Introduction

Botrytis cinerea [teleomorph: Botryotinia fuckeliana (de Bary) Whetzel] is a necrotrophic fungus with a wide host range (Jarvis 1977; Elad et al. 2004). Modern hybrid tomato (Solanum lycopersicum) cultivars are susceptible to B. cinerea although some cultivars show a certain level of quantitative resistance (ten Have et al. 2007). The presumed polygenic inheritance has limited the success of breeding for resistance to B. cinerea.

Quantitative resistance to B. cinerea has been identified in several wild relatives of S. lycopersicum (Urbasch 1986; Egashira et al. 2000; Nicot et al. 2002; ten Have et al. 2007). A stem bioassay suitable to quantify susceptibility of tomato to B. cinerea was used to screen a collection of wild tomato accessions (ten Have et al. 2007), and two parameters were calculated: the proportion of outgrowing lesions or disease incidence (DI) and lesion growth (LG) rate expressed as the increase in lesion size in mm/day. All four tested accessions of S. habrochaites showed quantitative resistance (ten Have et al. 2007). S. habrochaites LYC4 was used previously to study the genetic basis of this resistance (Finkers et al. 2007a, b) and a total of ten quantitative trait loci (QTL) were identified illustrating the genetic complexity of resistance to B. cinerea. Also S. neorickii G1.1601 showed a certain level of resistance (ten Have et al. 2007). A F2 mapping population of S. lycopersicum cv. Moneymaker × S. neorickii G1.1601, previously developed to identify QTLs for resistance to Oidium neolycopersici (Bai et al. 2003), was screened for susceptibility to B. cinerea. Since F2 seeds were no longer available, F3 families were used for the analysis. Segregating BC3S1 families and BC3S2 plants were generated in order to confirm the effects identified in the F3 analysis. We report the identification of (putative) QTLs, from S. neorickii G1.1601 involved in resistance to B. cinerea. Results of this study were compared to previously identified QTLs for resistance to B. cinerea in S. habrochaites LYC4 (Finkers et al. 2007a, b).

Materials and methods

Plant material

Three tomato accessions were used in this study: S. lycopersicum cv. Moneymaker (hereafter referred to as SL), S. neorickii G1.1601 (hereafter referred to as SN) and S. habrochaites LYC4 (hereafter referred to as SH). A cross between SL and SN was made and an F2 population (n = 209) developed as described in detail by Bai et al. (2003). From this F2 population, F3 seeds of each genotype were collected, but only 122 F2 plants produced enough F3 seeds for further study. Marker data for 75 of the 122 F2 plants used in this study were available.

For confirmation of the QTLs, three selected F3 plants were backcrossed to SL to obtain BC1 seeds. Two subsequent backcrosses to SL resulted in a BC3 progeny of 53 plants and marker assisted selection (MAS) was used to select three plants containing either QTL. Three BC3S1 families (n = 86 each) were genotyped to select plants homozygous SL or SN for the putative QTLs. A selected set of BC3S1 genotypes homozygous SN for the region of interest was grown to produce BC3S2 seeds.

Experimental setup and stem assay

For each of the 122 F3 families, five seedlings were grown and their susceptibility to B. cinerea was evaluated. For logistic reasons the disease assays were divided (at random) into 13 experiments with equal numbers of plants (50 plants/week). Eight SL controls were included in each experiment.

For the BC3S1 families, six replicates were grown by taking cuttings of each genotype including a set of SL, SN, and SH controls. To assess susceptibility to B. cinerea in the BC3S2 plants, two experiments were performed. In each experiment, three replicates, grown from seeds, of each genotype was tested.

The stem assay was performed according to ten Have et al. (2007). In short, stems of 6–8-week-old plants were cut into six pieces of five cm length and the top of each segment was inoculated with a droplet of 5 μl inoculum, containing ∼106 conidia per ml. Inoculum of B. cinerea strain B05.10 was prepared according to Benito et al. (1998). Incubations were performed at 15°C in the dark at 100% relative humidity. The infection progress was measured at day four and five after inoculation using a Vernier caliper. For each genotype, the percentage of successfully infected stem pieces was calculated (DI). The LG rate was calculated as the increase in lesion size between day four and five (mm/day) for the infected stem pieces.

DNA isolation and marker analysis

Twelve plants of each F3 family were grown and one leaf was harvested from each plant and pooled for DNA isolation in order to deduce the original F2 genotype. The AFLP™ and CAPS analysis of the F3, BC3, and BC3S1 populations were performed as described previously (Finkers et al. 2007a, b). The following ten AFLP primer combinations were used for genotyping: P14M48, P14M49, P14M50, P14M60, P14M61, P15M48, P18M50, P18M51, P22M50, and P22M51. AFLP primer nomenclature and adapter sequences have been described previously by Bai et al. (2003).

