confidence SNP discovery from the kudzu transcriptomes we sequenced and two previously sequenced P. montana var. lobata transcriptomes. Pairwise comparisons between P. montana var. lobata ranged from 358 to 24,475 SNPs, while comparisons between P. montana var. lobata and N. phaseoloides ranged from 5185 to 30,143 SNPs. (Continued on next page) * Correspondence: egana@si.edu; ashegan2@gmail.com Computational Biology Institute, Milken Institute School of Public Health, George Washington University, Washington, DC, USA Department of Botany, National Museum of Natural History, Smithsonian Institution, Washington, DC, USA Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Haynsen et al. BMC Genomics (2018) 19:439 Page 2 of 17 (Continued from previous page) Conclusions: The discovered molecular markers for kudzu provide a starting point for comparative genetic studies within phaseoloid legumes. This study both adds to the current genetic resources and presents the first available genomic resources for the invasive kudzu vine. Additionally, this study is the first to provide molecular evidence to support the hypothesis of Japan as a source of US kudzu and begins to narrow the origin of US kudzu to the central Japanese island of Honshu. Keywords: Pueraria montana var. lobata,Kudzu, Neustanthus phaseoloides, Transcriptome, Invasive, Molecular markers Background characterization and development of microsatellites Pueraria montana (Lour.) Merr. var. lobata (Willd.) (SSRs) and single nucleotide polymorphisms (SNPs). Maesen & Almeida ex Sanjappa and Pradeep (kudzu) and Transcriptome sequencing is currently one of the most Neustanthus phaseoloides (Roxburgh) Bentham (tropical popular applications of next-generation sequencing due kudzu), members of the phaseoloid clade of subfamily to its versatility, cost efficiency, and suitability for use on Papilionoideae of the Fabaceae family, are twining vines non-model organisms [16]. Transcriptomes are often native to Southeast Asia that have been introduced glo- mined for expressed simple sequence repeats (eSSRs) for bally for livestock forage, nitrogen soil enrichment, and marker development and genetic diversity studies. eSSRs erosion control [1]. Prior to recent molecular and taxo- have been shown to have greater transferability across nomic revision [2], Neustanthus was placed within taxa than traditional ‘anonymous’ SSRs [17, 18]. This in- Pueraria, along with ~ 17 additional species native to creased transferability can be utilized in multiple ways. southeast Asia [3]. A comprehensive molecular system- First, if a transcriptome is not available for the species of atic study of Pueraria sensu van der Maesen [4]con- interest, a closely related species whose transcriptome is firmed that its species, including several legumes of available can be used as a surrogate reference for micro- economic importance, comprise a polyphyletic assem- satellite development. Second, if a researcher is studying blage of separate evolutionary lineages spread across two closely related taxa and transcriptomes are available the phaseoloid clade [5]. for both, a single set of markers can be developed that Both kudzu and tropical kudzu share a penchant for in- work on both species to reduce costs. To this end, we vasiveness in their naturalized areas, the southeastern have compared the transcriptomes of kudzu and tropical United States (US) and the pantropics, respectively. Of the kudzu to identify shared eSSRs between the species in two taxa, kudzu is a far greater agricultural pest and has order to develop primers that can be used equally well garnered the majority of scientific inquiry. Kudzu was in- for population genetic studies of either species, and shed troduced into the US during the Centennial Exposition of light on the introduction history of the notorious inva- 1876 in Philadelphia, Pennsylvania [6]. The vine is cur- sive kudzu in the United States. rently found in 30 states and is considered an agricultural In the present study, three transcriptomes, two P. mon- pest throughout the southeastern US [7], costing millions tana var. lobata and one N. phaseoloides,were denovo of dollars in eradication and management measures annu- assembled and characterized. Intra- and inter-specific ally [8, 9]. A major aspect that could be influencing the in- comparisons were made between transcriptomes and two vasiveness and spread of kudzu are high levels of genetic sets of population genetic markers were identified: eSSRs variation observed across populations in the US. This and SNPs. The eSSRs were validated across Asian and could be due to multiple introductions from its native North American populations of P. montana. var. lobata range, either of a single genetically diverse population, or and used to explore population diversity and structure from multiple genetically distinct subpopulations, poten- across native and introduced ranges. The resulting data tially from different geographic regions or from more than provide genetic resources for future studies of kudzu and one of the taxonomically recognized varieties of Pueraria related genera through development of high-resolution montana. marker sets for genetic diversity assessment and popula- Several molecular markers have been used over the tion studies. past two decades to estimate the introduced and native genetic diversities of kudzu and two other Pueraria Results montana varieties: Pueraria montana var. montana and Transcriptome sequencing and quality control Pueraria montana var. thomsonii (Benth.) Wiersema ex Transcriptome sequencing produced between 279,109 D.B. Ward [10–15]. However, despite the ecological and and 423,426 reads per transcriptome (Table 1), with economic importance of kudzu, its modern molecular Neustanthus phaseoloides (hereafter CPP02) having the marker resources are limited, lagging particularly in the most reads produced. CPP02 and the greenhouse-raised Haynsen et al. BMC Genomics (2018) 19:439 Page 3 of 17 Table 1 Statistics following ConDeTri cleaning and Trinity assembly Accessions CPP27 Pmnk6 CPP02 Number of raw reads 279,109 396,022 423,426 Number of raw bases (bp) 112,337,841 247,596,818 158,214,933 Number of clean reads 257,015 381,166 348,529 Cleaned reads / Raw reads (%) 92.1% 71.0% 82.3% Number of clean bases (bp) 75,672,645 124,810,371 87,666,889 Mean clean read length (bp) 294 444 252 Number of aligned reads 99,248 116,524 119,452 Aligned read / Cleaned reads (%) 38.6% 41.4% 34.3% Number of contigs 18,325 15,736 18,412 Number of bases in contigs (bp) 11,703,977 15,640,762 11,892,992 Mean contig length (bp) 639 994 646 N50 (bp) 755 1256 759 Longest contig (bp) 4335 4815 6221 Number of singletons 60,869 45,306 73,994 Singletons / Cleaned reads (%) 23.7% 16.1% 21.2% Number of bases in singletons (bp) 17,591,281 20,431,176 18,048,611 Mean singleton length (bp) 289 451 244 Number of transcripts (contigs + singletons) 79,194 61,042 92,406 bp base pairs kudzu (hereafter CPP27) were sequenced on the same De novo assembly run and were multiplexed with two other transcriptomes Trinity used an average of 38.1% of the ConDeTri cleaned not reported here. While sequencing of CPP02 produced reads in its assemblies and produced an average of 17,491 the most reads, the mean read length before cleaning contigs. The mean contig lengths ranged from 639 bp to was shorter than that of CPP27 (373 bp vs. 402 bp, re- 994 bp (Table 1) and each of the accessions had contigs spectively), as was the mean read length after cleaning exceeding 3000 bp (Fig. 1). Additionally, Bowtie2 mapped (252 vs. 294, respectively). The tendency for shorter ~ 68% of each accession’s contigs back to their raw reads DNA fragments to be incorporated at the library con- (Additional file 1). Overall transcriptome contamination struction phase and sequencing stage may provide an was low, with fungal contamination ranging between 2.64 explanation for the difference in the number of raw and 3.53%, while prokaryote and viral contamination reads produced between CPP27 and CPP02. However, ranged from 0.5 to 1.32% (Additional file 2). Transcrip- following cleaning, the number of clean bases was tome completeness varied greatly, with a range of comparable between CPP02 and CPP27, as were all complete units from 164 to 361 and duplicate units simi- other downstream metrics (Table 1). While 454 pyro- larly showing a > 2× difference between transcriptomes sequencing was used for all three transcriptomes, the (Fig. 2). Specifically, transcriptome completeness was ap- chemistries between the two CPP transcriptomes and proximately 64, 77, and 70%, for CPP27, Pmnk6, and the wild-collected kudzu (hereafter Pmnk6) transcrip- CPP02, respectively. The reciprocal best BLAST hits tome differed, with the Pmnk6 transcriptome benefit- (RBH) of the transcriptomes showed that 1525 transcripts ing from an improved chemistry, as seen in the were shared among all three (Fig. 3). increased number of raw bases, the average read length before cleaning (625 bp) and the mean clean Functional annotation of transcriptomes read length (444; Table 1). These sequencing improve- In total, we have obtained 13,230, 18,446 and 24,447 asso- ments translated into improved assembly statistics, ciated GO IDs for CPP02, CPP27 and Pmnk6 transcrip- such as increased mean contig length (~ 1.5× that of tomes, respectively (Table 2) corresponding to the 33, 43 the CPP transcriptomes), higher N50 (1.65× CPP) and and 51% of original contigs in each transcriptome, while fewer singletons (Table 1). However, the improved only 9.6, 17 and 36% of the singletons had associated func- chemistry did not lead to differences in the number tional protein information (GO IDs). Therefore, more of aligned reads in the assembled transcriptomes than 90, 82 and 63% of singletons were discarded during (Additional file 1). the multiple searches, which is unfortunate because over Haynsen et al. BMC Genomics (2018) 19:439 Page 4 of 17 Fig. 1 Contig length distributions of Trinity using the ConDeTri dynamic read trimmer Fig. 2 Transcriptome completeness of transcripts quantified through 956 universal single-copy orthologs using BUSCO Haynsen et al. BMC Genomics (2018) 19:439 Page 5 of 17 tropical kudzu (CPP27 vs. CPP02, Pmnk6 vs. CPP02, and CPP27/Pmnk6 vs. CPP02, respectively). The over 30,000 SNPs identified between CPP27/Pmnk6 vs. CPP02 is greater than the sum of SNPs from the individ- ual comparisons of P. montana var. lobata to N. phaseo- loides because the merged transcripts offer a more complete snapshot of a US kudzu transcriptome which was used as the reference for SNP detection. Lastly, we found 24,475 SNPs within kudzu from among three countries (Japan vs. Pmnk6(US)/CPP27(US)/ China). The majority of high-confidence SNPs were found within contigs rather than singletons (Table 3), which is expected given the fact that more highly expressed genes will be more likely to be represented by > 20× coverage (one of our criteria for high confidence) and are most likely to assemble into contigs. Also of note, the transi- tion/transversion ratio varied from 1.41 to 1.73 (Table 3) with the higher ratios found between the intergeneric Fig. 3 The number of unique and shared, homologous transcripts comparisons than the intraspecific comparisons. among kudzu CPP27 and Pmnk6 and tropical kudzu CPP02 transcriptome assemblies as ascertained via reciprocal best BLAST hit analyses eSSR discovery and characterization The eSSR analysis of the transcripts detected 5255 and 54, 56 and 66% of final annotated transcripts belong to 4586 perfect eSSRs for CPP27 and CPP02, respectively. the singletons in CPP02, CPP27, and Pmnk6, respectively The majority (76.7 and 76.8%) of eSSRs were (Table 2). In all three transcriptomes, the highest top hit tri-nucleotide repeats (TNRs; Table 4). After vetting for species for the annotated proteins were Glycine max (L.) primer design, there were 1646 potential eSSRs identi- Merr., G. soja Siebold & Zucc. and Cajanus cajan (L.) fied in P. montana var. lobata and 1459 in N. phaseo- Millsp., respectively (Additional files 3, 4,and 5). Summar- loides. Looking only at TNRs (1458 for CPP27 and 1273 ies of the biological process, cellular components and mo- for CPP02), 25 matches were found between P. montana lecular function categories for each transcriptome are var. lobata and N. phaseoloides in which either the for- shown in Fig. 4. ward or reverse primers were identical, suggesting hom- ology. However, no sets of primer pairs (forward and SNP discovery reverse primers together) were found duplicated be- We conducted pairwise tests for high-confidence SNP tween transcriptomes. Alterations to the non-identical discovery of the kudzu transcriptomes (Table 3, Add- primer pair within the 25 matches allowed for the itional files 6, 7, 8, 9 and 10). Our conservative assess- creation of 17 identical primer pairs between CPP27 ment of SNPs reduced thousands of high-confidence and CPP02. These 17 shared primer pairs represent SNPs to a lower number (Table 3) that are 1) one-to-one inter-generic phaseoloid eSSRs. Additionally, 13 TNR point mutations without length variants, 2) have vari- primer pairs specific to P. montana var. lobata were ation frequency over 95%, and 3) have a repeat depth of also selected for screening. Of the 30 total eSSR pri- 20 or more. As such, we identified 358 SNPs between mer pairs, 21 pairs were advanced to the Culley et al. the two US kudzu transcriptomes (CPP27 vs. Pmnk6), [19] protocol; of the nine primer pairs that were and 5185, 19,028, and 30,143 SNPs between kudzu and eliminated, four did not amplify a product, four Table 2 Summary of gene ontology analysis Accessions Transcripts Orfs Predictions BLAST Hits Annotated GO IDs ECs CPP27 79,194 37,741 30,716 28,795 18,446 8039 (18,325/60869) (13,534/17182) (12,583/16212) (7958/10488) Pmnk6 61,042 50,320 42,386 39,366 24,447 6337 (15,736/45306) (14,821/27565) (12,705/26661) (8079/16368) CPP02 92,406 34,223 27,661 22,472 13,230 4064 (18,412/73994) (14,677/12984) (10,407/12065) (6085/7145) Orfs open reading frames, GO gene ontology, ECs enzyme codes. Parentheses: (contigs/singletons) Haynsen et al. BMC Genomics (2018) 19:439 Page 6 of 17 Fig. 4 Gene ontology classifications of kudzu and tropical kudzu annotated transcripts. Numbers indicate the number of sequences associated with the particular GO term in each category amplified in an unexpected size range, and one displayed Population structure and genetic diversity of kudzu double banding (Additional file 11). Of the 21 primer pairs Three genetic units were determined to be the optimal that were assessed with the Culley et al. [19]protocol, value of K in STRUCTURE across the 75 accessions (K = 3, seven were discarded due to multiple banding and four Fig. 5, Additional file 12). The US is primarily composed of for lack of amplification, whereas a further three were a single genetic unit, with a couple individuals assigned to a removed due to the presence of monomorphic alleles second unit; whereas, China and Japan are more heteroge- (Additional file 11). The final set of eSSR primer pairs neous in their composition, yet they are still composed of identified seven polymorphic loci displaying single the same 2 units found in the US. Thailand, however, is bands of expected sizes (Table 5). composed of a single genetic unit that is unique to that Table 3 Single nucleotide polymorphism detection among kudzu and tropical kudzu genotypes a b c Comparison HC SNPs SNPs > 95% SNPs >20x Total SNPs Ts/Tv Pmnk6 vs CPP27 10,417 6016 426 358 1.41 (7494/2923) (4125/1891) (252/174) CPP02 vs CPP27 99,584 86,626 5831 5185 1.60 (81,276/18308) (70,638/15988) (5091/740) CPP02 vs Pmnk6 220,739 174,884 21,258 19,028 1.73 (164,118/56621) (127,311/47573) (19,255/2003) CPP02 vs Pmnk6, CPP27 314,416 248,719 33,603 30,143 1.71 (229,163/85251) (178,102/70617) (29,812/3791) Japan vs Pmnk6, CPP27, China 494,234 79,088 27,108 24,475 1.47 (494,234/0) (79,088/0) (27,108/0) SNPs with the > 95% frequency SNPs with > 95% frequency and > 20x coverage One-to-one point mutations after exclusion of indels and length variants; HC: high confidence; parentheses: (contigs/singletons) Haynsen et al. BMC Genomics (2018) 19:439 Page 7 of 17 Table 4 Transcriptome eSSRs sampled. Genetic structuring as assessed by pairwise F showed differences among groups, particularly in CPP27 CPP02 st Thailand and southern China (China 3; Table 8), cor- Transcripts 79,194 92,406 roborated by the structuring of genetic units shown Raw eSSRs 5255 4586 in Fig. 5. As defined by Wright [20], Thailand showed Dinucleotide 770 670 very great genetic variation (F > 0.25) with respect to st Trinucleotide 4032 3524 all other subpopulations, except China 3, with which Tetranucleotide 180 137 it showed great variation (0.15 < F < 0.25). The rest st Pentanucleotide 106 79 of the comparisons resulted in little to moderate genetic variation (0 < F < 0.05 and 0.05 < F < 0.15, respectively). Hexanucleotide 167 176 st st The neighbor-joining distance tree supports the pairwise Primered eSSRs 1646 1459 F results (Fig. 6): 1) Thailand is a distantly related lineage st Dinucleotide 14 28 to the nine other subpopulations representing P. montana Trinucleotide 1458 1273 var. montana and var. thomsonii; 2) the Chinese subpopu- Tetranucleotide 62 54 lations are divided into three lineages; and 3) the US sub- Pentanucleotide 41 25 populations are more genetically similar to Japan 2. Hexanucleotide 71 79 Discussion Invasive species are increasingly widening their scope nation, which supports our classification of its accessions as across the globe, yet the genetic mechanisms underlying being different varieties of P. montana, specifically var. invasiveness or weediness remain a mystery. In the gen- thomsonii and var. montana. omics era, scientists have raised a clarion call to arms to The national populations exhibited from 20 to 43 al- build genomic resources to study invasive species [21]. leles across a total of seven loci (Table 6), while the sub- Understanding the introduction history and relative gen- populations exhibited from 20 to 36 total alleles etic diversity of invasive species is an important step to (Additional file 13). China was composed of the greatest gaining a foothold on management and control, a goal number of alleles, in particular, China 3 (southern), requiring the development of variable molecular markers while Thailand was composed of the fewest number of such as microsatellites or SNPs to assess genetic diver- alleles. sity and population structure. In this study, we have as- After Bonferroni correction, none of the subpopulations’ sembled and characterized multiple transcriptomes of observed and expected heterozygosities significantly dif- the invasive Kudzu vine, Pueraria montana var. lobata, fered (Table 7), supporting the hypothesis that all the sub- and for tropical kudzu, Neustanthus phaseoloides, a