Genetic drift precluded adaptation of an insect seed predator to a novel host plant in a long-term selection experiment

Host specialization is considered a primary driver of the enormous diversity of herbivorous insects. Trade-offs in host use are hypothesized to promote this specialization, but they have mostly been studied in generalist herbivores. We conducted a multi-generation selection experiment to examine the adaptation of the specialist seed-feeding bug, Lygaeus equestris, to three novel host plants (Helianthus annuus, Verbascum thapsus and Centaurea phrygia) and to test whether trade-offs promote specialization. During the selection experiment, body size of L. equestris increased more on the novel host plant H. annuus compared to the primary host plant, Vincetoxicum hirundinaria, but this effect was not observed in other fitness related traits. In addition to selection, genetic drift caused variation among the experimental herbivore populations in their ability to exploit the host plants. Microsatellite data indicated that the level of within-population genetic variation decreased and population differentiation increased more in the selection line feeding on H. annuus compared to V. hirundinaria. We found a negative correlation between genetic differentiation and heterozygosity at the end of the experiment, suggesting that differentiation was significantly affected by genetic drift. We did not find fitness trade-offs between L. equestris feeding on the four hosts. Thus, trade-offs do not seem to promote specialization in L. equestris. Our results suggest that this insect herbivore is not likely to adapt to a novel host species in a time-scale of 20 generations despite sufficient genetic variation and that genetic drift disrupted the response to selection.

total number of generations by using generation times specific to each selection line averaged between the beginning and the end of the experiment. Although the average generation time in Helianthus selection line was longer by one day, it did not result in fewer generations over the course of the 26 months.

Microsatellite discovery for Lygaeus equestris
Microsatellite discovery and all subsequent genotyping was performed by Center of Evolutionary Applications (University of Turku, Finland). Microsatellites were characterized using a NGS approach where potential microsatellite loci were first located in silico from 454 pyrosequencing reads, after which primers were designed and potential loci tested for PCR amplification and polymorphism. Genomic DNA was extracted from muscle tissue with NucleoSpin Tissue kit (Macherey-Nagel). Genomic sequence data was produced by GenoScreen (France) through microsatellite motif enriched shotgun sequencing on Roche's 454 GS-FLX platform (1/32 Titanium plate), using pooled DNA from 11 individuals. As a result 24789 reads were received, out of which 7866 reads had repeat motifs (see below for criteria) and primer design for amplification of microsatellite loci was successful in 434 reads. Primers were designed with software msatcommander 1.0.8 (Faircloth, 2008) for 72 loci containing a di-, tri-or tetranucleotide motif with minimum 8 repeats and amplification was tested using non labelled M13-tailed primers in combination with a dye labelled (FAM, VIC, NED, PET) M13 primer (see e.g. Boutin-Ganache et al., 2001). To improve the microsatellite peak profiles, a GTTT-tail was added to the 5' end of each reverse primer (Brownstein et al., 1996).
A test set of 8 samples was used and each loci was tested in three different PCR conditions with varying annealing temperature. Amplification tests were done in 8 µl reactions containing c. 50-100 ng DNA, 0.1 µM forward primer with M13 tail, 0.2 µM reverse primer, 0.4 µM labelled M13 primer and 1X Qiagen multiplex PCR mastermix. The following two phase PCR profile was used: 95 °C for 15 min, 15 cycles of 94°C for 30 s, annealing at 60, 58 or 56 °C for 1.5 min, 72 °C for 1 min, following with 25 cycles of 94 °C for 30 s, annealing 52 °C for 1.5 min, 72 °C for 1 min and a final extension 72 °C for 15 min. Amplifications were performed on PTC-100 (MJ Research) and AB 2720 (Applied Biosystems) thermal cyclers.
For fragment analysis four PCR products with different dyes were pooled, one microliter from each and diluted with 85 µl of sterile water. 2 µl of the pooled and diluted PCR product was combined with GS600LIZ size standard (Applied Biosystems) and HiDiformamide (Applied Biosystems). Samples were denatured at 98 °C for three minutes and the size of the fragments was determined by capillary electrophoresis on an ABI Prism TM 3130xl genetic analysis instrument. Peak profiles were visually checked using GeneMapper version 4.0 (Applied Biosystems) and each locus was categorized as having "good amplification, nonspecific amplification and weak/no amplification". Depending on assigned category, the annealing temperature for each locus was either increased or decreased for the second and similarly again for third test round.
A larger preliminary data set (N = 48) was created with all loci that could be confidently genotyped and had two or more alleles. Based on this dataset we chose 18 markers that could be combined in one panel, i.e. analyzed simultaneously in one Abi fragment analysis, and reordered the forward primers with labels to replace the M13-tailed primers used earlier. More testing was done to optimize a multiplexed PCR panel, including several PCR tests with minor modifications to equalize peak intensities, minimize nonspecific amplification and confirm repeatability. Two markers were left out due to difficulties in allele determination. The final microsatellite panel consisted of 16 markers and could be amplified in three multiplexed reactions and one singleplex reaction. We used Micro-Checker 2.2.3 (Van Oosterhout et al. 2004) to test for the presence and frequency of null alleles with the method of Brookfield 1. The results indicated presence of null alleles in four loci that were then excluded from further analysis.

Microsatellite genotyping
DNA was extracted from whole individuals stored in 70 % ethanol and kept in + 4 °C. The head and thorax were dissected and used for DNA extraction with a silica-based purification method modified from Elphinstone et al., (2003). DNA was diluted 1:7 with sterile water and samples were genotyped with 12 microsatellite markers (Lyga-04, Lyga-05,  Table S2). To improve the microsatellite peak profiles, a GTTT-tail was added to the 5' end of each nonlabelled primer (Brownstein et al., 1996).
PCR amplification was carried out in three 8 µl multiplexed reactions (MP1-3) and one singleplex (S) reaction using QIAGEN Multiplex PCR Kit (Qiagen Inc. Valencia, CA, USA) with the annealing temperature of 55, 58 and 60 °C. Primer concentrations varied from 0.08 to 0.3 µM. PCR profile was according to the manufacturer's standard protocol for microsatellites. Amplifications were performed on PTC-100 (MJ Research) and AB 2720 (Applied Biosystems) thermal cyclers. For electrophoresis the PCR products were pooled by combining 0.8 to 1.0 µl of each multi-or singleplexed PCR and diluted with 100 µl of sterile water. 2 µl of the pooled and diluted PCR product was combined with GS600LIZ size standard (Applied Biosystems) and HiDi-formamide (Applied Biosystems). Samples were denatured at 98 °C for three minutes and the size of the fragments was determined by capillary electrophoresis on an ABI PrismTM 3130xl genetic analysis instrument.
The genotypes were scored using GeneMapper version 4.0 (Applied Biosystems) and following visual inspection, exported to a spreadsheet program for downstream analyses. Genotyping error was assessed by repeating the PCR amplification of 16 individuals, genotyping these samples separately and then comparing to the original genotypes. Direct count genotyping error rate per allele per locus varied 0 -3.3 %, mean overall loci 0.2 %.