CAPS and SCAR primers were obtained from the “Solanaceae Genomics Website” (http://www.sgn.cornell.edu) or designed on sequences of genomic or cDNA clones available from the same source. Polymorphisms between SL and SN were determined using the CAPS digestion approach described by Bai et al. (2004). Markers, PCR conditions and restriction endonucleases used for genotyping are presented in Table 1.

Table 1 Primer sequences, lengths of PCR products and enzymes revealing a polymorphism for CAPS/SCAR markers
Full size table

Data analysis

Marker data were analyzed and a genetic linkage map was calculated with Joinmap® 3.0 (van Ooijen and Voorrips 2001). The susceptibility of the F2 genotype was estimated by taking the means of the replicated disease assays of five F3 plants. Phenotypic data of the BC3S1 and BC3S2 plants were analyzed using the general linear model (GLM) approach as implemented in SPSS 12.0 (SPSS Inc., Chicago, IL, USA). The minimal adequate models for DI and LG were determined independently in the data of the BC3S1 families and BC3S2 plants. This resulted in the following models: DI = constant + genotype + block and LG = constant + genotype + block + genotype × block.

Quantitative trait locus analysis of the F3 and BC3S1 populations were performed using the Kruskal–Wallis analysis as embedded in MapQTL® 5.0 (van Ooijen 2003). Data of the grouped BC3S2 plants were analyzed by comparing mean observations of each group to the mean observation of SL using a Dunnett test (Dunnett 1955). A probability of P < 0.05 was used to refer to a QTL as significant. Linkage maps were drawn using MapChart (Voorrips 2002). The correlation between traits was examined by interpreting Pearson correlation coefficients.

Results

Analysis of the F3 lines

F3 seeds of the cross between the susceptible cultivar S. lycopersicum cv. Moneymaker (SL) and S. neorickii G1.1601 (SN) were available (Bai et al. 2003) in sufficient quantity for 122 of the original 209 F2 plants. For 75 of these 122 F2 individuals, marker data were available; therefore we decided to (re) genotype all 122 F3 families. Ten AFLP primer combinations resulted in a total of 234 markers: 120 SL specific and 114 SN specific. 192 AFLP markers were placed on the paternal and maternal linkage maps (data not shown).

A quantitative B. cinerea disease assay on stem segments (ten Have et al. 2007) yielded data on DI and LG. The frequency distributions of both traits suggested normal, quantitative trait characteristics (data not shown). The susceptible control, SL, showed a DI and LG comparable to previous experiments (Table 2; Finkers et al. 2007a). Kruskal–Wallis (KW) analysis identified three linkage groups putatively containing a QTL (pQTL) for decreased susceptibility to B. cinerea. Based on the map of Bai et al. (2003) these linkage groups could be assigned to Chromosomes 3, 4, and 9 and the pQTLs will be referred to as pQTL3, pQTL4, and pQTL9 accordingly. To integrate the maternal and paternal linkage groups of Chromosome 3, 4, and 9, thirteen co-dominant CAPS markers were developed (Table 1). Using the integrated linkage maps, the effect of each pQTL was recalculated using KW analysis (Table 3). All three pQTL regions showed a skewed segregation resulting in a deficit of plants homozygous SL for pQTL3 and pQTL4 and a deficit of plants homozygous SN for pQTL9 (Table 3). In spite of the correlation between DI and LG (Pearson: r = 0.258 and P < 0.01), all pQTLs were associated with a single trait: pQTL3 and pQTL9 conferred a reduced LG while pQTL4 conferred a reduced DI.

Table 2 Mean disease incidence (DI) and lesion growth (LG) of the controls
Full size table
Table 3 Effect of the pQTLs identified in the F3 population
Full size table

Confirmation of the QTLs using BC3S1 and BC3S2 plants

Fifty-three BC3 plants were genotyped using AFLP and three BC3 plants were selected heterozygous for the alleles of either pQTL3, pQTL4 or pQTL9 in a genetic background as similar as possible to the recurrent parent SL (Table 4). Only one BC3 plant was identified containing the SN allele of pQTL9. However, this plant was also heterozygous for the pQTL3 and pQTL4 alleles. Three BC3S1 families were grown and MAS was used to obtain a more balanced test design (Table 4). In 3 of the 20 plants homozygous for the SN allele of pQTL9, SN alleles of pQTL3 and pQTL4 were absent. As a result, experiments aimed at confirming pQTL9 were only performed using BC3S2 lines.

