The ca. 120 species of goldenrod (Solidago L.; Asteraceae) are largely confined to North America and occupy an impressive array of habitats, including tundra, rock outcrops, bogs, sand dunes, prairies, barrens, rockhouses, and a variety of woodlands (Semple and Cook, 2006). This taxonomic and ecological diversity has led to Solidago's popularity as a study system in evolution and ecology. Microsatellite, or simple sequence repeat (SSR), markers could represent a valuable tool in many of these instances, for example, allowing for the estimation of kinship, the identification of invasive genotypes, and the estimation of gene flow among populations.

Microsatellite data could also help clarify Solidago species boundaries. The taxonomic complexity of the genus is widely recognized, a problem stemming from sheer species richness, low overall levels of genetic differentiation, occasional interspecific hybridization, and frequent polyploidy (Semple and Cook, 2006). An accurate delimitation of Solidago species would provide a robust account of biodiversity in the genus and enhance the evolutionary and ecological studies noted above. Given the low overall genetic divergence among Solidago species (Schilling et al., 2008), it should be possible to identify SSR loci that amplify in most species, providing a standard comparative genetic toolkit for the genus.

METHODS AND RESULTS

Silica-dried tissue from a diploid individual of S. gigantea Aiton (confirmed by a meiotic chromosome count) was collected in Chester County, Tennessee, USA. A voucher specimen for this collection (Beck 1258) has been deposited at the Wichita State University Herbarium (WICH). Total DNA was extracted with a DNeasy Plant Mini Kit (QIAGEN, Valencia, California, USA). An Illumina paired-end shotgun library was prepared by shearing 1 µg of DNA using a Covaris S220 ultrasonicator (Covaris, Woburn, Massachusetts, USA) and following the standard Illumina TruSeq DNA Library Kit protocol (Illumina, San Diego, California, USA) using a multiplex identifier adapter index. Sequencing was conducted on the Illumina HiSeq 2000 with 100-bp paired-end reads. Five million of the resulting reads were analyzed with the program PAL_FINDER_v0.02.03 (Castoe et al., 2012) to extract those reads that contained di-, tri-, tetra-, penta-, and hexanucleotide SSRs. Once positive reads were identified in PAL_FINDER_v0.02.03, they were batched to a local installation of Primer3 version 2.0.0 (Rozen and Skaletsky, 2000) for primer design. To avoid targeting multiple-copy loci, only those for which either primer sequence occurred one or two times in the 5 million reads were selected. A total of 1888 loci met this criterion.

To select a set of loci for initial screening, we focused on loci with tetra- and trinucleotide repeat motifs and with primer melting temperatures between 55°C and 65°C. Furthermore, loci were targeted for which only one of the paired-end reads sequenced into the repeat motif to avoid relatively small fragment sizes. Using these criteria, 80 loci were chosen for initial screening using a “CAG-tag” strategy similar to the M13 approach in Schuelke (2000). The forward primer from each locus was 5′ modified with an engineered “CAG-tag” sequence (5′-CAGTCGGGCGTCATCA-3′) to enable use of a third, fluorescently labeled primer (identical to the CAG-tag) in PCR. In addition, the “PIG-tail” sequence GTTT was added to the 5′ end of the reverse primer to reduce double peaks. Reactions (10 µL) included 1× Promega GoTaq Buffer (Promega Corporation, Fitchburg, Wisconsin, USA); 0.2 mM each dNTP; 2.5 mM MgCl₂; 0.025 µg bovine serum albumin (BSA); 0.5 U Promega GoTaq; 0.4 µM unlabeled primer; 0.04 µM CAG-labeled primer; 0.4 µM labeled CAG-tag, and ca. 30 ng DNA template. PCR amplification involved the touchdown cycling protocol outlined in Lance et al. (2010). CAG-tag screening included DNA extracted from eight herbarium specimens representing species from four subsections of Solidago sect. Solidago (Solidago subsect. Triplinerviae (Torrey & A. Gray) G. L. Nesom, Solidago subsect. Glomeruliflorae (Torrey & A. Gray) A. Gray, Solidago subsect. Squarrosae A. Gray, and Solidago subsect. Junceae (Rydb.) G. L. Nesom) and a sample of Brintonia discoidea (Elliott) Greene, representing a monotypic genus potentially sister to Solidago (Schilling et al., 2008). Full details for these eight specimens are provided in Appendix 1.

Fourteen loci (Table 1) were identified as variable, interpretable, and broadly amplifiable across the four tested Solidago subsections and outgroup Brintonia Greene. These loci were then further evaluated in a larger set of diploid individuals from Solidago subsect. Triplinerviae (47 samples representing 10 species), Solidago subsect. Squarrosae (47 samples representing 10 species), and Solidago subsect. Junceae (32 samples representing seven species). Full specimen details are provided in Appendix 1. All 126 samples were extracted from herbarium specimens archived at the University of Waterloo Herbarium (WAT), the University of Tennessee Herbarium (TENN), the Duke University Herbarium (DUKE), or the Missouri Botanical Garden Herbarium (MO) using the modified cetyltrimethylammonium bromide (CTAB) protocol detailed in Beck et al. (2012). Forward primers (minus the CAG-tag) were dye labeled with either 6-FAM or HEX, while reverse primers retained the PIG-tail for all but two loci (Table 1). Sets of two or three loci were simultaneously amplified using the multiplex PCR protocol described in Beck et al. (2012). Amplicons were sized using the GeneScan 500 LIZ Size Standard on an Applied Biosystems 3730xl DNA Analyzer (Life Technologies, Carlsbad, California, USA) at the University of Chicago Comprehensive Cancer Center DNA Sequencing and Genotyping Facility (Chicago, Illinois, USA). Alleles were determined using GeneMarker 1.9 (SoftGenetics, State College, Pennsylvania, USA).

The 14 loci were variable and generally transferable across the 27 species representing three Solidago subsections (Table 2). The number of alleles per locus ranged from seven to 51, and all loci (if amplifiable) were polymorphic in all three subsections. A null allele was inferred if no amplification was observed in all individuals of a given species, and seven of the 14 loci exhibited no evidence for null alleles in any of the 27 species (Table 2). Not surprisingly, the fewest null alleles were observed in subsect. Triplinerviae (11 of 140 locus/species combinations), the subsection to which S. gigantea belongs. In only one case did all species in a subsection exhibit a null allele for a given locus (Sg_7 in subsect. Squarrosae). Lineage-specific locus duplication was inferred in two cases based on the observation of more than two alleles per individual in multiple confirmed diploid samples (Sg_4 in subsect. Junceae and Sg_5 in subsect. Squarrosae).

CONCLUSIONS

The general transferability, single-copy status, and variability of these loci suggest that primers designed for a single Solidago species should be applicable across the genus. Screening of the 14 SSR loci described here and those previously reported for S. sempervirens L. (Wieczorek and Geber, 2002), S. canadensis L. (Zhao et al., 2012), and S. altissima L. (Sakata et al., 2013) should therefore provide a set of >20 informative SSR loci for any goldenrod species. These loci were also readily amplifiable from herbarium specimens of a wide age range (1932–2007, Appendix 1), creating opportunities for the broad inclusion of archived museum material in future studies.

Table 1.

Characteristics of 14 loci broadly amplifiable in Solidago.^a

TABLE 2.

Number of alleles, size range, and amplification success in three Solidago subsections. Loci successfully amplified in all taxa are shown in bold.