spe- populations were in Hardy-Weinberg equilibrium when cies until recently thought congeneric with kudzu [2, 5]. Table 5 Seven eSSR primers optimized and used to assess population genetics in kudzu accessions Locus Sequence Dye/Tail SSR Length (bp) PP2 F: 5′-TAG GAG TGC AGC AAG CAT ATG CCG CGG ATC TTT GAA AG-3’ VIC /M13A AAC 100–130 R: 5’-CAA ATT GGC CCT GTC CCA AT-3’ n/a PP4 F: 5′-TGT AAA ACG ACG GCC AGT CAT GCC CAC GTG CTT CAT AG-3’ 6FAM/M13 GCT 100–140 R: 5’-CTC TCA GAT CCA GGC CCA AA-3’ n/a PP10 F: 5′-TAG GAG TGC AGC AAG CAT GGC ATG TAG ATC CAG CTA AA-3’ VIC/M13A GGT 310–330 R: 5′-TTG ACA GAT TTC TGA TTC TTG G-3’ n/a PP13 F: 5′-TAG GAG TGC AGC AAG CAT GAT TGA GCA GGC ACG AGA AC-3’ VIC/M13A GCT 270–300 R: 5’-CAG TAG CAG GCA TGT GTT GG-3’ n/a PL1 F: 5’-CAC TGC TTA GAG CGA TGC TGT AAG CGT TCG TTC GTT GG-3’ PET/M13B CTT 400–440 R: 5’-TCA ACC TGG TGC TCT CTG AC-3’ n/a PL7 F: 5′-TGT AAA ACG ACG GCC AGT AGT GGC CTT GCT CTT CTT CC-3’ 6FAM/M13 CTT 80–140 R: 5′-GTG TCA TCT CAG CAC GTT GG-3’ n/a PL11 F: 5′-TGT AAA ACG ACG GCC AGT TGG CAT CAT CCT TCA ACC AC-3’ 6FAM/M13 ACC 300–330 R: 5′-ATT CGG GAA TAG TGG GTG GG-3’ n/a F forward primer, R reverse primer. Dyes VIC: 2′-chloro-7′phenyl-1,4-dichloro-6-carboxy-fluorescein; 6FAM: 6-carboxyfluorescein; PET: chemical structure currently unpublished as proporietary to Lifetech. Tail: see Culley et al. [19] for information about M13, M13A, and M13B Haynsen et al. BMC Genomics (2018) 19:439 Page 8 of 17 Fig. 5 STRUCTURE diagram of 75 P. montana accessions across four nations (K = 3) Kudzu is well known as an invasive species in both agri- consideration when dealing with potentially polyploid cultural and natural areas due to its fast growth, clonal plants [23, 25, 26]. Pueraria is descendent from an ancient habit, and extensive introductions outside its native range. polyploidy event that transpired 50–60 mya near the ori- Tropical kudzu is also known to be invasive in its intro- gin of the papilionoid subfamily [27, 28], creating a dupli- duced ranges, but to a lesser extent. We mined our tran- cated genomic complement that has fractionated over scriptomes of these two species for molecular markers time but whose signature still remains within descendent (eSSRs and SNPs), screened and validated eSSRs, and per- genomes. Longer reads are more likely to unambiguously formed functional annotations of the transcriptomes, im- assemble or align across homoeologues, duplicated genes proving the genetic resources available for kudzu and produced via allopolyploidy [29]. Furthermore, the longer tropical kudzu. reads result in the sequencing of more full-length mRNA transcripts, an outcome that argues for including single- Transcriptome characterization tons (those reads that do not assemble into contigs) in the Whether researching model or non-model organisms, se- overall transcript complement. Although pyrosequencing quencing the transcriptome of a species is a natural begin- produces fewer overall reads as compared to Illumina, its ning for genome-wide resource development and study ability to produce longer transcripts is advantageous, par- [22, 23], enabling the characterization of gene expression ticularly for allopolyploid species and other hybrids where profiles, genetic marker discovery, and phylogenetic infer- avoiding the assembly of chimeric sequences is important. ence [24]. Here, we characterize the transcriptomes of two The comparative results across our transcriptomes in accessions of kudzu, one wild-collected (Pmnk6) and one terms of the number of transcripts discovered and the partially inbred line propagated by the USDA agriculture relative overlap among pairwise comparisons provides research service (CPP27), as well as one of tropical kudzu some insights into the relative impact of environment vs. (CPP02). We chose to use 454 pyrosequencing technology shared ancestry. CPP02 had the highest number of tran- over Illumina due to the longer read lengths, an important scripts and the highest number of unique transcripts, with Pmnk6 having the least number of transcripts, even though it presents the best transcriptome in terms of Table 6 Allelic frequency for Pueraria national populations mean contig length, N50, and BUSCO results. One ex- Locus USA China Japan Thailand Mean SD Total planation involves the number of tissues used for se- N =25 N =21 N =22 N =7 quencing. CPP02 utilized three tissues (young leaves, PP2 8 7 6 4 6.25 1.71 9 young shoot tips, and buds) while CPP27 used two tis- PP4 4 5 7 3 4.75 1.71 9 sues (young leaves and young shoot tips), and Pmnk6 PP10 5 5 6 3 4.75 1.26 8 used a single tissue (young leaves). Given this informa- PP13 3 7 4 2 4.00 2.16 7 tion, it makes sense that the transcriptome that was composed of the greatest number of tissues resulted in PL1 4 4 2 4 3.50 1.00 9 the highest number of unique transcripts due to expres- PL7 8 8 11 3 7.50 3.32 15 sional differences across tissue types. CPP02 and CPP27 PL11 5 7 3 1 4.00 2.58 7 shared the highest number of reciprocal best BLAST hits Mean 5.29 6.14 5.57 2.86 4.96 1.96 9.14 (RBH). However, one would expect the two kudzu acces- SD 1.98 1.46 2.99 1.07 1.42 0.80 2.73 sions (CPP27 and Pmnk6) to share the greatest number of Total 37 43 39 20 34.75 13.73 64 overlapping transcripts due to shared ancestry. This could N number of accessions, SD standard deviation also be explained by the fact that the two transcriptomes Haynsen et al. BMC Genomics (2018) 19:439 Page 9 of 17 Table 7 Observed and expected heterozygosities for Pueraria subpopulations US 1 US 2 US 3 CN 1 CN 2 CN 3 JP 1 JP 2 JP 3 TH # Individuals 8 10 75887 877 Obs. Het. 0.717 0.552 0.472 0.611 0.378 0.632 0.396 0.506 0.656 0.594 Exp. Het. 0.643 0.503 0.547 0.648 0.589 0.763 0.579 0.572 0.661 0.583 HWE p-value 0.766 0.251 0.611 0.765 0.079 0.392 0.013 0.429 0.869 0.442 US United States, CN China, JP Japan, TH Thailand, Obs: Observed, Exp Expected, Het Heterozygosity, HWE Hardy-Weinberg Equilibrium that shared the most homologous tissues resulted in the have yet to be described, perhaps including species-specific highest number of shared transcripts. Alternatively, the “orphan” genes [34]. Catalytic activity, binding, metabolic seeming disparity in shared best BLAST hits could be ex- and cellular processes were among the most highly repre- plained by the relative impacts of a shared environment, sented groups regarding GO analysis (Fig. 4)acrossall three which often affects gene expression. Our two CPP tran- transcriptomes, as expected given that we used young tis- scriptomes were both grown in the same greenhouse en- sues that are undergoing extensive metabolic activities. vironment at the same time and so their gene expression profiles may be expected to be more similar than those of Single nucleotide polymorphism discovery the two P. montana var. lobata accessions, one of which SNPs are fast becoming the marker of choice due to was grown in the greenhouse (CPP27) and one in the wild their ease of discovery via next generation sequencing (Pmnk6). A similar finding was discovered across tran- technologies [35]. Additionally, the ease of mining SNPs scriptomes of Eutrema salsugineum (Pall.) Al-Shehbaz & from previously produced transcriptomes can provide a Warwick plants that were grown in field (uncontrolled en- new use for previously published data sets that may be vironment) vs. cabinet (controlled environment) conditions, sitting idle in online repositories. SNPs, though less with the plants grown in the controlled environment shar- polymorphic than SSRs, may provide higher resolution ing a higher number of expressed genes as compared to the assessment of genetic variation and identification of more geographically proximate plants grown in differing population structure [36]. We detected a near 100-fold environments [30]. increase in the number of SNPs detected between kudzu In this study, we were able to annotate over 13,000 tran- and tropical kudzu as compared to that detected within scripts from kudzu and tropical kudzu (Table 1). Our tran- kudzu. SNPs discovered between kudzu and tropical scriptomes do not provide a full gene complement due to kudzu may represent species level, fixed differences be- low sequencing depth as evidence by our BUSCO results tween these genera. Validation of these SNPs is beyond (Fig. 2). However, the level of unannotated transcripts in the scope of this paper; nevertheless, this list presents a this study is similar to results reported from other significant resource for future work in genetic diversity non-model legumes, like winged bean [31], chickpea [32], assessment, genetic mapping, genome-wide association and field pea [33]. The unidentified transcripts are likely mapping, or evolution-based studies of invasiveness, and due to 1) correspondence to non-coding regions or pseudo- marks the first SNP markers discovered to date in Puer- genes, 2) short length of transcripts, or 3) coding genes that aria and Neustanthus. Use of these SNP markers across Table 8 Subpopulation pairwise F st US 1 US 2 US 3 CN 1 CN 2 CN 3 JP 1 JP 2 JP 3 TH US 1 – 0.811 0.541 0.441 0.009 0.000* 0.297 0.126 0.099 0.000* US 2 −0.023 – 0.378 0.730 0.009 0.000* 0.432 0.108 0.072 0.000* US 3 − 0.011 − 0.008 – 0.306 0.009 0.000* 0.360 0.946 0.153 0.000* CN 1 − 0.009 − 0.022 0.024 – 0.108 0.009 0.901 0.162 0.108 0.000* CN 2 0.075 0.098 0.099 0.075 – 0.297 0.207 0.081 0.739 0.000* CN 3 0.077 0.107 0.120 0.073 0.022 – 0.063 0.000* 0.324 0.000* JP 1 0.015 −0.002 0.022 −0.035 0.051 0.064 – 0.207 0.486 0.000* JP 2 0.016 0.025 −0.030 0.049 0.078 0.085 0.042 – 0.135 0.000* JP 3 0.029 0.028 0.042 0.037 −0.014 0.006 0.002 0.036 – 0.000* TH 0.315 0.370 0.377 0.330 0.310 0.244 0.347 0.352 0.322 – Below diagonal pairwise F values, above diagonal p-values st US United States, CN China, JP Japan, TH Thailand * = significant under Bonferroni correction (p < 0.001) Haynsen et al. BMC Genomics (2018) 19:439 Page 10 of 17 Japan2 USA3 USA2 USA1 Japan1 China1 Japan3 46 China2 China3 Thailand 0.05 Fig. 6 Neighbor joining distance tree based on F values and 10,000 bootstraps. US = United States; CN = China; JP = Japan; and TH = Thailand st a wide population-level sampling throughout Asia would population every few meters, suggesting that they treated a enable a robust investigation into the introduction his- patch of kudzu as a population, whereas we sampled indi- tory of kudzu within the US. viduals no closer than ~ 1 km apart, and viewed a popula- tion as a regional area comprised of numerous, eSSR marker discovery and validation non-connected patches. With the abilities to grow over eSSRs are routinely developed from transcriptomic data, 12 in. per day and root at the nodes, a kudzu patch may providing a ready source for genetic diversity assessment likely represent only one or a few genets [43]. Therefore, through cost-effective means [37]. In spite of being derived the reported clonal sampling of Bentley and Mauricio [15] from coding DNA, which is evolutionarily conserved, may be the cause of the near 0.0 observed heterozygosities eSSRs have proven a variable and valuable resource for gen- and may not be indicative of the primers themselves. etic studies [18]. In our study, we detected ~ 5000 eSSRs each within kudzu and tropical kudzu.Overall,trinucleo- Genetic diversity of kudzu tide SSR motifs (TNRs) were the most abundant, as found For the past two decades, the genetic diversity of kudzu consistently in other plant studies [17, 38–42]. Presumably has been assessed with the various molecular markers of this is because TNRs will not affect the open reading frames the corresponding era. For instance, Pappert et al. [10] of coding regions [38]. We investigated the utility of 30 used 13 allozymes across 1000 US accessions to con- eSSR markers discovered in our data and optimized seven clude that introduced kudzu possessed considerable gen- for use across kudzu. When compared to the etic variation with a lack of geographic structuring. kudzu-derived SSR markers of Hoffberg et al. [14], similar- Similar conclusions were subsequently reached by Jewett ities and benefits are found. For instance, Hoffberg et al. et al. [11] using 18 random amplified polymorphic DNA [14] assessed their 15 genomic SSRs against 102 geograph- (RAPD) markers across 50 accessions from the US and ically dispersed individuals, finding that their alleles per China, and by Sun et al. [12] using 11 inter-simple se- locus ranged from 2 to 8, whereas our alleles per locus quence repeat (ISSR) markers across 108 accessions ranged from 7to15(Table 6). This comparison shows from the US and China. A decade later, Bentley and twice as many alleles within a smaller sample size, approxi- Mauricio [15], using 15 SSRs and one chloroplast mately two-thirds the size of Hoffberg et al. [14]. One ex- marker across 1747 US accessions, reported that the planation for the difference in allele numbers could be high levels of genetic diversity result from high clonal attributed to the differing sampling ranges, with our indi- reproduction in kudzu, as described by Ellstrand and viduals being collected from a greater global area. However, Roose [44], Balloux et al. [45], and Halkett et al. [46]. when Bentley and Mauricio [15] used the Hoffberg et al. Specifically, high levels of genetic variation are expected [14] primers on 1747 accessions of kudzu from solely the in clonal populations when the populations were US they identified 2–17 alleles per locus, which also repre- founded by sexual propagules [44], which can be the sents a doubling of alleles but in a smaller sampling area. case even if recruitment of sexual offspring into estab- Additionally, when our observed heterozygosities are com- lished populations is rare. This may be the case for pared to the primers of Hoffberg et al. [14], they ranged kudzu due to its deliberate introduction by landowners from 0.372–0.726 (Table 7), while Hoffberg et al. [14] into novel habitats from seed stock. Additionally, clonal ranged from 0.0–0.9 and Bentley and Mauricio ranged from populations are capable of maintaining higher genetic 0.004–0.741. The large difference in the heterozygosity diversity at each locus even though they support a lower comparisons, particularly when focusing on the low end, number of different genotypes [45, 46]. Our results cor- may be attributed to differences in sampling strategies. roborate the findings that introduced kudzu displays Bentley and Mauricio [15] report sampling kudzu within a high levels of genetic variation throughout the US (Table Haynsen et al. BMC Genomics (2018) 19:439 Page 11 of 17 6, Additional file 13); however, we still maintain that the and confidence for testing genetic associations between high genetic variation is possibly indicative of multiple introduced and native kudzu, efforts that are currently introductions from across its native range. underway. Population structure and introduction history of kudzu Conclusions Kudzu is said to have first been brought to the US by the This study produced critical genomic resources for the Japanese who planted it as an ornamental vine outside highly invasive kudzu vine by characterizing transcrip- their pavilion at the 1876 World’s Fair Centennial Exhib- tomes and producing marker databases for SNP and ition in Philadelphia [47]. Later, David Fairchild, a plant eSSR markers, foundational resources for understanding explorer for the United States Department of Agriculture, ecological adaptation that may enable future insights noted its uses, including as forage, in Japan and brought into invasiveness through gene discovery, marker-trait back some seeds to plant near his home in Washington, analyses, and further genetic diversity studies. We exem- D.C., as a trial. In the 1930’s, the US government began plified the utility of our marker databases by assessing planting millions of seedlings across the southeastern the genetic diversity of native and introduced popula- states as a means of erosion control. Whether the US gov- tions of kudzu using seven eSSRs. As a naturalized inva- ernment sourced these kudzu seedlings from one or mul- sive vine that was intentionally introduced throughout tiple native populations from Japan or elsewhere is not millions of acres of the southeastern US, kudzu presents known. unique challenges for management, especially given its Although there is consensus across most studies show- high genetic diversity across the US, a finding supported ing robust findings of high levels of genetic variation of by our genetic diversity analyses. The origin of this gen- kudzu in the US, most of the studies reported a lack of etic diversity remains a matter of speculation, however, geographic patterning of genotypes, and none included this study has begun to refine the proposed hypothesis wide sampling across Asia so as to enable an investiga- of single or multiple introductions from different genetic tion into source populations of US introduction(s). Our populations. This study is the first to provide molecular results include new clues in identifying the native origins evidence that indicates the island of Honshu, Japan as of US kudzu. The Thailand subpopulation is composed one source of US kudzu. Our analyses suggest either a of non-P. montana var. lobata individuals. With evi- single introduction from a highly diverse source popula- dence for strong genetic differentiation and zero popula- tion in Japan, or more likely multiple introductions from tion admixture between Thailand and other multiple sources, potentially also from northern Japan subpopulations, we can definitively rule Thailand out as (Island of Hokkaido) or northern China. Given the eco- a source of US kudzu introductions. It may also be pos- logical and economic devastation wrought by kudzu in sible to rule southern China out as an origin of US the United States, it is critical that we improve our un- kudzu introductions due to pairwise comparisons with derstanding of the history, process, origin(s), and im- the central and southern US, which showed moderate pacts of the U.S. kudzu invasion. We have assembled levels of genetic variation (Table 8), as well as the distant transcriptomes and mined them for eSSRs that we have placement of China 3 on the NJ tree (Fig. 6). provided as a resource for further genetic studies into Of particular interest in the investigation of source the origin(s) and range expansions of kudzu to that end. populations for the introduction of US kudzu is the NJ By increasing both the sample ranges and sizes it should tree clade composed of all the US subpopulations and be possible to identify more accurately the origin of Japan 2, the centrally located Japanese subpopulation introduction and the number of introductions with the (Fig. 6). With a bootstrap value of 50, these four subpop- markers we have developed, efforts that are currently ulations can be distinguished from the rest of the tree underway. and within this clade, Japan 2 and US 3, the southern US, are paired together with a support of 92. These find- Methods ings suggest that central Japan is a source of US kudzu. Plant material for transcriptome sequencing and Its association with US 3, the southern US populations, population genetics makes sense considering that this area was where kudzu Transcriptomic work in this study incorporated plant was first planted for soil erosion control and where tissues from two accessions of kudzu, P. montana var. farmers cultivated kudzu for fodder at the behest of the lobata, and one accession of tropical kudzu, N. phaseo- US government. Our study is the first to provide mo- loides [formerly Pueraria phaseoloides (Roxb.) Benth.]