Botrytis strains were isolated from infected plant tissue by streaking out conidia onto potato dextrose agar (PDA) augmented with 200 mg penicillin G and 200 mg streptomycin sulfate L⁻¹ (Weber and Hahn, 2011) or HA agar (10 g malt extract, 4 g glucose, 4 g yeast extract, 15 g agar per liter, pH 5.5) augmented with 5 mg tannic acid L⁻¹, to inhibit growth of several contaminating fungi. Purified strains were cultivated on HA, and subjected to long-term storage by freezing to −80°C.

Growth tests of B. cinerea strains were performed on agar plates containing either HA or Gamborg GB5 minimal medium containing 25 mM glucose as described by Plesken et al. (2015b). For determination of radial growth, three 10 μL droplets containing 10⁵ conidia mL⁻¹ were applied to the center of a 9 cm agar plate, and incubated for 72 h before evaluation of the colony diameter.

Fungicide sensitivity assays were performed by analysing germination and growth of conidia on agar plates containing discriminatory fungicide concentrations. Because these assays were started in 2010 and performed in two different labs, prior to the decision for a joint publication, two fungicide resistance assays were used. The method of Weber and Hahn (2011) was used for Botrytis isolates collected in Northern Germany, whereas all other isolates were examined according to Leroch et al. (2013). Quantitative evaluation of fungicide resistance levels was performed as described by Leroch et al. (2013). Active ingredients were obtained by using single-compound fungicides, as follows: fenhexamid from Teldor (Bayer CropScience, Monheim, Germany); trifloxystrobin from Flint (Bayer) or azoxystrobin from Ortiva (Syngenta, Basel, Switzerland); boscalid from Cantus (BASF, Limburgerhof, Germany); cyprodinil from Chorus (Syngenta); fludioxonil from Saphire (Syngenta); iprodione from Rovral WG (BASF) and thiophanate-methyl from Cercobin FL (BASF). The SDHI compound fluopyram was also sometimes tested but data are not reported here because this compound was not registered until the 2015 season (Weber et al., 2015).

For in planta assays, commercial fungicides were prepared as stock solutions in EtOH (1 mg mL⁻¹) and adjusted to the recommended doses in water. For inoculation experiments of freshly detached tomato leaves, these were sprayed with fungicides (or water in the controls) to leaf wetness, and dried in a sterile air flow cabinet. After inoculation with 20 μl droplets of spore suspension (10⁵ conidia mL⁻¹ in GB5 medium with 25 mM glucose), leaves were incubated in a moist chamber and checked for lesion development after 72 h.

For the greenhouse and outdoor inoculation experiments, strawberry plants (cv. Sonata) were raised in pots (9 cm diameter) filled with TKS1 peat-based substrate (Floragard, Oldenburg, Germany). The pots were placed in trays and watered by drop irrigation. After 6 weeks, plants carrying immature fruits were treated twice at an interval of 7–9 days with ~5 mL fungicide solution per plant using a hand-held spray bottle, each time followed 24 h later by spray inoculation with 1 mL of B. cinerea spore suspension (3 × 10⁵ conidia mL⁻¹) per plant. Fungicide applications were conducted with Teldor® (2 g L⁻¹; active ingredient fenhexamid), Signum® (1.8 g L⁻¹; boscalid + pyraclostrobin) or Switch® (1 g L⁻¹; cyprodinil + fludioxonil). Inoculation was performed with the fully sensitive B. cinerea strain B05.10 or the multiresistant strain MR-S4 (Table 2). Each experimental variant including unsprayed inoculated and unsprayed uninoculated controls comprised six plants placed at random. Fruits were harvested 5–17 days after the first inoculation. Rotting fruit were collected at harvest and again after incubation in a humid chamber for up to 1 week. From sporulating lesions of all infected fruits, conidia were recovered and tested on agar plates with discriminatory fungicide concentrations for their resistance against fenhexamid, a QoI fungicide (azoxystrobin or trifloxystrobin), boscalid, cyprodinil and fludioxonil. Isolates with resistance patterns that differed from the inoculated strain were considered as immigrated contaminant strains, and excluded from evaluation of the specific infection rate.

