Plant and insect materials
A total of 56 apple plants were grown from seeds and sampled for this study. Cultivated apple plants resulting from crosses between various cultivated apple varieties were used (M. domestica, referred to as “Dom”, N = 14, Table S1). The seeds were kindly provided by INRAE IRHS Angers that performed every year crosses for apple breeding programs. A total of 42 M. sylvestris plants were grown from field-collected seeds. These wild apple seeds originated from three out of the five known European wild apple populations (referred to as Danish: Syl_Dk, French: Syl_Fr and Romanian: Syl_Ro, N = 14 per population). Each population was represented by a single sampling site, and within each site, each seed was sampled on a single mother tree, so that each seedling has a different parental origin. Though M. domestica is usually grafted, new plants were grown from seed to eliminate the rootstock effect.
After field sampling, seeds were stored at -20 °C before vernalization for the experiment. Seeds were then vernalized for three months at 4 °C in the dark, then grown in controlled conditions for two months before being individually transferred to 3 L pots containing commercial sterilized potting soil. Potted plants were grown in a growth chamber for four weeks under the following conditions: 20 ± 1 °C, 75 ± 5% Relative Humidity (RH), and a 16:8 light:dark (L:D) photoperiod. The 56 plants were then genotyped using 13 previously published microsatellite markers (see below) to confirm their genetic status (i.e., belonging to one of the M. sylvestris European populations or crop-to-wild/wild-to-wild hybrid).
A single colony of D. plantaginea (Hemiptera: Aphididae) was used and provided by INRAE which were sampled as a population in spring 2018 from an apple tree at the Agrocampus Ouest orchard (Angers, France) (Philippe Robert, personal communication). This aphid population was mass reared without differentiating individual aphid clones on M. domestica cv. “Jonagold” plants obtained by in vitro multiplication21. Pots containing three plants (90 × 90 × 70 mm) were placed in a Plexiglas cube (50 cm). Mass rearing and experiments were performed in growth chambers under 20 ± 1 °C, 60 ± 5% RH, and a 16:8 L:D cycle.
Synchronized first instar nymphs were obtained by placing parthenogenetic adult females on plantlets for 24 h before removing them. They were then reared on M. domestica cv. “Jonagold” plants inside Plexiglas aerated boxes (36 × 24 × 14 cm) for ten days then used as the young adult RAA for the behavioral/performance experiments.
Apple population genetic diversity and structure
Genomic DNA was extracted with the NucleoSpin plant DNA extraction kit II (Macherey & Nagel, Düren, Germany) according to the manufacturer’s instructions. Microsatellites were amplified by multiplex PCR, with the Multiplex PCR Kit (QIAGEN, Inc.). We used 13 microsatellite markers, Ch01f02, Ch01f03, Ch01h01, Ch01h10, Ch02c06, Ch02c09, Ch02c11, Ch02d08, Ch03d07, Ch04c07, Ch05f06, GD12, and Hi02c07 in four multiplexes (MP01, MP02, MP03, MP04)4. PCR were performed in a final reaction volume of 15 ml (7.5 ml of QIAGEN Multiplex Master Mix, 10–20 mM of each primer, with the forward primer labelled with a fluorescent dye and 10 ng of template DNA) (See4 for more details). The final volume was achieved with distilled water. A touch-down PCR program (initial annealing temperature of 60 °C, decreasing by 1 °C per cycle down to 55 °C) was used. Genotyping was performed on the GENTYANE platform (INRAE Clermont-Ferrand) using an ABI PRISM X3730XL, with 2 ml of GS500LIZ size standard (Applied Biosystems). Alleles were scored with GENEMAPPER 4.0 software (Applied Biosystems). Only multilocus genotypes with < 10% missing data were retained.
The genetic status of each seedling was assessed using the individual-based Bayesian clustering method implemented in STRUCTURE 2.3.322. STRUCTURE makes use of Markov Chain Monte Carlo (MCMC) simulations to infer the proportion of ancestry of genotypes from K distinct clusters. The underlying algorithm attempts to minimize deviations from Hardy–Weinberg and linkage disequilibria. STRUCTURE was run from K = 1 to K = 8, ten independent runs were carried out for each K and 500,000 MCMC iterations were used after a burn-in of 50,000 steps. CLUMPAK (Greedy algorithm)23 was used to look for distinct modes among the 10 replicated runs of each K. STRUCTURE analyses were run for the full dataset (N = 55, DNA could not be extracted from one Romanian seedling), and included as well 40 M. domestica genotypes as a reference for the cultivated apple gene pool6. We determined the strongest level of genetic structure using ΔK24, as implemented in the online post processing software Structure Harvester25. However, the K identified by this criterion often does not correspond to the finest biologically relevant population structure6,7,26,27. A lack of consideration of intraspecies genetic structure in STRUCTURE analyses can bias the interpretation of introgression rates. We therefore visualized the bar plots and chose the K value for which all clusters had well assigned individuals while no further well-delimited and biogeographically relevant clusters could be identified for higher K values.