. lecular evidence to support the hypothesis of Japan as a One kudzu accession (noted here as Pmnk6) was wild genetic source of US kudzu. However, a wider sampling collected from Williamsburg, Virginia [voucher speci- across the native Asian range coupled with higher num- men G. Tate s.n. (WILLI) collected 8 July 2013]. Leaf tis- bers of genetic markers would increase statistical power sue was collected in RNALater and preserved at − 20 ° F Haynsen et al. BMC Genomics (2018) 19:439 Page 12 of 17 prior to RNA extraction. The other two plants were quality score thresholds (hq) were set to 25 and low grown from seed obtained from the United States De- quality score thresholds (lq) were set to 10; the fraction of partment of Agriculture (USDA) Germplasm Resources bases per read having to exceed hq were set to 0.8 and the Information Network seed bank: accession PI 434246 of minimum number of high quality bases (mh) and max- P. montana var. lobata (noted here as CPP27) was field imum number of low quality bases (ml) within the sliding collected in 1979 from the United States, locality un- window were set to 30 and 5, respectively. Cleaned reads known, and is maintained by the Coffeeville Plant Mate- were de novo assembled using Trinity (v2.0.6) [51]under rials Center, Soil Conservation Service, Coffeeville, MS; default parameters on two high-performance computing accession PI 470272 of N. phaseoloides (noted here as clusters: the Smithsonian Institution High Performance CPP02) was donated in 1981 from a field collection by Cluster (SI/HPC) and the George Washington University D.R. Bienz, 5 Jun 1981, Banjarbaru, S. Kalimantan, Colonial One Cluster. In order to minimize redundant Indonesia. Seeds were grown to maturity in the green- transcripts, a by-product of the assembly process, house at Cornell University (Ithaca, NY, US) for 3 years CD-HIT-EST was used with a threshold of 0.9 to obtain prior to RNA extraction. For eSSR screening and popu- unique transcripts [52]. To evaluate the quality of the as- lation genetic studies, we sampled 75 accessions repre- semblies, criteria including the number of aligned reads, senting all three varieties of P. montana throughout total number of contigs produced, mean contig length, their native and US introduced range: US (25), China N50, and transcript annotations were considered. RSEM (21), Japan (22) and Thailand (7) (Additional file 14). [53]and Bowtie2[54] were used to identify the number of Leaf material was immediately stored in silica for desic- aligned reads in the assembled transcriptomes. The KRA- cation. Genomic DNA was extracted from samples using KEN suite was utilized in conjunction with prokaryote and Autogen robotics (Autogen Inc.) and a modified CTAB fungal databases to identify potential contaminants within extraction protocol [48]. the transcriptomes [55]. BUSCO(v1.1b1), apipelineused to accurately annotate core genes in eukaryotic genomes, RNA extraction and transcriptome sequencing was used to determine the completeness of the assemblies For the two accessions raised in the greenhouse, tissues [56]. At the time of use, BUSCO utilized a plant core data- were flash frozen in liquid nitrogen prior to RNA extrac- base of 956 single copy genes that are shared between Ara- tion. Neustanthus phaseoloides (CPP02) was sampled for bidopsis, Oryza, Populus,and Vitis [57]. Reciprocal Best young leaves, young shoot tips, and buds. Unfortunately, BLAST Hits (RBH) between transcripts and among tran- kudzu never flowered in the greenhouse, so only young scripts were performed on a local installation of Galaxy shoot tips and young leaves were harvested for CPP27. [58–60]and Toolshed [61] to characterize the number of For the wild collected kudzu (Pmnk6), only young leaves shared, homologous transcripts recovered in each Trinity were harvested. RNA extraction, cDNA library construc- assembled transcriptomes [62, 63]. tion, and transcriptome sequencing were carried out as previously described [31]. cDNA libraries from CPP27 Functional annotation of transcriptomes and CPP02 were multiplexed with two other libraries We used transcripts (contigs + singletons) assembled by not reported here across one titer plate on the Roche Trinity to annotate our transcriptomes (CPP27, CPP02, 454 Genome Sequencer FLX platform using Titanium and Pmnk6). To identify candidate coding regions, we chemistry at the Brigham Young University Sequencing filtered sequences based on a minimum amino acid Center (Provo, UT, US). Pmnk6 was also multiplexed length of 100 using the TransDecoder program v2.0.1 with three other transcriptomes not reported here and [64] with the TransDecoder.LongOrfs command. BlastP sequenced using Roche 454 pyrosequencing, but using and Pfam searches were carried out to detect open Roche’s next improvement on the titanium chemistry reading frames (ORFs) with similarity to known pro- that produced reads ~ 800 bp long. The raw sequence teins and to maximize sensitivity for capturing ORFs data generated from CPP27, Pmnk6, and CPP02 were that may have functional significance. The BlastP deposited at the National Center for Biotechnology Infor- search was done using the Swissprot database with mation (NCBI) Sequence Read Archive (SRA) under acces- the E-value of 1E-5 and Pfam search was done using sion numbers SRR5925648, SRR5925647, and SRR5925649, HMMER [65]and thePfamdatabase[66]. Output respectively. files from the BlastP and Pfam searches were used to ensure that peptides with BLAST or domain hits were De novo transcriptome assemblies retained by running the TransDecoder.Predict com- Raw reads were assessed for quality with FastQC [49] mand. The peptide sequences from the final candidate and subsequently cleaned with ConDeTri [50], a ORFs were used to run BlastP searches against the content-dependent read trimmer under the following NCBI’s nonredundant (nr) database with the E-value settings: reads below 50 bp were removed, Phred high of 1E-5 on the SI/HPC. The BLAST results were then Haynsen et al. BMC Genomics (2018) 19:439 Page 13 of 17 Fig. 7 Sampling sites: (a) United States: US 1, US 2, US 3 (25); (b) Japan: JP 1, JP 2, JP 3 (22); and (c) China: CN 1, CN 2, CN 3 (21) and Thailand: TH (7) imported into the Blast2GO program v1.9.3 [67] to as- Single nucleotide polymorphism identification sign Gene Ontology (GO) terms. We ran mapping, anno- For SNP identification among the kudzu accessions, we tation and InterProScan analyses for the three used the transcripts (contigs + singletons) from our CPP27, transcriptomes separately. Pmnk6, and CPP02 assemblies and also incorporated two Haynsen et al. BMC Genomics (2018) 19:439 Page 14 of 17 publicly available P. montana var. lobata transcriptomes, [69] transcriptomes were not utilized for eSSR discovery SRX480408 from China derived from two tissues [68], and because none were available at the time eSSR mining took DRA001736 from Japan consisting of five pooled tissues place. Thirty potential eSSR primer pairs were chosen [69]. We assembled the public sequences using Trinity as from those discovered here and initially screened against a described above. Multiple pairwise comparisons between subset of accessions (Additional file 11). Seventeen of the transcriptomes were conducted to evaluate the distribu- 30 primer pairs represent putatively homologous eSSRs tion of SNPs between US kudzu samples (CPP27 vs. present in both P. montana var. lobata and N. phaseo- Pmnk6) and identify intergeneric SNPs between kudzu loides (primer pairs designated PP) while the rest are P. and N. phaseoloides (CPP27 vs. CPP02 and Pmnk6 vs. montana var. lobata specific (primer pairs designated PL). CPP02). Additionally, the two US kudzu samples were The method of Culley et al. [19] was used to screen, combined by concatenating the two transcript files such optimize and amplify eSSRs. Primer pairs were eliminated that the samples represent the diversity in ‘US kudzu’ and based on the Culley et al. [19] protocol if they produced subsequently compared to tropical kudzu to further iden- superfluous primer diming between the specific and tailed tify intergeneric SNPs (CPP27/Pmnk6 vs. CPP02). Lastly, primers or produced PCR products of unexpected size. SNPs were called via comparison of all four P. montana Primer pairs were further eliminated if 1) primers did not var. lobata transcriptomes, with the transcriptome from amplify viable product as seen via gel electrophoresis, 2) Japan used as reference (Japan vs. CPP27/Pmnk6/China). primers amplified more bands than expected, or 3) The Japan transcriptome was chosen as reference because primers were monomorphic. it incorporated the highest number of tissues, thus puta- Screening of primer pairs against a subset of seven ac- tively having the higher chance of capturing greater cessions ultimately yielded seven primer pairs that were expressed sequence diversity within the genome. To call characterized across all 75 accessions. Primers, fluores- SNPs, GS Reference Mapper v2.9 (454 Life Sciences, cent dyes, and Culley method tail adaptors used for each Roche, US) was used under default settings. The transcrip- of the seven eSSRs are listed in Table 5. Initial rounds of tome composed of the greatest number of tissues was used amplification across the entire sampling set were per- as the reference to which reads from the others were as- formed in 12 μL reactions containing 1X Biolase NH sembled against. We used only high-confidence variants buffer, 1.0 μL primer mix, 1.2 mM MgCl , 0.12 μLof (454HCDiffs, > 95%) in each comparison and further fil- 8 μM dNTPs, 0.35 U of Taq polymerase (Biolase), and tered these variants to those having 20× or greater cover- 5-80 ng DNA template. PCR was performed on an Ap- age. To ensure the highest SNP call quality, we discarded plied Biosystems 2720 thermocycler with settings of 95C any SNPs where 1) the reference or variant involved one or for 5 min, followed by 35 cycles of 95C for 30s, 50C for more N’s or 2) the reference or variant allele was a single 45 s, 72C for 30s, and a final 72C extension for 5 min. nucleotide insertion or deletion or did not include a point Annealing temperatures were adjusted between mutation in the length variant [70]. 51.5C-58C for primers PP13, PL1, PL11, and PP2. Prod- uct bands were resolved using 1.5% sodium borate gels Expressed simple sequence repeat (eSSR) loci discovery, containing GelRed stain and visualized under UV light. screening and characterization Accessions that failed to amplify after two or more initial The ConDeTri cleaned, Trinity assembled, and attempts were subsequently attempted with an adjusted redundancy-vetted transcripts of CPP27 and CPP02 concentration of 2.38 μg MgCl per reaction. Further were mined for di-, tri-, tetra-, penta-, and hexanucleo- failed amplifications were then tried using AmpliTaq Gold tide microsatellites with MSATCOMMANDER [71]. using reaction mix 1X AmpliTaq buffer, 1.0uL of primer Afterwards, MSATCOMMANDER and Primer3 [72] mix, 2.86 μg MgCl , 1.2uL of 8 μMdNTPs,0.375 Uof were used to design primer pairs for each species with AmpliTaq Gold Taq polymerase [0.075 μLof1000 U in an expected product size ranging from 100 to 450 bp. 200 μL], and 5-80 ng DNA template. Successful products Primer lengths were allowed to range from 18 to 22 bp, were genotyped using an ABI3730 sequencer at the annealing temperatures were optimized at 60 °C, and Smithsonian NMNH LAB facilities. Genotypes were called GC contents were held between 30 and 70%. Developed using GeneMapper (v5.0) [74]. primers for both species were then cross-compared to identify homologous primer regions, which could signify Examination of population structure and genetic diversity interspecies transferability. The corresponding tran- indices scripts for primers that were shared between P. lobata Genetic population structuring was assessed with and N. phaseoloides were blasted against the GenBank STRUCTURE v2.3.4 [75] and STRUCTURE HAR- nonredundant database using BLASTX [73]with an VESTER v0.6.94 [76]. The length of burnin period was − 10 E-value of 10 to determine the function of their associ- set to 100,000, while the number of MCMC reps after ated unigenes. Pmnk6, SRX480408 [68] and DRA001736 burnin was set to 900,000, resulting in a total of 1 Haynsen et al. BMC Genomics (2018) 19:439 Page 15 of 17 million generations. No LOCPRIOR information was with variant; # Fwd Total: number of forward-aligned reads total; # Rev. provided for the STRUCTURE runs. A job consisting of Total: number of reverse-aligned reads total. (XLSX 578 kb) 10 iterations, evaluating Ks from 1 to 10 for the 75 P. Additional file 7: SNPs_CPP02_vs_CPP27. High-confidence single nu- cleotide polymorphisms between tropical kudzu CPP02 (reference: Ref) montana accessions, was run and the results were and kudzu accession CPP27 (variant: Var). Abbreviations as described for uploaded to STRUCTURE Harvester for analysis. The Additional file 6. (XLSX 4932 kb) optimal K was assessed via the Evanno et al. [77] Additional file 8: SNPs_CPP02_vs_Pmnk6. High-confidence single nu- method. Individual and population files were loaded into cleotide polymorphisms between tropical kudzu CPP02 (reference: Ref) and kudzu accession Pmnk6 (variant: Var). Abbreviations as described for CLUMPP v1.1.2 [78] to address label switching and the Additional file 6. (XLSX 11073 kb) potential for multimodality across the 10 STRUCTURE Additional file 9: SNPs_CPP02_vs_Pmnk6_CPP27. High-confidence sin- iterations. The CLUMPP program utilized the FullSearch gle nucleotide polymorphisms between tropical kudzu CPP02 (reference: method, the number of individuals in each population Ref) and a composite transcriptome comprising reads from kudzu accessions CPP27 and Pmnk6 (variant: Var). Abbreviations as described for influenced weights, and the pairwise matrix similarity Additional file 6. (XLSX 11520 kb) statistic was set to G’. All additional options remained as Additional file 10: SNPs_Japan_vs_Pmnk6_CPP27_China. High- default settings. CLUMPP outputs for the individual and confidence single nucleotide polymorphisms among kudzu accessions population files were visualized with DISTRUCT v1.1 from Japan (reference: Ref) and reads from US kudzu (Pmnk6 and CPP27) and China (variants: Var). Abbreviations as described for Additional file 6. [79]. Genetic diversity statistics were calculated in Arle- (XLSX 30817 kb) quin v3.5.1.9 [80]. The default parameters of Arlequin Additional file 11: Table S3. Thirty primer pairs tested for polymorphic were used on our 75-individual data set that was subdi- amplification in Pueraria montana. Primers labeled PP were designed vided from the four sampled nations to 10 geographic- from kudzu and tropical kudzu transcriptomes whereas those designated PL were designed from kudzu only. Bold primers are those used for ally defined subpopulations: US (3), China (3), Japan (3), population genetic analyses in this study. F: forward primer; R: reverse and Thailand (1) (Fig. 7). The subpopulation designa- primer. (PDF 33 kb) tions were based primarily on geographic proximity that Additional file 12: Figure S4. Delta K of STRUCTURE run (K = 3). Plot of allowed for groupings of at-least five individuals along Delta K for STRUCTURE analyses from K = 2 through K = 9, with K = 3 seen as the optimal number of genetic clusters. (PDF 18 kb) similar latitudinal lines; however, due to the different Additional file 13: Table S4. Allele table for Pueraria subpopulations. scales of sampling done across nations, the ranges of the Number of alleles discovered for each locus within each subpopulation, latitudinal boundaries of the subpopulations differed. with mean and standard deviation (SD) for each subpopulation and each POPTREEW [81] was used to make a neighbor joining locus. (PDF 19 kb) (NJ) distance tree with F distances [82] for the above Additional file 14: Table S5. Plant material used for eSSR validation st and population genetics. Species determination, subpopulation listed subpopulations. Bootstrap support for the tree was designation (pop), country and state/province/island of origin within the calculated with 10,000 replicates. United States (US), China (CN), Japan (JP) or Thailand (TH), voucher information, accession number, and geographical coordinates for each of the 75 plants used in the population genetic analyses. (PDF 34 kb) Additional files Additional file 1: Table S1. Trinity contig reads mapped back to the Abbreviations raw and cleaned reads. Numbers of cleaned and raw reads mapped back BLAST: Basic local alignment search tool; bp: Base pair; to contigs via Bowtie2. (PDF 126 kb) BUSCO: Benchmarking universal single-copy orthologs; eSSR: Expressed simple sequence repeat; GO: Gene ontology; hq: High quality; lq: Low Additional file 2: Table S2. Contaminated reads as assessed by Kraken. quality; mh: Minimum high quality; ml: Maximum low quality; NCBI: National Number (percentage) of cleaned reads annotated by Kraken as Center for Biotechnology Information; nr: Nonredundant; ORF: Open reading prokaryotic or fungal. (PDF 126 kb) frame; RBH: Reciprocal best hits; RIN: RNA integrity; SI/HPC: Smithsonian Additional file 3: Figure S1. CPP27 Top-Hit Species Distribution. Top-hit Institution High Performance Cluster; SNP: Single nucleotide polymorphism; species distribution of CPP27 proteins annotated against NCBI’s SRA: Sequence read archive; SSR: Simple sequence repeat; TNR: Tri- non-redundant database showing the highest distribution of hits against nucleotide repeat legume species. (PDF 808 kb) Additional file 4: Figure S2. Pmnk6 Top-Hit Species Distribution. Top-hit species distribution of Pmnk6 proteins annotated against NCBI’s Acknowledgements non-redundant database showing the highest distribution of hits against We thank Susan Sherman-Broyles and Jane L. Doyle for help in sustaining legume species. (PDF 753 kb) plants in the greenhouse and to Beth Chambers and Gus Tate, Herbarium of the College of William and Mary, for help in obtaining voucher specimens Additional file 5: Figure S3. CPP02 Top-Hit Species Distribution. Top-hit for Pmnk6. Additionally, we thank Cheng-Xin Fu, Lu-Xian Liu, Xin-fen Gao species distribution of CPP02 proteins annotated against NCBI’s non- and Bo Xu for assistance collecting in China, Tetsukazu Yahara, Tadashi Kajita, redundant database showing the highest distribution of hits against leg- Firouzeh Javadi, Tomoko Otao and Yumi Kagawa for help in Japan, and ume species. (PDF 2607 kb) Voradol Chamchumroon, Kongkanda Chayamarit, Thaveechok Jumruschay Additional file 6: SNPs_Pmnk6_vs_CPP27. High-confidence single nu- Rumrada Meeboonya, Nannapat Pattharahirantricin, Rachun Pooma, Sukontip cleotide polymorphisms between US kudzu accessions Pmnk6 (variant: Sirimongkol, and Ruth P. Clark for help in Thailand. Computations were Var) and CPP27 (reference: Ref). Accno: contig in reference; Pos: position; completed in part on the Smithsonian Institution High Performance Cluster Nuc: nucleotide; Total Depth: number of variant reads aligned against the (SI/HPC) and the George Washington University Colonial One Cluster. We reference; Var Freq: frequency of variant SNP within aligned reads; # Fwd: also thank the Computational Biology Institute at the George Washington number of forward reads with variant; # Rev.: number of reverse reads University for graduate support for MSH. Haynsen et al. BMC Genomics (2018) 19:439 Page 16 of 17 Funding 6. Forseth IN Jr, Innis AF. Kudzu (Pueraria montana): history, physiology, and This research was supported by funding from the US National Science ecology combine to make a major ecosystem threat. Crit Rev Plant Sci. Foundation to ANE (DEB-1352217) and JJD (DEB-0948800). 