Asteraceae

We reconstructed a community phylogeny for LDFP using maximum likelihood (ML) and maximum parsimony (MP) algorithms. Three different marker combinations were examined for performance in phylogenetic reconstruction: rbcL+matK, rbcL+trnH-psbA, and rbcL+matK+trnH-psbA. For all combinations of markers, 142 species were included (Heterotrichum cynosum in the Asteraceae was excluded because only sequence data for trnH-psbA was available), with six sequences of rbcLa obtained from GenBank and used in conjunction with our barcode sequences. ML analyses were conducted using RAxML [30] via the CIPRES supercomputer cluster (www.phylo.org). The different locus combinations were partitioned for independent model assessment at each marker. For all combinations of markers a single most likely tree was estimated in addition to running 200–250 bootstrap replicates depending on the marker set. The same gene combinations were used in a MP using PAUP v.4.0 [31] and also run through a local cluster in which we implemented a modification of the parsimony ratchet [32] following Carolan et al. [33] (link). This resulted in a very large number of equally parsimonious trees, with over 350,000 trees produced for the rbcLa+trnH-psbA data set and over 250,000 produced for rbcLa+matK. The three gene matrix produced many fewer trees, approximately 25,000. For both ML and MP trees, a 50% majority tree was constructed and used to quantify overall levels of support for each node within the trees, the rates of well-supported monophylly for taxonomic hierarchies (genus, family, order) and concordance with expected topologies. Analyses were conducted both with and without use of a constraint tree, as described below.

Repeat elements were identified de novo with the RepeatModeler pipeline and masked with RepeatMasker software. Perfect, imperfect and compound SSRs were identified using the SciRoKo SSR-search module (http://kofler.or.at/bioinformatics/SciRoKo). Satellite motifs were discovered using TRF56 (link) and filtered for a minimum length of 80 bp and and 4 repeated monomers. Sequences were then clustered at 90% identity using CD-HIT57 (link) to retrieve representative monomers; only clusters with at least 5 elements were retained. Occurrences on the assembly were obtained via BLASTn alignment of representative sequences with a minimum identity of 90%; only independent non-consecutive matches with a minimum distance of 50 bp were considered for statistics.
Gene prediction utilized reiterative runs of the MAKER suite58 (link). Both EST sequences and RNAseq data were used to guide gene annotation. RNAseq data of eight globe artichoke and cardoon genotypes was retrieved from SRA archive (PRJNA72327). EST sequences for C. cardunculus and other available Compositae species were downloaded from the NCBI as well as the non-redundant (nr) protein database for Viridiplantae. RNAseq reads were aligned to the reference assembly using TopHat2 aligner and de novo transcripts were assembled using the Cufflinks package with default parameters. A first run of MAKER annotation pipeline was carried out by employing only transcript assemblies along with ESTs and protein alignments to retrieve candidate gene predictions. After filtering for high quality preliminary predictions, HMM models for Augustus and SNAP ab initio gene prediction algorithms were produced. Then, by utilizing these ad-hoc HMM models, along with EST and proteins alignments as supporting evidence, final gene models were obtained in a second run of MAKER.
Predicted protein sequences were functionally annotated using InterproScan559 (link) against all the available databases (ProDom-2006.1, Panther-7.2, SMART-6.2, PrositeProfiles-20.89, TIGRFAM-12.0, PrositePatterns-20.89, PfamA-26.0, SuperFamily-1.75, PRINTS-42.0, Gene3d-3.5.0, PIRSF-2.83, HAMAP-201207.4, Coils-2.2). In parallel, the same protein set was clustered using OrthoMCL60 with default parameters. Putative functions were assigned to each protein cluster based in the InterproScan functional predictions for all the members in the cluster. Protein datasets for Arabidposis thaliana, Brassica rapa, Fragaria vesca and Solanum lycopersicum (September 2014), were download from Phytozome V961 (link). A predicted proteome was used for Lactuca sativa (unpublished data). All the proteins were clustered with the C. cardunculus using OrthoMCL to generate orthologous clusters with default parameters.
OrthoMCL Clusters, with species having expanded number of genes, were mined using a chi-square test comparing the counts of genes/species against an expected value. Only clusters with a mean number of genes above five were considered. These clusters were analysed for a significant deviation from mean gene count for species adopting a Bonferroni corrected p-value (p < 0.05). GO enrichment in the globe artichoke specific clusters was calculated with AgriGo (http://bioinfo.cau.edu.cn/agriGO) and visualized with the REVIGO suite (http://revigo.irb.hr). Genomic regions carrying cluster of genes were highlighted by aligning genomic full-length genes on the reference genome using BWA (bwa-sw algorithm; http://bio-bwa.sourceforge.net).

