We analyzed 107 Fragaria accessions belonging to 12 species during an initial screen in 2015. The species dominating this study was Fragaria vesca, represented by 50 accessions, followed by 19 Fragaria moschata, 14 Fragaria viridis, nine Fragaria chiloensis, five Fragaria nilgerrensis, three Fragaria × bifera, and two Fragaria orientalis accessions. Five species—Fragaria tibetica, Fragaria moupinensis, Fragaria virginiana, Fragaria cascadensis, and Fragaria nipponica—were represented by a single accession.

Supplementary Table S1 lists information pertaining to each of the tested accession. These accessions comprise cultivars and worldwide collections, most of them from Europe (Figure 1). Among them, F. vesca tetra and F. vesca poly 3 are artificial tetraploid descendants from diploid F. vesca ssp. vesca obtained via mutation breeding (Olbricht et al., 2014).

All Fragaria accessions analyzed in 2015 were maintained by Hansabred GmbH & Co. KG in Dresden, Germany. For each accession, at least three plants, which had been clonally propagated from stolons, were grown in a mixture of peat and sand in clay pots (20 cm in diameter) in a frost-free glasshouse in Dresden. The harvest and subsequent screening took place between late May and early July 2015. Ripe berries with the pedicel attached were carefully harvested between 06:00 and 08:00 h (UTC+2), packaged in soft tissue to prevent transportation damage and immediately sent to Munich, Germany, where the fruits were exposed to D. suzukii on the same day. All fruits that were bruised or damaged during transportation were excluded from further experiments.

Eighteen accessions were chosen to be analyzed again in 2016 (Supplementary Table S2). These accessions were maintained in the glasshouse (16 h:8 h, light:dark; 15–22°C; 60% humidity) at the Biocenter of the LMU Munich, Germany. Strawberry flowers were hand-pollinated and ripe strawberries were harvested between 14:00 and 15:00 h (UTC +2). Fruits were immediately used in the infestation assays.

The widest diameter of the fruit, as a representation of the fruit size, was measured using a size chart displaying circles with diameters ranging from 9 to 39 mm in 1 mm steps. The same fruits were then tested in the infestation assay (see below). For 9 out of 681 fruits in year 2015, the diameters were smaller than 9 mm. These outliers were not included in statistical analysis concerning fruit diameter. To accommodate all possible fruit diameters in 2016, circles with diameters ranging from 5 to 39 mm in 1 mm steps were used.

The D. suzukii line used in this study was initially established by Nicolas Gompel from a single female collected in the French Alpes-Maritimes (43°52′45″N 7°26′34″E) in July 2011. The flies were maintained and expanded at 22°C on standard cornmeal medium in a climate chamber (50% humidity; 12 h:12 h, light:dark; light from 08:00 h). The infestation assays were carried out with 5- to 8-day-old fertilized D. suzukii females (Supplementary Method S1), which were isolated under brief anesthesia on a chill table (BioQuip Products Inc., Compton, CA, USA), and allowed to recover for 5 h at room temperature. An independent population of female D. suzukii was generated for each infestation assay.

Single strawberries were placed in separate polypropylene vials (28 mm × 85 mm; Semadeni, Ostermundigen, Germany) with a crumpled piece of tissue paper (125 mm × 125 mm) filling the bottom (Figure 2A). The tissue was soaked with 3 ml of Milli-Q filtered water, and vials were closed with Rayon foam stoppers (28 mm; K-TK, Retzstadt, Germany). Three D. suzukii females were transferred into each vial and incubated with the strawberry between 16:00 and 17:00 h in 2015, or 15:00 and 17:00 h (UTC +2) in 2016. The exposure time was therefore 1 and 2 h in 2015 and 2016, respectively. The time to adult fly emergence was measured in days post-exposure (DPE), with day 1 being the day of exposure itself. The vials were kept in the same climate chamber in which the flies had been maintained until 17 DPE. Emergence of adult flies was checked once per day. Fly emergence was defined as the total number of D. suzukii adults that emerged from an individual fruit until 17 DPE. Additionally, the number of eggs deposited into the individual fruits was determined at 2 DPE in 2016, except for the white fruits from accessions 214, 220 and 223, the color of which prevented accurate determination of egg numbers.

The infestation assays were conducted once per week for five successive weeks in 2015. At least four fruits were tested per accession for 102 of the 107 tested accessions. For the remaining five accessions, the limited fruit production only permitted analysis of three fruits (Supplementary Table S1). Four infestation assays were conducted in three successive weeks in 2016. In both years, each accession was tested in a variable number of independent infestation assays (Supplementary Tables S1 and S2).

Because soft fruits are the site of oviposition and larval development, we tested how well D. suzukii propagated in a total of 681 fruits representing 107 accessions of 12 Fragaria species. We exposed ripe fruits to a controlled number of fertilized D. suzukii female flies and counted the number of adult flies emerging from each berry (Figure 2A; see Materials and Methods).

The offspring typically needed 13–17 days to emerge. Rare cases of fly emergence beyond 17 DPE were observed (10/681 fruits), however, this late emergence phenomenon was not consistently observed in any particular accession. At 17 DPE, fungal growth was visible on all strawberries, yet with a variable extent of mycelium growth. No visually significant correlation between fungal growth and fly emergence was observed.