2004;23:401–13. 7. Follak S. Potential distribution and environmental threat of Pueraria lobata. Availability of data and materials Cent Eur J of Biol. 2011;6:457–69. The transcriptomes generated and analyzed during the current study are 8. Westbrooks R. Invasive plants, changing the landscape of America: fact available in the NCBI repository, [Study PRJNA397892, accessions: book. In: Federal Interagency Committee for the Management of Noxiuous SRR5925647, SRR5925648, and SRR5925649, http://www.ncbi.nlm.nih.gov/ and Exotic Weeds: Washington; 1998. bioproject/397892, release date 30 June 2018]. The SNP data generated 9. Kudzu SD. In: Simberloff D, Rejmanek D, editors. Encyclopedia of biological during this study are included in this published article’s Additional files 6, 7, invasions. California: University of California Press; 2011. p. 396–9. 8, 9 and 10 however, the SNPs contained in Additional file 10 are not 10. Pappert RA, Hamrick JL, Donovan LA. Genetic variation in Pueraria lobata publicly available due to file size restrictions but are available from the (Fabaceae), an introduced, clonal, invasive plant of the southereastern corresponding author upon reasonable request. United States. Am J Bot. 2000;87:1240–5. 11. Jewett DK, Jiang CJ, Britton KO, Sun JH, Tang J. Characterizing specimens of Authors’ contributions kudzu and related taxa with RAPDs. Castanea. 2003;68:254–60. All authors contributed to various aspects of this work (ordered by degree of 12. Sun JH, Li Z-C, Jewett DK, Britton KO, Ye WH, Ge X-J. Genetic diversity of contribution): conceived the study (ANE, MSH); aided in experimental design Pueraria lobata (kudzu) and closely related taxa as revealed by inter-simple (MSH, ANE); obtained research funds (ANE, JJD); coordinated activities (ANE, sequence repeat analysis. Weed Res. 2005;45:255–60. MSH); obtained and grew plants (ANE, MSH, JJD); RNA Isolation and Library 13. Heider B, Fischer E, Berndl T, Schultze-Kraft R. Analysis of genetic variation Prep (ANE); transcriptome assembly and analyses (MSH, MV, ANE); among accessions of Pueraria montana (Lour.) Merr. var. lobata and Pueraria microsatellite primer design (MSH); microsatellite primer validation (MSH, GM, phaseoloides (Roxb.) Benth. based on RAPD markers. Genet Resour Crop DZ, RZMR); prepared figures (MSH, MV, ANE); contributed to preparation of Evol. 2007;54:529–42. the manuscript (MSH, ANE, MV, JJD, KAC). All authors edited and approved 14. Hoffberg SL,Bentley KE, Lee JB,Myhre KE, Iwao K,Glenn TC, etal. the final manuscript. Characterization of 15 microsatellite loci in kudzu (Pueraria montana var. lobata) from the native and introduced ranges. Conserv Genet Resour. 2015;7:403–5. Ethics approval and consent to participate 15. Bentley K, Mauricio R. High degree of clonal reproduction and lack of large- All plant material was collected in accordance with institutional, national, scale geographic patterning mark the introduced range of the invasive vine, and international guidelines and under appropriate permits. Permits and kudzu (Pueraria montana var. lobata) in North America. Am J Bot. 2016;103: voucher specimens are deposited at the US National Herbarium (US) with all 1499–507. specimens determined by Dr. Ashley N. Egan. 16. Strickler SR, Bombarely A, Mueller LA. Designing a transcriptome next- generation sequencing project for a nonmodel plant species. Am J Bot. Competing interests 2012;99:257–66. The authors declare that they have no competing interests. 17. Varshney RK, Sigmund Rm Borner A, Korzun V, Stein N, Sorrells ME, et al. Interspecific transferability and comparative mapping of barley EST-SSR markers in wheat, rye, and rice. Plant Sci. 2005;168:195–202. Publisher’sNote 18. Ellis J, Burke J. EST-SSRs as a resource for population genetic analyses. Springer Nature remains neutral with regard to jurisdictional claims in Heredity. 2007;99:125–32. published maps and institutional affiliations. 19. Culley TM, Stamper TI, Stokes RL, Brzyski JR, Hardiman NA, Klooster MR, et al. An efficient technique for primer development and application that integrates Author details 1 fluorescent labeling and multiplex PCR. Appl Plant Sci. 2013;1:1–10. Department of Biology, George Washington University, Washington, DC, 2 20. Wright S. Evolution and the genetics of populations. Vol. 4. Variability within USA. Computational Biology Institute, Milken Institute School of Public 3 and among natural populations. Chicago: University of Chicago Press; 1978. Health, George Washington University, Washington, DC, USA. Department of 21. Stewart CN Jr, Tranel PJ, Horvath DP, Anderson JV, Rieseberg LH, Westwood Botany, National Museum of Natural History, Smithsonian Institution, JH, et al. Evolution of weediness and invasiveness: charting the course for Washington, DC, USA. Present address: College of Engineering, Oregon weed genomics. Weed Sci. 2009;57:451–62. State University, Corvallis, OR, USA. Present address: Department of Biology, 22. Ekblom R, Galindo J. Applications of next-generation sequencing in Washington University in St. Louis, St. Louis, MO, USA. Present address: molecular ecology of non-model organisms. Heredity. 2011;107:1–15. Department of Biology, Indiana University Bloomington, Bloomington, IN, 23. Egan AN, Schlueter J, Spooner DM. Applications of next-generation USA. School of Integrated Plant Science, Plant Breeding and Genetics sequencing in plant biology. Am J Bot. 2012;99:175–85. Section, Cornell University, Ithaca, NY, USA. Department of Invertebrate 24. Wen J, Egan AN, Dikow RB, Zimmer EA. Utility of transcriptome sequencing Zoology, National Museum of Natural History, Smithsonian Institution, for phylogenetic inference and character evolution. In: Hörandl E, Washington, DC, USA. Appelhans MS, editors. Next-generation sequencing in plant systematics. Königstein: Koeltz scientific books; 2015. p. 51–91. Received: 25 April 2018 Accepted: 15 May 2018 25. Ilut DC, Coate JE, Luciano AK, Owens TG, May GD, Farmer A, et al. A comparative transcriptomic study of an allotetraploid and its diploid progenitors illustrates the unique advantages and challenges of RNA-Seq in References plant species. Am J Bot. 2012;99:383–96. 1. van der Maesen LJG. Pueraria: Botanical characteristics. In: Keung WM, editor. 26. Grover CE, Salmon A, Wendel JF. Targeted sequence capture as a powerful Pueraria:the genus Pueraria. New York: Taylor and Francis; 2002. p. 1–28. tool for evolutionary analysis. Am J Bot. 2012;99:312–9. 2. Egan AN, Pan B. Resolution of polyphyly in Pueraria (Leguminosae, 27. Egan AN, Doyle J. A comparison of global, gene-specific, and relaxed clock Papilionoideae): the creation of two new genera, Haymondia and methods in a comparative genomics framework: dating the polyploid Toxicopueraria, the resurrection of Neustanthus, and a new combination in history of soybean (Glycine max). Syst Biol. 2010;59:534–47. Teyleria. Phytotaxa. 2015;218:201–26. 28. Cannon SB, McKain MR, Harkess A, Nelson MN, Dash S, Deyholos MK, et al. 3. van der Maesen LJG. Revision of the genus Pueraria DC. with some notes Multiple polyploid events in the early radiation of nodulating and on Teyleria Backer: Leguminosae. Wag Ag Un P. 1985;85:1–132. nonnodulating legumes. Mol. Biol. Evol. 2015;32:193–210. 4. van der Maesen LJG. Pueraria, the kudzu and its relatives, an update of the taxonomy. In: Sorensen M, editor. Proceedings of the first international 29. Glover NM, Redestig H, Dessimoz C. Homoeologs: what are they and how symposium on tuberous legumes, Gualdeloupe, FWI. Denmark: DSR do we infer them? Trends Plant Sci. 2016;21:609–21. Boghandel; 1994. p. 55–86. 30. Champigny MJ, Sung WW, Catana V, Salwan R, Summers PS, Dudley SA, et 5. Egan AN, Vatanparast M, Cagle W. Parsing polyphyletic Pueraria: delimiting al. RNA-Seq effectively monitors gene expression in Eutrema salsugineum distinct evolutionary lineages through phylogeny. Mol. Phylogenet. Evol. plants growing in an extreme natural habitat and in controlled growth 2016;104:44–59. cabinet conditions. BMC Genomics. 2013;14:578. Haynsen et al. BMC Genomics (2018) 19:439 Page 17 of 17 31. Vatanparast M, Shetty P, Chopra SP, Doyle JJ, Sathyanarayana N, Egan AN. 57. Duarte JM, Wall PK, Edger PP, Landherr LL, Ma H, Pires PK, et al. Transcriptome sequencing and marker development in winged bean Identification of shared single copy nuclear genes in Arabidopsis, Populus, (Psophocarpus tetragonolobus; Leguminosae). Sci Rep. 2016;6:29070. Vitis, and Oryza and their phylogenetic utility across various taxonomic 32. Kudapa H, Azam S, Sharpe AG, Taran B, Li R, Deonovic B, et al. Comprehensive levels. BMC Evol Biol. 2010;10:61. transcriptome assembly of chickpea (Cicer arietinum L.) using sanger and next 58. Goecks J, Nekrutenko A, Taylor J. The galaxy team. Galaxy: a comprehensive generation sequencing platforms: development and applications. PLoS One. approach for supporting accessible, reproducible, and transparent 2014;9:e86039. computational research in the life sciences. Genome Biol. 2010;11:R86. 33. Sudheesh S, Sawbridge TI, Cogan NO, Kennedy P, Forster JW, Kaur S. De 59. Blankenberg D, Von Kuster G, Coraor N, Ananda G, Lazarus R, Mangan M, et novo assembly and characterization of the field pea transcriptome using al. Galaxy: a web-based genome analysis tool for experimentalists. Curr RNA-Seq. BMC Genomics. 2015;16:611. Protoc Mol Biol. 2010;Unit-19.1021. 60. Giardine B, Riemer C, Hardison RC, Burhans R, Elnitski L, Shah P, et al. 34. Khalturin K, Hemmrich G, Fraune S, Augustin R, Bosch TCG. More than just Galaxy: a platform for interactive large-scale genome analysis. Genome orphans: are taxonomically-restricted genes important in evolution? Trends Res. 2005;15:1451–5. Genet. 2009;25:404–13. 61. Blankenberg D, Von Kuster G, Bouvier E, Baker D, Afgan E, Stoler N, et al. 35. Kumar S, Banks TW, Cloutier S. SNP discovery through next-generation Dissemination of scientific software with galaxy toolshed. Genome Biol. sequencing and its applications. Int J Plant Genomics. 2012; https://doi.org/ 2014;15:403. 10.1155/2012/831460. 62. Cock PJA, Chilton JM, Grüning B, Johnson JE, Soranzo N. NCBI BLAST+ 36. Varshney RK, Chabane K, Hendre PS, Aggarwal RK, Graner A. Comparative integrated into galaxy. GigaScience. 2015;4:39. assessment of EST-SSR, EST-SNP and AFLP markers for evaluation of genetic 63. Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos, Bealer K, et al. diversity and conservation of genetic resources using wild, cultivated and BLAST+: architecture and applications. BMC Bioinformatics. 2009;10:421. elite barleys. Plant Sci. 2007;173:638–49. 64. Haas B, Papanicolaou A. TransDecoder https://transdecoder.github.io. 37. Zalapa JE, Cuevas H, Zhu H, Steffan S, Senalik D, Zeldin E, et al. Using next- Accessed 14 Aug 2017. generation sequencing approaches to isolate simple sequence repeat (SSR) 65. Finn RD, Clements J, Eddy SR. HMMER web server: interactive sequence loci in the plant sciences. Am J Bot. 2012;99:193–208. similarity searching. Nucleic Acids Res. 2011;39:W29–37. 38. Morgante M, Hanafey M, Powell W. Microsatellites are preferentially associated 66. Sonnhammer ELL, Eddy SR, Durbin R. Pfam: a comprehensive database of with nonrepetitive DNA in plant genomes. Nat Genet. 2002;30:194–200. protein doman families based on seed alignments. Proteins. 1997;28:405–20. 39. Lopez L, Barreiro R, Fischer M, Koch MA. Mining microsatellite markers from 67. Conesa A, Götz S, García-Gómez J, Terol J, Talón M, Robles M. Blast2GO: a public expressed sequence tags databases for the study of threatened universal tool for annotation, visualization and analysis in functional plants. BMC Genomics. 2015;16:781. genomics research. Bioinformatics. 2005;21:3674–6. 40. Thumilan BM, Sajeevan RS, Biradar J, Madhuri T, Nataraja KN, Sreeman SM. 68. Wang X, Li S, Li J, Li C, Zhang Y. De novo transcriptome sequencing in Development and characterization of genic SSR markers from Indian Pueraria lobata to identify putative genes involved in isoflavones mulberry transcriptome and their transferability to related species of biosynthesis. Plant Cell Rep. 2015;34:733–43. Moraceae. PLoS One. 2016;11:e0162909. 69. Han R, Takahashi H, Nakamura M, Yoshimoto N, Suzuki H, Shibata D, et al. 41. Wang P, Yang L, Zhang E, Qin Z, Wang H, Liao Y, et al. Characterization and Transcriptomic landscape of Pueraria lobata demonstrates potential for development of EST-SSR markers from a cold-stressed transcriptome of phytochemical study. Front Plant Sci. 2015;6:426. Centipedegrass by Illumina paired-end sequencing. Plant Mol Biol Rep. 70. Schmutz J, McClean PE, Mamidi S, Wu GA, Cannon SB, Grimwood J, et al. A 2017;35:215–23. reference genome for common bean and genome-wide analysis of dual 42. Yang Z, Peng ZS, Yang H. Identification of novel and useful EST-SSR markers domestications. Nat Genet. 2014;46:707–13. from de novo transcriptome sequence of wheat (Triticum aestivum L.). 71. Faircloth BC. MSATCOMANDER: detection of microsatellite repeat arrays and Genet Mol Res. 2016;15:1. automated, locus-specific primer design. Mol Ecol Resour. 2008;8:92–4. 43. Kartzinel TR, Hamrick JL, Wang C, Bowsher AW, Quigley BG. Heterogeneity 72. Rozen S, Skaletsky H. Primer3 on the WWW for general users and for of clonal patterns among patches of kudzu, Pueraria montana var. lobata, biologist programmers. Methods Mol Biol. 2000;132:365–86. an invasive plant. Ann Bot. 2015;116:739–50. 73. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W, et al. 44. Ellstrand NC, Roose ML. Patterns of genotypic diversity in clonal plant Gapped BLAST and PSI-BLAST: a new generation of protein database search species. Am J Bot. 1987;74:123–31. programs. Nucleic Acids Res. 1997;25:3389–402. 45. Balloux F, Lehmann L, de Meeûs T. The population genetics of clonal and 74. Chatterji S, Pachter L. Reference based annotation with GeneMapper. partially clonal diploids. Genetics. 2003;164:1635–44. Genome Biol. 2006;7:R29. 46. Halkett FJ, Simon JC, Balloux FO. Tackling the population genetics of clonal 75. Pritchard JK, Stephens M, Donnelly P. Inference of population structure and partially clonal organisms. Trends Ecol. Evol. 2005;20:194–201. using multilocus genotype data. Genetics. 2000;155:945–59. 47. Shurtleff W, Aoyagi A. The book of kudzu: a culinary and healing guide. 76. Earl DA, vonHoldt BM. STRUCTURE HARVESTER: a website and program for Brookline: Autumn Press; 1997. visualizing STRUCTURE output and implementing the Evanno method. 48. Doyle JJ, Doyle JL. A rapid DNA isolation procedure for small quantities of Conserv Genet Resour. 2012;4:359–61. fresh leaf tissue. Phytochem Bull. 1987;19:11–5. 77. Evanno G, Regnaut S, Goudet J. Detecting the number of clusters of 49. Andrews S. Fast QC. A quality control tool for high throuput sequence data. individuals using the software STRUCTURE: a simulation study. Mol Ecol. http://www.bioinformatics.babraham.ac.uk/projects/fastqc/. Accessed 14 Aug 2017. 2005;14:2611–20. 50. Smeds L, Künstner A. ConDeTri – a content dependent read trimmer for 78. Jakobsson M, Rosenberg NA. CLUMPP: a cluster matching and permutation Illumina data. PLoS One. 2011;6:e26314. program for dealing with label switching and multimodality in analysis of 51. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. Full- population structure. Bioinformatics. 2007;23:1801–6. length transcriptome assembly from RNA-Seq data without a reference 79. Rosenberg NA. DISTRUCT: a program for the graphical display of population genome. Nat Biotechnol. 2011;29:644–52. structure. Mol Ecol Notes. 2004;4:137–8. 52. Fu L, Niu B, Zhu Z, Wu S, Li W. CD-HIT: accelerated for clustering the next- 80. Excoffier L, Laval G, Schneider S. Arlequin (version 3.0): an integrated software generation sequencing data. Bioinformatics. 2012;28:3150–2. package for population genetics data analysis. Evol Bioinforma. 2005;1:47–50. 53. Li B, Dewey CN. RSEM: accurate transcript quantification from RNA-Seq data 81. Takezaki N, Nei M, Tamura K. POPTREEW: Web version of POPTREE for with or without a reference genome. BMC Bioinformatics. 2011;12:323. constructing population trees from allele frequency data and computing 54. Langmead B, Salzberg SL. Fast gapped-read alignment with bowtie 2. Nat other population statistics. Mol Biol Evol. 2014;31:1622–4. Methods. 2012;9:357–9. 82. Latter BDH. Selection in finite populations with multiple alleles. III. Genetic 55. Davis MPA, van Dongen S, Abreu-Goodger C, Bartonicek N, Enright AJ. Kraken: divergence with centripetal selection and mutation. Genetics. 1972;70:475–90. a set of tools for quality control and analysis of high-throughput sequence data. Methods. 2013;63:41–9. 56. Simão FA, Waterhouse RM, Ioannidis P, Kriventseva EV, Zdobnov EM. BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics. 2015;31:3210–2. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Genomics Springer Journals

De novo transcriptome assembly of Pueraria montana var. lobata and Neustanthus phaseoloides for the development of eSSR and SNP markers: narrowing the US origin(s) of the invasive kudzu

Free
17 pages

Loading next page...
 
/lp/springer_journal/de-novo-transcriptome-assembly-of-pueraria-montana-var-lobata-and-TKMH48yDcn
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s).
Subject
Life Sciences; Life Sciences, general; Microarrays; Proteomics; Animal Genetics and Genomics; Microbial Genetics and Genomics; Plant Genetics and Genomics
eISSN
1471-2164
D.O.I.