A database was established representing a COS of 1,704 tomato unigenes (1612 Sl, 29 S. habrochaites and 63 S. pennellii) from 113,932 ESTs [15 ]. From these, a single and longest EST was chosen to design primers. Using the tools developed for Compositae Genome database, the position of introns was first estimated using the procedures above. A set of 1,268 primers were designed to amplify across estimated intron sites with primers 50–100 bp from the intron. Amplification of primers was tested on a single line, M82.
Primers that successfully amplified a product were tested for polymorphism using sequencing in a series of three pools representing different degrees of diversity. The design has complementary pools representing each class (fresh market, processing, other) with one diverse line from an alternate class to maximize the chance of detecting a polymorphism within or among pools. Using a series of empirical tests with lines with known SNPs in ratios of 1:7, 1:5, 1:3 and 1:1, we determined that an unknown polymorphism can be reliably detected with sequencing with a 1:3 dilution. Pool 1 consisted of O 9242, FL7600, Ha7998, PI114490; Pool 2 included M82, O 8245, O 88119, NC84173 and; Pool 3 consisted of Sun1642, Heinz1706, O 9242, FL7600 (Table 2). DNA was extracted from each line and was combined in equi-molar concentrations.
For all sequencing reactions, forward and reverse primers were tailed with M13 sequences and sequenced using standard protocols for Sanger sequencing (Applied Biosystems, Foster City, CA) in forward and reverse directions using a ABI 3730 (Applied Biosystems, Foster City, CA). Trace files were trimmed with Phred options -trim_cutoff 0.02" which translates to Phred 17 score. [29 (link)]. Assembly was achieved with Phrap/Consed and options were set at " -retainduplicates and -forcelevel 5". These options were optimized to give the best trim and assembly parameters for calling SNPs. Stringent trim parameters are favored in this case to minimize the high number of false SNPs associated with poor sequence on the ends. Amplicon sizes were estimated and included in Additional file 1, Tables S2 and S3. To calculate a more accurate estimate than from gel electrophoresis, the sequenced contig(s) size was used as a minimum. When greater than one contig per locus was obtained as a result of unpredictably large introns, the forward and reverse contig sizes were added.
SNPs were first identified semi-manually using Polyphred as heterozygotes within pools or homozygous differences among pools. The line, M82, was used as reference to screen amplicons for single copy number. Amplicons with putative SNPs were then amplified in the individual 12 lines (Table 2) and sequenced as described above. Only SNPs showing both homozygous alleles were called. Data was extracted from Polyphred using custom scripts ([30 ] See Additional file 1). Similarly, data for indels were extracted from Polyphred. SSRs (di to tetra repeats) were extracted from all sequenced loci for M82, our reference line, and the various genotypes and reported for all sequenced individuals. The sequence database was analyzed for all known repeats for tomato [1 ]. All loci were cross-referenced to the SGN COSII for tomato, pepper, potato and coffee and associated maps [14 (link)].

The Artemisia vulgaris L. used in this experiment was collected from Tangyin, Henan Province. A. vulgaris L. belongs to the Artemisia genus of the Compositae family. It is a perennial herb or slightly semishrubby plant with a strong fragrance. The taproot of the plant is obvious and slightly long, with many lateral roots, and the plant is 1.5 cm in diameter. This plant often has recumbent rhizomes and vegetative shoots. The stem of the plant grows singly or clustered, with distinct longitudinal ribs; their color is brown or tawny brown, the stem is slightly woody at the base, grassy above, with a few short branches, and the length is 3–5 cm. The stems and branches have gray arachnid-like hair. The leaves are thick, covered with gray–white pubescence above, with white glandular spots and small concave spots, and densely covered with gray–white spider silky hairs on the abaxial surface; the basal leaves have long stalks, fading at the anthesis. The shape of the lower stem leaves is suborbicular or broadly ovate, pinnate and deeply lobed, with 2–3 lobes on each side; the lobes’ shapes are oval or obovate and long elliptic, with 2–3 small teeth on each of them. The main and lateral veins on the back of the stem are dark brown or rusty, and the length of petiole is 0.5–0.8 cm. Middle leaves are ovate, triangular-ovate or subrhomboid, with one (to two) pinnate deeply lobed to hemiclobed, with 2 or 3 lobes on each side; the shape of lobes is ovate, ovate-lanceolate or lanceolate, the leaf base is broadly cuneate tapering into a short petiole, and the leaf veins are conspicuous, bulging on the back, with the leaf stalk base usually without false stipes or very small false stipes. Upper leaves and bracts are pinnatifid lobed, deeply 3-lobed, 3-lobed, or not divided, but the shape is elliptic, long elliptic lanceolate, lanceolate or linear-lanceolate.
The shape of the capitulum is oval, sessile or subsessile, with several to more than 10 of them arranged in small spikes or compound spikes on branches, and usually reconstituted in narrow pinnacle-shaped panicles on the stems, with capitulum decumbent after flowers. Involucral bracts have 3–4 layers, arranged in a imbricate shape, the outer involucral bracts are small, in ovate or narrowly ovate shapes, abaxially densely covered with grayish white silky spider hair, and the edge is membranous; the middle involucral bracts are longer than the outer ones, with a long ovate shape, abaxially covered with silky spider silky hair, and the inner involucral bracts are thin, abaxially nearly glabrous. The inflorescence torr is small, (6–10 in female flowers). The purple corolla is narrowly tubular, 2-lobed on the leaf, and the style is slender, 2-forked at the apex. The inflorescence of bisexual flowers is 8–12, the corolla is tubular or goblet shaped, with glandular spots on outside, and the leaves are purple. The anthers are narrowly linear, with long triangular and pointed apical appendages and inconspicuous pinnacles on the base of anther. The achene has a long ovate or oblong shape. The flowering and fruiting period is from July to October.

The datasets of translated representative transcripts were processed by aligning the contigs against the NCBI NR protein database, which was provided on Oct. 18, 2022, with the default parameter (E-value < 1e-30). In addition, the metagenomic community was visualized using Krona (version 2.8.1) (Ondov et al., 2011 (link)). Further, the major transcripts related to phenylpropanoid biosynthesis were translated and compared with the proteome sequences of six species of Asteraceae. The same E-value threshold (1e-30) was applied to remove spurious hits.