Narcissus L. (Amaryllidaceae) is the single most important ornamental crop for both the cut flower and the bulb trade combined. Complex breeding programs of daffodils over the past 150 yr have resulted in more than 30,000 registered cultivars (Könyves et al., 2011), but this makes the description and commercialization of new cultivars increasingly complex. Naming new cultivars requires the identification and description of discriminating features, and molecular markers, such as microsatellites, could provide fast, cheap, and easily searchable data to achieve this (Culham and Grant, 1999). To assess the use of microsatellites in Narcissus for cultivar identification and for taxonomic revision, we developed new microsatellite markers for Narcissus sect. Bulbocodii DC. (hoop-petticoat daffodils) as a test case. This section is an excellent study group due to its distinct floral morphology, having a large funnel-shaped corona; its limited distribution, ranging from southern Morocco to southwest France; and its long history in cultivation (David and Könyves, 2013). The section exhibits natural variation in both morphology and in chromosome number, ranging from diploid to octoploid (Fernandes, 1963), and the taxa frequently hybridize, resulting in four to 33 taxa from species down to varietal ranks, depending on taxonomic treatment. The microsatellite markers described here were developed using material from a naturally occurring population and screened using a combination of wild and cultivated plants to establish the extent of genetic variation.
METHODS AND RESULTS
Material from the wild was collected across the natural distribution of Narcissus sect. Bulbocodii (Könyves, 2014). In total, 44 populations were sampled (Appendix 1). Total genomic DNA was extracted using a modified cetyltrimethylammonium bromide (CTAB) protocol (Doyle and Doyle, 1987). Microsatellite library development and primer design were carried out by Genoscreen (Lille, France). An equimolar DNA mix of 10 individuals of population KET (Appendix 1) was enriched with eight microsatellite probes (TG, TC, AAC, AAG, AGG, ACG, ACAT, ACTC) and sequenced according to the GS FLX protocol by Malausa et al. (2011). The resulting library consisted of 37,979 raw sequences. Of these, 5765 contained microsatellites, and primers were designed for 351 using QDD (Meglécz et al., 2010) following Malausa et al. (2011).
Resources allowed test PCR amplification of 67 primer pairs from the 351 developed. The primers were chosen to maximize the variation in length of amplicon, motif repeat sequence, and motif length. Test amplification of primers used one sample each from populations CAT and V, and four samples from an existing living collection (accession no. SJ20597, SJ001999, BD96/198, and Narcissus ‘Golden Bells’, the most widely available cultivar in this section; Appendix 2), with the equimolar DNA mix of population KET used as a positive control. PCR reactions were performed in a 10-µL volume containing final concentrations of 1× Bioline Biomix (Bioline Reagents Ltd., London, United Kingdom), 0.1–0.2 µM of each primer, and 10 ng of DNA template. Cycling conditions were 94°C for 120 s; 40 cycles of 94°C for 45 s, 48–63°C for 30 s, 72°C for 45 s; and finally 72°C for 10 min (see Table 1). The PCR products were separated on 2% w/v agarose gels in 1× TAE buffer (pH 8.0) stained with ethidium bromide with accompanying HyperLadder 100bp (Bioline Reagents Ltd.) as a marker. Gels were photographed under ultraviolet illumination to record the presence of PCR products. Of the 67 primer pairs selected for initial trial, 39 primer pairs amplified the expected target fragments. Microsatellite variability was tested with an equimolar DNA mix of 19 samples (marked with a/b in Appendix 1 and 2) by ligating the PCR products with the M13 promoter and labeling the products with 6-FAM according to Cryer et al. (2005). Fragment analysis of amplicons was carried out by Source BioScience (Nottingham, United Kingdom). The electropherograms were analyzed using GeneMapper version 4.0 (Applied Biosystems by Life Technologies, Carlsbad, California, USA). Thirty-three of the tested primer pairs amplified multiple clean peaks. Of these, the best 24, based on the overall quality of the electropherograms, were used to genotype seven samples (two samples each of populations KET and TIG, one each of populations CAT and POR, and one of Narcissus ‘Golden Bells’). Forward primers were labeled with fluorescent dyes 6-FAM, HEX (Sigma-Aldrich, St. Louis, Missouri, USA), NED, PET, or VIC (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Readable electropherograms were obtained for 19 primer pairs (Table 1). Of these, resources allowed the 11 best markers (most length-variable and reproducible) to be used to genotype 317 samples of hoop-petticoat daffodils across the natural distribution range to assess the degree of polymorphism in nature (Table 2).