Figure 2B shows the fly emergence from individual strawberries sorted by accession. The median value of fly emergence for the accessions ranged from 0 to 16, indicating strong variation between the accessions tested.

To test whether the variation of fly emergence observed between berries correlated with their accession or their species, or both, we applied generalized linear mixed models (GLMMs; Supplementary Method S2; McCullagh and Nelder, 1989; Pinheiro and Bates, 2000; Bates et al., 2015; R Core Team, 2015; Supplementary Figures S2A,C). The accessions and species of the berry were both significant factors contributing to the emergence probability of a given accession (GLMM likelihood profile analysis, P < 0.01).

The data were collected from experiments carried out on five different days within a period of 5 weeks. We therefore tested whether the observed variation in emergence was due to uncontrolled variables that influenced the outcome at different days of exposure, such as variation in female fly oviposition behavior on different days. When we added the date of the individual experiment as a co-factor to the GLMM, it was not significant, but the species and the accession still were (Supplementary Method S2; P < 0.01), demonstrating that harvest times had no influence on the emergence probability.

Altogether, this dataset unveils a tractable range of variation in D. suzukii capability to emerge from the berries across the genus Fragaria.

Considering that major developmental steps of D. suzukii larvae take place inside the fruits, the developmental process might be influenced by the amount of available food (strawberry tissue) or physical conditions, or both, thus influencing the overall fly emergence from a berry. We therefore tested whether the fruit size, represented by the diameter of a fruit, was linked to fly emergence.

The diameter of individual berries ranged from 9 to 29 mm, and variations were observed among the individual fruits of each species (Supplementary Figure S1) and the average fruit diameter of accessions (Supplementary Table S1). We observed a significant positive correlation between the average emergence probability and average fruit diameter of Fragaria species (P = 0.024; Figure 3).

When accession, species and fruit diameter were analyzed in a single GLMM, the significance of the species vanished, whereas the accession number and fruit diameter remained significant factors (P < 0.001) for the emergence probability.

The geographical origins of the tested accessions are scattered in several continents, including East Asia, from where D. suzukii originates (Figure 1). Due to the possible co-evolution between Fragaria species and D. suzukii, we initially suspected genotypes resulting in low fly emergence to also originate from East Asia. Whenever geographical information about the accessions (i.e., country of origin, longitude, latitude, or the distance to the equator) was added to the GLMM model (Supplementary Method S2), however, no significant effect was discovered.

We estimated the effect of accessions on emergence probability as well as fly emergence in separate GLMMs. In both cases we corrected for possible effects of fruit diameter. Ten accessions were predicted to have the strongest aversive effects in at least one model (Supplementary Figures S2B,C). These accessions were considered the candidate accessions with reduced fly emergence and were selected to be tested again in 2016, the year following the initial screen.

Five accessions that represented the other extreme end of the fly emergence and emergence probability, i.e., showing highest values of both traits, in the 2015 screen were included as controls (hereafter referred to as reference accessions). These reference accessions had similar fruit diameters as, and could be crossed with, the candidate accession belonging to the same species (Supplementary Table S2), thus representing potential crossing parents for positional cloning approaches to identify genes or QTLs determining fly emergence. We could not find suitable reference accession for accessions 1 and 118. Additionally, we re-tested three accessions, 327, 337 and 330, that had shown a low fly emergence, but were not among the top 10 candidates in the models (Figure 2; Supplementary Figures S2A,B; Supplementary Table S2). We excluded the accessions 1, 411 and 177 from the re-test analysis, due to limitations in fruit production.

In our experiments in 2016, we increased the number of fruits for each accession and counted the number of eggs deposited into individual fruits (other than those of accessions 214, 220 and 223). Fruits devoid of eggs (10.1%; 17/168) were excluded from further determination of fly emergence and emergence probability (Supplementary Figure S3; Supplementary Table S2). The re-test experiments of 18 accessions resulted in similar overall variation in fly emergence and emergence probability that had been observed in the initial 2015 screen (Supplementary Figure S4; Supplementary Table S2).

Of the five accessions that showed high fly emergence and emergence probability in 2015, only three, 206, 214 and 349, performed consistently in both years (Supplementary Figure S4; Supplementary Table S2), whereas 436 and 441 did not show similar fly emergence compared to 2015. Three candidate accessions, 223, 220 and 300, reproduced the low fly emergence phenotype, while other candidate accessions (185, 209, 118 and 340) failed to do so (Supplementary Figure S4). Similarly, only candidate accessions 300 and 340 showed relatively low emergence probability whereas other candidates did not (Supplementary Table S2).

Amongst the confirmed candidates, accession 300 was of particular interest. The number of eggs deposited into berries of accession 300 was not statistically different from that of the reference accession of F. vesca, 349 (Supplementary Figure S3; paired t-test, P > 0.05). However, only 30% of fruits of accession 300 supported fly emergence, a rate significantly lower than in accession 349 which had over 90% of its fruits supporting fly emergence (Supplementary Table S2; Fisher’s exact test, P < 0.005). In the few cases where D. suzukii did emerge from fruits of accession 300, a maximum of two flies emerged, also significantly fewer than from accession 349 (permutation test, P < 0.05), consistent with the phenotype of low fly emergence observed in 2015 (Figure 2; Supplementary Figure S4).

Overall, we found three accessions with reproducibly reduced fly emergence, of which accession 300 is particularly strong in countering fly emergence.