Once the best K chosen, wild plants assigned to the cultivated gene pool with a membership coefficient > 0.1 were classified as crop-to-wild hybrids (i.e., introgressed by M. domestica). Once crop-wild hybrids removed, plants assigned to a given wild gene pool with a cumulated membership coefficient > 0.9 were defined as “pure wild” individuals. Plants assigned to the wild gene pool with a cumulated membership coefficient < 0.9 to a given wild apple gene pool were defined as wild-wild-hybrids The pure, crop-to-wild and wild-wild hybrids were included as factors in the statistical analyses. Pure seedlings were then assigned to a population (i.e., group of plants with a cumulated membership coefficient of up to 0.90 for a given wild apple cluster). Pure populations from the same geographic origin (i.e., Romania or France or Denmark) which showed (1) weak genetic differentiation with other wild populations (2) low number of individuals were merged. The “population” was then used as a factor for statistical analyses on physiological and behavioral assays. Population genetics statistics were estimated with Genodive28 for each “pure” wild apple population including expected and observed heterozygosities, Weir and Cockerham F-statistics, Jost’s D, and deviations from Hardy–Weinberg equilibrium.
Dysaphis plantaginea feeding behavior
The feeding behavior of the RAA was investigated using the electrical penetration graph (EPG) method29. Individual aphids were connected to the Giga-8 DC-EPG amplifier, each being placed on the abaxial side of a new growing leaf of an individual plant. The recordings were performed continuously for 8 h during the photophase inside a Faraday cage. Acquisition and analysis of the EPG waveforms were carried out using the PROBE 3.5 software (EPG Systems, www.epgsystems.eu). Parameters from the recorded waveforms were calculated with the EPG-Calc 6.1.7 software30. They were based on different EPG waveforms31 corresponding to: (Pr) stylet activity within plant tissues; (C) stylet pathways in plant tissues except phloem and xylem; (E1) salivation in phloem elements; (E2) passive phloem sap ingestion; (G) active xylem sap ingestion; and (F) derailed stylet mechanics. A total of eight plants per M. sylvestris population (Syl_Dk, Syl_Fr, Syl_Ro) or M. domestica (Dom) genetic group were used for the EPG measurements. EPG records were obtained from 25 aphids for M. domestica, and from 27 aphids for each M. sylvestris population.
Dysaphis plantaginea performance
Two to three clip-cages were installed on 12 plants per genetic group identified in this study. Each cage contained an individual, synchronized aphid and was enclosed on a newly grown leaf. For each synchronized adult, observations were assessed every 24 h for 10 days. Survival (i.e., the duration of adult survival over the period of 10 days) and daily fecundity (i.e., the number of newly larviposited nymphs) were collected for 25 adults for M. domestica, and 28 to 29 adults for each of the three M. sylvestris populations.
To measure aphid weight, newly larviposited nymphs were enclosed for nine days in clip-cages on newly grown leaves similar to the above. For each plant genetic background, up to 20 aphids (i.e. young adults that had not larviposited yet) were then collected and stored in a freezer at – 80 °C. Each individual aphid was weighed using an electronic precision balance (Mettler M3, class 1, Max: 3 g Low: 1 µg, T = − 3G [dd] = 1 µg).
Statistical analyses
All statistical analyses were performed using the R software version 3.6.2 (The R Foundation, https://www.r-project.org/). Generalized linear models (GLM) with a likelihood ratio and Chi-square test were used to assess whether there was an effect of the host plant on aphid feeding behavior and performance. The apple tree genotype was included as the main factor. Data on daily aphid fecundity and some EPG parameters describing the number of occurrences of a particular phase (e.g. “n_E2”) were not normally distributed (count data), accordingly a GLM was carried out using respectively a quasi-Poisson and a Poisson distribution; a quasi-likelihood function was used to correct for over-dispersion, and Log was specified as the link function in the model. EPG data on feeding phase durations (e.g. duration of phloem sap ingestion “s_E2”) and aphid weight were not normally distributed, so a GLM using a Gamma (link = “inverse”) distribution was carried out. Analysis of the time before the first probe (“t.1Pr”) and before the first phloem sap ingestion (“t.1E2”) and adult survival has been carried out using the Cox proportional hazards (CPH) regression model, which is adapted to treat time-dependent parameters. Absence of an EPG reading were treated as missing values. The assumption of validity of proportional hazards was validated using the function “coxph” (package R: “survival”, version 3.1.8: https://github.com/therneau/survival). To assess whether the crop-to-wild hybrid status had a plant-mediated effect on RAA feeding behavior and performance, the same statistical tests were carried out with the hybrid statuts (i.e. “wild pure” or “wild-crop hybrid”) as the fixed factor while restricting the data set to the M. sylvestris populations only.
Finally, because of their close genetic relatedness, the Danish and French wild apples plants (Syl_Dk and Syl_Fr, respectively) were grouped together, as well as the M. domestica and the Romanian wild apple plants (Dom and Syl_Ro, respectively). A Monte-Carlo permutation test (999 replicates) was conducted to test for the significance of the differences of median of EPG phases duration, daily fecundity and weight between aphids submitted to these two host groups. Analysis of the time before the first probe (“t.1Pr”) and before the first phloem sap ingestion (“t.1E2”) and adult survival has been carried out using the CPH regression model. The function “randtest” (package R: “ade4”: https://cran.r-project.org/web/packages/ade4/ade4.pdf) was run to access the significance of the observed differences.
The fit of all GLM was controlled by a visual evaluation of residuals and QQ plots. Concerning QQ plots, the distribution of the series were considered to follow the chosen theoretical law if the points of the graph were roughly aligned on a straight line. Any other structuring of the points (curvature(s), many distant points, etc.) indicated the opposite. GLM post-hoc comparisons were carried out by pairwise comparisons using estimated marginal means (package R: “emmeans”, https://cran.r-project.org/web/packages/emmeans/emmeans.pdf).
Ethics approval
The article does not contain any studies with human participants or vertebrate animals.
Source: Ecology - nature.com