10.1186/s12864-018-4798-3
Publisher site
See Article on Publisher Site

Abstract

Background: Kudzu, Pueraria montana var. lobata, is a woody vine native to Southeast Asia that has been introduced globally for cattle forage and erosion control. The vine is highly invasive in its introduced areas, including the southeastern US. Modern molecular marker resources are limited for the species, despite its importance. Transcriptomes for P. montana var. lobata and a second phaseoloid legume taxon previously ascribed to genus Pueraria, Neustanthus phaseoloides, were generated and mined for microsatellites and single nucleotide polymorphisms. Results: Roche 454 sequencing of P. montana var. lobata and N. phaseoloides transcriptomes produced read numbers ranging from ~ 280,000 to ~ 420,000. Trinity assemblies produced an average of 17,491 contigs with mean lengths ranging from 639 bp to 994 bp. Transcriptome completeness, according to BUSCO, ranged between 64 and 77%. After vetting for primer design, there were 1646 expressed simple sequence repeats (eSSRs) identified in P. montana var. lobata and 1459 in N. phaseoloides. From these eSSRs, 17 identical primer pairs, representing inter-generic phaseoloid eSSRs, were created. Additionally, 13 primer pairs specific to P. montana var. lobata were also created. From these 30 primer pairs, a final set of seven primer pairs were used on 68 individuals of P. montana var. lobata for characterization across the US, China, and Japan. The populations exhibited from 20 to 43 alleles across the seven loci. We also conducted pairwise tests for high-confidence SNP discovery from the kudzu transcriptomes we sequenced and two previously sequenced P. montana var. lobata transcriptomes. Pairwise comparisons between P. montana var. lobata ranged from 358 to 24,475 SNPs, while comparisons between P. montana var. lobata and N. phaseoloides ranged from 5185 to 30,143 SNPs. (Continued on next page) * Correspondence: egana@si.edu; ashegan2@gmail.com Computational Biology Institute, Milken Institute School of Public Health, George Washington University, Washington, DC, USA Department of Botany, National Museum of Natural History, Smithsonian Institution, Washington, DC, USA Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Haynsen et al. BMC Genomics (2018) 19:439 Page 2 of 17 (Continued from previous page) Conclusions: The discovered molecular markers for kudzu provide a starting point for comparative genetic studies within phaseoloid legumes. This study both adds to the current genetic resources and presents the first available genomic resources for the invasive kudzu vine. Additionally, this study is the first to provide molecular evidence to support the hypothesis of Japan as a source of US kudzu and begins to narrow the origin of US kudzu to the central Japanese island of Honshu. Keywords: Pueraria montana var. lobata,Kudzu, Neustanthus phaseoloides, Transcriptome, Invasive, Molecular markers Background characterization and development of microsatellites Pueraria montana (Lour.) Merr. var. lobata (Willd.) (SSRs) and single nucleotide polymorphisms (SNPs). Maesen & Almeida ex Sanjappa and Pradeep (kudzu) and Transcriptome sequencing is currently one of the most Neustanthus phaseoloides (Roxburgh) Bentham (tropical popular applications of next-generation sequencing due kudzu), members of the phaseoloid clade of subfamily to its versatility, cost efficiency, and suitability for use on Papilionoideae of the Fabaceae family, are twining vines non-model organisms [16]. Transcriptomes are often native to Southeast Asia that have been introduced glo- mined for expressed simple sequence repeats (eSSRs) for bally for livestock forage, nitrogen soil enrichment, and marker development and genetic diversity studies. eSSRs erosion control [1]. Prior to recent molecular and taxo- have been shown to have greater transferability across nomic revision [2], Neustanthus was placed within taxa than traditional ‘anonymous’ SSRs [17, 18]. This in- Pueraria, along with ~ 17 additional species native to creased transferability can be utilized in multiple ways. southeast Asia [3]. A comprehensive molecular system- First, if a transcriptome is not available for the species of atic study of Pueraria sensu van der Maesen [4]con- interest, a closely related species whose transcriptome is firmed that its species, including several legumes of available can be used as a surrogate reference for micro- economic importance, comprise a polyphyletic assem- satellite development. Second, if a researcher is studying blage of separate evolutionary lineages spread across two closely related taxa and transcriptomes are available the phaseoloid clade [5]. for both, a single set of markers can be developed that Both kudzu and tropical kudzu share a penchant for in- work on both species to reduce costs. To this end, we vasiveness in their naturalized areas, the southeastern have compared the transcriptomes of kudzu and tropical United States (US) and the pantropics, respectively. Of the kudzu to identify shared eSSRs between the species in two taxa, kudzu is a far greater agricultural pest and has order to develop primers that can be used equally well garnered the majority of scientific inquiry. Kudzu was in- for population genetic studies of either species, and shed troduced into the US during the Centennial Exposition of light on the introduction history of the notorious inva- 1876 in Philadelphia, Pennsylvania [6]. The vine is cur- sive kudzu in the United States. rently found in 30 states and is considered an agricultural In the present study, three transcriptomes, two P. mon- pest throughout the southeastern US [7], costing millions tana var. lobata and one N. phaseoloides,were denovo of dollars in eradication and management measures annu- assembled and characterized. Intra- and inter-specific ally [8, 9]. A major aspect that could be influencing the in- comparisons were made between transcriptomes and two vasiveness and spread of kudzu are high levels of genetic sets of population genetic markers were identified: eSSRs variation observed across populations in the US. This and SNPs. The eSSRs were validated across Asian and could be due to multiple introductions from its native North American populations of P. montana. var. lobata range, either of a single genetically diverse population, or and used to explore population diversity and structure from multiple genetically distinct subpopulations, poten- across native and introduced ranges. The resulting data tially from different geographic regions or from more than provide genetic resources for future studies of kudzu and one of the taxonomically recognized varieties of Pueraria related genera through development of high-resolution montana. marker sets for genetic diversity assessment and popula- Several molecular markers have been used over the tion studies. past two decades to estimate the introduced and native genetic diversities of kudzu and two other Pueraria Results montana varieties: Pueraria montana var. montana and Transcriptome sequencing and quality control Pueraria montana var. thomsonii (Benth.) Wiersema ex Transcriptome sequencing produced between 279,109 D.B. Ward [10–15]. However, despite the ecological and and 423,426 reads per transcriptome (Table 1), with economic importance of kudzu, its modern molecular Neustanthus phaseoloides (hereafter CPP02) having the marker resources are limited, lagging particularly in the most reads produced. CPP02 and the greenhouse-raised Haynsen et al. BMC Genomics (2018) 19:439 Page 3 of 17 Table 1 Statistics following ConDeTri cleaning and Trinity assembly Accessions CPP27 Pmnk6 CPP02 Number of raw reads 279,109 396,022 423,426 Number of raw bases (bp) 112,337,841 247,596,818 158,214,933 Number of clean reads 257,015 381,166 348,529 Cleaned reads / Raw reads (%) 92.1% 71.0% 82.3% Number of clean bases (bp) 75,672,645 124,810,371 87,666,889 Mean clean read length (bp) 294 444 252 Number of aligned reads 99,248 116,524 119,452 Aligned read / Cleaned reads (%) 38.6% 41.4% 34.3% Number of contigs 18,325 15,736 18,412 Number of bases in contigs (bp) 11,703,977 15,640,762 11,892,992 Mean contig length (bp) 639 994 646 N50 (bp) 755 1256 759 Longest contig (bp) 4335 4815 6221 Number of singletons 60,869 45,306 73,994 Singletons / Cleaned reads (%) 23.7% 16.1% 21.2% Number of bases in singletons (bp) 17,591,281 20,431,176 18,048,611 Mean singleton length (bp) 289 451 244 Number of transcripts (contigs + singletons) 79,194 61,042 92,406 bp base pairs kudzu (hereafter CPP27) were sequenced on the same De novo assembly run and were multiplexed with two other transcriptomes Trinity used an average of 38.1% of the ConDeTri cleaned not reported here. While sequencing of CPP02 produced reads in its assemblies and produced an average of 17,491 the most reads, the mean read length before cleaning contigs. The mean contig lengths ranged from 639 bp to was shorter than that of CPP27 (373 bp vs. 402 bp, re- 994 bp (Table 1) and each of the accessions had contigs spectively), as was the mean read length after cleaning exceeding 3000 bp (Fig. 1). Additionally, Bowtie2 mapped (252 vs. 294, respectively). The tendency for shorter ~ 68% of each accession’s contigs back to their raw reads DNA fragments to be incorporated at the library con- (Additional file 1). Overall transcriptome contamination struction phase and sequencing stage may provide an was low, with fungal contamination ranging between 2.64 explanation for the difference in the number of raw and 3.53%, while prokaryote and viral contamination reads produced between CPP27 and CPP02. However, ranged from 0.5 to 1.32% (Additional file 2). Transcrip- following cleaning, the number of clean bases was tome completeness varied greatly, with a range of comparable between CPP02 and CPP27, as were all complete units from 164 to 361 and duplicate units simi- other downstream metrics (Table 1). While 454 pyro- larly showing a > 2× difference between transcriptomes sequencing was used for all three transcriptomes, the (Fig. 2). Specifically, transcriptome completeness was ap- chemistries between the two CPP transcriptomes and proximately 64, 77, and 70%, for CPP27, Pmnk6, and the wild-collected kudzu (hereafter Pmnk6) transcrip- CPP02, respectively. The reciprocal best BLAST hits tome differed, with the Pmnk6 transcriptome benefit- (RBH) of the transcriptomes showed that 1525 transcripts ing from an improved chemistry, as seen in the were shared among all three (Fig. 3). increased number of raw bases, the average read length before cleaning (625 bp) and the mean clean Functional annotation of transcriptomes read length (444; Table 1). These sequencing improve- In total, we have obtained 13,230, 18,446 and 24,447 asso- ments translated into improved assembly statistics, ciated GO IDs for CPP02, CPP27 and Pmnk6 transcrip- such as increased mean contig length (~ 1.5× that of tomes, respectively (Table 2) corresponding to the 33, 43 the CPP transcriptomes), higher N50 (1.65× CPP) and and 51% of original contigs in each transcriptome, while fewer singletons (Table 1). However, the improved only 9.6, 17 and 36% of the singletons had associated func- chemistry did not lead to differences in the number tional protein information (GO IDs). Therefore, more of aligned reads in the assembled transcriptomes than 90, 82 and 63% of singletons were discarded during (Additional file 1). the multiple searches, which is unfortunate because over Haynsen et al. BMC Genomics (2018) 19:439 Page 4 of 17 Fig. 1 Contig length distributions of Trinity using the ConDeTri dynamic read trimmer Fig. 2 Transcriptome completeness of transcripts quantified through 956 universal single-copy orthologs using BUSCO Haynsen et al. BMC Genomics (2018) 19:439 Page 5 of 17 tropical kudzu (CPP27 vs. CPP02, Pmnk6 vs. CPP02, and CPP27/Pmnk6 vs. CPP02, respectively). The over 30,000 SNPs identified between CPP27/Pmnk6 vs. CPP02 is greater than the sum of SNPs from the individ- ual comparisons of P. montana var. lobata to N. phaseo- loides because the merged transcripts offer a more complete snapshot of a US kudzu transcriptome which was used as the reference for SNP detection. Lastly, we found 24,475 SNPs within kudzu from among three countries (Japan vs. Pmnk6(US)/CPP27(US)/ China). The majority of high-confidence SNPs were found within contigs rather than singletons (Table 3), which is expected given the fact that more highly expressed genes will be more likely to be represented by > 20× coverage (one of our criteria for high confidence) and are most likely to assemble into contigs. Also of note, the transi- tion/transversion ratio varied from 1.41 to 1.73 (Table 3) with the higher ratios found between the intergeneric Fig. 3 The number of unique and shared, homologous transcripts comparisons than the intraspecific comparisons. among kudzu CPP27 and Pmnk6 and tropical kudzu CPP02 transcriptome assemblies as ascertained via reciprocal best BLAST hit analyses eSSR discovery and characterization The eSSR analysis of the transcripts detected 5255 and 54, 56 and 66% of final annotated transcripts belong to 4586 perfect eSSRs for CPP27 and CPP02, respectively. the singletons in CPP02, CPP27, and Pmnk6, respectively The majority (76.7 and 76.8%) of eSSRs were (Table 2). In all three transcriptomes, the highest top hit tri-nucleotide repeats (TNRs; Table 4). After vetting for species for the annotated proteins were Glycine max (L.) primer design, there were 1646 potential eSSRs identi- Merr., G. soja Siebold & Zucc. and Cajanus cajan (L.) fied in P. montana var. lobata and 1459 in N. phaseo- Millsp., respectively (Additional files 3, 4,and 5). Summar- loides. Looking only at TNRs (1458 for CPP27 and 1273 ies of the biological process, cellular components and mo- for CPP02), 25 matches were found between P. montana lecular function categories for each transcriptome are var. lobata and N. phaseoloides in which either the for- shown in Fig. 4. ward or reverse primers were identical, suggesting hom- ology. However, no sets of primer pairs (forward and SNP discovery reverse primers together) were found duplicated be- We conducted pairwise tests for high-confidence SNP tween transcriptomes. Alterations to the non-identical discovery of the kudzu transcriptomes (Table 3, Add- primer pair within the 25 matches allowed for the itional files 6, 7, 8, 9 and 10). Our conservative assess- creation of 17 identical primer pairs between CPP27 ment of SNPs reduced thousands of high-confidence and CPP02. These 17 shared primer pairs represent SNPs to a lower number (Table 3) that are 1) one-to-one inter-generic phaseoloid eSSRs. Additionally, 13 TNR point mutations without length variants, 2) have vari- primer pairs specific to P. montana var. lobata were ation frequency over 95%, and 3) have a repeat depth of also selected for screening. Of the 30 total eSSR pri- 20 or more. As such, we identified 358 SNPs between mer pairs, 21 pairs were advanced to the Culley et al. the two US kudzu transcriptomes (CPP27 vs. Pmnk6), [19] protocol; of the nine primer pairs that were and 5185, 19,028, and 30,143 SNPs between kudzu and eliminated, four did not amplify a product, four Table 2 Summary of gene ontology analysis Accessions Transcripts Orfs Predictions BLAST Hits Annotated GO IDs ECs CPP27 79,194 37,741 30,716 28,795 18,446 8039 (18,325/60869) (13,534/17182) (12,583/16212) (7958/10488) Pmnk6 61,042 50,320 42,386 39,366 24,447 6337 (15,736/45306) (14,821/27565) (12,705/26661) (8079/16368) CPP02 92,406 34,223 27,661 22,472 13,230 4064 (18,412/73994) (14,677/12984) (10,407/12065) (6085/7145) Orfs open reading frames, GO gene ontology, ECs enzyme codes. Parentheses: (contigs/singletons) Haynsen et al. BMC Genomics (2018) 19:439 Page 6 of 17 Fig. 4 Gene ontology classifications of kudzu and tropical kudzu annotated transcripts. Numbers indicate the number of sequences associated with the particular GO term in each category amplified in an unexpected size range, and one displayed Population structure and genetic diversity of kudzu double banding (Additional file 11). Of the 21 primer pairs Three genetic units were determined to be the optimal that were assessed with the Culley et al. [19]protocol, value of K in STRUCTURE across the 75 accessions (K = 3, seven were discarded due to multiple banding and four Fig. 5, Additional file 12). The US is primarily composed of for lack of amplification, whereas a further three were a single genetic unit, with a couple individuals assigned to a removed due to the presence of monomorphic alleles second unit; whereas, China and Japan are more heteroge- (Additional file 11). The final set of eSSR primer pairs neous in their composition, yet they are still composed of identified seven polymorphic loci displaying single the same 2 units found in the US. Thailand, however, is bands of expected sizes (Table 5). composed of a single genetic unit that is unique to that Table 3 Single nucleotide polymorphism detection among kudzu and tropical kudzu genotypes a b c Comparison HC SNPs SNPs > 95% SNPs >20x Total SNPs Ts/Tv Pmnk6 vs CPP27 10,417 6016 426 358 1.41 (7494/2923) (4125/1891) (252/174) CPP02 vs CPP27 99,584 86,626 5831 5185 1.60 (81,276/18308) (70,638/15988) (5091/740) CPP02 vs Pmnk6 220,739 174,884 21,258 19,028 1.73 (164,118/56621) (127,311/47573) (19,255/2003) CPP02 vs Pmnk6, CPP27 314,416 248,719 33,603 30,143 1.71 (229,163/85251) (178,102/70617) (29,812/3791) Japan vs Pmnk6, CPP27, China 494,234 79,088 27,108 24,475 1.47 (494,234/0) (79,088/0) (27,108/0) SNPs with the > 95% frequency SNPs with > 95% frequency and > 20x coverage One-to-one point mutations after exclusion of indels and length variants; HC: high confidence; parentheses: (contigs/singletons) Haynsen et al. BMC Genomics (2018) 19:439 Page 7 of 17 Table 4 Transcriptome eSSRs sampled. Genetic structuring as assessed by pairwise F showed differences among groups, particularly in CPP27 CPP02 st Thailand and southern China (China 3; Table 8), cor- Transcripts 79,194 92,406 roborated by the structuring of genetic units shown Raw eSSRs 5255 4586 in Fig. 5. As defined by Wright [20], Thailand showed Dinucleotide 770 670 very great genetic variation (F > 0.25) with respect to st Trinucleotide 4032 3524 all other subpopulations, except China 3, with which Tetranucleotide 180 137 it showed great variation (0.15 < F < 0.25). The rest st Pentanucleotide 106 79 of the comparisons resulted in little to moderate genetic variation (0 < F < 0.05 and 0.05 < F < 0.15, respectively). Hexanucleotide 167 176 st st The neighbor-joining distance tree supports the pairwise Primered eSSRs 1646 1459 F results (Fig. 6): 1) Thailand is a distantly related lineage st Dinucleotide 14 28 to the nine other subpopulations representing P. montana Trinucleotide 1458 1273 var. montana and var. thomsonii; 2) the Chinese subpopu- Tetranucleotide 62 54 lations are divided into three lineages; and 3) the US sub- Pentanucleotide 41 25 populations are more genetically similar to Japan 2. Hexanucleotide 71 79 Discussion Invasive species are increasingly widening their scope nation, which supports our classification of its accessions