Table 1.
Characteristics of 19 microsatellite loci developed for Narcissus sect. Bulbocodii.
Table 2.
Summary statistics of the chosen 11 microsatellites based on 312–317 hoop-petticoat daffodil samples.
PCR amplifications were performed as single reactions according to the previously detailed cycling conditions. The PCR products were combined for multiplex fragment analysis. Unambiguously identifying microsatellite alleles in polyploids can be challenging, as identifying stutter peaks in samples of unknown ploidy is difficult and can lead to inclusion of noise in a data set. To avoid this, alleles were scored according to the MANUAL 8 scoring routine described by Pfeiffer et al. (2011). Moreover, as the allele dosage of polyploids is unknown, traditional population genetic techniques (e.g., deviation from Hardy–Weinberg equilibrium) cannot readily be applied. Therefore, we used a presence-absence scoring of peaks to estimate polymorphism, similar to a dominant marker (e.g., amplified fragment length polymorphism [AFLP]) data set.
The number of alleles per locus ranged from four to 21, the observed heterozygosity (Ho) ranged from 0.101 to 0.297, and allelic diversity (calculated as: 
, where pi is the frequency with which the ith allele was detected) ranged from 0.279 to 0.842. In addition to the allelic variation, there were null genotypes for six of the markers, with frequency ranging from 0.002 to 0.121, confirmed by repeating PCR amplifications. The presence of null genotypes was expected due to incomplete transferability of these markers in section Bulbocodii. However, in a presence-absence data set these are valuable characters that allow samples with a null-allele data set for some individual markers to be included.
Two populations each of the most widely sampled species (N. bulbocodium L. [CAT, ALD], N. cantabricus DC. [SDF, HOR], and N. romieuxii Braun-Blanq. & Maire [KET, OUL]) were used to calculate genotypic diversity estimators (Table 3). The total number of different alleles per population across all loci (A) ranged from 26 to 42, the number of private alleles per population across all loci (Ap) was between zero and three, proportion of observed heterozygotes averaged per locus (Ho) was from 0.19 to 0.35, proportion of null genotypes carried by each individual averaged across all loci (Fg0) ranged from 0 to 0.21, and the genotypic richness was 0.94 or 1 (calculated as: R = G − 1/N − 1; where G is the number of multilocus genotypes and N is the number of genotyped samples; Dorken and Eckert, 2001).
Broader transferability of these markers was tested using 18 species belonging to seven of the nine (Blanchard, 1990) other Narcissus sections. The success of the transfer was assessed using fragment analysis. The 11 markers were all transferable to other Narcissus sections to some degree, ranging from 39% to 100% (Table 4).
CONCLUSIONS
The microsatellite markers developed in this study are sufficiently variable to allow species-level and population-level variation of hoop-petticoat daffodils to be investigated. The markers show potential to be used to develop molecular identification tools for daffodil cultivars, and to contribute toward the taxonomic revision of section Bulbocodii. The high degree of transferability suggests that these markers have the potential to distinguish many Narcissus cultivars in most sections of the genus.
Table 3.
Results of initial genotypic variability screening among populations of Narcissus sect. Bulbocodii.
Table 4.
Transferability of the chosen 11 microsatellite loci in 18 Narcissus species.