Competitive infection assays with apples were performed as follows: Apples (cv. Golden Delicious) were surface-sterilized by submersion for 5 min in a 1% sodium hypochlorite solution, rinsed three times with sterile water, and dried in a sterile air-flow cabinet. Fruits were wounded at two or four equidistant sites above the fruit equator by removing the peel with a sterile 6 mm diameter stainless steel cork borer. Wounds were inoculated with 20 μL of a conidial suspension of B. cinerea. The suspension was prepared by adjusting conidia suspended in GB5 medium containing 25 mM glucose to a concentration of 2 × 10⁵ mL⁻¹, and mixing either with the suspension of another B. cinerea strain or with medium alone (controls). Inoculated apples were incubated in moist plastic boxes for 7 days. Subsequently, the fruit was divided into two or four pieces, and from each side a tangential cut was made in order to expose the rotten area to the air and to promote sporulation. Spores were then collected and the spore suspension was transferred onto HA agar plates supplemented either with 10 ppm fenhexamid to distinguish between sensitive and MR strains, or with 0.2 ppm or 1 ppm fludioxonil to distinguish between strains with or without partial fludioxonil resistance due to the MDR1(h) efflux mechanism.

Sclerotium formation was induced by incubation of HA plates with mycelium-containing agar plaques for 4 weeks at 15°C in the dark. Viability of sclerotia was tested by storing the plates for 8 months at 10–14°C, followed by transfer of five individual sclerotia per strain onto a fresh HA agar plate and incubation at room temperature. After 3 days, development of mycelium from the sclerotia was checked.

For determination of fungicide resistance mutations of selected B. cinerea isolates, total DNA was isolated as described by Leroch et al. (2013). Mutations were identified by amplification of target gene fragments, followed by either sequencing or primer-induced restriction analysis (PIRA)-PCR (Haliassos et al., 1989) to identify specific mutations. The F412S mutations in erg27, the G143A mutation in cytB, the H272R and the H272Y mutations in sdhB, the E198A mutation in tubA, and the ΔL497 mutation in mrr1 were identified by PIRA-PCR, using the primers and enzymes shown in Table 1. Restriction enzymes were purchased from NEB Biolabs (Frankfurt, Germany). Mutations in bos1 conferring iprodione resistance were identified by pyrosequencing (Grabke and Stammler, 2015). All other mutations were identified by sequencing.

For statistical analysis of the results, a two-sided unpaired t-test was performed (Prism 5 for Windows, Version 5.01).

Gray mold isolates from a variety of host plants were collected and analyzed for their sensitivity or resistance to the main classes of site-specific fungicides with activity against Botrytis. Our assays permitted the detection of medium and/or high resistance to all fungicides tested, including the five registered fungicide classes QoI (azoxystrobin or trifloxystrobin), boscalid, fenhexamid, cyprodinil and fludioxonil. In the following, strains with medium and high resistance were rated as resistant. Strains simultaneously resistant to four or all five registered fungicides are referred to as MR (multiresistant) strains.

Raspberries receive repeated fungicide sprays against Botrytis during flowering. Plants are kept either as annual crops or cultivated for 10 years or more, thereby permitting long-term surveys of fungicide resistance development. One such field was planted in 2002 and examined in 2010–2015. In 2010 resistance frequencies were below 10% for fenhexamid and fludioxonil and between 15 and 55% for the other fungicides. In the following years, resistance frequencies steadily increased to more than 85% for all fungicides except for benzimidazoles, resulting in an almost completely resistant population (Figure 1). In line with this development, strains with 5-fold resistance to all registered fungicide classes were first recorded at harvest 2012 when they comprised 15% of the population, and their share rose dramatically to 87% in 2015 (data not shown). Similar trends were seen in two other raspberry fields (Table S1). Strawberries are grown in Germany in different cultivation forms for 2 to 3 years. Frequent fungicide treatments are carried out during the flowering period, and high resistance frequencies are common in these fields (Weber, 2011; Leroch et al., 2013). The data shown here are from isolates obtained in the 2013 seasons in six fields in Western and Southern Germany, which had all received four or more sprays with site-specific fungicides. Resistance frequencies approached 100% for all five classes of current Botrytis fungicides (Table S1). Everbearing strawberries with an extended harvest period from mid-summer to autumn are increasingly being cultivated in Germany for local or regional sale. These cultures are usually sprayed weekly against gray mold, raising an enormous selection pressure on the pathogen. Botrytis isolates from a field with everbearing strawberries in Southern Germany in 2014, treated for 8 weeks alternately with cyprodinil + fludioxonil or with fenhexamid, showed very high resistance frequencies. In an untreated plot within the same field, somewhat lower resistance frequencies were observed (Figure 2). Isolates from other fields of everbearing strawberries in Northern Germany also showed very high resistance levels, a high proportion of strains possessing 5-fold resistance to all fungicides registered against Botrytis (Table S1).

