Camissoniopsis cheiranthifolia (Hornem. ex Spreng.) W. L. Wagner & Hoch (Onagraceae) is a diploid, bee-pollinated, short-lived perennial endemic to the Pacific coastal dunes of Baja California, California, and Oregon (Raven, 1969; Wagner et al., 2007). Being restricted to coastal dunes, it is continuously distributed along a near-linear, easily accessed geographic range, providing opportunities for studying the ecology and evolution of geographic range limits (Samis and Eckert, 2007, 2009). This species also exhibits striking variation in floral traits and the relative importance of outcrossing vs. self-fertilization, providing opportunities to investigate the evolution of mating systems (Eckert et al., 2006; Button et al., 2012). Dart et al. (2012) showed that populations in southern California are large-flowered (LF), predominantly outcrossing, and either largely self-incompatible (SI) or self-compatible (SC). Populations in Baja California toward the southern range limit, on the Channel Islands off California and north of Point Conception, California, to the northern range limit in southern Oregon are small-flowered (SF), SC, and predominantly selfing. The proportion of seeds outcrossed estimated at the population level from the segregation of allozyme polymorphism in progeny arrays ranged from 0.0–1.0 and correlated positively with flower size. Lineages within Camissoniopsis W. L. Wagner & Hoch and closely related Eulobus Nutt. ex Torr. & A. Gray and Camissonia Link appear to have undergone speciation via polyploidization involving hybridization (Raven, 1969; Wagner et al., 2007). In Camissoniopsis, five of 14 species are polyploid, predominantly selfing, and were likely derived through hybridization. Camissoniopsis cheiranthifolia and C. bistorta (Nutt. ex Torr. & A. Gray) W. L. Wagner & Hoch are the only two species that include outcrossing populations. Throughout the genus, species' ranges frequently overlap, and ongoing hybridization may be maintaining high morphological variation within and low differentiation among species. We developed microsatellite markers for C. cheiranthifolia that would cross-amplify in related taxa to better investigate mating system evolution, the genetic structure of geographic ranges, and the ecology and genetics of hybridization.
METHODS AND RESULTS
A microsatellite-enriched genomic library was developed following Glenn and Schable (2005) and Hamilton et al. (1999). Using silica-dried leaf tissue from one plant from each of two populations (Appendix 1), total DNA was isolated using cetyltrimethylammonium bromide (CTAB) extraction (Doyle and Doyle, 1987). We digested 5 µg of pooled DNA at 37°C overnight with AluI + HaeIII + RsaI restriction enzymes. Digested DNA was dephosphorylated using 0.01 unit calf intestinal alkaline phosphatase per picomole ends of DNA at 50°C for 1 h, purified using an equal volume of 25:24:1 phenol : chloroform : isoamyl alcohol, precipitated using 2.5 volumes of cold 100% ethanol and 3 M sodium acetate (NaOAc), and then resuspended in TE buffer (10 mM Tris [pH 8.0], 1 mM EDTA). DNA quality and size were evaluated on 1.5% agarose gels (fragments ranged from 200–1000 bp).
Characteristics of 24 microsatellite primer pairs developed for Camissoniopsis cheiranthifolia.
Digested DNA was ligated to 1 µM of SNX double-stranded linkers using T4 DNA ligase (Invitrogen, Burlington, Ontario, Canada) and 20 units XmnI (New England Biolabs, Whitby, Ontario, Canada) overnight at 16°C. Linker ligation was tested using PCR amplification with SNX forward primer (5′-CTAAGGCCTTGCTAGCAGAAGC-3′) in a reaction with 1× buffer, 2.0 mM MgCl2, 150 µM dNTPs (Roche Diagnostics, Laval, Quebec, Canada), 0.5 µM primer, 1 µg/µL bovine serum albumin (BSA), and 1 unit Taq polymerase (reagents from Invitrogen except dNTPs).
Linker-ligated DNA was hybridized to 3′ biotinylated (AC)13 and (AG)13 probes for 4 h at 70°C after 10 min at 95°C. Enriched DNA was captured using streptavidin beads (DynaBeads M-280 Streptavidin, Invitrogen) and verified with PCR as above. Approximately 20 ng/µL of amplified DNA was used in transformation with the TOPO TA Cloning Kit (Invitrogen) and grown on Luria-Bertani plates with 50 ng/mL ampicillin. About 350 colonies were screened for microsatellites using fluorescent DIG probes (Roche Diagnostics). For positive clones, insert sizes were estimated with PCR using M13 primers and verified on 1% agarose gels. DNA was extracted from 115 positive clones with appropriate insert sizes and PCR products were sequenced at Genome Quebec (McGill University, Montreal, Quebec, Canada) or Robarts Research Institute (University of Western Ontario, London, Ontario, Canada). Ninety-three of these clones contained a total of 90 unique microsatellite regions. Primer pairs were designed for the 32 clones that had both linkers, suitable flanking region at both ends, and a minimum of eight repeats. We used Primer3web version 4.0.0 (Koressaar and Remm, 2007; Untergasser et al., 2012) and Amplifix 1.5.4 ( http://crn2m.univ-mrs.fr/AmplifX) to design primer pairs optimized to contain 18–22 bases, 40–60% GC content, 50–60°C melting temperature, and yield 100–350 bp PCR products.