as across the globe, yet the genetic mechanisms underlying being different varieties of P. montana, specifically var. invasiveness or weediness remain a mystery. In the gen- thomsonii and var. montana. omics era, scientists have raised a clarion call to arms to The national populations exhibited from 20 to 43 al- build genomic resources to study invasive species [21]. leles across a total of seven loci (Table 6), while the sub- Understanding the introduction history and relative gen- populations exhibited from 20 to 36 total alleles etic diversity of invasive species is an important step to (Additional file 13). China was composed of the greatest gaining a foothold on management and control, a goal number of alleles, in particular, China 3 (southern), requiring the development of variable molecular markers while Thailand was composed of the fewest number of such as microsatellites or SNPs to assess genetic diver- alleles. sity and population structure. In this study, we have as- After Bonferroni correction, none of the subpopulations’ sembled and characterized multiple transcriptomes of observed and expected heterozygosities significantly dif- the invasive Kudzu vine, Pueraria montana var. lobata, fered (Table 7), supporting the hypothesis that all the sub- and for tropical kudzu, Neustanthus phaseoloides, a spe- populations were in Hardy-Weinberg equilibrium when cies until recently thought congeneric with kudzu [2, 5]. Table 5 Seven eSSR primers optimized and used to assess population genetics in kudzu accessions Locus Sequence Dye/Tail SSR Length (bp) PP2 F: 5′-TAG GAG TGC AGC AAG CAT ATG CCG CGG ATC TTT GAA AG-3’ VIC /M13A AAC 100–130 R: 5’-CAA ATT GGC CCT GTC CCA AT-3’ n/a PP4 F: 5′-TGT AAA ACG ACG GCC AGT CAT GCC CAC GTG CTT CAT AG-3’ 6FAM/M13 GCT 100–140 R: 5’-CTC TCA GAT CCA GGC CCA AA-3’ n/a PP10 F: 5′-TAG GAG TGC AGC AAG CAT GGC ATG TAG ATC CAG CTA AA-3’ VIC/M13A GGT 310–330 R: 5′-TTG ACA GAT TTC TGA TTC TTG G-3’ n/a PP13 F: 5′-TAG GAG TGC AGC AAG CAT GAT TGA GCA GGC ACG AGA AC-3’ VIC/M13A GCT 270–300 R: 5’-CAG TAG CAG GCA TGT GTT GG-3’ n/a PL1 F: 5’-CAC TGC TTA GAG CGA TGC TGT AAG CGT TCG TTC GTT GG-3’ PET/M13B CTT 400–440 R: 5’-TCA ACC TGG TGC TCT CTG AC-3’ n/a PL7 F: 5′-TGT AAA ACG ACG GCC AGT AGT GGC CTT GCT CTT CTT CC-3’ 6FAM/M13 CTT 80–140 R: 5′-GTG TCA TCT CAG CAC GTT GG-3’ n/a PL11 F: 5′-TGT AAA ACG ACG GCC AGT TGG CAT CAT CCT TCA ACC AC-3’ 6FAM/M13 ACC 300–330 R: 5′-ATT CGG GAA TAG TGG GTG GG-3’ n/a F forward primer, R reverse primer. Dyes VIC: 2′-chloro-7′phenyl-1,4-dichloro-6-carboxy-fluorescein; 6FAM: 6-carboxyfluorescein; PET: chemical structure currently unpublished as proporietary to Lifetech. Tail: see Culley et al. [19] for information about M13, M13A, and M13B Haynsen et al. BMC Genomics (2018) 19:439 Page 8 of 17 Fig. 5 STRUCTURE diagram of 75 P. montana accessions across four nations (K = 3) Kudzu is well known as an invasive species in both agri- consideration when dealing with potentially polyploid cultural and natural areas due to its fast growth, clonal plants [23, 25, 26]. Pueraria is descendent from an ancient habit, and extensive introductions outside its native range. polyploidy event that transpired 50–60 mya near the ori- Tropical kudzu is also known to be invasive in its intro- gin of the papilionoid subfamily [27, 28], creating a dupli- duced ranges, but to a lesser extent. We mined our tran- cated genomic complement that has fractionated over scriptomes of these two species for molecular markers time but whose signature still remains within descendent (eSSRs and SNPs), screened and validated eSSRs, and per- genomes. Longer reads are more likely to unambiguously formed functional annotations of the transcriptomes, im- assemble or align across homoeologues, duplicated genes proving the genetic resources available for kudzu and produced via allopolyploidy [29]. Furthermore, the longer tropical kudzu. reads result in the sequencing of more full-length mRNA transcripts, an outcome that argues for including single- Transcriptome characterization tons (those reads that do not assemble into contigs) in the Whether researching model or non-model organisms, se- overall transcript complement. Although pyrosequencing quencing the transcriptome of a species is a natural begin- produces fewer overall reads as compared to Illumina, its ning for genome-wide resource development and study ability to produce longer transcripts is advantageous, par- [22, 23], enabling the characterization of gene expression ticularly for allopolyploid species and other hybrids where profiles, genetic marker discovery, and phylogenetic infer- avoiding the assembly of chimeric sequences is important. ence [24]. Here, we characterize the transcriptomes of two The comparative results across our transcriptomes in accessions of kudzu, one wild-collected (Pmnk6) and one terms of the number of transcripts discovered and the partially inbred line propagated by the USDA agriculture relative overlap among pairwise comparisons provides research service (CPP27), as well as one of tropical kudzu some insights into the relative impact of environment vs. (CPP02). We chose to use 454 pyrosequencing technology shared ancestry. CPP02 had the highest number of tran- over Illumina due to the longer read lengths, an important scripts and the highest number of unique transcripts, with Pmnk6 having the least number of transcripts, even though it presents the best transcriptome in terms of Table 6 Allelic frequency for Pueraria national populations mean contig length, N50, and BUSCO results. One ex- Locus USA China Japan Thailand Mean SD Total planation involves the number of tissues used for se- N =25 N =21 N =22 N =7 quencing. CPP02 utilized three tissues (young leaves, PP2 8 7 6 4 6.25 1.71 9 young shoot tips, and buds) while CPP27 used two tis- PP4 4 5 7 3 4.75 1.71 9 sues (young leaves and young shoot tips), and Pmnk6 PP10 5 5 6 3 4.75 1.26 8 used a single tissue (young leaves). Given this informa- PP13 3 7 4 2 4.00 2.16 7 tion, it makes sense that the transcriptome that was composed of the greatest number of tissues resulted in PL1 4 4 2 4 3.50 1.00 9 the highest number of unique transcripts due to expres- PL7 8 8 11 3 7.50 3.32 15 sional differences across tissue types. CPP02 and CPP27 PL11 5 7 3 1 4.00 2.58 7 shared the highest number of reciprocal best BLAST hits Mean 5.29 6.14 5.57 2.86 4.96 1.96 9.14 (RBH). However, one would expect the two kudzu acces- SD 1.98 1.46 2.99 1.07 1.42 0.80 2.73 sions (CPP27 and Pmnk6) to share the greatest number of Total 37 43 39 20 34.75 13.73 64 overlapping transcripts due to shared ancestry. This could N number of accessions, SD standard deviation also be explained by the fact that the two transcriptomes Haynsen et al. BMC Genomics (2018) 19:439 Page 9 of 17 Table 7 Observed and expected heterozygosities for Pueraria subpopulations US 1 US 2 US 3 CN 1 CN 2 CN 3 JP 1 JP 2 JP 3 TH # Individuals 8 10 75887 877 Obs. Het. 0.717 0.552 0.472 0.611 0.378 0.632 0.396 0.506 0.656 0.594 Exp. Het. 0.643 0.503 0.547 0.648 0.589 0.763 0.579 0.572 0.661 0.583 HWE p-value 0.766 0.251 0.611 0.765 0.079 0.392 0.013 0.429 0.869 0.442 US United States, CN China, JP Japan, TH Thailand, Obs: Observed, Exp Expected, Het Heterozygosity, HWE Hardy-Weinberg Equilibrium that shared the most homologous tissues resulted in the have yet to be described, perhaps including species-specific highest number of shared transcripts. Alternatively, the “orphan” genes [34]. Catalytic activity, binding, metabolic seeming disparity in shared best BLAST hits could be ex- and cellular processes were among the most highly repre- plained by the relative impacts of a shared environment, sented groups regarding GO analysis (Fig. 4)acrossall three which often affects gene expression. Our two CPP tran- transcriptomes, as expected given that we used young tis- scriptomes were both grown in the same greenhouse en- sues that are undergoing extensive metabolic activities. vironment at the same time and so their gene expression profiles may be expected to be more similar than those of Single nucleotide polymorphism discovery the two P. montana var. lobata accessions, one of which SNPs are fast becoming the marker of choice due to was grown in the greenhouse (CPP27) and one in the wild their ease of discovery via next generation sequencing (Pmnk6). A similar finding was discovered across tran- technologies [35]. Additionally, the ease of mining SNPs scriptomes of Eutrema salsugineum (Pall.) Al-Shehbaz & from previously produced transcriptomes can provide a Warwick plants that were grown in field (uncontrolled en- new use for previously published data sets that may be vironment) vs. cabinet (controlled environment) conditions, sitting idle in online repositories. SNPs, though less with the plants grown in the controlled environment shar- polymorphic than SSRs, may provide higher resolution ing a higher number of expressed genes as compared to the assessment of genetic variation and identification of more geographically proximate plants grown in differing population structure [36]. We detected a near 100-fold environments [30]. increase in the number of SNPs detected between kudzu In this study, we were able to annotate over 13,000 tran- and tropical kudzu as compared to that detected within scripts from kudzu and tropical kudzu (Table 1). Our tran- kudzu. SNPs discovered between kudzu and tropical scriptomes do not provide a full gene complement due to kudzu may represent species level, fixed differences be- low sequencing depth as evidence by our BUSCO results tween these genera. Validation of these SNPs is beyond (Fig. 2). However, the level of unannotated transcripts in the scope of this paper; nevertheless, this list presents a this study is similar to results reported from other significant resource for future work in genetic diversity non-model legumes, like winged bean [31], chickpea [32], assessment, genetic mapping, genome-wide association and field pea [33]. The unidentified transcripts are likely mapping, or evolution-based studies of invasiveness, and due to 1) correspondence to non-coding regions or pseudo- marks the first SNP markers discovered to date in Puer- genes, 2) short length of transcripts, or 3) coding genes that aria and Neustanthus. Use of these SNP markers across Table 8 Subpopulation pairwise F st US 1 US 2 US 3 CN 1 CN 2 CN 3 JP 1 JP 2 JP 3 TH US 1 – 0.811 0.541 0.441 0.009 0.000* 0.297 0.126 0.099 0.000* US 2 −0.023 – 0.378 0.730 0.009 0.000* 0.432 0.108 0.072 0.000* US 3 − 0.011 − 0.008 – 0.306 0.009 0.000* 0.360 0.946 0.153 0.000* CN 1 − 0.009 − 0.022 0.024 – 0.108 0.009 0.901 0.162 0.108 0.000* CN 2 0.075 0.098 0.099 0.075 – 0.297 0.207 0.081 0.739 0.000* CN 3 0.077 0.107 0.120 0.073 0.022 – 0.063 0.000* 0.324 0.000* JP 1 0.015 −0.002 0.022 −0.035 0.051 0.064 – 0.207 0.486 0.000* JP 2 0.016 0.025 −0.030 0.049 0.078 0.085 0.042 – 0.135 0.000* JP 3 0.029 0.028 0.042 0.037 −0.014 0.006 0.002 0.036 – 0.000* TH 0.315 0.370 0.377 0.330 0.310 0.244 0.347 0.352 0.322 – Below diagonal pairwise F values, above diagonal p-values st US United States, CN China, JP Japan, TH Thailand * = significant under Bonferroni correction (p < 0.001) Haynsen et al. BMC Genomics (2018) 19:439 Page 10 of 17 Japan2 USA3 USA2 USA1 Japan1 China1 Japan3 46 China2 China3 Thailand 0.05 Fig. 6 Neighbor joining distance tree based on F values and 10,000 bootstraps. US = United States; CN = China; JP = Japan; and TH = Thailand st a wide population-level sampling throughout Asia would population every few meters, suggesting that they treated a enable a robust investigation into the introduction his- patch of kudzu as a population, whereas we sampled indi- tory of kudzu within the US. viduals no closer than ~ 1 km apart, and viewed a popula- tion as a regional area comprised of numerous, eSSR marker discovery and validation non-connected patches. With the abilities to grow over eSSRs are routinely developed from transcriptomic data, 12 in. per day and root at the nodes, a kudzu patch may providing a ready source for genetic diversity assessment likely represent only one or a few genets [43]. Therefore, through cost-effective means [37]. In spite of being derived the reported clonal sampling of Bentley and Mauricio [15] from coding DNA, which is evolutionarily conserved, may be the cause of the near 0.0 observed heterozygosities eSSRs have proven a variable and valuable resource for gen- and may not be indicative of the primers themselves. etic studies [18]. In our study, we detected ~ 5000 eSSRs each within kudzu and tropical kudzu.Overall,trinucleo- Genetic diversity of kudzu tide SSR motifs (TNRs) were the most abundant, as found For the past two decades, the genetic diversity of kudzu consistently in other plant studies [17, 38–42]. Presumably has been assessed with the various molecular markers of this is because TNRs will not affect the open reading frames the corresponding era. For instance, Pappert et al. [10] of coding regions [38]. We investigated the utility of 30 used 13 allozymes across 1000 US accessions to con- eSSR markers discovered in our data and optimized seven clude that introduced kudzu possessed considerable gen- for use across kudzu. When compared to the etic variation with a lack of geographic structuring. kudzu-derived SSR markers of Hoffberg et al. [14], similar- Similar conclusions were subsequently reached by Jewett ities and benefits are found. For instance, Hoffberg et al. et al. [11] using 18 random amplified polymorphic DNA [14] assessed their 15 genomic SSRs against 102 geograph- (RAPD) markers across 50 accessions from the US and ically dispersed individuals, finding that their alleles per China, and by Sun et al. [12] using 11 inter-simple se- locus ranged from 2 to 8, whereas our alleles per locus quence repeat (ISSR) markers across 108 accessions ranged from 7to15(Table 6). This comparison shows from the US and China. A decade later, Bentley and twice as many alleles within a smaller sample size, approxi- Mauricio [15], using 15 SSRs and one chloroplast mately two-thirds the size of Hoffberg et al. [14]. One ex- marker across 1747 US accessions, reported that the planation for the difference in allele numbers could be high levels of genetic diversity result from high clonal attributed to the differing sampling ranges, with our indi- reproduction in kudzu, as described by Ellstrand and viduals being collected from a greater global area. However, Roose [44], Balloux et al. [45], and Halkett et al. [46]. when Bentley and Mauricio [15] used the Hoffberg et al. Specifically, high levels of genetic variation are expected [14] primers on 1747 accessions of kudzu from solely the in clonal populations when the populations were US they identified 2–17 alleles per locus, which also repre- founded by sexual propagules [44], which can be the sents a doubling of alleles but in a smaller sampling area. case even if recruitment of sexual offspring into estab- Additionally, when our observed heterozygosities are com- lished populations is rare. This may be the case for pared to the primers of Hoffberg et al. [14], they ranged kudzu due to its deliberate introduction by landowners from 0.372–0.726 (Table 7), while Hoffberg et al. [14] into novel habitats from seed stock. Additionally, clonal ranged from 0.0–0.9 and Bentley and Mauricio ranged from populations are capable of maintaining higher genetic 0.004–0.741. The large difference in the heterozygosity diversity at each locus even though they support a lower comparisons, particularly when focusing on the low end, number of different genotypes [45, 46]. Our results cor- may be attributed to differences in sampling strategies. roborate the findings that introduced kudzu displays Bentley and Mauricio [15] report sampling kudzu within a high levels of genetic variation throughout the US (Table Haynsen et al. BMC Genomics (2018) 19:439 Page 11 of 17 6, Additional file 13); however, we still maintain that the and confidence for testing genetic associations between high genetic variation is possibly indicative of multiple introduced and native kudzu, efforts that are currently introductions from across its native range. underway. Population structure and introduction history of kudzu Conclusions Kudzu is said to have first been brought to the US by the This study produced critical genomic resources for the Japanese who planted it as an ornamental vine outside highly invasive kudzu vine by characterizing transcrip- their pavilion at the 1876 World’s Fair Centennial Exhib- tomes and producing marker databases for SNP and ition in Philadelphia [47]. Later, David Fairchild, a plant eSSR markers, foundational resources for understanding explorer for the United States Department of Agriculture, ecological adaptation that may enable future insights noted its uses, including as forage, in Japan and brought into invasiveness through gene discovery, marker-trait back some seeds to plant near his home in Washington, analyses, and further genetic diversity studies. We exem- D.C., as a trial. In the 1930’s, the US government began plified the utility of our marker databases by assessing planting millions of seedlings across the southeastern the genetic diversity of native and introduced popula- states as a means of erosion control. Whether the US gov- tions of kudzu using seven eSSRs. As a naturalized inva- ernment sourced these kudzu seedlings from one or mul- sive vine that was intentionally introduced throughout tiple native populations from Japan or elsewhere is not millions of acres of the southeastern US, kudzu presents known. unique challenges for management, especially given its Although there is consensus across most studies show- high genetic diversity across the US, a finding supported ing robust findings of high levels of genetic variation of by our genetic diversity analyses. The origin of this gen- kudzu in the US, most of the studies reported a lack of etic diversity remains a matter of speculation, however, geographic patterning of genotypes, and none included this study has begun to refine the proposed hypothesis wide sampling across Asia so as to enable an investiga- of single or multiple introductions from different genetic tion into source populations of US introduction(s). Our populations. This study is the first to provide molecular results include new clues in identifying the native origins evidence that indicates the island of Honshu, Japan as of US kudzu. The Thailand subpopulation is composed one source of US kudzu. Our analyses suggest either a of non-P. montana var. lobata individuals. With evi- single introduction from a highly diverse source popula- dence for strong genetic differentiation and zero popula- tion in Japan, or more likely multiple introductions from tion admixture between Thailand and other multiple sources, potentially also from northern Japan subpopulations, we can definitively rule Thailand out as (Island of Hokkaido) or northern China. Given the eco- a source of US kudzu introductions. It may also be pos- logical and economic devastation wrought by kudzu in sible to rule southern China out as an origin of US the United States, it is critical that we improve our un- kudzu introductions due to pairwise comparisons with derstanding of the history, process, origin(s), and im- the central and southern US, which showed moderate pacts of the U.S. kudzu invasion. We have assembled levels of genetic variation (Table 8), as well as the distant transcriptomes and mined them for eSSRs that we have placement of China 3 on the NJ tree (Fig. 6). provided as a resource for further genetic studies into Of particular interest in the investigation of source the origin(s) and range expansions of kudzu to that end. populations for the introduction of US kudzu is the NJ By increasing both the sample ranges and sizes it should tree clade composed of all the US subpopulations and be possible to identify more accurately the origin of Japan 2, the centrally located Japanese subpopulation introduction and the number of introductions with the (Fig. 6). With a bootstrap value of 50, these four subpop- markers we have developed, efforts that are currently ulations can be distinguished from the rest of the tree underway. and within this clade, Japan 2 and US 3, the southern US, are paired together with a support of 92. These find- Methods ings suggest that central Japan is a source of US kudzu. Plant material for transcriptome sequencing and Its association with US 3, the southern US populations, population genetics makes sense considering that this area was where kudzu Transcriptomic work in this study incorporated plant was first planted for soil erosion control and where tissues from two accessions of kudzu, P. montana var. farmers cultivated kudzu for fodder at the behest of the lobata, and one accession of tropical kudzu, N. phaseo- US government. Our study is the first to provide mo- loides [formerly Pueraria phaseoloides (Roxb.) Benth.]