In the Wine Road Region of Germany, gray mold isolates from grapevine berries have been previously found to show moderate resistance frequencies to the registered fungicides (Leroch et al., 2011). A survey of isolates harvested in 2011 and 2014 revealed that resistance mainly occurred against QoI, whereas no or rather low levels of resistance existed against the other registered fungicides. Remarkably, the proportion of boscalid resistant isolates had increased from 4 to 35% between 2001 and 2014, probably due to the increased use of this fungicide (Table S1). None of the grapevine isolates had accumulated resistance to more than three registered fungicides. To test whether a transfer of MR strains from heavily treated strawberry fields into vineyards could occur, we analyzed a vineyard and an adjacent strawberry field in a fruit growing region in Southern Germany. Resistance frequencies in the strawberry and the grapevine populations were similar and much higher than resistance frequencies observed in the German Wine Road region (Figure 3).

In Germany, gray mold on stone fruit such as cherries and plums is less important than brown rot caused by Monilinia spp., and fungicide sprays are less frequent than in soft fruit fields. A long-term survey of a commercial cherry orchard conducted for 6 consecutive years revealed almost no resistance of any of the tested isolates in 2010. In 2011, a sudden increase in resistant isolates and the appearance of MR isolates was observed, which suggested their immigration from another field (Figure 4). The elevated resistance levels were largely maintained during subsequent years except for a transient decrease in 2014. Isolates obtained from four other cherry orchards in 2012, 2013, and 2014 showed intermediate to high resistance levels, and between 7 and 33% of strains were resistant to all five registered fungicides (Table S1). Isolates from a plum orchard in the same region sampled in 2012 showed high resistance frequencies, with 40% of all isolates being 5-fold resistant (Table S1). We also obtained a low number of isolates from tree seedlings raised in a tree nursery in Northern Germany, all of which belonged to the MR type (Table S1).

Botrytis isolates from ornamental flowers were obtained from greenhouses at 50 different sites in Eastern and Southern Germany. Frequencies of fungicide resistance were highly variable (Table S2). Isolates with 4-fold or 5-fold resistance to current fungicides were recovered from 17 sites. The average fungicide resistance frequencies of flower isolates ranged from 20% against fludioxonil to 75% against QoI (Figure 5). Resistance to benzimidazoles was also very common, even though this fungicide has not been in use for many years. Genetic characterization with species-specific PCR primers (Plesken et al., 2015a) revealed eight isolates to belong to B. pseudocinerea; six of them were sensitive to all fungicides and two were resistant only to benzimidazoles. Isolates belonging to B. cinerea groups N and S showed a random distribution, with no evidence of any host preference or differences in fungicide resistance frequencies (Table S2). In glasshouses for which no fungicide treatments were reported, MR strains were observed at similar frequencies as in the treated locations (data not shown).