Table 4 Description of the three BC3S1 families
Full size table

While assessing susceptibility to B. cinerea in the BC3S1 families, only the first experiment showed a sufficient level of infection (mean DI of 63%; observed 4 days post inoculation). Replicate experiments in subsequent weeks showed a low level of infection (mean DI < 20%) and were therefore discarded. In the first assay, the control SL showed the expected level of susceptibility (Table 2; Finkers et al. 2007a). However, the susceptibility of the controls SN and S. habrochaites LYC4 (SH) was higher than previously reported (Finkers et al. 2007a; ten Have et al. 2007) suggesting an overall high disease pressure in this experiment.

Susceptibility to B. cinerea in BC3S2 lines was assessed in two independent experiments. The mean DI of the two experiments was 29 and 62%, respectively, and yielded, on average, five independent observations for each BC3S2 line. The susceptible control SL and the partial resistant control SH showed the expected level of susceptibility (Table 2) but SN was more susceptible than previously reported (ten Have et al. 2007).

Experiments aimed at confirming pQTL3, in the BC3S1 population, did not lead to identification of a significant reduction of LG, yet the presence of the homozygous SN resistance allele resulted in a reduced LG (Table 5; P = 0.273). A significant reduction was observed while testing the BC3S2 lines (Table 6; group baa; P < 0.001). pQTL4, reducing DI, could not significantly be confirmed using either BC3S1 (Table 5; P = 0.413) or BC3S2 lines (Table 6; group aba; P = 0.645). However, in both BC3S1 population and BC3S2 lines, a reduction in DI was observed in the presence of the SN resistance allele. The LG reducing effect of pQTL9 was not confirmed in the BC3S2 lines (Table 6; group aab; P = 1.000). Instead, a 9% reduction in DI was observed for lines homozygous for the SN allele of pQTL9 (P = 0.665).

Table 5 Effect of pQTL3 or pQTL4 in its respective BC3S1 family
Full size table
Table 6 Estimated mean values for disease incidence (DI) and lesion growth (LG) of the BC3S2 lines
Full size table

In addition, BC3S2 lines were tested in which a combination of loci were present (Table 6). Two plants were homozygous for the SN alleles of pQTL4 and pQTL9 and showed a lower DI than plants containing either SN alleles of pQTL4 or pQTL9. The 22% lower DI than SL of these plants was, however, not significant (group abb; P = 0.056). Plants containing a combination of the SN alleles of pQTL3 (higher DI and lower LG compared to SL) and pQTL9 (lower DI and similar LG compared to SL; group bab) were as susceptible as SL. However, the observed mean was not deviating from the mean estimated from lines containing each QTL separately.

Discussion

While performing the B. cinerea stem assays, susceptibility of the controls varied between experiments (Table 2). In experiments with a harsher infection, quantitative resistance of SN is less robust than the quantitative resistance of SH. The higher resistance of SH suggests the presence of a larger number of QTLs, or more effective QTLs. Stem morphology and vascular development of SH stems might also play a role in resistance (Coaker et al. 2002). Variation in the bioassays and environment influences the confirmation of QTLs for resistance to B. cinerea (Finkers et al. 2007a). External influences can be minimized by growing plants in climate rooms but this is logistically not feasible for experiments on this scale.

A correlation between DI and LG was observed in data of the F3 families (r = 0.258; P < 0.01), but no significant correlation was observed in the SH F2 population (r = 0.173; P > 0.05; Finkers et al. 2007a). These correlations are lower than the correlation observed while testing the SH IL population (r = 0.65; P < 0.01; Finkers et al. 2007b). The heterogeneous genetic background of F2 and F3 plants and/or the type of bioassay used may obscure the obvious relationship which was observed between DI and LG in the SH IL population (Finkers et al. 2007b).

Initially, three pQTLs were identified while analyzing the F3 families. The resistance allele from each pQTL was derived from SN. The LG reducing effect of the pQTL3 resistance allele could be confirmed using BC3S2 lines. Because of this, pQTL3 will now be referred to as QTL3. No significant confirmation was obtained for the DI reducing effect of the pQTL4 resistance allele. Nevertheless, the presence of the pQTL4 resistance allele reduced DI in all three tested populations. For this reason, this QTL is still regarded as interesting for commercial tomato resistance breeding. Criteria for significance were not met due to the lack of replications (BC3S1 and BC3S2) or the lack of the number of lines tested per group (BC3S2). The LG reducing effect of the pQTL9 resistance allele was not confirmed, but instead we observed a decreased DI. Additional experiments are required to confirm pQTL4 and pQTL9.