The forward primer of each pair was labeled with a D4 red-labeled M13 tail (5′-CACGACGTTGTAAAACGA-3′) (Sigma-Aldrich Canada, Oakville, Ontario, Canada). The number and the identity of samples used for an initial testing of each pair varied. We used one to seven DNA samples from five to 42 (mean = 30.6) populations covering the entire geographic range of C. cheiranthifolia and one to six DNA samples from three to 12 (mean = 9.7) populations of C. bistorta (Appendix 1). Each sample was genotyped twice in single-locus 5-µL PCR reactions containing 0.5 µL of DNA template (10 µg/µL), 2.5 µL of Multiplex PCR Master Mix (QIAGEN, Toronto, Ontario, Canada), 0.1 µL of each forward and reverse primers (10 µM), 1.1 µL of M13taq (1 µM; Sigma-Aldrich Canada), 0.2 µL of Q-Solution, and 0.5 µL of sterile double-distilled water. PCR involved 15 min of denaturation at 94°C, followed by 35 cycles of 20 s at 94°C, 30 s at 55°C or 57°C, and 40 s at 72°C, with a 10-min final extension at 72°C. PCR product was diluted with double-distilled water to a final volume of 15 µL, and fragments were sized using a GenomeLab GeXP with the CEQ 8000 Genetic Analysis System version 9.0 (Beckman Coulter, Mississauga, Ontario, Canada).
Of 32 primer pairs, 24 yielded variable fragments of expected size and two of these amplified within two other loci (Table 1). For these two loci (A31b and C135b), a second primer pair (A31c and C135c) was redesigned to improve consistency of amplification in some C. cheiranthifolia but mainly in C. bistorta populations. For each locus, we estimated the number of alleles (A), observed (H o) and expected (H e) heterozygosities in one LF-SI population, one LF-SC population, two SF-SC populations (southern and northern parts of the range), and one LF-SI C. bistorta population using GenAlEx version 6.5 (Peakall and Smouse, 2012). We did not test for deviations from Hardy–Weinberg equilibrium because all populations of C. cheiranthifolia, including LF-SI populations, can exhibit some self-fertilization (Dart et al., 2012), so that H o is less than H e in many cases reported below.
Estimation of population genetic parameters for 21 microsatellite loci in four Camissoniopsis cheiranthifolia populations representing each geographic region and mating type, plus one population of the sister species C. bistorta. Population codes (in parentheses) are provided in Appendix 1.a
Within populations, A ranged from one to 12 across loci (mean = 4.3) and was highest in the LF-SI populations compared to the LF-SC population and the two SF-SC populations (Table 2). Using only 13 loci for which the same individuals were genotyped, we detected 130 alleles total, of which 56 were found only in C. cheiranthifolia (mean ±1 SE = 4.30 ± 0.49 private alleles per locus) and 10 only in C. bistorta (0.77 ± 0.26 private alleles per locus), suggesting that these markers could be useful to detect hybridization between these species, although a broader sample is required to determined which are diagnostic. Ho and He were highly variable but predictable based on the mating system, as both were highest in the two LF-SI populations, lower in the mixed-mating LF-SC population, and lower still in the two SF-SC populations (Table 2), thereby verifying the potential of these markers for studying the genetic consequences of mating system differentiation. Although cross-amplification often failed in samples from the eight related taxa, there were many loci at which amplification was successful (Appendix 1, Table 3). Of the 24 loci developed for C. cheiranthifolia, 17 were tested in C. micrantha (Hornem. ex Spreng.) W. L. Wagner & Hoch, C. lewisii (P. H. Raven) W. L. Wagner & Hoch, Eulobus angelorum (S. Watson) W. L. Wagner & Hoch, E. californicus Nutt. ex Torr. & A. Gray, and E. crassifolius (Greene) W. L. Wagner & Hoch, and successful amplification occurred for 17, 15, nine, and nine loci, respectively. Dick et al. (2014) tested 16 of these 24 loci in the serpentine endemic Camissonia benitensis P. H. Raven and its two widespread congeners C. strigulosa (Fisch. & C. A. Mey.) P. H. Raven and C. contorta (Douglas) Kearney and found six variable loci, which they used to quantify patterns of genetic diversity.
All 24 microsatellite loci were variable in C. cheiranthifolia and C. bistorta, and a number of them also amplified in eight closely related taxa, providing opportunities to test a broad range of ecological and evolutionary questions within species and across taxa. These markers will facilitate our ongoing studies of mating system evolution and geographic range limits in C. cheiranthifolia, as well as the genetic and ecological consequences of hybridization between C. cheiranthifolia and C. historta. The high frequency of cross-amplification in related taxa provides opportunities for comparative studies investigating the genetic consequences of variation in life history and mating system, and ongoing hybridization in this morphologically and ecologically variable group.
Cross-amplification and allele sizes of 24 microsatellite primer pairs developed for Camissoniopsis cheiranthifolia and screened in C. bistorta, C. micrantha, C. lewisii, Eulobus crassifolius, E. californicus, E. angelorum, Camissonia benitensis, C. strigulosa, and C. contorta.a,b
 The authors thank Jon P. Rebman (San Diego Natural History Museum), Sula Vanderplank (Botanical Research Institute of Texas), and Michael Simpson (San Diego State University) for help with plant identification; Zhengxin Sun for help with marker development and optimization; Katharina Bremer for help with primer design; and Kent Holsinger for comments on the manuscript. Funding is acknowledged from the Natural Sciences and Engineering Research Council of Canada (NSERC) for Discovery Grants to K.E.S. and C.G.E., and the Consejo Nacional de Ciencia y Tecnología (CONACYT) for a scholarship to A.L.V.