For detailed genetic and phenotypic analysis, 19 MR strains (10 of B. cinerea group S and 9 of B. cinerea group N) were selected from different fields. Most of these strains were also resistant against iprodione and carbendazim (Table 2). Nine MR strains were able to grow on 1 ppm fludioxonil due the ΔL497 deletion mutation in mrr1 conferring the MDR1h phenotype; the other MR strains grew on 0.2 ppm due to uncharacterized MDR1-related mutations (Leroch et al., 2013). Twenty sensitive strains served as controls, 10 each of B. cinerea N and B. cinerea S. Analysis of the target genes of the fungicides (except cyprodinil, for which the target site is unknown) revealed common mutations for boscalid (sdhB: H272R; H272Y), fenhexamid (erg27: F412S, F412I, and T63I), fludioxonil (intermediate resistance due to MDR1h-related mutation ΔL497; Leroch et al., 2013), QoI fungicides (cytB: G143A), iprodione (bos1: I365N, I365S, and V368F/Q369H) and benzimidazoles (tubA: E198A and F200Y) (Hahn, 2014).

For each of the 19 strains shown in Table 2 and for 20 sensitive strains, vegetative growth, sporulation, and infection behavior was compared (Figure 6). Radial growth and sporulation efficiency on HA medium were similar for MR and sensitive strains. In the presence of high salt concentrations (1 M NaCl), growth of MR and sensitive strains was inhibited to a similar extent. Unexpectedly, 10 mM H₂O₂ inhibited sensitive strains significantly more strongly than MR strains. Sclerotia were formed by MR and sensitive strains in variable numbers. To compare the viability of sclerotia, two batches of five sclerotia of each of 19 MR strains listed in Table 3, 18 sensitive strains were incubated for 8 months at 10–14°C and then tested for survival. All 370 sclerotia, except one sclerotium from an MR strain, germinated and formed a mycelium. When infection tests were performed on detached tomato leaves, average lesion diameters of MR strains were slightly but significantly smaller than those induced by sensitive strains (Figure 6D). In contrast, lesion formation and sporulation efficiency on inoculated apples were indistinguishable between MR and sensitive strains (Figure 6E).

For evaluation of their competitiveness during infection, wounded apples were inoculated with mixtures of three MR and three sensitive strains. We also compared MR strains with and without the MDR1 or MDR1h phenotypes to evaluate the effects of the efflux-based fludioxonil resistance on competitiveness. Conidia recovered from sporulating apple tissue were tested for sensitivity or resistance to a diagnostic fungicide (Table 3). In all experiments, MR strains were at least as competitive as the sensitive strains. Comparison of strain mixtures with and without MDR1 and MDR1h revealed no clear competitive advantage for either group of strains.

To estimate the loss of efficacy of fungicides against strains with resistance mutations, inoculation experiments of fungicide-treated host tissues were performed. In a laboratory experiment, detached tomato leaves were treated with fungicides and subsequently inoculated with either a sensitive or an MR strain of B. cinerea. While all strains induced large expanding lesions in unsprayed leaf tissue after 72 h (not shown), the sensitive strains were almost completely inhibited by all fungicides even at a 1:10 dilution of the recommended dose. In contrast, the MR strains produced almost normal lesions in the presence of fenhexamid and boscalid, but were strongly inhibited by recommended doses of fludioxonil and cyprodinil. In the presence of cyprodinil and/or fludioxonil at 1:10 dilution, MR strains but not the sensitive strains were able to cause lesions (Table 4).

The efficacy of fungicide treatments against a sensitive and an MR (MDR1h) strain was also tested with potted strawberry plants cultivated in a greenhouse (2014) and outdoors (2015). In these experiments, plants bearing immature fruits were sprayed with fungicides and subsequently inoculated. To identify the strains, conidia isolated from rotting fruit were tested for their fungicide resistance profiles. Strawberries inoculated with the sensitive strain were infected not only by this strain but also by strains with one or more resistances in 59% and 90% of fruits in the 2014 and 2015 experiments, respectively. In contrast, infections of strawberries inoculated with the MR-S4 strain were caused in >80% (2014) and >90% (2015) of the cases by this strain. After subtracting the infections caused by non-inoculated strains, the differential fungicide efficacies against the sensitive and the MR B. cinerea strain were evident in both experiments. Treatments with fenhexamid, boscalid + pyraclostrobin or cyprodinil + fludioxonil led to a significant or complete suppression of gray mold rot. In contrast, none of the three fungicides provided any protection against infection by the MR strain in either experiment (Figure 7).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.