. lecular evidence to support the hypothesis of Japan as a One kudzu accession (noted here as Pmnk6) was wild genetic source of US kudzu. However, a wider sampling collected from Williamsburg, Virginia [voucher speci- across the native Asian range coupled with higher num- men G. Tate s.n. (WILLI) collected 8 July 2013]. Leaf tis- bers of genetic markers would increase statistical power sue was collected in RNALater and preserved at − 20 ° F Haynsen et al. BMC Genomics (2018) 19:439 Page 12 of 17 prior to RNA extraction. The other two plants were quality score thresholds (hq) were set to 25 and low grown from seed obtained from the United States De- quality score thresholds (lq) were set to 10; the fraction of partment of Agriculture (USDA) Germplasm Resources bases per read having to exceed hq were set to 0.8 and the Information Network seed bank: accession PI 434246 of minimum number of high quality bases (mh) and max- P. montana var. lobata (noted here as CPP27) was field imum number of low quality bases (ml) within the sliding collected in 1979 from the United States, locality un- window were set to 30 and 5, respectively. Cleaned reads known, and is maintained by the Coffeeville Plant Mate- were de novo assembled using Trinity (v2.0.6) [51]under rials Center, Soil Conservation Service, Coffeeville, MS; default parameters on two high-performance computing accession PI 470272 of N. phaseoloides (noted here as clusters: the Smithsonian Institution High Performance CPP02) was donated in 1981 from a field collection by Cluster (SI/HPC) and the George Washington University D.R. Bienz, 5 Jun 1981, Banjarbaru, S. Kalimantan, Colonial One Cluster. In order to minimize redundant Indonesia. Seeds were grown to maturity in the green- transcripts, a by-product of the assembly process, house at Cornell University (Ithaca, NY, US) for 3 years CD-HIT-EST was used with a threshold of 0.9 to obtain prior to RNA extraction. For eSSR screening and popu- unique transcripts [52]. To evaluate the quality of the as- lation genetic studies, we sampled 75 accessions repre- semblies, criteria including the number of aligned reads, senting all three varieties of P. montana throughout total number of contigs produced, mean contig length, their native and US introduced range: US (25), China N50, and transcript annotations were considered. RSEM (21), Japan (22) and Thailand (7) (Additional file 14). [53]and Bowtie2[54] were used to identify the number of Leaf material was immediately stored in silica for desic- aligned reads in the assembled transcriptomes. The KRA- cation. Genomic DNA was extracted from samples using KEN suite was utilized in conjunction with prokaryote and Autogen robotics (Autogen Inc.) and a modified CTAB fungal databases to identify potential contaminants within extraction protocol [48]. the transcriptomes [55]. BUSCO(v1.1b1), apipelineused to accurately annotate core genes in eukaryotic genomes, RNA extraction and transcriptome sequencing was used to determine the completeness of the assemblies For the two accessions raised in the greenhouse, tissues [56]. At the time of use, BUSCO utilized a plant core data- were flash frozen in liquid nitrogen prior to RNA extrac- base of 956 single copy genes that are shared between Ara- tion. Neustanthus phaseoloides (CPP02) was sampled for bidopsis, Oryza, Populus,and Vitis [57]. Reciprocal Best young leaves, young shoot tips, and buds. Unfortunately, BLAST Hits (RBH) between transcripts and among tran- kudzu never flowered in the greenhouse, so only young scripts were performed on a local installation of Galaxy shoot tips and young leaves were harvested for CPP27. [58–60]and Toolshed [61] to characterize the number of For the wild collected kudzu (Pmnk6), only young leaves shared, homologous transcripts recovered in each Trinity were harvested. RNA extraction, cDNA library construc- assembled transcriptomes [62, 63]. tion, and transcriptome sequencing were carried out as previously described [31]. cDNA libraries from CPP27 Functional annotation of transcriptomes and CPP02 were multiplexed with two other libraries We used transcripts (contigs + singletons) assembled by not reported here across one titer plate on the Roche Trinity to annotate our transcriptomes (CPP27, CPP02, 454 Genome Sequencer FLX platform using Titanium and Pmnk6). To identify candidate coding regions, we chemistry at the Brigham Young University Sequencing filtered sequences based on a minimum amino acid Center (Provo, UT, US). Pmnk6 was also multiplexed length of 100 using the TransDecoder program v2.0.1 with three other transcriptomes not reported here and [64] with the TransDecoder.LongOrfs command. BlastP sequenced using Roche 454 pyrosequencing, but using and Pfam searches were carried out to detect open Roche’s next improvement on the titanium chemistry reading frames (ORFs) with similarity to known pro- that produced reads ~ 800 bp long. The raw sequence teins and to maximize sensitivity for capturing ORFs data generated from CPP27, Pmnk6, and CPP02 were that may have functional significance. The BlastP deposited at the National Center for Biotechnology Infor- search was done using the Swissprot database with mation (NCBI) Sequence Read Archive (SRA) under acces- the E-value of 1E-5 and Pfam search was done using sion numbers SRR5925648, SRR5925647, and SRR5925649, HMMER [65]and thePfamdatabase[66]. Output respectively. files from the BlastP and Pfam searches were used to ensure that peptides with BLAST or domain hits were De novo transcriptome assemblies retained by running the TransDecoder.Predict com- Raw reads were assessed for quality with FastQC [49] mand. The peptide sequences from the final candidate and subsequently cleaned with ConDeTri [50], a ORFs were used to run BlastP searches against the content-dependent read trimmer under the following NCBI’s nonredundant (nr) database with the E-value settings: reads below 50 bp were removed, Phred high of 1E-5 on the SI/HPC. The BLAST results were then Haynsen et al. BMC Genomics (2018) 19:439 Page 13 of 17 Fig. 7 Sampling sites: (a) United States: US 1, US 2, US 3 (25); (b) Japan: JP 1, JP 2, JP 3 (22); and (c) China: CN 1, CN 2, CN 3 (21) and Thailand: TH (7) imported into the Blast2GO program v1.9.3 [67] to as- Single nucleotide polymorphism identification sign Gene Ontology (GO) terms. We ran mapping, anno- For SNP identification among the kudzu accessions, we tation and InterProScan analyses for the three used the transcripts (contigs + singletons) from our CPP27, transcriptomes separately. Pmnk6, and CPP02 assemblies and also incorporated two Haynsen et al. BMC Genomics (2018) 19:439 Page 14 of 17 publicly available P. montana var. lobata transcriptomes, [69] transcriptomes were not utilized for eSSR discovery SRX480408 from China derived from two tissues [68], and because none were available at the time eSSR mining took DRA001736 from Japan consisting of five pooled tissues place. Thirty potential eSSR primer pairs were chosen [69]. We assembled the public sequences using Trinity as from those discovered here and initially screened against a described above. Multiple pairwise comparisons between subset of accessions (Additional file 11). Seventeen of the transcriptomes were conducted to evaluate the distribu- 30 primer pairs represent putatively homologous eSSRs tion of SNPs between US kudzu samples (CPP27 vs. present in both P. montana var. lobata and N. phaseo- Pmnk6) and identify intergeneric SNPs between kudzu loides (primer pairs designated PP) while the rest are P. and N. phaseoloides (CPP27 vs. CPP02 and Pmnk6 vs. montana var. lobata specific (primer pairs designated PL). CPP02). Additionally, the two US kudzu samples were The method of Culley et al. [19] was used to screen, combined by concatenating the two transcript files such optimize and amplify eSSRs. Primer pairs were eliminated that the samples represent the diversity in ‘US kudzu’ and based on the Culley et al. [19] protocol if they produced subsequently compared to tropical kudzu to further iden- superfluous primer diming between the specific and tailed tify intergeneric SNPs (CPP27/Pmnk6 vs. CPP02). Lastly, primers or produced PCR products of unexpected size. SNPs were called via comparison of all four P. montana Primer pairs were further eliminated if 1) primers did not var. lobata transcriptomes, with the transcriptome from amplify viable product as seen via gel electrophoresis, 2) Japan used as reference (Japan vs. CPP27/Pmnk6/China). primers amplified more bands than expected, or 3) The Japan transcriptome was chosen as reference because primers were monomorphic. it incorporated the highest number of tissues, thus puta- Screening of primer pairs against a subset of seven ac- tively having the higher chance of capturing greater cessions ultimately yielded seven primer pairs that were expressed sequence diversity within the genome. To call characterized across all 75 accessions. Primers, fluores- SNPs, GS Reference Mapper v2.9 (454 Life Sciences, cent dyes, and Culley method tail adaptors used for each Roche, US) was used under default settings. The transcrip- of the seven eSSRs are listed in Table 5. Initial rounds of tome composed of the greatest number of tissues was used amplification across the entire sampling set were per- as the reference to which reads from the others were as- formed in 12 μL reactions containing 1X Biolase NH sembled against. We used only high-confidence variants buffer, 1.0 μL primer mix, 1.2 mM MgCl , 0.12 μLof (454HCDiffs, > 95%) in each comparison and further fil- 8 μM dNTPs, 0.35 U of Taq polymerase (Biolase), and tered these variants to those having 20× or greater cover- 5-80 ng DNA template. PCR was performed on an Ap- age. To ensure the highest SNP call quality, we discarded plied Biosystems 2720 thermocycler with settings of 95C any SNPs where 1) the reference or variant involved one or for 5 min, followed by 35 cycles of 95C for 30s, 50C for more N’s or 2) the reference or variant allele was a single 45 s, 72C for 30s, and a final 72C extension for 5 min. nucleotide insertion or deletion or did not include a point Annealing temperatures were adjusted between mutation in the length variant [70]. 51.5C-58C for primers PP13, PL1, PL11, and PP2. Prod- uct bands were resolved using 1.5% sodium borate gels Expressed simple sequence repeat (eSSR) loci discovery, containing GelRed stain and visualized under UV light. screening and characterization Accessions that failed to amplify after two or more initial The ConDeTri cleaned, Trinity assembled, and attempts were subsequently attempted with an adjusted redundancy-vetted transcripts of CPP27 and CPP02 concentration of 2.38 μg MgCl per reaction. Further were mined for di-, tri-, tetra-, penta-, and hexanucleo- failed amplifications were then tried using AmpliTaq Gold tide microsatellites with MSATCOMMANDER [71]. using reaction mix 1X AmpliTaq buffer, 1.0uL of primer Afterwards, MSATCOMMANDER and Primer3 [72] mix, 2.86 μg MgCl , 1.2uL of 8 μMdNTPs,0.375 Uof were used to design primer pairs for each species with AmpliTaq Gold Taq polymerase [0.075 μLof1000 U in an expected product size ranging from 100 to 450 bp. 200 μL], and 5-80 ng DNA template. Successful products Primer lengths were allowed to range from 18 to 22 bp, were genotyped using an ABI3730 sequencer at the annealing temperatures were optimized at 60 °C, and Smithsonian NMNH LAB facilities. Genotypes were called GC contents were held between 30 and 70%. Developed using GeneMapper (v5.0) [74]. primers for both species were then cross-compared to identify homologous primer regions, which could signify Examination of population structure and genetic diversity interspecies transferability. The corresponding tran- indices scripts for primers that were shared between P. lobata Genetic population structuring was assessed with and N. phaseoloides were blasted against the GenBank STRUCTURE v2.3.4 [75] and STRUCTURE HAR- nonredundant database using BLASTX [73]with an VESTER v0.6.94 [76]. The length of burnin period was − 10 E-value of 10 to determine the function of their associ- set to 100,000, while the number of MCMC reps after ated unigenes. Pmnk6, SRX480408 [68] and DRA001736 burnin was set to 900,000, resulting in a total of 1 Haynsen et al. BMC Genomics (2018) 19:439 Page 15 of 17 million generations. No LOCPRIOR information was with variant; # Fwd Total: number of forward-aligned reads total; # Rev. provided for the STRUCTURE runs. A job consisting of Total: number of reverse-aligned reads total. (XLSX 578 kb) 10 iterations, evaluating Ks from 1 to 10 for the 75 P. Additional file 7: SNPs_CPP02_vs_CPP27. High-confidence single nu- cleotide polymorphisms between tropical kudzu CPP02 (reference: Ref) montana accessions, was run and the results were and kudzu accession CPP27 (variant: Var). Abbreviations as described for uploaded to STRUCTURE Harvester for analysis. The Additional file 6. (XLSX 4932 kb) optimal K was assessed via the Evanno et al. [77] Additional file 8: SNPs_CPP02_vs_Pmnk6. High-confidence single nu- method. Individual and population files were loaded into cleotide polymorphisms between tropical kudzu CPP02 (reference: Ref) and kudzu accession Pmnk6 (variant: Var). Abbreviations as described for CLUMPP v1.1.2 [78] to address label switching and the Additional file 6. (XLSX 11073 kb) potential for multimodality across the 10 STRUCTURE Additional file 9: SNPs_CPP02_vs_Pmnk6_CPP27. High-confidence sin- iterations. The CLUMPP program utilized the FullSearch gle nucleotide polymorphisms between tropical kudzu CPP02 (reference: method, the number of individuals in each population Ref) and a composite transcriptome comprising reads from kudzu accessions CPP27 and Pmnk6 (variant: Var). Abbreviations as described for influenced weights, and the pairwise matrix similarity Additional file 6. (XLSX 11520 kb) statistic was set to G’. All additional options remained as Additional file 10: SNPs_Japan_vs_Pmnk6_CPP27_China. High- default settings. CLUMPP outputs for the individual and confidence single nucleotide polymorphisms among kudzu accessions population files were visualized with DISTRUCT v1.1 from Japan (reference: Ref) and reads from US kudzu (Pmnk6 and CPP27) and China (variants: Var). Abbreviations as described for Additional file 6. [79]. Genetic diversity statistics were calculated in Arle- (XLSX 30817 kb) quin v3.5.1.9 [80]. The default parameters of Arlequin Additional file 11: Table S3. Thirty primer pairs tested for polymorphic were used on our 75-individual data set that was subdi- amplification in Pueraria montana. Primers labeled PP were designed vided from the four sampled nations to 10 geographic- from kudzu and tropical kudzu transcriptomes whereas those designated PL were designed from kudzu only. Bold primers are those used for ally defined subpopulations: US (3), China (3), Japan (3), population genetic analyses in this study. F: forward primer; R: reverse and Thailand (1) (Fig. 7). The subpopulation designa- primer. (PDF 33 kb) tions were based primarily on geographic proximity that Additional file 12: Figure S4. Delta K of STRUCTURE run (K = 3). Plot of allowed for groupings of at-least five individuals along Delta K for STRUCTURE analyses from K = 2 through K = 9, with K = 3 seen as the optimal number of genetic clusters. (PDF 18 kb) similar latitudinal lines; however, due to the different Additional file 13: Table S4. Allele table for Pueraria subpopulations. scales of sampling done across nations, the ranges of the Number of alleles discovered for each locus within each subpopulation, latitudinal boundaries of the subpopulations differed. with mean and standard deviation (SD) for each subpopulation and each POPTREEW [81] was used to make a neighbor joining locus. (PDF 19 kb) (NJ) distance tree with F distances [82] for the above Additional file 14: Table S5. Plant material used for eSSR validation st and population genetics. Species determination, subpopulation listed subpopulations. Bootstrap support for the tree was designation (pop), country and state/province/island of origin within the calculated with 10,000 replicates. United States (US), China (CN), Japan (JP) or Thailand (TH), voucher information, accession number, and geographical coordinates for each of the 75 plants used in the population genetic analyses. (PDF 34 kb) Additional files Additional file 1: Table S1. Trinity contig reads mapped back to the Abbreviations raw and cleaned reads. Numbers of cleaned and raw reads mapped back BLAST: Basic local alignment search tool; bp: Base pair; to contigs via Bowtie2. (PDF 126 kb) BUSCO: Benchmarking universal single-copy orthologs; eSSR: Expressed simple sequence repeat; GO: Gene ontology; hq: High quality; lq: Low Additional file 2: Table S2. Contaminated reads as assessed by Kraken. quality; mh: Minimum high quality; ml: Maximum low quality; NCBI: National Number (percentage) of cleaned reads annotated by Kraken as Center for Biotechnology Information; nr: Nonredundant; ORF: Open reading prokaryotic or fungal. (PDF 126 kb) frame; RBH: Reciprocal best hits; RIN: RNA integrity; SI/HPC: Smithsonian Additional file 3: Figure S1. CPP27 Top-Hit Species Distribution. Top-hit Institution High Performance Cluster; SNP: Single nucleotide polymorphism; species distribution of CPP27 proteins annotated against NCBI’s SRA: Sequence read archive; SSR: Simple sequence repeat; TNR: Tri- non-redundant database showing the highest distribution of hits against nucleotide repeat legume species. (PDF 808 kb) Additional file 4: Figure S2. Pmnk6 Top-Hit Species Distribution. Top-hit species distribution of Pmnk6 proteins annotated against NCBI’s Acknowledgements non-redundant database showing the highest distribution of hits against We thank Susan Sherman-Broyles and Jane L. Doyle for help in sustaining legume species. (PDF 753 kb) plants in the greenhouse and to Beth Chambers and Gus Tate, Herbarium of the College of William and Mary, for help in obtaining voucher specimens Additional file 5: Figure S3. CPP02 Top-Hit Species Distribution. Top-hit for Pmnk6. Additionally, we thank Cheng-Xin Fu, Lu-Xian Liu, Xin-fen Gao species distribution of CPP02 proteins annotated against NCBI’s non- and Bo Xu for assistance collecting in China, Tetsukazu Yahara, Tadashi Kajita, redundant database showing the highest distribution of hits against leg- Firouzeh Javadi, Tomoko Otao and Yumi Kagawa for help in Japan, and ume species. (PDF 2607 kb) Voradol Chamchumroon, Kongkanda Chayamarit, Thaveechok Jumruschay Additional file 6: SNPs_Pmnk6_vs_CPP27. High-confidence single nu- Rumrada Meeboonya, Nannapat Pattharahirantricin, Rachun Pooma, Sukontip cleotide polymorphisms between US kudzu accessions Pmnk6 (variant: Sirimongkol, and Ruth P. Clark for help in Thailand. Computations were Var) and CPP27 (reference: Ref). Accno: contig in reference; Pos: position; completed in part on the Smithsonian Institution High Performance Cluster Nuc: nucleotide; Total Depth: number of variant reads aligned against the (SI/HPC) and the George Washington University Colonial One Cluster. We reference; Var Freq: frequency of variant SNP within aligned reads; # Fwd: also thank the Computational Biology Institute at the George Washington number of forward reads with variant; # Rev.: number of reverse reads University for graduate support for MSH. Haynsen et al. BMC Genomics (2018) 19:439 Page 16 of 17 Funding 6. Forseth IN Jr, Innis AF. Kudzu (Pueraria montana): history, physiology, and This research was supported by funding from the US National Science ecology combine to make a major ecosystem threat. Crit Rev Plant Sci. Foundation to ANE (DEB-1352217) and JJD (DEB-0948800). 