The position of each SN QTL was compared to previously mapped QTLs, conferring resistance to B. cinerea, from SH (Finkers et al. 2007a, b). Interestingly, both pQTL4 and pQTL9 from SN are located at positions homologous to the SH QTLs Rbcq4a and Rbcq9b (Fig. 1). CAPS analysis of the SH introgression lines (IL) containing Rbcq3 (Finkers et al. 2007b) does not exclude the possibility that SN QTL3 is at a homologous position. We postulate that the three SN (p)QTLs identified in this study have QTLs at homologous positions in SH.

Fig. 1
figure 1

Overview of resistance loci mapped on tomato Chromosomes 3, 4, and 9: markers are from the core RFLP map (Tanksley et al. 1992). Closed blocks within the bars show the approximate locations of mapped RGAs (Zhang et al. 2002). The approximate locations of monogenic resistance genes (R genes) and quantitative trait loci (QTL, underlined) for disease-resistance are shown. For clarity, some loci were renamed for ease of display. Explanation of abbreviations of the R genes or QTLs is presented in Tables 7 and 8, respectively. The approximate locations of (p)QTLs identified in this study are presented as bars on the right hand side of each chromosome

Full size image

Experiments aimed at confirming the LG reducing effect of pQTL9 unexpectedly resulted in the identification of a pQTL reducing DI. Previous analysis of the SH IL population, resulted in identification of two QTLs on Chromosome 9: Rbcq9a reducing LG and Rbcq9b reducing DI (Finkers et al. 2007b). The region homologous to Rbcq9a is homozygous SL in this set of BC3S2 plants and may explain why only a reduced DI was observed. Future experiments aimed at confirming the reduction in LG should focus on testing lines containing SN alleles in the region homologous to Rbcq9a.

The SN pQTL9 did not confer a reduction in DI in the SL × SN F3 population (Table 3), as was also observed for the SH Rbcq9b in the SL × SH F2 population (Finkers et al. 2007a). Segregation of multiple DI-reducing loci combined with the under-representation of SN alleles for Chromosome 9 might have obscured identification of this locus. Several BC3S2 lines contained multiple QTLs. Two plants contained both pQTL4 and pQTL9 (group abb) and the additional reduction in DI of these lines suggests a fully additive model, showing the potential of pyramiding these two pQTLs.

Figure 1 shows a comparison between (p)QTL positions identified in this study and previously mapped QTLs conferring resistance to B. cinerea (Finkers et al. 2007b), mapped positions of resistance genes (R-genes), R-gene analogs mapped by Zhang et al. (2002) and QTLs conferring resistance to other diseases (Tables 7, 8). The positions of B. cinerea QTLs may be homologous with the positions of R-genes or QTLs conferring resistance to Phytophtora infestans and Xanthomonas campestris (Chromosome 3); Ralstonia solanacearum, P. infestans, and O. neolycopersici (Chromosome 4); Alternaria solani, Fusarium oxysporum, and TMV (Chromosome 9). It remains to be determined whether loci at homologues positions involve the same genes triggering a general defense mechanism such as papillae formation or phytoalexin production or whether the observed homologous positions are coincidental. The recessive gene ol-2, identified in S. lycopersicum var. cerasiforme and mapped at a position homologous to pQTL4, is involved in papillae formation (Bai et al. 2005). The observation that pQTL4 acts dominant implies that resistance to B. cinerea and to O. neolycopersici cannot be conferred by the same gene. However, it cannot be excluded that each species contains a similar ancestral gene, which has diverged into alleles conferring specificity to different pathogens. The isolation of B. cinerea R-genes, followed by complementation and subsequent testing of these plants for resistance to multiple pathogens, might resolve such questions.

Table 7 Qualitative resistance genes mapped on the tomato Chromosomes 3, 4, and 9
Full size table
Table 8 Quantitative resistance loci with at least one locus mapped on the tomato Chromosomes 3, 4 or 9
Full size table

The resistance alleles of pQTL4 and Rbcq4a act dominantly in reducing DI (Finkers et al. 2007a). Over-dominance of pQTL4 was observed in the BC3S1 population, but an independent confirmation is needed. Dominant QTLs for resistance to B. cinerea are advantageous in commercial F1 hybrid cultivar development. Disease tests, using the stem assay, resulted in identification of QTLs either reducing DI or reducing LG (Finkers et al. 2007a). QTLs generally contributed to both a lower DI and LG in a greenhouse assay on mature plants (Finkers et al. 2007b). QTL3 and pQTL4 might therefore be effective in reducing both DI and LG when used in commercial breeding programs. Besides S. habrochaites LYC4, S. neorickii G1.1601 is an alternative source for QTLs conferring resistance to B. cinerea.