Morphologically variable complexes of species represent formidable challenges for species delimitation and for understanding the mechanisms involved in their evolution. In these cases, genetic markers such as microsatellites can augment traditional systematic approaches by revealing patterns of gene flow, reproductive isolation, and potential cryptic variation between either sequence-invariant or morphologically indistinguishable groups (e.g., Frullania Raddi; Ramaiya et al., 2010). One taxonomically difficult group is Camassia Lindl., a monophyletic clade within the Agavaceae (Smith et al., 2008). This genus consists of six North American species of bulbous, spring perennials that are insect pollinated, ecologically important in diverse habitats, and culturally valuable as food plants of native peoples (Tomimatsu et al., 2009). Despite taxonomic distinctions supported by allozyme, morphological, and phenological criteria (Ranker and Schnabel, 1986; Uyeda and Kephart, 2006), species delimitation has been difficult, given high morphological variability and poor cpDNA resolution among both sympatric and geographically isolated taxa. Where the ranges of the most common western species (C. quamash (Pursh) Greene and C. leichtlinii (Baker) S. Watson) overlap, a total of 10 subspecies exist; potential hybridization among these taxa contributes additional complexity to species definition (Fishbein et al., 2010). To untangle the intricate taxonomic relationships within this difficult species complex, we developed 16 microsatellite markers for all Camassia species using a traditional approach. We aim to use these markers to examine the evolutionary processes and patterns within this genus, clarify taxonomic distinctions among morphologically similar taxa, and inform management decisions where Camassia is used in restoration and reintroduction efforts.
METHODS AND RESULTS
Microsatellite markers were developed in Camassia using the nonradioactive method of Glenn and Schable (2005) with individual samples from three different species. DNA was isolated with either cetyltrimethylammonium bromide (CTAB; following Doyle and Doyle, 1987) or a QIAGEN DNeasy Plant Mini Kit (QIAGEN, Culver City, California, USA) from three individuals of the following species: C. howellii S. Watson (sample HRA 5) from Hugo Rest Area in Grants Pass, Oregon (latitude 42.518433°, longitude −123.363021°), C. leichtlinii subsp. suksdorfii (Greenm.) Gould (sample BFV-12) from Butterfly Valley in Quincy, California (40.012283°, −120.9941190°), and C. quamash subsp. maxima Gould (sample BPP 99.85) from Bush Pasture Park in Salem, Oregon (44.929083°, −123.035666°). All DNA was treated with RNase, digested with the RsaI restriction enzyme (New England Biolabs, Ipswich, Massachusetts, USA), ligated to the SuperSNX adapter pair (see Glenn and Schable, 2005), and bound to biotinylated oligos CA15, GA15, CAC10, CTC10, and CAG10 via PCR, before fragments were recovered with Dynabeads (Invitrogen, Carlsbad, California, USA), and then ligated and transformed into E. coli TOP10 competent cells using the TOPO TA Cloning Kit (Invitrogen). Of the several hundred colonies obtained, 236 tested positive for containing inserts; these were amplified via PCR, using plastic toothpicks to transfer each colony directly to the PCR reaction wells. The PCR products were screened for size on 4% agarose gels; 192 products between 450 and 1200 bp were sequenced at the High-Throughput Genomics Center (Seattle, Washington, USA). Of these, 27 (14%) sequences contained a central repeat region and were deemed suitable for primer development. Primer pairs were then designed for each of these sequences using the Primer3 software program (Rozen and Skaletsky, 2000) with the following criteria: primer size 16–22 bp, 30–60% GC content, melting temperature 50– 60°C, and fragment size 90–100 bp.
Microsatellite loci information based on 16 primer pairs developed in Camassia. a
Descriptive statistics for individual microsatellite loci within each of the six sampled Camassia species.a
Amplification success of all microsatellite primers across the following taxa of Camassia and Hastingsia: C. angusta (CAAN), C. cusickii (CACU), C. howellii (CAHO), C. leichtlinii (CALE), C. quamash (CAQU), C. scilloides (CASC), H. alba (HAAL), H. atropurpurea (HAAT), H. bracteosa (HABR), and H. serpentinicola (HASE).a
The 27 primer pairs were tested separately using three individuals representing different Camassia species in 10-µL PCR reactions composed of the following: 5 µL QIAGEN Master Mix (QIAGEN), 1 µL Primer mix (composed of 2 µM of each unlabeled forward and reverse primers), 3.8 µL of dH2O, and 0.2 µL DNA (∼10 ng). The PCR conditions, as suggested for the QIAGEN kit, were as follows: 95°C for 15 min, followed by 35 cycles each of 94°C for 30 s, 57°C for 45 s, and 72°C for 60 s, and then a final extension of 72°C for 30 min. Samples were analyzed at Cornell University's Life Sciences Core Laboratory Center on a 3730xl sequencer (Applied Biosystems, Carlsbad, California, USA) using the 500 LIZ internal size standard. Fragment analysis was then conducted with GeneMarker version 1.85 software (SoftGenetics, State College, Pennsylvania, USA). Nine of the 27 primers failed to amplify across all individuals or exhibited multiple banding patterns and were excluded from further analysis.
