Mistletoes are parasitic plants dispersed by animals, mainly birds (Watson, 2001). The mistletoe Phoradendron californicum Nutt. has a broad distribution in North America and is a native parasitic plant of leguminous plants in the Sonoran Desert, mainly Prosopis articulata S. Watson and Cercidium microphyllum (Torr.) Rose & I. M. Johnst. The interactions among mistletoes, their host plants, and their animal dispersers are currently under study in both natural and fragmented landscapes in the desert of the Baja California peninsula. Mistletoe–host plants' antagonistic interactions, along with animal dispersers, may form complex networks whose function and structure can be influenced by fragmentation at different scales, e.g., molecular or populations. Thus, the mistletoe-host interaction is ideal for studies on the effects of landscape fragmentation on plantanimal interactions.

Our aim is to determine to what extent landscape fragmentation affects mistletoe-animal interactions, the dispersal process of seeds, and the resulting networks of spatial genetic variability. For these purposes, we isolated and characterized nuclear microsatellite markers that are being successfully applied to describe spatial patterns of genetic structure. To date, microsatellite primers have not been developed for this mistletoe species.

METHODS AND RESULTS

We extracted genomic DNA using a DNeasy Plant Mini Kit (QIAGEN, Valencia, California, USA) and used ∼ 5 µg from one P. californicum individual collected in Santo Domingo Valley, Baja California Sur, Mexico (see Patch 33 population in Appendix 1 for GPS coordinates and voucher), to construct an enriched genomic library digested with RsaI (Sigma-Aldrich Corporation, St. Louis, Missouri, USA) for microsatellite loci using the oligonucleotide probes (GT)₁₅, (CT)₁₅, (GATA)₁₀, (GACA)₈, and (GATGT)₅ (Glenn and Schable, 2005). We sequenced a total of 288 Escherichia coli clones, from which 65 (22.5%) contained microsatellite loci but only 17 (5.9%) had flanking sequences adequate for the design of PCR primers, using Primer3 software (Rozen and Skaletsky, 2000). From these, only one polymorphic locus (Phca63B) was successfully amplified and scored. To overcome the low efficiency of enrichment and cloning methods, we used ∼ 5 µg of genomic DNA from the same individual to construct a shotgun genomic library that was sequenced on 1/8th of a plate using 454 GS FLX Titanium chemistry (Roche Applied Science, Indianapolis, Indiana, USA) at the University of Arizona Genetics Core (Tucson, Arizona, USA). We generated 70.1 Mb of quality-filtered data, distributed over 196,512 unique reads with an average length of 357 bp after quality filtering (quality score [Q] ≥ 20 using a 10-bp sliding window). We located microsatellite loci containing at least 10 perfect repeats and designed primers using the software QDD (Meglécz et al., 2010). We used unique sequence reads and consensus sequences within contigs that grouped sequences with a similarity ≥95% in regions ≥30 bp flanking the repeats to design primers. This step eliminated duplicated loci that have diverged in the flanking regions (except recent duplicates) and reduced null alleles by assembling sequence data from regions with coverage ≥1× for anchoring primers.

TABLE 1.

Characteristics of 18 microsatellite markers isolated from populations of Phoradendron californicum.

We found 115 di-, 376 tri-, and 13 tetranucleotide loci that met our criteria, and selected all of the tetranucleotides, 24 dinucleotides, and 24 trinucleotides with the largest number of repeats for primer synthesis. For primer testing, DNA was isolated from silica-dried stems using a modified cetyltrimethylammonium bromide (CTAB) extraction method (Milligan, 1998) that included tissue grinding in a Mixer Mill MM301 (Retsch, Haan, Germany) and TLE resuspension (10 mM Tris-HCl [pH 8.0], 0.1 mM EDTA). We sampled a total of 98 P. californicum individuals from three natural populations located in the Sonoran Desert in the Baja California peninsula, Mexico (Appendix 1). Sampling was designed to capture at least one sample from each parasitized host tree in each patch. Hosts were more aggregated in patch 32 compared to other patches.

PCR amplifications were performed in a 20-µL final volume containing 1× buffer (67 mM Tris-HCL [pH 8.8], 16 mM (NH₄)₂SO₄, 0.01% Tween-20), 2.5 mM MgCl₂ (1.5 mM for locus PhcaGQL), 0.01% bovine serum albumin (BSA; Roche Diagnostics, Barcelona, Spain), 0.25 mM dNTP, 0.40 µM dye-labeled M13 primer, 0.40 µM “pig-tailed” reverse primer, 0.04 µM M13-tailed forward primer (see M13 and “pig-tail” sequences in Table 1), 0.5 U Taq DNA polymerase (Bioline, London, United Kingdom), and approximately 50–70 ng genomic DNA. Reactions were undertaken in a “touchdown” PCR protocol in a Bio-Rad DNA Engine Peltier Thermal Cycler (Bio-Rad Laboratories, Hercules, California, USA), with an initial 2 min of denaturation at 94°C; 17 cycles at 92°C for 30 s, annealing at 60–44°C for 30 s (1°C decrease in each cycle), and extension at 72°C for 30 s; 25 cycles at 92°C for 30 s, 44°C for 30 s, and 72°C for 30 s; and a final extension of 5 min at 72°C. Amplified fragments were analyzed on an ABI 3130xl Genetic Analyzer (Applied Biosystems, Foster City, California, USA) and sized using GeneMapper 4.0 (Applied Biosystems) and GeneScan 500 LIZ size standard (Applied Biosystems). No multiplexing was attempted at the PCR stage (see mix for capillary electrophoresis in Table 1).