2004;23:401–13. 7. Follak S. Potential distribution and environmental threat of Pueraria lobata. Availability of data and materials Cent Eur J of Biol. 2011;6:457–69. The transcriptomes generated and analyzed during the current study are 8. Westbrooks R. Invasive plants, changing the landscape of America: fact available in the NCBI repository, [Study PRJNA397892, accessions: book. In: Federal Interagency Committee for the Management of Noxiuous SRR5925647, SRR5925648, and SRR5925649, http://www.ncbi.nlm.nih.gov/ and Exotic Weeds: Washington; 1998. bioproject/397892, release date 30 June 2018]. The SNP data generated 9. Kudzu SD. In: Simberloff D, Rejmanek D, editors. Encyclopedia of biological during this study are included in this published article’s Additional files 6, 7, invasions. California: University of California Press; 2011. p. 396–9. 8, 9 and 10 however, the SNPs contained in Additional file 10 are not 10. Pappert RA, Hamrick JL, Donovan LA. Genetic variation in Pueraria lobata publicly available due to file size restrictions but are available from the (Fabaceae), an introduced, clonal, invasive plant of the southereastern corresponding author upon reasonable request. United States. Am J Bot. 2000;87:1240–5. 11. Jewett DK, Jiang CJ, Britton KO, Sun JH, Tang J. Characterizing specimens of Authors’ contributions kudzu and related taxa with RAPDs. Castanea. 2003;68:254–60. All authors contributed to various aspects of this work (ordered by degree of 12. Sun JH, Li Z-C, Jewett DK, Britton KO, Ye WH, Ge X-J. Genetic diversity of contribution): conceived the study (ANE, MSH); aided in experimental design Pueraria lobata (kudzu) and closely related taxa as revealed by inter-simple (MSH, ANE); obtained research funds (ANE, JJD); coordinated activities (ANE, sequence repeat analysis. Weed Res. 2005;45:255–60. MSH); obtained and grew plants (ANE, MSH, JJD); RNA Isolation and Library 13. Heider B, Fischer E, Berndl T, Schultze-Kraft R. Analysis of genetic variation Prep (ANE); transcriptome assembly and analyses (MSH, MV, ANE); among accessions of Pueraria montana (Lour.) Merr. var. lobata and Pueraria microsatellite primer design (MSH); microsatellite primer validation (MSH, GM, phaseoloides (Roxb.) Benth. based on RAPD markers. Genet Resour Crop DZ, RZMR); prepared figures (MSH, MV, ANE); contributed to preparation of Evol. 2007;54:529–42. the manuscript (MSH, ANE, MV, JJD, KAC). All authors edited and approved 14. Hoffberg SL,Bentley KE, Lee JB,Myhre KE, Iwao K,Glenn TC, etal. the final manuscript. Characterization of 15 microsatellite loci in kudzu (Pueraria montana var. lobata) from the native and introduced ranges. Conserv Genet Resour. 2015;7:403–5. Ethics approval and consent to participate 15. Bentley K, Mauricio R. High degree of clonal reproduction and lack of large- All plant material was collected in accordance with institutional, national, scale geographic patterning mark the introduced range of the invasive vine, and international guidelines and under appropriate permits. Permits and kudzu (Pueraria montana var. lobata) in North America. Am J Bot. 2016;103: voucher specimens are deposited at the US National Herbarium (US) with all 1499–507. specimens determined by Dr. Ashley N. Egan. 16. Strickler SR, Bombarely A, Mueller LA. Designing a transcriptome next- generation sequencing project for a nonmodel plant species. Am J Bot. Competing interests 2012;99:257–66. The authors declare that they have no competing interests. 17. Varshney RK, Sigmund Rm Borner A, Korzun V, Stein N, Sorrells ME, et al. Interspecific transferability and comparative mapping of barley EST-SSR markers in wheat, rye, and rice. Plant Sci. 2005;168:195–202. Publisher’sNote 18. Ellis J, Burke J. EST-SSRs as a resource for population genetic analyses. Springer Nature remains neutral with regard to jurisdictional claims in Heredity. 2007;99:125–32. published maps and institutional affiliations. 19. Culley TM, Stamper TI, Stokes RL, Brzyski JR, Hardiman NA, Klooster MR, et al. An efficient technique for primer development and application that integrates Author details 1 fluorescent labeling and multiplex PCR. Appl Plant Sci. 2013;1:1–10. Department of Biology, George Washington University, Washington, DC, 2 20. Wright S. Evolution and the genetics of populations. Vol. 4. Variability within USA. Computational Biology Institute, Milken Institute School of Public 3 and among natural populations. Chicago: University of Chicago Press; 1978. Health, George Washington University, Washington, DC, USA. Department of 21. Stewart CN Jr, Tranel PJ, Horvath DP, Anderson JV, Rieseberg LH, Westwood Botany, National Museum of Natural History, Smithsonian Institution, JH, et al. Evolution of weediness and invasiveness: charting the course for Washington, DC, USA. Present address: College of Engineering, Oregon weed genomics. Weed Sci. 2009;57:451–62. State University, Corvallis, OR, USA. Present address: Department of Biology, 22. Ekblom R, Galindo J. Applications of next-generation sequencing in Washington University in St. Louis, St. Louis, MO, USA. Present address: molecular ecology of non-model organisms. Heredity. 2011;107:1–15. Department of Biology, Indiana University Bloomington, Bloomington, IN, 23. Egan AN, Schlueter J, Spooner DM. Applications of next-generation USA. School of Integrated Plant Science, Plant Breeding and Genetics sequencing in plant biology. Am J Bot. 2012;99:175–85. Section, Cornell University, Ithaca, NY, USA. Department of Invertebrate 24. Wen J, Egan AN, Dikow RB, Zimmer EA. Utility of transcriptome sequencing Zoology, National Museum of Natural History, Smithsonian Institution, for phylogenetic inference and character evolution. In: Hörandl E, Washington, DC, USA. Appelhans MS, editors. Next-generation sequencing in plant systematics. Königstein: Koeltz scientific books; 2015. p. 51–91. Received: 25 April 2018 Accepted: 15 May 2018 25. Ilut DC, Coate JE, Luciano AK, Owens TG, May GD, Farmer A, et al. A comparative transcriptomic study of an allotetraploid and its diploid progenitors illustrates the unique advantages and challenges of RNA-Seq in References plant species. Am J Bot. 2012;99:383–96. 1. van der Maesen LJG. Pueraria: Botanical characteristics. In: Keung WM, editor. 26. Grover CE, Salmon A, Wendel JF. Targeted sequence capture as a powerful Pueraria:the genus Pueraria. New York: Taylor and Francis; 2002. p. 1–28. tool for evolutionary analysis. Am J Bot. 2012;99:312–9. 2. Egan AN, Pan B. Resolution of polyphyly in Pueraria (Leguminosae, 27. Egan AN, Doyle J. A comparison of global, gene-specific, and relaxed clock Papilionoideae): the creation of two new genera, Haymondia and methods in a comparative genomics framework: dating the polyploid Toxicopueraria, the resurrection of Neustanthus, and a new combination in history of soybean (Glycine max). Syst Biol. 2010;59:534–47. Teyleria. Phytotaxa. 2015;218:201–26. 28. Cannon SB, McKain MR, Harkess A, Nelson MN, Dash S, Deyholos MK, et al. 3. van der Maesen LJG. Revision of the genus Pueraria DC. with some notes Multiple polyploid events in the early radiation of nodulating and on Teyleria Backer: Leguminosae. Wag Ag Un P. 1985;85:1–132. nonnodulating legumes. Mol. Biol. Evol. 2015;32:193–210. 4. van der Maesen LJG. Pueraria, the kudzu and its relatives, an update of the taxonomy. In: Sorensen M, editor. Proceedings of the first international 29. Glover NM, Redestig H, Dessimoz C. Homoeologs: what are they and how symposium on tuberous legumes, Gualdeloupe, FWI. Denmark: DSR do we infer them? Trends Plant Sci. 2016;21:609–21. Boghandel; 1994. p. 55–86. 30. Champigny MJ, Sung WW, Catana V, Salwan R, Summers PS, Dudley SA, et 5. Egan AN, Vatanparast M, Cagle W. Parsing polyphyletic Pueraria: delimiting al. RNA-Seq effectively monitors gene expression in Eutrema salsugineum distinct evolutionary lineages through phylogeny. Mol. Phylogenet. Evol. plants growing in an extreme natural habitat and in controlled growth 2016;104:44–59. cabinet conditions. BMC Genomics. 2013;14:578. Haynsen et al. BMC Genomics (2018) 19:439 Page 17 of 17 31. Vatanparast M, Shetty P, Chopra SP, Doyle JJ, Sathyanarayana N, Egan AN. 57. Duarte JM, Wall PK, Edger PP, Landherr LL, Ma H, Pires PK, et al. Transcriptome sequencing and marker development in winged bean Identification of shared single copy nuclear genes in Arabidopsis, Populus, (Psophocarpus tetragonolobus; Leguminosae). Sci Rep. 2016;6:29070. Vitis, and Oryza and their phylogenetic utility across various taxonomic 32. Kudapa H, Azam S, Sharpe AG, Taran B, Li R, Deonovic B, et al. Comprehensive levels. BMC Evol Biol. 2010;10:61. transcriptome assembly of chickpea (Cicer arietinum L.) using sanger and next 58. Goecks J, Nekrutenko A, Taylor J. The galaxy team. Galaxy: a comprehensive generation sequencing platforms: development and applications. PLoS One. approach for supporting accessible, reproducible, and transparent 2014;9:e86039. computational research in the life sciences. Genome Biol. 2010;11:R86. 33. Sudheesh S, Sawbridge TI, Cogan NO, Kennedy P, Forster JW, Kaur S. De 59. Blankenberg D, Von Kuster G, Coraor N, Ananda G, Lazarus R, Mangan M, et novo assembly and characterization of the field pea transcriptome using al. Galaxy: a web-based genome analysis tool for experimentalists. Curr RNA-Seq. BMC Genomics. 2015;16:611. Protoc Mol Biol. 2010;Unit-19.1021. 60. Giardine B, Riemer C, Hardison RC, Burhans R, Elnitski L, Shah P, et al. 34. Khalturin K, Hemmrich G, Fraune S, Augustin R, Bosch TCG. More than just Galaxy: a platform for interactive large-scale genome analysis. Genome orphans: are taxonomically-restricted genes important in evolution? Trends Res. 2005;15:1451–5. Genet. 2009;25:404–13. 61. Blankenberg D, Von Kuster G, Bouvier E, Baker D, Afgan E, Stoler N, et al. 35. Kumar S, Banks TW, Cloutier S. SNP discovery through next-generation Dissemination of scientific software with galaxy toolshed. Genome Biol. sequencing and its applications. Int J Plant Genomics. 2012; https://doi.org/ 2014;15:403. 10.1155/2012/831460. 62. Cock PJA, Chilton JM, Grüning B, Johnson JE, Soranzo N. NCBI BLAST+ 36. Varshney RK, Chabane K, Hendre PS, Aggarwal RK, Graner A. Comparative integrated into galaxy. GigaScience. 2015;4:39. assessment of EST-SSR, EST-SNP and AFLP markers for evaluation of genetic 63. Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos, Bealer K, et al. diversity and conservation of genetic resources using wild, cultivated and BLAST+: architecture and applications. BMC Bioinformatics. 2009;10:421. elite barleys. Plant Sci. 2007;173:638–49. 64. Haas B, Papanicolaou A. TransDecoder https://transdecoder.github.io. 37. Zalapa JE, Cuevas H, Zhu H, Steffan S, Senalik D, Zeldin E, et al. Using next- Accessed 14 Aug 2017. generation sequencing approaches to isolate simple sequence repeat (SSR) 65. Finn RD, Clements J, Eddy SR. HMMER web server: interactive sequence loci in the plant sciences. Am J Bot. 2012;99:193–208. similarity searching. Nucleic Acids Res. 2011;39:W29–37. 38. Morgante M, Hanafey M, Powell W. Microsatellites are preferentially associated 66. Sonnhammer ELL, Eddy SR, Durbin R. Pfam: a comprehensive database of with nonrepetitive DNA in plant genomes. Nat Genet. 2002;30:194–200. protein doman families based on seed alignments. Proteins. 1997;28:405–20. 39. Lopez L, Barreiro R, Fischer M, Koch MA. Mining microsatellite markers from 67. Conesa A, Götz S, García-Gómez J, Terol J, Talón M, Robles M. Blast2GO: a public expressed sequence tags databases for the study of threatened universal tool for annotation, visualization and analysis in functional plants. BMC Genomics. 2015;16:781. genomics research. Bioinformatics. 2005;21:3674–6. 40. Thumilan BM, Sajeevan RS, Biradar J, Madhuri T, Nataraja KN, Sreeman SM. 68. Wang X, Li S, Li J, Li C, Zhang Y. De novo transcriptome sequencing in Development and characterization of genic SSR markers from Indian Pueraria lobata to identify putative genes involved in isoflavones mulberry transcriptome and their transferability to related species of biosynthesis. Plant Cell Rep. 2015;34:733–43. Moraceae. PLoS One. 2016;11:e0162909. 69. Han R, Takahashi H, Nakamura M, Yoshimoto N, Suzuki H, Shibata D, et al. 41. Wang P, Yang L, Zhang E, Qin Z, Wang H, Liao Y, et al. Characterization and Transcriptomic landscape of Pueraria lobata demonstrates potential for development of EST-SSR markers from a cold-stressed transcriptome of phytochemical study. Front Plant Sci. 2015;6:426. Centipedegrass by Illumina paired-end sequencing. Plant Mol Biol Rep. 70. Schmutz J, McClean PE, Mamidi S, Wu GA, Cannon SB, Grimwood J, et al. A 2017;35:215–23. reference genome for common bean and genome-wide analysis of dual 42. Yang Z, Peng ZS, Yang H. Identification of novel and useful EST-SSR markers domestications. Nat Genet. 2014;46:707–13. from de novo transcriptome sequence of wheat (Triticum aestivum L.). 71. Faircloth BC. MSATCOMANDER: detection of microsatellite repeat arrays and Genet Mol Res. 2016;15:1. automated, locus-specific primer design. Mol Ecol Resour. 2008;8:92–4. 43. Kartzinel TR, Hamrick JL, Wang C, Bowsher AW, Quigley BG. Heterogeneity 72. Rozen S, Skaletsky H. Primer3 on the WWW for general users and for of clonal patterns among patches of kudzu, Pueraria montana var. lobata, biologist programmers. Methods Mol Biol. 2000;132:365–86. an invasive plant. Ann Bot. 2015;116:739–50. 73. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W, et al. 44. Ellstrand NC, Roose ML. Patterns of genotypic diversity in clonal plant Gapped BLAST and PSI-BLAST: a new generation of protein database search species. Am J Bot. 1987;74:123–31. programs. Nucleic Acids Res. 1997;25:3389–402. 45. Balloux F, Lehmann L, de Meeûs T. The population genetics of clonal and 74. Chatterji S, Pachter L. Reference based annotation with GeneMapper. partially clonal diploids. Genetics. 2003;164:1635–44. Genome Biol. 2006;7:R29. 46. Halkett FJ, Simon JC, Balloux FO. Tackling the population genetics of clonal 75. Pritchard JK, Stephens M, Donnelly P. Inference of population structure and partially clonal organisms. Trends Ecol. Evol. 2005;20:194–201. using multilocus genotype data. Genetics. 2000;155:945–59. 47. Shurtleff W, Aoyagi A. The book of kudzu: a culinary and healing guide. 76. Earl DA, vonHoldt BM. STRUCTURE HARVESTER: a website and program for Brookline: Autumn Press; 1997. visualizing STRUCTURE output and implementing the Evanno method. 48. Doyle JJ, Doyle JL. A rapid DNA isolation procedure for small quantities of Conserv Genet Resour. 2012;4:359–61. fresh leaf tissue. Phytochem Bull. 1987;19:11–5. 77. Evanno G, Regnaut S, Goudet J. Detecting the number of clusters of 49. Andrews S. Fast QC. A quality control tool for high throuput sequence data. individuals using the software STRUCTURE: a simulation study. Mol Ecol. http://www.bioinformatics.babraham.ac.uk/projects/fastqc/. Accessed 14 Aug 2017. 2005;14:2611–20. 50. Smeds L, Künstner A. ConDeTri – a content dependent read trimmer for 78. Jakobsson M, Rosenberg NA. CLUMPP: a cluster matching and permutation Illumina data. PLoS One. 2011;6:e26314. program for dealing with label switching and multimodality in analysis of 51. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. Full- population structure. Bioinformatics. 2007;23:1801–6. length transcriptome assembly from RNA-Seq data without a reference 79. Rosenberg NA. DISTRUCT: a program for the graphical display of population genome. Nat Biotechnol. 2011;29:644–52. structure. Mol Ecol Notes. 2004;4:137–8. 52. Fu L, Niu B, Zhu Z, Wu S, Li W. CD-HIT: accelerated for clustering the next- 80. Excoffier L, Laval G, Schneider S. Arlequin (version 3.0): an integrated software generation sequencing data. Bioinformatics. 2012;28:3150–2. package for population genetics data analysis. Evol Bioinforma. 2005;1:47–50. 53. Li B, Dewey CN. RSEM: accurate transcript quantification from RNA-Seq data 81. Takezaki N, Nei M, Tamura K. POPTREEW: Web version of POPTREE for with or without a reference genome. BMC Bioinformatics. 2011;12:323. constructing population trees from allele frequency data and computing 54. Langmead B, Salzberg SL. Fast gapped-read alignment with bowtie 2. Nat other population statistics. Mol Biol Evol. 2014;31:1622–4. Methods. 2012;9:357–9. 82. Latter BDH. Selection in finite populations with multiple alleles. III. Genetic 55. Davis MPA, van Dongen S, Abreu-Goodger C, Bartonicek N, Enright AJ. Kraken: divergence with centripetal selection and mutation. Genetics. 1972;70:475–90. a set of tools for quality control and analysis of high-throughput sequence data. Methods. 2013;63:41–9. 56. Simão FA, Waterhouse RM, Ioannidis P, Kriventseva EV, Zdobnov EM. BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics. 2015;31:3210–2.

Journal

BMC GenomicsSpringer Journals

Published: Jun 5, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off