The resulting 18 primer pairs representing di-, tri-, and hexamer nucleotide repeats (Table 1) were then tested for each of the following six species of Camassia (Appendix 1), using samples from at least two populations whenever possible: C. angusta (Engelm. & A. Gray) Blank, (n = 24), C. cusickii S. Watson (10), C. howellii (19), C. leichtlinii (24), C. quamash (23), and C. scilloides (Raf.) Cory (35). Primers were also tested in related Hastingsia S. Watson species (Appendix 1), consisting of H. alba (Durand) S. Watson (2), H. atropurpurea Becking (5), H. bracteosa S. Watson (10), and H. serpentinicola Becking (13). For this analysis, primers were individually labeled with 6-FAM, PET, NED, or VIC, and placed in one of three primer mix solutions (see Table 1) for multiplex PCR using the QIAGEN Multiplexing Kit with the same PCR reaction amounts and thermocycler program as outlined above for testing individual primers. PCR products were sent to Cornell University with the same conditions as described above. At this point, another primer pair was discarded due to inconsistent amplification success. A second primer pair was also removed because it was based on the same library sequence as an existing primer (Camas33), exhibiting identical banding patterns but of different sizes (ca. 400 bp for Camas35 compared to ca. 250 bp for Camas33). Similarly, primers Camas34 and Camas79 were found to be based on the same library sequence but, as they amplified different portions and were not in linkage disequilibrium (see below), they were both included. Analysis of genetic variation of the 16 primer pairs was conducted in Genetic Data Analysis (Lewis and Zaykin, 2001) to quantify the number of alleles (A) per locus, as well as observed (Ho) and expected (He) values of heterozygosity, and tests of Hardy—Weinberg equilibrium and linkage disequilibrium.
All 16 primer pairs amplified in at least one Camassia species (Tables 2, 3), and three primer pairs amplified strongly across both genera (Camas45, Camas56, and Camas83), with an additional two pairs performing consistently across all Camassia taxa (Camas20 and Camas90). Each primer amplified to some degree within the species from which it was originally developed (Table 3), except for Camas34 which exhibited only very small peaks in C. leichtlinii but performed well within other taxa. Within Camassia (Table 2), the number of alleles per locus ranged from one to 18, averaging 3.94 per species. Levels of heterozygosity for polymorphic loci in Camassia ranged from 0.000 to 1.00 and from 0.033 to 0.917 for Ho and He, respectively. Up to eight loci were out of Hardy—Weinberg equilibrium in at least one population, reflecting in part the limited sampling, but a given locus did not deviate across all populations. The only primers in linkage disequilibrium were found in C. angusta (Camas29/56, Camas33/49, Camas29/49, and Camas62/90) and C. leichtlinii (Camas13/51, Camas20/51, Camas33/90, and Camas51/56). We are now using these primers to examine levels of genetic variation and population divergence within and among the entire Camassia/Hastingsia species complex.
Given the morphological and genetic variability as well as taxonomic challenges present in Camassia and Hastingsia, we anticipate that these microsatellite markers will be instrumental in helping to elucidate species delimitation, hybridization, and lineage differentiation within the complex. Consistent amplification of several markers across at least two genera suggest that they may be useful to further spur research in closely related taxa of the Chlorogaloideae and Agavaceae groups.
 The authors thank the following student researchers for contributions: Adam Kotaich, Kate Halpin, Katie Hoesly, Tyler Runyon, Mike Wilder, and Sunita Yadav. Colleagues Jenny Archibald, Chris Smith, Mark Fishbein, Briana Gross, Geraldine Allen, Julie Morgan, Andy Schnabel, and Matt White gave valuable expertise or time. This research was supported by the Earthwatch Institute, the M. J. Murdock Trust, and the National Science Foundation (grant DEB-1146531 to S.R.K.). Of many providing site access, we particularly thank the Indiana Division of State Preserves and Parks, the Nature Conservancy, and the U.S. Forest Service.