From a total of 61 loci tested from the next-generation library, four were monomorphic, 27 showed complex or nonspecific amplification, six showed high frequencies of null alleles, and 11 failed to amplify. Therefore, we obtained a total of 13 polymorphic loci, adding to a total of 14 polymorphic loci among both libraries (Table 1). We observed a total of 187 alleles for our P. californicum sample, averaging overall 13.3 alleles per locus (range 4–28; see Table 2 for heterozygosities). The combined probability of individual identity based on the 14 polymorphic loci was estimated at 1.42 × 10⁻¹⁷ with CERVUS (Kalinowski et al., 2007).

TABLE 2.

Results of screening of 14 polymorphic markers in three populations (Patches 32, 33, and 59) of Phoradendron californicum.^a

In addition, we tested the 14 polymorphic and the four additional monomorphic loci on 12 and eight individuals of two additional species of Phoradendron: P. diguetianum Tiegh. and P. juniperinum A. Gray (Appendix 1). These species were selected because (1) they belong to the same genus; (2) they are also distributed in the Sonoran Desert, and P. diguetianum occurs in the same sampling area; and (3) both species are common and, therefore, easy to collect. Sampling was designed to capture a wider range of potential variation among Phoradendron species, covering a more extensive geographic area. DNA isolation of P. diguetianum samples was performed as described for P. californicum, whereas P. juniperinum samples were extracted with the DNeasy Plant Mini Kit (QIAGEN). PCR conditions were as described above for P. californicum except for labeling PCR products of the 14 polymorphic loci (see Table 3). Three positive and one negative control were run with each PCR to ensure accurate scoring. Phoradendron diguetianum showed four loci (Phca0GH, PhcaCC3, Phca63B, and PhcaW15) with ambiguous locus-specific amplification, and the loci PhcaGQL, PhcaK4A, and PhcaSRK were monomorphic; P. juniperinum showed no PCR amplifications.

Table 2 summarizes the features of the 14 loci after inspecting their number of alleles, observed and expected heterozygosities (CERVUS 3.0.3; Kalinowski et al., 2007), and testing for deviations from Hardy–Weinberg equilibrium (HWE; Arlequin 3.5.1.3; Excoffier and Lischer, 2010), gametic disequilibrium (GENEPOP 4.1.4; Rousset, 2008), and the presence of null alleles according to the van Oosterhout method (MICRO-CHECKER 2.2.3; van Oosterhout et al., 2004). We used Bonferroni-corrected P values to assess the significance of the results obtained. Several loci among populations showed a significant deviation from HWE (Bonferroni-corrected P < 0.05/14 = 0.0036) due to a deficit of heterozygote individuals (Table 2). Only loci PhcaZ7Y and Phca1G8 showed this deviation in all three populations, and Phca0Q7, Phca0GH, Phca63B, Phca1HP, and PhcaGQL showed deviations in two populations. Significant gametic disequilibrium (P < 0.0036) was detected for two pairs of loci, but patterns were not consistent among populations (only Phca0GH-PhcaCC3 in population 33 and Phca0GH-Phca54T in population 59 were significant). Signs for the presence of null alleles were suggested by the general excess of homozygotes in eight loci for each population (Table 2). Thus PhcaZ7Y, Phca1G8, and PhcaGQL consistently showed null allele frequencies between 0.128–0.320 in all three populations. In contrast, loci Phca0Q7, Phca0GH, Phca63B, Phca1HP, and PhcaCC3 showed evidence of null alleles at frequencies ≤0.2576 in two populations. These results were not unexpected and could stem from the characteristic pattern of aggregation of parasitic plants, imposing limitations to pollen- and seed-mediated gene flow within aggregations and reduced genetic differentiation among individuals within these clumps. Consistent and significant deviations from HWE and gametic disequilibrium within but not among populations are unlikely to be caused by null alleles or physical proximity between pairs of loci, and could indicate the presence of marked genetic structure within sampled populations (Francois and Durand, 2010).

TABLE 3.

Cross-species amplification testing of Phoradendron californicum microsatellite markers in P. juniperinum and P. diguetianum. ^a

CONCLUSIONS

Observed levels of polymorphism suggest that the reported markers are adequate for characterizing spatial genetic structure at reduced spatial scales and for studying aggregation patterns in Phoradendron host-parasite relationships. Cross-species amplifications were unsuccessful in P. juniperinum and ambiguous in P. diguetianum, most likely due to high genetic divergence among the studied Phoradendron species.